Telethon’s Documentation

from telethon.sync import TelegramClient, events

with TelegramClient('name', api_id, api_hash) as client:
   client.send_message('me', 'Hello, myself!')
   print(client.download_profile_photo('me'))

   @client.on(events.NewMessage(pattern='(?i).*Hello'))
   async def handler(event):
      await event.reply('Hey!')

   client.run_until_disconnected()

What is this?

Telegram is a popular messaging application. This library is meant to make it easy for you to write Python programs that can interact with Telegram. Think of it as a wrapper that has already done the heavy job for you, so you can focus on developing an application.

How should I use the documentation?

If you are getting started with the library, you should follow the documentation in order by pressing the “Next” button at the bottom-right of every page.

You can also use the menu on the left to quickly skip over sections.

Installation

Telethon is a Python library, which means you need to download and install Python from https://www.python.org/downloads/ if you haven’t already. Once you have Python installed, upgrade pip and run:

python3 -m pip install --upgrade pip
python3 -m pip install --upgrade telethon

…to install or upgrade the library to the latest version.

Installing Development Versions

If you want the latest unreleased changes, you can run the following command instead:

python3 -m pip install --upgrade https://github.com/LonamiWebs/Telethon/archive/v1.zip

Note

The development version may have bugs and is not recommended for production use. However, when you are reporting a library bug, you should try if the bug still occurs in this version.

Verification

To verify that the library is installed correctly, run the following command:

python3 -c "import telethon; print(telethon.__version__)"

The version number of the library should show in the output.

Optional Dependencies

If cryptg is installed, the library will work a lot faster, since encryption and decryption will be made in C instead of Python. If your code deals with a lot of updates or you are downloading/uploading a lot of files, you will notice a considerable speed-up (from a hundred kilobytes per second to several megabytes per second, if your connection allows it). If it’s not installed, pyaes will be used (which is pure Python, so it’s much slower).

If pillow is installed, large images will be automatically resized when sending photos to prevent Telegram from failing with “invalid image”. Official clients also do this.

If aiohttp is installed, the library will be able to download WebDocument media files (otherwise you will get an error).

If hachoir is installed, it will be used to extract metadata from files when sending documents. Telegram uses this information to show the song’s performer, artist, title, duration, and for videos too (including size). Otherwise, they will default to empty values, and you can set the attributes manually.

Note

Some of the modules may require additional dependencies before being installed through pip. If you have an apt-based system, consider installing the most commonly missing dependencies (with the right pip):

apt update
apt install clang lib{jpeg-turbo,webp}-dev python{,-dev} zlib-dev
pip install -U --user setuptools
pip install -U --user telethon cryptg pillow

Thanks to @bb010g for writing down this nice list.

Signing In

Before working with Telegram’s API, you need to get your own API ID and hash:

  1. Login to your Telegram account with the phone number of the developer account to use.
  2. Click under API Development tools.
  3. A Create new application window will appear. Fill in your application details. There is no need to enter any URL, and only the first two fields (App title and Short name) can currently be changed later.
  4. Click on Create application at the end. Remember that your API hash is secret and Telegram won’t let you revoke it. Don’t post it anywhere!

Note

This API ID and hash is the one used by your application, not your phone number. You can use this API ID and hash with any phone number or even for bot accounts.

Editing the Code

This is a little introduction for those new to Python programming in general.

We will write our code inside hello.py, so you can use any text editor that you like. To run the code, use python3 hello.py from the terminal.

Important

Don’t call your script telethon.py! Python will try to import the client from there and it will fail with an error such as “ImportError: cannot import name ‘TelegramClient’ …”.

Signing In

We can finally write some code to log into our account!

from telethon import TelegramClient

# Use your own values from my.telegram.org
api_id = 12345
api_hash = '0123456789abcdef0123456789abcdef'

# The first parameter is the .session file name (absolute paths allowed)
with TelegramClient('anon', api_id, api_hash) as client:
    client.loop.run_until_complete(client.send_message('me', 'Hello, myself!'))

In the first line, we import the class name so we can create an instance of the client. Then, we define variables to store our API ID and hash conveniently.

At last, we create a new TelegramClient instance and call it client. We can now use the client variable for anything that we want, such as sending a message to ourselves.

Note

Since Telethon is an asynchronous library, you need to await coroutine functions to have them run (or otherwise, run the loop until they are complete). In this tiny example, we don’t bother making an async def main().

See Mastering asyncio to find out more.

Using a with block is the preferred way to use the library. It will automatically start() the client, logging or signing up if necessary.

If the .session file already existed, it will not login again, so be aware of this if you move or rename the file!

Signing In as a Bot Account

You can also use Telethon for your bots (normal bot accounts, not users). You will still need an API ID and hash, but the process is very similar:

from telethon.sync import TelegramClient

api_id = 12345
api_hash = '0123456789abcdef0123456789abcdef'
bot_token = '12345:0123456789abcdef0123456789abcdef'

# We have to manually call "start" if we want an explicit bot token
bot = TelegramClient('bot', api_id, api_hash).start(bot_token=bot_token)

# But then we can use the client instance as usual
with bot:
    ...

To get a bot account, you need to talk with @BotFather.

Signing In behind a Proxy

If you need to use a proxy to access Telegram, you will need to either:

and then change

TelegramClient('anon', api_id, api_hash)

with

TelegramClient('anon', api_id, api_hash, proxy=("socks5", '127.0.0.1', 4444))

(of course, replacing the protocol, IP and port with the protocol, IP and port of the proxy).

The proxy= argument should be a dict (or tuple, for backwards compatibility), consisting of parameters described in PySocks usage.

The allowed values for the argument proxy_type are:

  • For Python <= 3.5:
    • socks.SOCKS5 or 'socks5'
    • socks.SOCKS4 or 'socks4'
    • socks.HTTP or 'http'
  • For Python >= 3.6:
    • All of the above
    • python_socks.ProxyType.SOCKS5
    • python_socks.ProxyType.SOCKS4
    • python_socks.ProxyType.HTTP

Example:

proxy = {
    'proxy_type': 'socks5', # (mandatory) protocol to use (see above)
    'addr': '1.1.1.1',      # (mandatory) proxy IP address
    'port': 5555,           # (mandatory) proxy port number
    'username': 'foo',      # (optional) username if the proxy requires auth
    'password': 'bar',      # (optional) password if the proxy requires auth
    'rdns': True            # (optional) whether to use remote or local resolve, default remote
}

For backwards compatibility with PySocks the following format is possible (but discouraged):

proxy = (socks.SOCKS5, '1.1.1.1', 5555, True, 'foo', 'bar')

Using MTProto Proxies

MTProto Proxies are Telegram’s alternative to normal proxies, and work a bit differently. The following protocols are available:

  • ConnectionTcpMTProxyAbridged
  • ConnectionTcpMTProxyIntermediate
  • ConnectionTcpMTProxyRandomizedIntermediate (preferred)

For now, you need to manually specify these special connection modes if you want to use a MTProto Proxy. Your code would look like this:

from telethon import TelegramClient, connection
#   we need to change the connection ^^^^^^^^^^

client = TelegramClient(
    'anon',
    api_id,
    api_hash,

    # Use one of the available connection modes.
    # Normally, this one works with most proxies.
    connection=connection.ConnectionTcpMTProxyRandomizedIntermediate,

    # Then, pass the proxy details as a tuple:
    #     (host name, port, proxy secret)
    #
    # If the proxy has no secret, the secret must be:
    #     '00000000000000000000000000000000'
    proxy=('mtproxy.example.com', 2002, 'secret')
)

In future updates, we may make it easier to use MTProto Proxies (such as avoiding the need to manually pass connection=).

In short, the same code above but without comments to make it clearer:

from telethon import TelegramClient, connection

client = TelegramClient(
    'anon', api_id, api_hash,
    connection=connection.ConnectionTcpMTProxyRandomizedIntermediate,
    proxy=('mtproxy.example.com', 2002, 'secret')
)

Quick-Start

Let’s see a longer example to learn some of the methods that the library has to offer. These are known as “friendly methods”, and you should always use these if possible.

from telethon import TelegramClient

# Remember to use your own values from my.telegram.org!
api_id = 12345
api_hash = '0123456789abcdef0123456789abcdef'
client = TelegramClient('anon', api_id, api_hash)

async def main():
    # Getting information about yourself
    me = await client.get_me()

    # "me" is a user object. You can pretty-print
    # any Telegram object with the "stringify" method:
    print(me.stringify())

    # When you print something, you see a representation of it.
    # You can access all attributes of Telegram objects with
    # the dot operator. For example, to get the username:
    username = me.username
    print(username)
    print(me.phone)

    # You can print all the dialogs/conversations that you are part of:
    async for dialog in client.iter_dialogs():
        print(dialog.name, 'has ID', dialog.id)

    # You can send messages to yourself...
    await client.send_message('me', 'Hello, myself!')
    # ...to some chat ID
    await client.send_message(-100123456, 'Hello, group!')
    # ...to your contacts
    await client.send_message('+34600123123', 'Hello, friend!')
    # ...or even to any username
    await client.send_message('username', 'Testing Telethon!')

    # You can, of course, use markdown in your messages:
    message = await client.send_message(
        'me',
        'This message has **bold**, `code`, __italics__ and '
        'a [nice website](https://example.com)!',
        link_preview=False
    )

    # Sending a message returns the sent message object, which you can use
    print(message.raw_text)

    # You can reply to messages directly if you have a message object
    await message.reply('Cool!')

    # Or send files, songs, documents, albums...
    await client.send_file('me', '/home/me/Pictures/holidays.jpg')

    # You can print the message history of any chat:
    async for message in client.iter_messages('me'):
        print(message.id, message.text)

        # You can download media from messages, too!
        # The method will return the path where the file was saved.
        if message.photo:
            path = await message.download_media()
            print('File saved to', path)  # printed after download is done

with client:
    client.loop.run_until_complete(main())

Here, we show how to sign in, get information about yourself, send messages, files, getting chats, printing messages, and downloading files.

You should make sure that you understand what the code shown here does, take note on how methods are called and used and so on before proceeding. We will see all the available methods later on.

Important

Note that Telethon is an asynchronous library, and as such, you should get used to it and learn a bit of basic asyncio. This will help a lot. As a quick start, this means you generally want to write all your code inside some async def like so:

client = ...

async def do_something(me):
    ...

async def main():
    # Most of your code should go here.
    # You can of course make and use your own async def (do_something).
    # They only need to be async if they need to await things.
    me = await client.get_me()
    await do_something(me)

with client:
    client.loop.run_until_complete(main())

After you understand this, you may use the telethon.sync hack if you want do so (see Compatibility and Convenience), but note you may run into other issues (iPython, Anaconda, etc. have some issues with it).

Updates

Updates are an important topic in a messaging platform like Telegram. After all, you want to be notified when a new message arrives, when a member joins, when someone starts typing, etc. For that, you can use events.

Important

It is strongly advised to enable logging when working with events, since exceptions in event handlers are hidden by default. Please add the following snippet to the very top of your file:

import logging
logging.basicConfig(format='[%(levelname) 5s/%(asctime)s] %(name)s: %(message)s',
                    level=logging.WARNING)

Getting Started

Let’s start things with an example to automate replies:

from telethon import TelegramClient, events

client = TelegramClient('anon', api_id, api_hash)

@client.on(events.NewMessage)
async def my_event_handler(event):
    if 'hello' in event.raw_text:
        await event.reply('hi!')

client.start()
client.run_until_disconnected()

This code isn’t much, but there might be some things unclear. Let’s break it down:

from telethon import TelegramClient, events

client = TelegramClient('anon', api_id, api_hash)

This is normal creation (of course, pass session name, API ID and hash). Nothing we don’t know already.

@client.on(events.NewMessage)

This Python decorator will attach itself to the my_event_handler definition, and basically means that on a NewMessage event, the callback function you’re about to define will be called:

async def my_event_handler(event):
    if 'hello' in event.raw_text:
        await event.reply('hi!')

If a NewMessage event occurs, and 'hello' is in the text of the message, we reply() to the event with a 'hi!' message.

Note

Event handlers must be async def. After all, Telethon is an asynchronous library based on asyncio, which is a safer and often faster approach to threads.

You must await all method calls that use network requests, which is most of them.

More Examples

Replying to messages with hello is fun, but, can we do more?

@client.on(events.NewMessage(outgoing=True, pattern=r'\.save'))
async def handler(event):
    if event.is_reply:
        replied = await event.get_reply_message()
        sender = replied.sender
        await client.download_profile_photo(sender)
        await event.respond('Saved your photo {}'.format(sender.username))

We could also get replies. This event filters outgoing messages (only those that we send will trigger the method), then we filter by the regex r'\.save', which will match messages starting with ".save".

Inside the method, we check whether the event is replying to another message or not. If it is, we get the reply message and the sender of that message, and download their profile photo.

Let’s delete messages which contain “heck”. We don’t allow swearing here.

@client.on(events.NewMessage(pattern=r'(?i).*heck'))
async def handler(event):
    await event.delete()

With the r'(?i).*heck' regex, we match case-insensitive “heck” anywhere in the message. Regex is very powerful and you can learn more at https://regexone.com/.

So far, we have only seen the NewMessage, but there are many more which will be covered later. This is only a small introduction to updates.

Entities

When you need the user or chat where an event occurred, you must use the following methods:

async def handler(event):
    # Good
    chat = await event.get_chat()
    sender = await event.get_sender()
    chat_id = event.chat_id
    sender_id = event.sender_id

    # BAD. Don't do this
    chat = event.chat
    sender = event.sender
    chat_id = event.chat.id
    sender_id = event.sender.id

Events are like messages, but don’t have all the information a message has! When you manually get a message, it will have all the information it needs. When you receive an update about a message, it won’t have all the information, so you have to use the methods, not the properties.

Make sure you understand the code seen here before continuing! As a rule of thumb, remember that new message events behave just like message objects, so you can do with them everything you can do with a message object.

Next Steps

These basic first steps should have gotten you started with the library.

By now, you should know how to call friendly methods and how to work with the returned objects, how things work inside event handlers, etc.

Next, we will see a quick reference summary of all the methods and properties that you will need when using the library. If you follow the links there, you will expand the documentation for the method and property, with more examples on how to use them.

Therefore, you can find an example on every method of the client to learn how to use it, as well as a description of all the arguments.

After that, we will go in-depth with some other important concepts that are worth learning and understanding.

From now on, you can keep pressing the “Next” button if you want, or use the menu on the left, since some pages are quite lengthy.

A note on developing applications

If you’re using the library to make an actual application (and not just automate things), you should make sure to comply with the ToS:

[…] when logging in as an existing user, apps are supposed to call [GetTermsOfServiceUpdate] to check for any updates to the Terms of Service; this call should be repeated after expires seconds have elapsed. If an update to the Terms Of Service is available, clients are supposed to show a consent popup; if accepted, clients should call [AcceptTermsOfService], providing the termsOfService id JSON object; in case of denial, clients are to delete the account using [DeleteAccount], providing Decline ToS update as deletion reason.

However, if you use the library to automate or enhance your Telegram experience, it’s very likely that you are using other applications doing this check for you (so you wouldn’t run the risk of violating the ToS).

The library itself will not automatically perform this check or accept the ToS because it should require user action (the only exception is during sign-up).

FAQ

Let’s start the quick references section with some useful tips to keep in mind, with the hope that you will understand why certain things work the way that they do.

Code without errors doesn’t work

Then it probably has errors, but you haven’t enabled logging yet. To enable logging, at the following code to the top of your main file:

import logging
logging.basicConfig(format='[%(levelname) 5s/%(asctime)s] %(name)s: %(message)s',
                    level=logging.WARNING)

You can change the logging level to be something different, from less to more information:

level=logging.CRITICAL  # won't show errors (same as disabled)
level=logging.ERROR     # will only show errors that you didn't handle
level=logging.WARNING   # will also show messages with medium severity, such as internal Telegram issues
level=logging.INFO      # will also show informational messages, such as connection or disconnections
level=logging.DEBUG     # will show a lot of output to help debugging issues in the library

See the official Python documentation for more information on logging.

How can I except FloodWaitError?

You can use all errors from the API by importing:

from telethon import errors

And except them as such:

try:
    await client.send_message(chat, 'Hi')
except errors.FloodWaitError as e:
    # e.seconds is how many seconds you have
    # to wait before making the request again.
    print('Flood for', e.seconds)

My account was deleted/limited when using the library

First and foremost, this is not a problem exclusive to Telethon. Any third-party library is prone to cause the accounts to appear banned. Even official applications can make Telegram ban an account under certain circumstances. Third-party libraries such as Telethon are a lot easier to use, and as such, they are misused to spam, which causes Telegram to learn certain patterns and ban suspicious activity.

There is no point in Telethon trying to circumvent this. Even if it succeeded, spammers would then abuse the library again, and the cycle would repeat.

The library will only do things that you tell it to do. If you use the library with bad intentions, Telegram will hopefully ban you.

However, you may also be part of a limited country, such as Iran or Russia. In that case, we have bad news for you. Telegram is much more likely to ban these numbers, as they are often used to spam other accounts, likely through the use of libraries like this one. The best advice we can give you is to not abuse the API, like calling many requests really quickly.

We have also had reports from Kazakhstan and China, where connecting would fail. To solve these connection problems, you should use a proxy.

Telegram may also ban virtual (VoIP) phone numbers, as again, they’re likely to be used for spam.

More recently (year 2023 onwards), Telegram has started putting a lot more measures to prevent spam (with even additions such as anonymous participants in groups or the inability to fetch group members at all). This means some of the anti-spam measures have gotten more aggressive.

The recommendation has usually been to use the library only on well-established accounts (and not an account you just created), and to not perform actions that could be seen as abuse. Telegram decides what those actions are, and they’re free to change how they operate at any time.

If you want to check if your account has been limited, simply send a private message to @SpamBot through Telegram itself. You should notice this by getting errors like PeerFloodError, which means you’re limited, for instance, when sending a message to some accounts but not others.

For more discussion, please see issue 297.

How can I use a proxy?

This was one of the first things described in Signing In.

How do I access a field?

This is basic Python knowledge. You should use the dot operator:

me = await client.get_me()
print(me.username)
#       ^ we used the dot operator to access the username attribute

result = await client(functions.photos.GetUserPhotosRequest(
    user_id='me',
    offset=0,
    max_id=0,
    limit=100
))

# Working with list is also pretty basic
print(result.photos[0].sizes[-1].type)
#           ^       ^ ^       ^ ^
#           |       | |       | \ type
#           |       | |       \ last size
#           |       | \ list of sizes
#  access   |       \ first photo from the list
#  the...   \ list of photos
#
# To print all, you could do (or mix-and-match):
for photo in result.photos:
    for size in photo.sizes:
        print(size.type)

AttributeError: ‘coroutine’ object has no attribute ‘id’

You either forgot to:

import telethon.sync
#              ^^^^^ import sync

Or:

async def handler(event):
    me = await client.get_me()
    #    ^^^^^ note the await
    print(me.username)

sqlite3.OperationalError: database is locked

An older process is still running and is using the same 'session' file.

This error occurs when two or more clients use the same session, that is, when you write the same session name to be used in the client:

  • You have an older process using the same session file.
  • You have two different scripts running (interactive sessions count too).
  • You have two clients in the same script running at the same time.

The solution is, if you need two clients, use two sessions. If the problem persists and you’re on Linux, you can use fuser my.session to find out the process locking the file. As a last resort, you can reboot your system.

If you really dislike SQLite, use a different session storage. There is an entire section covering that at Session Files.

event.chat or event.sender is None

Telegram doesn’t always send this information in order to save bandwidth. If you need the information, you should fetch it yourself, since the library won’t do unnecessary work unless you need to:

async def handler(event):
    chat = await event.get_chat()
    sender = await event.get_sender()

File download is slow or sending files takes too long

The communication with Telegram is encrypted. Encryption requires a lot of math, and doing it in pure Python is very slow. cryptg is a library which containns the encryption functions used by Telethon. If it is installed (via pip install cryptg), it will automatically be used and should provide a considerable speed boost. You can know whether it’s used by configuring logging (at INFO level or lower) before importing telethon.

Note that the library does not download or upload files in parallel, which can also help with the speed of downloading or uploading a single file. There are snippets online implementing that. The reason why this is not built-in is because the limiting factor in the long run are FloodWaitError, and using parallel download or uploads only makes them occur sooner.

What does “Server sent a very new message with ID” mean?

You may also see this error as “Server sent a very old message with ID”.

This is a security feature from Telethon that cannot be disabled and is meant to protect you against replay attacks.

When this message is incorrectly reported as a “bug”, the most common patterns seem to be:

  • Your system time is incorrect.
  • The proxy you’re using may be interfering somehow.
  • The Telethon session is being used or has been used from somewhere else. Make sure that you created the session from Telethon, and are not using the same session anywhere else. If you need to use the same account from multiple places, login and use a different session for each place you need.

What does “Server replied with a wrong session ID” mean?

This is a security feature from Telethon that cannot be disabled and is meant to protect you against unwanted session reuse.

When this message is reported as a “bug”, the most common patterns seem to be:

  • The proxy you’re using may be interfering somehow.
  • The Telethon session is being used or has been used from somewhere else. Make sure that you created the session from Telethon, and are not using the same session anywhere else. If you need to use the same account from multiple places, login and use a different session for each place you need.
  • You may be using multiple connections to the Telegram server, which seems to confuse Telegram.

Most of the time it should be safe to ignore this warning. If the library still doesn’t behave correctly, make sure to check if any of the above bullet points applies in your case and try to work around it.

If the issue persists and there is a way to reliably reproduce this error, please add a comment with any additional details you can provide to issue 3759, and perhaps some additional investigation can be done (but it’s unlikely, as Telegram is sending unexpected data).

What does “Could not find a matching Constructor ID for the TLObject” mean?

Telegram uses “layers”, which you can think of as “versions” of the API they offer. When Telethon reads responses that the Telegram servers send, these need to be deserialized (into what Telethon calls “TLObjects”).

Every Telethon version understands a single Telegram layer. When Telethon connects to Telegram, both agree on the layer to use. If the layers don’t match, Telegram may send certain objects which Telethon no longer understands.

When this message is reported as a “bug”, the most common patterns seem to be that he Telethon session is being used or has been used from somewhere else. Make sure that you created the session from Telethon, and are not using the same session anywhere else. If you need to use the same account from multiple places, login and use a different session for each place you need.

What does “Task was destroyed but it is pending” mean?

Your script likely finished abruptly, the asyncio event loop got destroyed, and the library did not get a chance to properly close the connection and close the session.

Make sure you’re either using the context manager for the client or always call await client.disconnect() (by e.g. using a try/finally).

What does “The asyncio event loop must not change after connection” mean?

Telethon uses asyncio, and makes use of things like tasks and queues internally to manage the connection to the server and match responses to the requests you make. Most of them are initialized after the client is connected.

For example, if the library expects a result to a request made in loop A, but you attempt to get that result in loop B, you will very likely find a deadlock. To avoid a deadlock, the library checks to make sure the loop in use is the same as the one used to initialize everything, and if not, it throws an error.

The most common cause is asyncio.run, since it creates a new event loop. If you asyncio.run a function to create the client and set it up, and then you asyncio.run another function to do work, things won’t work, so the library throws an error early to let you know something is wrong.

Instead, it’s often a good idea to have a single async def main and simply asyncio.run() it and do all the work there. From it, you’re also able to call other async def without having to touch asyncio.run again:

# It's fine to create the client outside as long as you don't connect
client = TelegramClient(...)

async def main():
    # Now the client will connect, so the loop must not change from now on.
    # But as long as you do all the work inside main, including calling
    # other async functions, things will work.
    async with client:
        ....

if __name__ == '__main__':
    asyncio.run(main())

Be sure to read the asyncio documentation if you want a better understanding of event loop, tasks, and what functions you can use.

What does “bases ChatGetter” mean?

In Python, classes can base others. This is called inheritance. What it means is that “if a class bases another, you can use the other’s methods too”.

For example, Message bases ChatGetter. In turn, ChatGetter defines things like obj.chat_id.

So if you have a message, you can access that too:

# ChatGetter has a chat_id property, and Message bases ChatGetter.
# Thus you can use ChatGetter properties and methods from Message
print(message.chat_id)

Telegram has a lot to offer, and inheritance helps the library reduce boilerplate, so it’s important to know this concept. For newcomers, this may be a problem, so we explain what it means here in the FAQ.

Can I send files by ID?

When people talk about IDs, they often refer to one of two things: the integer ID inside media, and a random-looking long string.

You cannot use the integer ID to send media. Generally speaking, sending media requires a combination of ID, access_hash and file_reference. The first two are integers, while the last one is a random bytes sequence.

  • The integer id will always be the same for every account, so every user or bot looking at a particular media file, will see a consistent ID.
  • The access_hash will always be the same for a given account, but different accounts will each see their own, different access_hash. This makes it impossible to get media object from one account and use it in another. The other account must fetch the media object itself.
  • The file_reference is random for everyone and will only work for a few hours before it expires. It must be refetched before the media can be used (to either resend the media or download it).

The second type of “file ID” people refer to is a concept from the HTTP Bot API. It’s a custom format which encodes enough information to use the media.

Telethon provides an old version of these HTTP Bot API-style file IDs via message.file.id, however, this feature is no longer maintained, so it may not work. It will be removed in future versions. Nonetheless, it is possible to find a different Python package (or write your own) to parse these file IDs and construct the necessary input file objects to send or download the media.

Can I use Flask with the library?

Yes, if you know what you are doing. However, you will probably have a lot of headaches to get threads and asyncio to work together. Instead, consider using Quart, an asyncio-based alternative to Flask.

Check out quart_login.py for an example web-application based on Quart.

Can I use Anaconda/Spyder/IPython with the library?

Yes, but these interpreters run the asyncio event loop implicitly, which interferes with the telethon.sync magic module.

If you use them, you should not import sync:

# Change any of these...:
from telethon import TelegramClient, sync, ...
from telethon.sync import TelegramClient, ...

# ...with this:
from telethon import TelegramClient, ...

You are also more likely to get “sqlite3.OperationalError: database is locked” with them. If they cause too much trouble, just write your code in a .py file and run that, or use the normal python interpreter.

Client Reference

This page contains a summary of all the important methods and properties that you may need when using Telethon. They are sorted by relevance and are not in alphabetical order.

You should use this page to learn about which methods are available, and if you need a usage example or further description of the arguments, be sure to follow the links.

TelegramClient

This is a summary of the methods and properties you will find at TelegramClient.

Auth
start Starts the client (connects and logs in if necessary).
send_code_request Sends the Telegram code needed to login to the given phone number.
sign_in Logs in to Telegram to an existing user or bot account.
qr_login Initiates the QR login procedure.
log_out Logs out Telegram and deletes the current *.session file.
edit_2fa Changes the 2FA settings of the logged in user.
Base
connect Connects to Telegram.
disconnect Disconnects from Telegram.
is_connected Returns True if the user has connected.
disconnected Property with a Future that resolves upon disconnection.
loop Property with the asyncio event loop used by this client.
set_proxy Changes the proxy which will be used on next (re)connection.
Messages
send_message Sends a message to the specified user, chat or channel.
edit_message Edits the given message to change its text or media.
delete_messages Deletes the given messages, optionally “for everyone”.
forward_messages Forwards the given messages to the specified entity.
iter_messages Iterator over the messages for the given chat.
get_messages Same as iter_messages(), but returns a TotalList instead.
pin_message Pins a message in a chat.
unpin_message Unpins a message in a chat.
send_read_acknowledge Marks messages as read and optionally clears mentions.
Uploads
send_file Sends message with the given file to the specified entity.
upload_file Uploads a file to Telegram’s servers, without sending it.
Downloads
download_media Downloads the given media from a message object.
download_profile_photo Downloads the profile photo from the given user, chat or channel.
download_file Low-level method to download files from their input location.
iter_download Iterates over a file download, yielding chunks of the file.
Dialogs
iter_dialogs Iterator over the dialogs (open conversations/subscribed channels).
get_dialogs Same as iter_dialogs(), but returns a TotalList instead.
edit_folder Edits the folder used by one or more dialogs to archive them.
iter_drafts Iterator over draft messages.
get_drafts Same as iter_drafts(), but returns a list instead.
delete_dialog Deletes a dialog (leaves a chat or channel).
conversation Creates a Conversation with the given entity.
Users
get_me Gets “me”, the current User who is logged in.
is_bot Return True if the signed-in user is a bot, False otherwise.
is_user_authorized Returns True if the user is authorized (logged in).
get_entity Turns the given entity into a valid Telegram User, Chat or Channel.
get_input_entity Turns the given entity into its input entity version.
get_peer_id Gets the ID for the given entity.
Chats
iter_participants Iterator over the participants belonging to the specified chat.
get_participants Same as iter_participants(), but returns a TotalList instead.
kick_participant Kicks a user from a chat.
iter_admin_log Iterator over the admin log for the specified channel.
get_admin_log Same as iter_admin_log(), but returns a list instead.
iter_profile_photos Iterator over a user’s profile photos or a chat’s photos.
get_profile_photos Same as iter_profile_photos(), but returns a TotalList instead.
edit_admin Edits admin permissions for someone in a chat.
edit_permissions Edits user restrictions in a chat.
get_permissions Fetches the permissions of a user in a specific chat or channel or get Default Restricted Rights of Chat or Channel.
get_stats Retrieves statistics from the given megagroup or broadcast channel.
action Returns a context-manager object to represent a “chat action”.
Parse Mode
parse_mode This property is the default parse mode used when sending messages.
Updates
on Decorator used to add_event_handler more conveniently.
run_until_disconnected Runs the event loop until the library is disconnected.
add_event_handler Registers a new event handler callback.
remove_event_handler Inverse operation of add_event_handler().
list_event_handlers Lists all registered event handlers.
catch_up “Catches up” on the missed updates while the client was offline.
set_receive_updates Change the value of receive_updates.
Bots
inline_query Makes an inline query to the specified bot (@vote New Poll).
Account
takeout Returns a TelegramClient which calls methods behind a takeout session.
end_takeout Finishes the current takeout session.

Events Reference

Here you will find a quick summary of all the methods and properties that you can access when working with events.

You can access the client that creates this event by doing event.client, and you should view the description of the events to find out what arguments it allows on creation and its attributes (the properties will be shown here).

Important

Remember that all events base ChatGetter! Please see FAQ if you don’t know what this means or the implications of it.

NewMessage

Occurs whenever a new text message or a message with media arrives.

Note

The new message event should be treated as a normal Message, with the following exceptions:

  • pattern_match is the match object returned by pattern=.
  • message is not the message string. It’s the Message object.

Remember, this event is just a proxy over the message, so while you won’t see its attributes and properties, you can still access them. Please see the full documentation for examples.

Full documentation for the NewMessage.

MessageEdited

Occurs whenever a message is edited. Just like NewMessage, you should treat this event as a Message.

Full documentation for the MessageEdited.

MessageDeleted

Occurs whenever a message is deleted. Note that this event isn’t 100% reliable, since Telegram doesn’t always notify the clients that a message was deleted.

It only has the deleted_id and deleted_ids attributes (in addition to the chat if the deletion happened in a channel).

Full documentation for the MessageDeleted.

MessageRead

Occurs whenever one or more messages are read in a chat.

Full documentation for the MessageRead.

inbox True if you have read someone else’s messages.
message_ids The IDs of the messages which contents’ were read.
get_messages Returns the list of Message which contents’ were read.
is_read Returns True if the given message (or its ID) has been read.

ChatAction

Occurs on certain chat actions, such as chat title changes, user join or leaves, pinned messages, photo changes, etc.

Full documentation for the ChatAction.

added_by The user who added users, if applicable (None otherwise).
kicked_by The user who kicked users, if applicable (None otherwise).
user The first user that takes part in this action.
input_user Input version of the self.user property.
user_id Returns the marked signed ID of the first user, if any.
users A list of users that take part in this action.
input_users Input version of the self.users property.
user_ids Returns the marked signed ID of the users, if any.
respond Responds to the chat action message (not as a reply).
reply Replies to the chat action message (as a reply).
delete Deletes the chat action message.
get_pinned_message If new_pin is True, this returns the Message object that was pinned.
get_added_by Returns added_by but will make an API call if necessary.
get_kicked_by Returns kicked_by but will make an API call if necessary.
get_user Returns user but will make an API call if necessary.
get_input_user Returns input_user but will make an API call if necessary.
get_users Returns users but will make an API call if necessary.
get_input_users Returns input_users but will make an API call if necessary.

UserUpdate

Occurs whenever a user goes online, starts typing, etc.

Full documentation for the UserUpdate.

user Alias for sender.
input_user Alias for input_sender.
user_id Alias for sender_id.
get_user Alias for get_sender.
get_input_user Alias for get_input_sender.
typing True if the action is typing a message.
uploading True if the action is uploading something.
recording True if the action is recording something.
playing True if the action is playing a game.
cancel True if the action was cancelling other actions.
geo True if what’s being uploaded is a geo.
audio True if what’s being recorded/uploaded is an audio.
round True if what’s being recorded/uploaded is a round video.
video True if what’s being recorded/uploaded is an video.
contact True if what’s being uploaded (selected) is a contact.
document True if what’s being uploaded is document.
photo True if what’s being uploaded is a photo.
last_seen Exact datetime.datetime when the user was last seen if known.
until The datetime.datetime until when the user should appear online.
online True if the user is currently online,
recently True if the user was seen within a day.
within_weeks True if the user was seen within 7 days.
within_months True if the user was seen within 30 days.

CallbackQuery

Occurs whenever you sign in as a bot and a user clicks one of the inline buttons on your messages.

Full documentation for the CallbackQuery.

id Returns the query ID.
message_id Returns the message ID to which the clicked inline button belongs.
data Returns the data payload from the original inline button.
chat_instance Unique identifier for the chat where the callback occurred.
via_inline Whether this callback was generated from an inline button sent via an inline query or not.
respond Responds to the message (not as a reply).
reply Replies to the message (as a reply).
edit Edits the message.
delete Deletes the message.
answer Answers the callback query (and stops the loading circle).
get_message Returns the message to which the clicked inline button belongs.

InlineQuery

Occurs whenever you sign in as a bot and a user sends an inline query such as @bot query.

Full documentation for the InlineQuery.

id Returns the unique identifier for the query ID.
text Returns the text the user used to make the inline query.
offset The string the user’s client used as an offset for the query.
geo If the user location is requested when using inline mode and the user’s device is able to send it, this will return the GeoPoint with the position of the user.
builder Returns a new InlineBuilder instance.
answer Answers the inline query with the given results.

Album

Occurs whenever you receive an entire album.

Full documentation for the Album.

grouped_id The shared grouped_id between all the messages.
text The message text of the first photo with a caption, formatted using the client’s default parse mode.
raw_text The raw message text of the first photo with a caption, ignoring any formatting.
is_reply True if the album is a reply to some other message.
forward The Forward information for the first message in the album if it was forwarded.
get_reply_message The Message that this album is replying to, or None.
respond Responds to the album (not as a reply).
reply Replies to the first photo in the album (as a reply).
forward_to Forwards the entire album.
edit Edits the first caption or the message, or the first messages’ caption if no caption is set, iff it’s outgoing.
delete Deletes the entire album.
mark_read Marks the entire album as read.
pin Pins the first photo in the album.

Raw

Raw events are not actual events. Instead, they are the raw Update object that Telegram sends. You normally shouldn’t need these.

Objects Reference

This is the quick reference for those objects returned by client methods or other useful modules that the library has to offer. They are kept in a separate page to help finding and discovering them.

Remember that this page only shows properties and methods, not attributes. Make sure to open the full documentation to find out about the attributes.

ChatGetter

All events base ChatGetter, and some of the objects below do too, so it’s important to know its methods.

chat Returns the User, Chat or Channel where this object belongs to.
input_chat This InputPeer is the input version of the chat where the message was sent.
chat_id Returns the marked chat integer ID.
is_private True if the message was sent as a private message.
is_group True if the message was sent on a group or megagroup.
is_channel True if the message was sent on a megagroup or channel.
get_chat Returns chat, but will make an API call to find the chat unless it’s already cached.
get_input_chat Returns input_chat, but will make an API call to find the input chat unless it’s already cached.

SenderGetter

Similar to ChatGetter, a SenderGetter is the same, but it works for senders instead.

sender Returns the User or Channel that sent this object.
input_sender This InputPeer is the input version of the user/channel who sent the message.
sender_id Returns the marked sender integer ID, if present.
get_sender Returns sender, but will make an API call to find the sender unless it’s already cached.
get_input_sender Returns input_sender, but will make an API call to find the input sender unless it’s already cached.

Message

The Message type is very important, mostly because we are working with a library for a messaging platform, so messages are widely used: in events, when fetching history, replies, etc.

It bases ChatGetter and SenderGetter.

Properties

Note

We document custom properties here, not all the attributes of the Message (which is the information Telegram actually returns).

text The message text, formatted using the client’s default parse mode.
raw_text The raw message text, ignoring any formatting.
is_reply True if the message is a reply to some other message.
forward The Forward information if this message is a forwarded message.
buttons Returns a list of lists of MessageButton, if any.
button_count Returns the total button count (sum of all buttons rows).
file Returns a File wrapping the photo or document in this message.
photo The Photo media in this message, if any.
document The Document media in this message, if any.
web_preview The WebPage media in this message, if any.
audio The Document media in this message, if it’s an audio file.
voice The Document media in this message, if it’s a voice note.
video The Document media in this message, if it’s a video.
video_note The Document media in this message, if it’s a video note.
gif The Document media in this message, if it’s a “gif”.
sticker The Document media in this message, if it’s a sticker.
contact The MessageMediaContact in this message, if it’s a contact.
game The Game media in this message, if it’s a game.
geo The GeoPoint media in this message, if it has a location.
invoice The MessageMediaInvoice in this message, if it’s an invoice.
poll The MessageMediaPoll in this message, if it’s a poll.
venue The MessageMediaVenue in this message, if it’s a venue.
action_entities Returns a list of entities that took part in this action.
via_bot The bot User if the message was sent via said bot.
via_input_bot Returns the input variant of via_bot.
client Returns the TelegramClient that patched this message.
Methods
respond Responds to the message (not as a reply).
reply Replies to the message (as a reply).
forward_to Forwards the message.
edit Edits the message if it’s outgoing.
delete Deletes the message.
get_reply_message The Message that this message is replying to, or None.
click Calls SendVote with the specified poll option or button.click on the specified button.
mark_read Marks the message as read.
pin Pins the message.
download_media Downloads the media contained in the message, if any.
get_entities_text Returns a list of (markup entity, inner text) (like bold or italics).
get_buttons Returns buttons when that property fails (this is rarely needed).

File

The File type is a wrapper object returned by Message.file, and you can use it to easily access a document’s attributes, such as its name, bot-API style file ID, etc.

id The old bot-API style file_id representing this file.
name The file name of this document.
ext The extension from the mime type of this file.
mime_type The mime-type of this file.
width The width in pixels of this media if it’s a photo or a video.
height The height in pixels of this media if it’s a photo or a video.
size The size in bytes of this file.
duration The duration in seconds of the audio or video.
title The title of the song.
performer The performer of the song.
emoji A string with all emoji that represent the current sticker.
sticker_set The InputStickerSet to which the sticker file belongs.

Conversation

The Conversation object is returned by the client.conversation() method to easily send and receive responses like a normal conversation.

It bases ChatGetter.

send_message Sends a message in the context of this conversation.
send_file Sends a file in the context of this conversation.
mark_read Marks as read the latest received message if message is None.
get_response Gets the next message that responds to a previous one.
get_reply Gets the next message that explicitly replies to a previous one.
get_edit Awaits for an edit after the last message to arrive.
wait_read Awaits for the sent message to be marked as read.
wait_event Waits for a custom event to occur.
cancel Cancels the current conversation.
cancel_all Calls cancel on all conversations in this chat.

AdminLogEvent

The AdminLogEvent object is returned by the client.iter_admin_log() method to easily iterate over past “events” (deleted messages, edits, title changes, leaving members…)

These are all the properties you can find in it:

id The ID of this event.
date The date when this event occurred.
user_id The ID of the user that triggered this event.
action The original ChannelAdminLogEventAction.
old The old value from the event.
new The new value present in the event.
changed_about Whether the channel’s about was changed or not.
changed_title Whether the channel’s title was changed or not.
changed_username Whether the channel’s username was changed or not.
changed_photo Whether the channel’s photo was changed or not.
changed_sticker_set Whether the channel’s sticker set was changed or not.
changed_message Whether a message in this channel was edited or not.
deleted_message Whether a message in this channel was deleted or not.
changed_admin Whether the permissions for an admin in this channel changed or not.
changed_restrictions Whether a message in this channel was edited or not.
changed_invites Whether the invites in the channel were toggled or not.
joined Whether user joined through the channel’s public username or not.
joined_invite Whether a new user joined through an invite link to the channel or not.
left Whether user left the channel or not.
changed_hide_history Whether hiding the previous message history for new members in the channel was toggled or not.
changed_signatures Whether the message signatures in the channel were toggled or not.
changed_pin Whether a new message in this channel was pinned or not.
changed_default_banned_rights Whether the default banned rights were changed or not.
stopped_poll Whether a poll was stopped or not.

Button

The Button class is used when you login as a bot account to send messages with reply markup, such as inline buttons or custom keyboards.

These are the static methods you can use to create instances of the markup:

inline Creates a new inline button with some payload data in it.
switch_inline Creates a new inline button to switch to inline query.
url Creates a new inline button to open the desired URL on click.
auth Creates a new inline button to authorize the user at the given URL.
text Creates a new keyboard button with the given text.
request_location Creates a new keyboard button to request the user’s location on click.
request_phone Creates a new keyboard button to request the user’s phone on click.
request_poll Creates a new keyboard button to request the user to create a poll.
clear Clears all keyboard buttons after sending a message with this markup.
force_reply Forces a reply to the message with this markup.

InlineResult

The InlineResult object is returned inside a list by the client.inline_query() method to make an inline query to a bot that supports being used in inline mode, such as @like.

Note that the list returned is in fact a subclass of a list called InlineResults, which, in addition of being a list (iterator, indexed access, etc.), has extra attributes and methods.

These are the constants for the types, properties and methods you can find the individual results:

ARTICLE
PHOTO
GIF
VIDEO
VIDEO_GIF
AUDIO
DOCUMENT
LOCATION
VENUE
CONTACT
GAME
type The always-present type of this result.
message The always-present BotInlineMessage that will be sent if click is called on this result.
title The title for this inline result.
description The description for this inline result.
url The URL present in this inline results.
photo Returns either the WebDocument thumbnail for normal results or the Photo for media results.
document Returns either the WebDocument content for normal results or the Document for media results.
click Clicks this result and sends the associated message.
download_media Downloads the media in this result (if there is a document, the document will be downloaded; otherwise, the photo will if present).

Dialog

The Dialog object is returned when you call client.iter_dialogs().

send_message Sends a message to this dialog.
archive Archives (or un-archives) this dialog.
delete Deletes the dialog from your dialog list.

Draft

The Draft object is returned when you call client.iter_drafts().

entity The entity that belongs to this dialog (user, chat or channel).
input_entity Input version of the entity.
get_entity Returns entity but will make an API call if necessary.
get_input_entity Returns input_entity but will make an API call if necessary.
text The markdown text contained in the draft.
raw_text The raw (text without formatting) contained in the draft.
is_empty Convenience bool to determine if the draft is empty or not.
set_message Changes the draft message on the Telegram servers.
send Sends the contents of this draft to the dialog.
delete Deletes this draft, and returns True on success.

Utils

The telethon.utils module has plenty of methods that make using the library a lot easier. Only the interesting ones will be listed here.

get_display_name Gets the display name for the given User, Chat or Channel.
get_extension Gets the corresponding extension for any Telegram media.
get_inner_text Gets the inner text that’s surrounded by the given entities.
get_peer_id Convert the given peer into its marked ID by default.
resolve_id Given a marked ID, returns the original ID and its Peer type.
pack_bot_file_id Inverse operation for resolve_bot_file_id.
resolve_bot_file_id Given a Bot API-style file_id, returns the media it represents.
resolve_invite_link Resolves the given invite link.

String-based Debugging

Debugging is really important. Telegram’s API is really big and there are a lot of things that you should know. Such as, what attributes or fields does a result have? Well, the easiest thing to do is printing it:

entity = await client.get_entity('username')
print(entity)

That will show a huge string similar to the following:

Channel(id=1066197625, title='Telegram Usernames', photo=ChatPhotoEmpty(), date=datetime.datetime(2016, 12, 16, 15, 15, 43, tzinfo=datetime.timezone.utc), version=0, creator=False, left=True, broadcast=True, verified=True, megagroup=False, restricted=False, signatures=False, min=False, scam=False, has_link=False, has_geo=False, slowmode_enabled=False, access_hash=-6309373984955162244, username='username', restriction_reason=[], admin_rights=None, banned_rights=None, default_banned_rights=None, participants_count=None)

That’s a lot of text. But as you can see, all the properties are there. So if you want the title you don’t use regex or anything like splitting str(entity) to get what you want. You just access the attribute you need:

title = entity.title

Can we get better than the shown string, though? Yes!

print(entity.stringify())

Will show a much better representation:

Channel(
    id=1066197625,
    title='Telegram Usernames',
    photo=ChatPhotoEmpty(
    ),
    date=datetime.datetime(2016, 12, 16, 15, 15, 43, tzinfo=datetime.timezone.utc),
    version=0,
    creator=False,
    left=True,
    broadcast=True,
    verified=True,
    megagroup=False,
    restricted=False,
    signatures=False,
    min=False,
    scam=False,
    has_link=False,
    has_geo=False,
    slowmode_enabled=False,
    access_hash=-6309373984955162244,
    username='username',
    restriction_reason=[
    ],
    admin_rights=None,
    banned_rights=None,
    default_banned_rights=None,
    participants_count=None
)

Now it’s easy to see how we could get, for example, the year value. It’s inside date:

channel_year = entity.date.year

You don’t need to print everything to see what all the possible values can be. You can just search in http://tl.telethon.dev/.

Remember that you can use Python’s isinstance to check the type of something. For example:

from telethon import types

if isinstance(entity.photo, types.ChatPhotoEmpty):
    print('Channel has no photo')

Entities

The library widely uses the concept of “entities”. An entity will refer to any User, Chat or Channel object that the API may return in response to certain methods, such as GetUsersRequest.

Note

When something “entity-like” is required, it means that you need to provide something that can be turned into an entity. These things include, but are not limited to, usernames, exact titles, IDs, Peer objects, or even entire User, Chat and Channel objects and even phone numbers from people you have in your contact list.

To “encounter” an ID, you would have to “find it” like you would in the normal app. If the peer is in your dialogs, you would need to client.get_dialogs(). If the peer is someone in a group, you would similarly client.get_participants(group).

Once you have encountered an ID, the library will (by default) have saved their access_hash for you, which is needed to invoke most methods. This is why sometimes you might encounter this error when working with the library. You should except ValueError and run code that you know should work to find the entity.

What is an Entity?

A lot of methods and requests require entities to work. For example, you send a message to an entity, get the username of an entity, and so on.

There are a lot of things that work as entities: usernames, phone numbers, chat links, invite links, IDs, and the types themselves. That is, you can use any of those when you see an “entity” is needed.

Note

Remember that the phone number must be in your contact list before you can use it.

You should use, from better to worse:

  1. Input entities. For example, event.input_chat, message.input_sender, or caching an entity you will use a lot with entity = await client.get_input_entity(...).
  2. Entities. For example, if you had to get someone’s username, you can just use user or channel. It will work. Only use this option if you already have the entity!
  3. IDs. This will always look the entity up from the cache (the *.session file caches seen entities).
  4. Usernames, phone numbers and links. The cache will be used too (unless you force a client.get_entity()), but may make a request if the username, phone or link has not been found yet.

In recent versions of the library, the following two are equivalent:

async def handler(event):
    await client.send_message(event.sender_id, 'Hi')
    await client.send_message(event.input_sender, 'Hi')

If you need to be 99% sure that the code will work (sometimes it’s simply impossible for the library to find the input entity), or if you will reuse the chat a lot, consider using the following instead:

async def handler(event):
    # This method may make a network request to find the input sender.
    # Properties can't make network requests, so we need a method.
    sender = await event.get_input_sender()
    await client.send_message(sender, 'Hi')
    await client.send_message(sender, 'Hi')

Getting Entities

Through the use of the Session Files, the library will automatically remember the ID and hash pair, along with some extra information, so you’re able to just do this:

# (These examples assume you are inside an "async def")
#
# Dialogs are the "conversations you have open".
# This method returns a list of Dialog, which
# has the .entity attribute and other information.
#
# This part is IMPORTANT, because it fills the entity cache.
dialogs = await client.get_dialogs()

# All of these work and do the same.
username = await client.get_entity('username')
username = await client.get_entity('t.me/username')
username = await client.get_entity('https://telegram.dog/username')

# Other kind of entities.
channel = await client.get_entity('telegram.me/joinchat/AAAAAEkk2WdoDrB4-Q8-gg')
contact = await client.get_entity('+34xxxxxxxxx')
friend  = await client.get_entity(friend_id)

# Getting entities through their ID (User, Chat or Channel)
entity = await client.get_entity(some_id)

# You can be more explicit about the type for said ID by wrapping
# it inside a Peer instance. This is recommended but not necessary.
from telethon.tl.types import PeerUser, PeerChat, PeerChannel

my_user    = await client.get_entity(PeerUser(some_id))
my_chat    = await client.get_entity(PeerChat(some_id))
my_channel = await client.get_entity(PeerChannel(some_id))

Note

You don’t need to get the entity before using it! Just let the library do its job. Use a phone from your contacts, username, ID or input entity (preferred but not necessary), whatever you already have.

All methods in the TelegramClient call .get_input_entity() prior to sending the request to save you from the hassle of doing so manually. That way, convenience calls such as client.send_message('username', 'hi!') become possible.

Every entity the library encounters (in any response to any call) will by default be cached in the .session file (an SQLite database), to avoid performing unnecessary API calls. If the entity cannot be found, additonal calls like ResolveUsernameRequest or GetContactsRequest may be made to obtain the required information.

Entities vs. Input Entities

Note

This section is informative, but worth reading. The library will transparently handle all of these details for you.

On top of the normal types, the API also make use of what they call their Input* versions of objects. The input version of an entity (e.g. InputPeerUser, InputChat, etc.) only contains the minimum information that’s required from Telegram to be able to identify who you’re referring to: a Peer’s ID and hash. They are named like this because they are input parameters in the requests.

Entities’ ID are the same for all user and bot accounts, however, the access hash is different for each account, so trying to reuse the access hash from one account in another will not work.

Sometimes, Telegram only needs to indicate the type of the entity along with their ID. For this purpose, Peer versions of the entities also exist, which just have the ID. You cannot get the hash out of them since you should not be needing it. The library probably has cached it before.

Peers are enough to identify an entity, but they are not enough to make a request with them. You need to know their hash before you can “use them”, and to know the hash you need to “encounter” them, let it be in your dialogs, participants, message forwards, etc.

Note

You can use peers with the library. Behind the scenes, they are replaced with the input variant. Peers “aren’t enough” on their own but the library will do some more work to use the right type.

As we just mentioned, API calls don’t need to know the whole information about the entities, only their ID and hash. For this reason, another method, client.get_input_entity() is available. This will always use the cache while possible, making zero API calls most of the time. When a request is made, if you provided the full entity, e.g. an User, the library will convert it to the required InputPeer automatically for you.

You should always favour client.get_input_entity() over client.get_entity() for this reason! Calling the latter will always make an API call to get the most recent information about said entity, but invoking requests don’t need this information, just the InputPeer. Only use client.get_entity() if you need to get actual information, like the username, name, title, etc. of the entity.

To further simplify the workflow, since the version 0.16.2 of the library, the raw requests you make to the API are also able to call client.get_input_entity() wherever needed, so you can even do things like:

await client(SendMessageRequest('username', 'hello'))

The library will call the .resolve() method of the request, which will resolve 'username' with the appropriated InputPeer. Don’t worry if you don’t get this yet, but remember some of the details here are important.

Full Entities

In addition to PeerUser, InputPeerUser, User (and its variants for chats and channels), there is also the concept of UserFull.

This full variant has additional information such as whether the user is blocked, its notification settings, the bio or about of the user, etc.

There is also messages.ChatFull which is the equivalent of full entities for chats and channels, with also the about section of the channel. Note that the users field only contains bots for the channel (so that clients can suggest commands to use).

You can get both of these by invoking GetFullUser, GetFullChat and GetFullChannel respectively.

Accessing Entities

Although it’s explicitly noted in the documentation that messages subclass ChatGetter and SenderGetter, some people still don’t get inheritance.

When the documentation says “Bases: telethon.tl.custom.chatgetter.ChatGetter” it means that the class you’re looking at, also can act as the class it bases. In this case, ChatGetter knows how to get the chat where a thing belongs to.

So, a Message is a ChatGetter. That means you can do this:

message.is_private
message.chat_id
await message.get_chat()
# ...etc

SenderGetter is similar:

message.user_id
await message.get_input_user()
message.user
# ...etc

Quite a few things implement them, so it makes sense to reuse the code. For example, all events (except raw updates) implement ChatGetter since all events occur in some chat.

Summary

TL;DR; If you’re here because of “Could not find the input entity for”, you must ask yourself “how did I find this entity through official applications”? Now do the same with the library. Use what applies:

# (These examples assume you are inside an "async def")
async with client:
    # Does it have a username? Use it!
    entity = await client.get_entity(username)

    # Do you have a conversation open with them? Get dialogs.
    await client.get_dialogs()

    # Are they participant of some group? Get them.
    await client.get_participants('username')

    # Is the entity the original sender of a forwarded message? Get it.
    await client.get_messages('username', 100)

    # NOW you can use the ID, anywhere!
    await client.send_message(123456, 'Hi!')

    entity = await client.get_entity(123456)
    print(entity)

Once the library has “seen” the entity, you can use their integer ID. You can’t use entities from IDs the library hasn’t seen. You must make the library see them at least once and disconnect properly. You know where the entities are and you must tell the library. It won’t guess for you.

Chats vs Channels

Telegram’s raw API can get very confusing sometimes, in particular when it comes to talking about “chats”, “channels”, “groups”, “megagroups”, and all those concepts.

This section will try to explain what each of these concepts are.

Chats

A Chat can be used to talk about either the common “subclass” that both chats and channels share, or the concrete Chat type.

Technically, both Chat and Channel are a form of the Chat type.

Most of the time, the term Chat is used to talk about small group chats. When you create a group through an official application, this is the type that you get. Official applications refer to these as “Group”.

Both the bot API and Telethon will add a minus sign (negate) the real chat ID so that you can tell at a glance, with just a number, the entity type.

For example, if you create a chat with CreateChatRequest, the real chat ID might be something like 123. If you try printing it from a message.chat_id you will see -123. This ID helps Telethon know you’re talking about a Chat.

Channels

Official applications create a broadcast channel when you create a new channel (used to broadcast messages, only administrators can post messages).

Official applications implicitly migrate an existing Chat to a megagroup Channel when you perform certain actions (exceed user limit, add a public username, set certain permissions, etc.).

A Channel can be created directly with CreateChannelRequest, as either a megagroup or broadcast.

Official applications use the term “channel” only for broadcast channels.

The API refers to the different types of Channel with certain attributes:

  • A broadcast channel is a Channel with the channel.broadcast attribute set to True.
  • A megagroup channel is a Channel with the channel.megagroup attribute set to True. Official applications refer to this as “supergroup”.
  • A gigagroup channel is a Channel with the channel.gigagroup attribute set to True. Official applications refer to this as “broadcast groups”, and is used when a megagroup becomes very large and administrators want to transform it into something where only they can post messages.

Both the bot API and Telethon will “concatenate” -100 to the real chat ID so that you can tell at a glance, with just a number, the entity type.

For example, if you create a new broadcast channel, the real channel ID might be something like 456. If you try printing it from a message.chat_id you will see -1000000000456. This ID helps Telethon know you’re talking about a Channel.

Converting IDs

You can convert between the “marked” identifiers (prefixed with a minus sign) and the real ones with utils.resolve_id. It will return a tuple with the real ID, and the peer type (the class):

from telethon import utils
real_id, peer_type = utils.resolve_id(-1000000000456)

print(real_id)  # 456
print(peer_type)  # <class 'telethon.tl.types.PeerChannel'>

peer = peer_type(real_id)
print(peer)  # PeerChannel(channel_id=456)

The reverse operation can be done with utils.get_peer_id:

print(utils.get_peer_id(types.PeerChannel(456)))  # -1000000000456

Note that this function can also work with other types, like Chat or Channel instances.

If you need to convert other types like usernames which might need to perform API calls to find out the identifier, you can use client.get_peer_id:

print(await client.get_peer_id('me'))  # your id

If there is no “mark” (no minus sign), Telethon will assume your identifier refers to a User. If this is not the case, you can manually fix it:

from telethon import types
await client.send_message(types.PeerChannel(456), 'hello')
#                         ^^^^^^^^^^^^^^^^^ explicit peer type

A note on raw API

Certain methods only work on a Chat, and some others only work on a Channel (and these may only work in broadcast, or megagroup). Your code likely knows what it’s working with, so it shouldn’t be too much of an issue.

If you need to find the Channel from a Chat that migrated to it, access the migrated_to property:

# chat is a Chat
channel = await client.get_entity(chat.migrated_to)
# channel is now a Channel

Channels do not have a “migrated_from”, but a ChannelFull does. You can use GetFullChannelRequest to obtain this:

from telethon import functions
full = await client(functions.channels.GetFullChannelRequest(your_channel))
full_channel = full.full_chat
# full_channel is a ChannelFull
print(full_channel.migrated_from_chat_id)

This way, you can also access the linked discussion megagroup of a broadcast channel:

print(full_channel.linked_chat_id)  # prints ID of linked discussion group or None

You do not need to use client.get_entity to access the migrated_from_chat_id Chat or the linked_chat_id Channel. They are in the full.chats attribute:

if full_channel.migrated_from_chat_id:
    migrated_from_chat = next(c for c in full.chats if c.id == full_channel.migrated_from_chat_id)
    print(migrated_from_chat.title)

if full_channel.linked_chat_id:
    linked_group = next(c for c in full.chats if c.id == full_channel.linked_chat_id)
    print(linked_group.username)

Updates in Depth

Properties vs. Methods

The event shown above acts just like a custom.Message, which means you can access all the properties it has, like .sender.

However events are different to other methods in the client, like client.get_messages. Events may not send information about the sender or chat, which means it can be None, but all the methods defined in the client always have this information so it doesn’t need to be re-fetched. For this reason, you have get_ methods, which will make a network call if necessary.

In short, you should do this:

@client.on(events.NewMessage)
async def handler(event):
    # event.input_chat may be None, use event.get_input_chat()
    chat = await event.get_input_chat()
    sender = await event.get_sender()
    buttons = await event.get_buttons()

async def main():
    async for message in client.iter_messages('me', 10):
        # Methods from the client always have these properties ready
        chat = message.input_chat
        sender = message.sender
        buttons = message.buttons

Notice, properties (message.sender) don’t need an await, but methods (message.get_sender) do need an await, and you should use methods in events for these properties that may need network.

Events Without the client

The code of your application starts getting big, so you decide to separate the handlers into different files. But how can you access the client from these files? You don’t need to! Just events.register them:

# handlers/welcome.py
from telethon import events

@events.register(events.NewMessage('(?i)hello'))
async def handler(event):
    client = event.client
    await event.respond('Hey!')
    await client.send_message('me', 'I said hello to someone')

Registering events is a way of saying “this method is an event handler”. You can use telethon.events.is_handler to check if any method is a handler. You can think of them as a different approach to Flask’s blueprints.

It’s important to note that this does not add the handler to any client! You never specified the client on which the handler should be used. You only declared that it is a handler, and its type.

To actually use the handler, you need to client.add_event_handler to the client (or clients) where they should be added to:

# main.py
from telethon import TelegramClient
import handlers.welcome

with TelegramClient(...) as client:
    client.add_event_handler(handlers.welcome.handler)
    client.run_until_disconnected()

This also means that you can register an event handler once and then add it to many clients without re-declaring the event.

Events Without Decorators

If for any reason you don’t want to use telethon.events.register, you can explicitly pass the event handler to use to the mentioned client.add_event_handler:

from telethon import TelegramClient, events

async def handler(event):
    ...

with TelegramClient(...) as client:
    client.add_event_handler(handler, events.NewMessage)
    client.run_until_disconnected()

Similarly, you also have client.remove_event_handler and client.list_event_handlers.

The event argument is optional in all three methods and defaults to events.Raw for adding, and None when removing (so all callbacks would be removed).

Note

The event type is ignored in client.add_event_handler if you have used telethon.events.register on the callback before, since that’s the point of using such method at all.

Stopping Propagation of Updates

There might be cases when an event handler is supposed to be used solitary and it makes no sense to process any other handlers in the chain. For this case, it is possible to raise a telethon.events.StopPropagation exception which will cause the propagation of the update through your handlers to stop:

from telethon.events import StopPropagation

@client.on(events.NewMessage)
async def _(event):
    # ... some conditions
    await event.delete()

    # Other handlers won't have an event to work with
    raise StopPropagation

@client.on(events.NewMessage)
async def _(event):
    # Will never be reached, because it is the second handler
    # in the chain.
    pass

Remember to check Update Events if you’re looking for the methods reference.

Understanding asyncio

With asyncio, the library has several tasks running in the background. One task is used for sending requests, another task is used to receive them, and a third one is used to handle updates.

To handle updates, you must keep your script running. You can do this in several ways. For instance, if you are not running asyncio’s event loop, you should use client.run_until_disconnected:

import asyncio
from telethon import TelegramClient

client = TelegramClient(...)
...
client.run_until_disconnected()

Behind the scenes, this method is await’ing on the client.disconnected property, so the code above and the following are equivalent:

import asyncio
from telethon import TelegramClient

client = TelegramClient(...)

async def main():
    await client.disconnected

asyncio.run(main())

You could also run client.disconnected until it completed.

But if you don’t want to await, then you should know what you want to be doing instead! What matters is that you shouldn’t let your script die. If you don’t care about updates, you don’t need any of this.

Notice that unlike client.disconnected, client.run_until_disconnected will handle KeyboardInterrupt for you. This method is special and can also be ran while the loop is running, so you can do this:

async def main():
    await client.run_until_disconnected()

loop.run_until_complete(main())

Sequential Updates

If you need to process updates sequentially (i.e. not in parallel), you should set sequential_updates=True when creating the client:

with TelegramClient(..., sequential_updates=True) as client:
    ...

Session Files

They are an important part for the library to be efficient, such as caching and handling your authorization key (or you would have to login every time!).

What are Sessions?

The first parameter you pass to the constructor of the TelegramClient is the session, and defaults to be the session name (or full path). That is, if you create a TelegramClient('anon') instance and connect, an anon.session file will be created in the working directory.

Note that if you pass a string it will be a file in the current working directory, although you can also pass absolute paths.

The session file contains enough information for you to login without re-sending the code, so if you have to enter the code more than once, maybe you’re changing the working directory, renaming or removing the file, or using random names.

These database files using sqlite3 contain the required information to talk to the Telegram servers, such as to which IP the client should connect, port, authorization key so that messages can be encrypted, and so on.

These files will by default also save all the input entities that you’ve seen, so that you can get information about a user or channel by just their ID. Telegram will not send their access_hash required to retrieve more information about them, if it thinks you have already seem them. For this reason, the library needs to store this information offline.

The library will by default too save all the entities (chats and channels with their name and username, and users with the phone too) in the session file, so that you can quickly access them by username or phone number.

If you’re not going to work with updates, or don’t need to cache the access_hash associated with the entities’ ID, you can disable this by setting client.session.save_entities = False.

Different Session Storage

If you don’t want to use the default SQLite session storage, you can also use one of the other implementations or implement your own storage.

While it’s often not the case, it’s possible that SQLite is slow enough to be noticeable, in which case you can also use a different storage. Note that this is rare and most people won’t have this issue, but it’s worth a mention.

To use a custom session storage, simply pass the custom session instance to TelegramClient instead of the session name.

Telethon contains three implementations of the abstract Session class:

  • MemorySession: stores session data within memory.
  • SQLiteSession: stores sessions within on-disk SQLite databases. Default.
  • StringSession: stores session data within memory, but can be saved as a string.

You can import these from telethon.sessions. For example, using the StringSession is done as follows:

from telethon.sync import TelegramClient
from telethon.sessions import StringSession

with TelegramClient(StringSession(string), api_id, api_hash) as client:
    ...  # use the client

    # Save the string session as a string; you should decide how
    # you want to save this information (over a socket, remote
    # database, print it and then paste the string in the code,
    # etc.); the advantage is that you don't need to save it
    # on the current disk as a separate file, and can be reused
    # anywhere else once you log in.
    string = client.session.save()

# Note that it's also possible to save any other session type
# as a string by using ``StringSession.save(session_instance)``:
client = TelegramClient('sqlite-session', api_id, api_hash)
string = StringSession.save(client.session)

There are other community-maintained implementations available:

  • SQLAlchemy: stores all sessions in a single database via SQLAlchemy.
  • Redis: stores all sessions in a single Redis data store.
  • MongoDB: stores the current session in a MongoDB database.

Creating your Own Storage

The easiest way to create your own storage implementation is to use MemorySession as the base and check out how SQLiteSession or one of the community-maintained implementations work. You can find the relevant Python files under the sessions/ directory in the Telethon’s repository.

After you have made your own implementation, you can add it to the community-maintained session implementation list above with a pull request.

String Sessions

StringSession are a convenient way to embed your login credentials directly into your code for extremely easy portability, since all they take is a string to be able to login without asking for your phone and code (or faster start if you’re using a bot token).

The easiest way to generate a string session is as follows:

from telethon.sync import TelegramClient
from telethon.sessions import StringSession

with TelegramClient(StringSession(), api_id, api_hash) as client:
    print(client.session.save())

Think of this as a way to export your authorization key (what’s needed to login into your account). This will print a string in the standard output (likely your terminal).

Warning

Keep this string safe! Anyone with this string can use it to login into your account and do anything they want to.

This is similar to leaking your *.session files online, but it is easier to leak a string than it is to leak a file.

Once you have the string (which is a bit long), load it into your script somehow. You can use a normal text file and open(...).read() it or you can save it in a variable directly:

string = '1aaNk8EX-YRfwoRsebUkugFvht6DUPi_Q25UOCzOAqzc...'
with TelegramClient(StringSession(string), api_id, api_hash) as client:
    client.loop.run_until_complete(client.send_message('me', 'Hi'))

These strings are really convenient for using in places like Heroku since their ephemeral filesystem will delete external files once your application is over.

The Full API

Important

While you have access to this, you should always use the friendly methods listed on Client Reference unless you have a better reason not to, like a method not existing or you wanting more control.

Introduction

The TelegramClient doesn’t offer a method for every single request the Telegram API supports. However, it’s very simple to call or invoke any request defined in Telegram’s API.

This section will teach you how to use what Telethon calls the TL reference. The linked page contains a list and a way to search through all types generated from the definition of Telegram’s API (in .tl file format, hence the name). These types include requests and constructors.

Note

The reason to keep both https://tl.telethon.dev and this documentation alive is that the former allows instant search results as you type, and a “Copy import” button. If you like namespaces, you can also do from telethon.tl import types, functions. Both work.

Telegram makes these .tl files public, which other implementations, such as Telethon, can also use to generate code. These files are versioned under what’s called “layers”. .tl files consist of thousands of definitions, and newer layers often add, change, or remove them. Each definition refers to either a Remote Procedure Call (RPC) function, or a type (which the TL reference calls “constructors”, as they construct particular type instances).

As such, the TL reference is a good place to go to learn about all possible requests, types, and what they look like. If you’re curious about what’s been changed between layers, you can refer to the TL diff site.

Using the TL reference

After you’ve found a request you want to send, a good start would be to simply copy and paste the autogenerated example into your script. Then you can simply tweak it to your needs.

If you want to do it from scratch, first, make sure to import the request into your code (either using the “Copy import” button near the top, or by manually spelling out the package under telethon.tl.functions.*).

Then, start reading the parameters one by one. If the parameter cannot be omitted, you will need to specify it, so make sure to spell it out as an input parameter when constructing the request instance. Let’s look at PingRequest for example. First, we copy the import:

from telethon.tl.functions import PingRequest

Then, we look at the parameters:

ping_id - long

A single parameter, and it’s a long (a integer number with a large range of values). It doesn’t say it can be omitted, so we must provide it, like so:

PingRequest(
    ping_id=48641868471
)

(In this case, the ping ID is a random number. You often have to guess what the parameter needs just by looking at the name.)

Now that we have our request, we can invoke it:

response = await client(PingRequest(
    ping_id=48641868471
))

To find out what response looks like, we can do as the autogenerated example suggests and “stringify” the result as a pretty-printed string:

print(result.stringify())

This will print out the following:

Pong(
    msg_id=781875678118,
    ping_id=48641868471
)

Which is a very easy way to get a feel for a response. You should nearly always print the stringified result, at least once, when trying out requests, to get a feel for what the response may look like.

But of course, you don’t need to do that. Without writing any code, you could have navigated through the “Returns” link to learn PingRequest returns a Pong, which only has one constructor, and the constructor has two members, msg_id and ping_id.

If you wanted to create your own Pong, you would use both members as input parameters:

my_pong = Pong(
    msg_id=781875678118,
    ping_id=48641868471
)

(Yes, constructing object instances can use the same code that .stringify would return!)

And if you wanted to access the msg_id member, you would simply access it like any other attribute access in Python:

print(response.msg_id)

Example walkthrough

Say client.send_message() didn’t exist, we could use the search to look for “message”. There we would find SendMessageRequest, which we can work with.

Every request is a Python class, and has the parameters needed for you to invoke it. You can also call help(request) for information on what input parameters it takes. Remember to “Copy import to the clipboard”, or your script won’t be aware of this class! Now we have:

from telethon.tl.functions.messages import SendMessageRequest

If you’re going to use a lot of these, you may do:

from telethon.tl import types, functions
# We now have access to 'functions.messages.SendMessageRequest'

We see that this request must take at least two parameters, a peer of type InputPeer, and a message which is just a Python string.

How can we retrieve this InputPeer? We have two options. We manually construct one, for instance:

from telethon.tl.types import InputPeerUser

peer = InputPeerUser(user_id, user_hash)

Or we call client.get_input_entity():

import telethon

async def main():
    peer = await client.get_input_entity('someone')

client.loop.run_until_complete(main())

Note

Remember that await must occur inside an async def. Every full API example assumes you already know and do this.

When you’re going to invoke an API method, most require you to pass an InputUser, InputChat, or so on, this is why using client.get_input_entity() is more straightforward (and often immediate, if you’ve seen the user before, know their ID, etc.). If you also need to have information about the whole user, use client.get_entity() instead:

entity = await client.get_entity('someone')

In the later case, when you use the entity, the library will cast it to its “input” version for you. If you already have the complete user and want to cache its input version so the library doesn’t have to do this every time its used, simply call telethon.utils.get_input_peer:

from telethon import utils
peer = utils.get_input_peer(entity)

Note

Since v0.16.2 this is further simplified. The Request itself will call client.get_input_entity for you when required, but it’s good to remember what’s happening.

After this small parenthesis about client.get_entity versus client.get_input_entity(), we have everything we need. To invoke our request we do:

result = await client(SendMessageRequest(peer, 'Hello there!'))

Message sent! Of course, this is only an example. There are over 250 methods available as of layer 80, and you can use every single of them as you wish. Remember to use the right types! To sum up:

result = await client(SendMessageRequest(
    await client.get_input_entity('username'), 'Hello there!'
))

This can further be simplified to:

result = await client(SendMessageRequest('username', 'Hello there!'))
# Or even
result = await client(SendMessageRequest(PeerChannel(id), 'Hello there!'))

Note

Note that some requests have a “hash” parameter. This is not your api_hash! It likely isn’t your self-user .access_hash either.

It’s a special hash used by Telegram to only send a difference of new data that you don’t already have with that request, so you can leave it to 0, and it should work (which means no hash is known yet).

For those requests having a “limit” parameter, you can often set it to zero to signify “return default amount”. This won’t work for all of them though, for instance, in “messages.search” it will actually return 0 items.

Requests in Parallel

The library will automatically merge outgoing requests into a single container. Telegram’s API supports sending multiple requests in a single container, which is faster because it has less overhead and the server can run them without waiting for others. You can also force using a container manually:

async def main():

    # Letting the library do it behind the scenes
    await asyncio.wait([
        client.send_message('me', 'Hello'),
        client.send_message('me', ','),
        client.send_message('me', 'World'),
        client.send_message('me', '.')
    ])

    # Manually invoking many requests at once
    await client([
        SendMessageRequest('me', 'Hello'),
        SendMessageRequest('me', ', '),
        SendMessageRequest('me', 'World'),
        SendMessageRequest('me', '.')
    ])

Note that you cannot guarantee the order in which they are run. Try running the above code more than one time. You will see the order in which the messages arrive is different.

If you use the raw API (the first option), you can use ordered to tell the server that it should run the requests sequentially. This will still be faster than going one by one, since the server knows all requests directly:

await client([
    SendMessageRequest('me', 'Hello'),
    SendMessageRequest('me', ', '),
    SendMessageRequest('me', 'World'),
    SendMessageRequest('me', '.')
], ordered=True)

If any of the requests fails with a Telegram error (not connection errors or any other unexpected events), the library will raise telethon.errors.common.MultiError. You can except this and still access the successful results:

from telethon.errors import MultiError

try:
    await client([
        SendMessageRequest('me', 'Hello'),
        SendMessageRequest('me', ''),
        SendMessageRequest('me', 'World')
    ], ordered=True)
except MultiError as e:
    # The first and third requests worked.
    first = e.results[0]
    third = e.results[2]
    # The second request failed.
    second = e.exceptions[1]

RPC Errors

RPC stands for Remote Procedure Call, and when the library raises a RPCError, it’s because you have invoked some of the API methods incorrectly (wrong parameters, wrong permissions, or even something went wrong on Telegram’s server).

You should import the errors from telethon.errors like so:

from telethon import errors

try:
    async with client.takeout() as takeout:
        ...

except errors.TakeoutInitDelayError as e:
    #  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^ here we except TAKEOUT_INIT_DELAY
    print('Must wait', e.seconds, 'before takeout')

There isn’t any official list of all possible RPC errors, so the list of known errors is provided on a best-effort basis. When new methods are available, the list may be lacking since we simply don’t know what errors can raise from them.

Once we do find out about a new error and what causes it, the list is updated, so if you see an error without a specific class, do report it (and what method caused it)!.

This list is used to generate documentation for the raw API page. For example, if we want to know what errors can occur from messages.sendMessage we can simply navigate to its raw API page and find it has 24 known RPC errors at the time of writing.

Base Errors

All the “base” errors are listed in API Errors. Any other more specific error will be a subclass of these.

If the library isn’t aware of a specific error just yet, it will instead raise one of these superclasses. This means you may find stuff like this:

telethon.errors.rpcbaseerrors.BadRequestError: RPCError 400: MESSAGE_POLL_CLOSED (caused by SendVoteRequest)

If you do, make sure to open an issue or send a pull request to update the list of known errors.

Common Errors

These are some of the errors you may normally need to deal with:

  • FloodWaitError (420), the same request was repeated many times. Must wait .seconds (you can access this attribute). For example:

    ...
from telethon import errors

try:
    messages = await client.get_messages(chat)
    print(messages[0].text)
except errors.FloodWaitError as e:
    print('Have to sleep', e.seconds, 'seconds')
    time.sleep(e.seconds)

  • SessionPasswordNeededError, if you have setup two-steps verification on Telegram and are trying to sign in.

  • FilePartMissingError, if you have tried to upload an empty file.

  • ChatAdminRequiredError, you don’t have permissions to perform said operation on a chat or channel. Try avoiding filters, i.e. when searching messages.

The generic classes for different error codes are:

  • InvalidDCError (303), the request must be repeated on another DC.
  • BadRequestError (400), the request contained errors.
  • UnauthorizedError (401), the user is not authorized yet.
  • ForbiddenError (403), privacy violation error.
  • NotFoundError (404), make sure you’re invoking Request’s!

If the error is not recognised, it will only be an RPCError.

You can refer to all errors from Python through the telethon.errors module. If you don’t know what attributes they have, try printing their dir (like print(dir(e))).

Attributes

Some of the errors carry additional data in them. When they look like EMAIL_UNCONFIRMED_X, the _X value will be accessible from the error instance. The current list of errors that do this is the following:

  • EmailUnconfirmedError has .code_length.
  • FileMigrateError has .new_dc.
  • FilePartMissingError has .which.
  • FloodTestPhoneWaitError has .seconds.
  • FloodWaitError has .seconds.
  • InterdcCallErrorError has .dc.
  • InterdcCallRichErrorError has .dc.
  • NetworkMigrateError has .new_dc.
  • PhoneMigrateError has .new_dc.
  • SlowModeWaitError has .seconds.
  • TakeoutInitDelayError has .seconds.
  • UserMigrateError has .new_dc.

Avoiding Limits

Don’t spam. You won’t get FloodWaitError or your account banned or deleted if you use the library for legit use cases. Make cool tools. Don’t spam! Nobody knows the exact limits for all requests since they depend on a lot of factors, so don’t bother asking.

Still, if you do have a legit use case and still get those errors, the library will automatically sleep when they are smaller than 60 seconds by default. You can set different “auto-sleep” thresholds:

client.flood_sleep_threshold = 0  # Don't auto-sleep
client.flood_sleep_threshold = 24 * 60 * 60  # Sleep always

You can also except it and act as you prefer:

from telethon.errors import FloodWaitError
try:
    ...
except FloodWaitError as e:
    print('Flood waited for', e.seconds)
    quit(1)

VoIP numbers are very limited, and some countries are more limited too.

HTTP Bot API vs MTProto

Telethon is more than just another viable alternative when developing bots for Telegram. If you haven’t decided which wrapper library for bots to use yet, using Telethon from the beginning may save you some headaches later.

What is Bot API?

The Telegram Bot API, also known as HTTP Bot API and from now on referred to as simply “Bot API” is Telegram’s official way for developers to control their own Telegram bots. Quoting their main page:

The Bot API is an HTTP-based interface created for developers keen on building bots for Telegram.

To learn how to create and set up a bot, please consult our Introduction to Bots and Bot FAQ.

Bot API is simply an HTTP endpoint which translates your requests to it into MTProto calls through tdlib, their bot backend.

Configuration of your bot, such as its available commands and auto-completion, is configured through @BotFather.

What is MTProto?

MTProto is Telegram’s own protocol to communicate with their API when you connect to their servers.

Telethon is an alternative MTProto-based backend written entirely in Python and much easier to setup and use.

Both official applications and third-party clients (like your own applications) logged in as either user or bots can use MTProto to communicate directly with Telegram’s API (which is not the HTTP bot API).

When we talk about MTProto, we often mean “MTProto-based clients”.

Advantages of MTProto over Bot API

MTProto clients (like Telethon) connect directly to Telegram’s servers, which means there is no HTTP connection, no “polling” or “web hooks”. This means less overhead, since the protocol used between you and the server is much more compact than HTTP requests with responses in wasteful JSON.

Since there is a direct connection to Telegram’s servers, even if their Bot API endpoint is down, you can still have connection to Telegram directly.

Using a MTProto client, you are also not limited to the public API that they expose, and instead, you have full control of what your bot can do. Telethon offers you all the power with often much easier usage than any of the available Python Bot API wrappers.

If your application ever needs user features because bots cannot do certain things, you will be able to easily login as a user and even keep your bot without having to learn a new library.

If less overhead and full control didn’t convince you to use Telethon yet, check out the wiki page MTProto vs HTTP Bot API with a more exhaustive and up-to-date list of differences.

Migrating from Bot API to Telethon

It doesn’t matter if you wrote your bot with requests and you were making API requests manually, or if you used a wrapper library like python-telegram-bot or pyTelegramBotAPI. It’s never too late to migrate to Telethon!

If you were using an asynchronous library like aiohttp or a wrapper like aiogram or dumbot, it will be even easier, because Telethon is also an asynchronous library.

Next, we will see some examples from the most popular libraries.

Migrating from python-telegram-bot

Let’s take their echobot.py example and shorten it a bit:

from telegram.ext import Updater, CommandHandler, MessageHandler, Filters

def start(update, context):
    """Send a message when the command /start is issued."""
    update.message.reply_text('Hi!')

def echo(update, context):
    """Echo the user message."""
    update.message.reply_text(update.message.text)

def main():
    """Start the bot."""
    updater = Updater("TOKEN")
    dp = updater.dispatcher
    dp.add_handler(CommandHandler("start", start))
    dp.add_handler(MessageHandler(Filters.text & ~Filters.command, echo))

    updater.start_polling()

    updater.idle()

if __name__ == '__main__':
    main()

After using Telethon:

from telethon import TelegramClient, events

bot = TelegramClient('bot', 11111, 'a1b2c3d4').start(bot_token='TOKEN')

@bot.on(events.NewMessage(pattern='/start'))
async def start(event):
    """Send a message when the command /start is issued."""
    await event.respond('Hi!')
    raise events.StopPropagation

@bot.on(events.NewMessage)
async def echo(event):
    """Echo the user message."""
    await event.respond(event.text)

def main():
    """Start the bot."""
    bot.run_until_disconnected()

if __name__ == '__main__':
    main()

Key differences:

  • The recommended way to do it imports fewer things.
  • All handlers trigger by default, so we need events.StopPropagation.
  • Adding handlers, responding and running is a lot less verbose.
  • Telethon needs async def and await.
  • The bot isn’t hidden away by Updater or Dispatcher.
Migrating from pyTelegramBotAPI

Let’s show another echobot from their README:

import telebot

bot = telebot.TeleBot("TOKEN")

@bot.message_handler(commands=['start'])
def send_welcome(message):
    bot.reply_to(message, "Howdy, how are you doing?")

@bot.message_handler(func=lambda m: True)
def echo_all(message):
    bot.reply_to(message, message.text)

bot.polling()

Now we rewrite it to use Telethon:

from telethon import TelegramClient, events

bot = TelegramClient('bot', 11111, 'a1b2c3d4').start(bot_token='TOKEN')

@bot.on(events.NewMessage(pattern='/start'))
async def send_welcome(event):
    await event.reply('Howdy, how are you doing?')

@bot.on(events.NewMessage)
async def echo_all(event):
    await event.reply(event.text)

bot.run_until_disconnected()

Key differences:

  • Instead of doing bot.reply_to(message), we can do event.reply. Note that the event behaves just like their message.
  • Telethon also supports func=lambda m: True, but it’s not necessary.
Migrating from aiogram

From their GitHub:

from aiogram import Bot, Dispatcher, executor, types

API_TOKEN = 'BOT TOKEN HERE'

# Initialize bot and dispatcher
bot = Bot(token=API_TOKEN)
dp = Dispatcher(bot)

@dp.message_handler(commands=['start'])
async def send_welcome(message: types.Message):
    """
    This handler will be called when client send `/start` command.
    """
    await message.reply("Hi!\nI'm EchoBot!\nPowered by aiogram.")

@dp.message_handler(regexp='(^cat[s]?$|puss)')
async def cats(message: types.Message):
    with open('data/cats.jpg', 'rb') as photo:
        await bot.send_photo(message.chat.id, photo, caption='Cats is here 😺',
                             reply_to_message_id=message.message_id)

@dp.message_handler()
async def echo(message: types.Message):
    await bot.send_message(message.chat.id, message.text)

if __name__ == '__main__':
    executor.start_polling(dp, skip_updates=True)

After rewrite:

from telethon import TelegramClient, events

# Initialize bot and... just the bot!
bot = TelegramClient('bot', 11111, 'a1b2c3d4').start(bot_token='TOKEN')

@bot.on(events.NewMessage(pattern='/start'))
async def send_welcome(event):
    await event.reply('Howdy, how are you doing?')

@bot.on(events.NewMessage(pattern='(^cat[s]?$|puss)'))
async def cats(event):
    await event.reply('Cats is here 😺', file='data/cats.jpg')

@bot.on(events.NewMessage)
async def echo_all(event):
    await event.reply(event.text)

if __name__ == '__main__':
    bot.run_until_disconnected()

Key differences:

  • Telethon offers convenience methods to avoid retyping bot.send_photo(message.chat.id, ...) all the time, and instead let you type event.reply.
  • Sending files is a lot easier. The methods for sending photos, documents, audios, etc. are all the same!
Migrating from dumbot

Showcasing their subclassing example:

from dumbot import Bot

class Subbot(Bot):
    async def init(self):
        self.me = await self.getMe()

    async def on_update(self, update):
        await self.sendMessage(
            chat_id=update.message.chat.id,
            text='i am {}'.format(self.me.username)
        )

Subbot(token).run()

After rewriting:

from telethon import TelegramClient, events

class Subbot(TelegramClient):
    def __init__(self, *a, **kw):
        super().__init__(*a, **kw)
        self.add_event_handler(self.on_update, events.NewMessage)

    async def connect():
        await super().connect()
        self.me = await self.get_me()

    async def on_update(event):
        await event.reply('i am {}'.format(self.me.username))

bot = Subbot('bot', 11111, 'a1b2c3d4').start(bot_token='TOKEN')
bot.run_until_disconnected()

Key differences:

  • Telethon method names are snake_case.
  • dumbot does not offer friendly methods like update.reply.
  • Telethon does not have an implicit on_update handler, so we need to manually register one.

Mastering asyncio

What’s asyncio?

asyncio is a Python 3’s built-in library. This means it’s already installed if you have Python 3. Since Python 3.5, it is convenient to work with asynchronous code. Before (Python 3.4) we didn’t have async or await, but now we do.

asyncio stands for Asynchronous Input Output. This is a very powerful concept to use whenever you work IO. Interacting with the web or external APIs such as Telegram’s makes a lot of sense this way.

Why asyncio?

Asynchronous IO makes a lot of sense in a library like Telethon. You send a request to the server (such as “get some message”), and thanks to asyncio, your code won’t block while a response arrives.

The alternative would be to spawn a thread for each update so that other code can run while the response arrives. That is a lot more expensive.

The code will also run faster, because instead of switching back and forth between the OS and your script, your script can handle it all. Avoiding switching saves quite a bit of time, in Python or any other language that supports asynchronous IO. It will also be cheaper, because tasks are smaller than threads, which are smaller than processes.

What are asyncio basics?

The code samples below assume that you have Python 3.7 or greater installed.

# First we need the asyncio library
import asyncio

# We also need something to run
async def main():
    for char in 'Hello, world!\n':
        print(char, end='', flush=True)
        await asyncio.sleep(0.2)

# Then, we can create a new asyncio loop and use it to run our coroutine.
# The creation and tear-down of the loop is hidden away from us.
asyncio.run(main())

What does telethon.sync do?

The moment you import any of these:

from telethon import sync, ...
# or
from telethon.sync import ...
# or
import telethon.sync

The sync module rewrites most async def methods in Telethon to something similar to this:

def new_method():
    result = original_method()
    if loop.is_running():
        # the loop is already running, return the await-able to the user
        return result
    else:
        # the loop is not running yet, so we can run it for the user
        return loop.run_until_complete(result)

That means you can do this:

print(client.get_me().username)

Instead of this:

me = client.loop.run_until_complete(client.get_me())
print(me.username)

# or, using asyncio's default loop (it's the same)
import asyncio
loop = asyncio.get_running_loop()  # == client.loop
me = loop.run_until_complete(client.get_me())
print(me.username)

As you can see, it’s a lot of boilerplate and noise having to type run_until_complete all the time, so you can let the magic module to rewrite it for you. But notice the comment above: it won’t run the loop if it’s already running, because it can’t. That means this:

async def main():
    # 3. the loop is running here
    print(
        client.get_me()  # 4. this will return a coroutine!
        .username  # 5. this fails, coroutines don't have usernames
    )

loop.run_until_complete(  # 2. run the loop and the ``main()`` coroutine
    main()  # 1. calling ``async def`` "returns" a coroutine
)

Will fail. So if you’re inside an async def, then the loop is running, and if the loop is running, you must await things yourself:

async def main():
    print((await client.get_me()).username)

loop.run_until_complete(main())

What are async, await and coroutines?

The async keyword lets you define asynchronous functions, also known as coroutines, and also iterate over asynchronous loops or use async with:

import asyncio

async def main():
    # ^ this declares the main() coroutine function

    async with client:
        # ^ this is an asynchronous with block

        async for message in client.iter_messages(chat):
            # ^ this is a for loop over an asynchronous generator

            print(message.sender.username)

asyncio.run(main())
# ^ this will create a new asyncio loop behind the scenes and tear it down
#   once the function returns. It will run the loop untiil main finishes.
#   You should only use this function if there is no other loop running.

The await keyword blocks the current task, and the loop can run other tasks. Tasks can be thought of as “threads”, since many can run concurrently:

import asyncio

async def hello(delay):
    await asyncio.sleep(delay)  # await tells the loop this task is "busy"
    print('hello')  # eventually the loop resumes the code here

async def world(delay):
    # the loop decides this method should run first
    await asyncio.sleep(delay)  # await tells the loop this task is "busy"
    print('world')  # eventually the loop finishes all tasks

async def main():
    asyncio.create_task(world(2))  # create the world task, passing 2 as delay
    asyncio.create_task(hello(delay=1))  # another task, but with delay 1
    await asyncio.sleep(3)  # wait for three seconds before exiting

try:
    # create a new temporary asyncio loop and use it to run main
    asyncio.run(main())
except KeyboardInterrupt:
    pass

The same example, but without the comment noise:

import asyncio

async def hello(delay):
    await asyncio.sleep(delay)
    print('hello')

async def world(delay):
    await asyncio.sleep(delay)
    print('world')

async def main():
    asyncio.create_task(world(2))
    asyncio.create_task(hello(delay=1))
    await asyncio.sleep(3)

try:
    asyncio.run(main())
except KeyboardInterrupt:
    pass

Can I use threads?

Yes, you can, but you must understand that the loops themselves are not thread safe. and you must be sure to know what is happening. The easiest and cleanest option is to use asyncio.run to create and manage the new event loop for you:

import asyncio
import threading

async def actual_work():
    client = TelegramClient(..., loop=loop)
    ...  # can use `await` here

def go():
    asyncio.run(actual_work())

threading.Thread(target=go).start()

Generally, you don’t need threads unless you know what you’re doing. Just create another task, as shown above. If you’re using the Telethon with a library that uses threads, you must be careful to use threading.Lock whenever you use the client, or enable the compatible mode. For that, see Compatibility and Convenience.

You may have seen this error:

RuntimeError: There is no current event loop in thread 'Thread-1'.

It just means you didn’t create a loop for that thread. Please refer to the asyncio documentation to correctly learn how to set the event loop for non-main threads.

client.run_until_disconnected() blocks!

All of what client.run_until_disconnected() does is run the asyncio’s event loop until the client is disconnected. That means the loop is running. And if the loop is running, it will run all the tasks in it. So if you want to run other code, create tasks for it:

from datetime import datetime

async def clock():
    while True:
        print('The time:', datetime.now())
        await asyncio.sleep(1)

loop.create_task(clock())
...
client.run_until_disconnected()

This creates a task for a clock that prints the time every second. You don’t need to use client.run_until_disconnected() either! You just need to make the loop is running, somehow. loop.run_forever() and loop.run_until_complete() can also be used to run the loop, and Telethon will be happy with any approach.

Of course, there are better tools to run code hourly or daily, see below.

What else can asyncio do?

Asynchronous IO is a really powerful tool, as we’ve seen. There are plenty of other useful libraries that also use asyncio and that you can integrate with Telethon.

  • aiohttp is like the infamous requests but asynchronous.
  • quart is an asynchronous alternative to Flask.
  • aiocron lets you schedule things to run things at a desired time, or run some tasks hourly, daily, etc.

And of course, asyncio itself! It has a lot of methods that let you do nice things. For example, you can run requests in parallel:

async def main():
    last, sent, download_path = await asyncio.gather(
        client.get_messages('telegram', 10),
        client.send_message('me', 'Using asyncio!'),
        client.download_profile_photo('telegram')
    )

loop.run_until_complete(main())

This code will get the 10 last messages from @telegram, send one to the chat with yourself, and also download the profile photo of the channel. asyncio will run all these three tasks at the same time. You can run all the tasks you want this way.

A different way would be:

loop.create_task(client.get_messages('telegram', 10))
loop.create_task(client.send_message('me', 'Using asyncio!'))
loop.create_task(client.download_profile_photo('telegram'))

They will run in the background as long as the loop is running too.

You can also start an asyncio server in the main script, and from another script, connect to it to achieve Inter-Process Communication. You can get as creative as you want. You can program anything you want. When you use a library, you’re not limited to use only its methods. You can combine all the libraries you want. People seem to forget this simple fact!

Why does client.start() work outside async?

Because it’s so common that it’s really convenient to offer said functionality by default. This means you can set up all your event handlers and start the client without worrying about loops at all.

Using the client in a with block, start, run_until_disconnected, and disconnect all support this.

Where can I read more?

Check out my blog post about asyncio, which has some more examples and pictures to help you understand what happens when the loop runs.

A Word of Warning

Full API is not how you are intended to use the library. You should always prefer the Client Reference. However, not everything is implemented as a friendly method, so full API is your last resort.

If you select a method in Client Reference, you will most likely find an example for that method. This is how you are intended to use the library.

Full API will break between different minor versions of the library, since Telegram changes very often. The friendly methods will be kept compatible between major versions.

If you need to see real-world examples, please refer to the wiki page of projects using Telethon.

Working with Chats and Channels

Note

These examples assume you have read The Full API.

Joining a chat or channel

Note that Chat are normal groups, and Channel are a special form of Chat, which can also be super-groups if their megagroup member is True.

Joining a public channel

Once you have the entity of the channel you want to join to, you can make use of the JoinChannelRequest to join such channel:

from telethon.tl.functions.channels import JoinChannelRequest
await client(JoinChannelRequest(channel))

# In the same way, you can also leave such channel
from telethon.tl.functions.channels import LeaveChannelRequest
await client(LeaveChannelRequest(input_channel))

For more on channels, check the channels namespace.

Joining a private chat or channel

If all you have is a link like this one: https://t.me/joinchat/AAAAAFFszQPyPEZ7wgxLtd, you already have enough information to join! The part after the https://t.me/joinchat/, this is, AAAAAFFszQPyPEZ7wgxLtd on this example, is the hash of the chat or channel. Now you can use ImportChatInviteRequest as follows:

from telethon.tl.functions.messages import ImportChatInviteRequest
updates = await client(ImportChatInviteRequest('AAAAAEHbEkejzxUjAUCfYg'))

Adding someone else to such chat or channel

If you don’t want to add yourself, maybe because you’re already in, you can always add someone else with the AddChatUserRequest, which use is very straightforward, or InviteToChannelRequest for channels:

# For normal chats
from telethon.tl.functions.messages import AddChatUserRequest

# Note that ``user_to_add`` is NOT the name of the parameter.
# It's the user you want to add (``user_id=user_to_add``).
await client(AddChatUserRequest(
    chat_id,
    user_to_add,
    fwd_limit=10  # Allow the user to see the 10 last messages
))

# For channels (which includes megagroups)
from telethon.tl.functions.channels import InviteToChannelRequest

await client(InviteToChannelRequest(
    channel,
    [users_to_add]
))

Note that this method will only really work for friends or bot accounts. Trying to mass-add users with this approach will not work, and can put both your account and group to risk, possibly being flagged as spam and limited.

Increasing View Count in a Channel

It has been asked quite a few times (really, many), and while I don’t understand why so many people ask this, the solution is to use GetMessagesViewsRequest, setting increment=True:

# Obtain `channel' through dialogs or through client.get_entity() or anyhow.
# Obtain `msg_ids' through `.get_messages()` or anyhow. Must be a list.

await client(GetMessagesViewsRequest(
    peer=channel,
    id=msg_ids,
    increment=True
))

Note that you can only do this once or twice a day per account, running this in a loop will obviously not increase the views forever unless you wait a day between each iteration. If you run it any sooner than that, the views simply won’t be increased.

Users

Note

These examples assume you have read The Full API.

Retrieving full information

If you need to retrieve the bio, biography or about information for a user you should use GetFullUser:

from telethon.tl.functions.users import GetFullUserRequest

full = await client(GetFullUserRequest(user))
# or even
full = await client(GetFullUserRequest('username'))

bio = full.full_user.about

See UserFull to know what other fields you can access.

Updating your name and/or bio

The first name, last name and bio (about) can all be changed with the same request. Omitted fields won’t change after invoking UpdateProfile:

from telethon.tl.functions.account import UpdateProfileRequest

await client(UpdateProfileRequest(
    about='This is a test from Telethon'
))

Updating your username

You need to use account.UpdateUsername:

from telethon.tl.functions.account import UpdateUsernameRequest

await client(UpdateUsernameRequest('new_username'))

Updating your profile photo

The easiest way is to upload a new file and use that as the profile photo through UploadProfilePhoto:

from telethon.tl.functions.photos import UploadProfilePhotoRequest

await client(UploadProfilePhotoRequest(
    await client.upload_file('/path/to/some/file')
))

Working with messages

Note

These examples assume you have read The Full API.

This section has been moved to the wiki, where it can be easily edited as new features arrive and the API changes. Please refer to the linked page to learn how to send spoilers, custom emoji, stickers, react to messages, and more things.

Philosophy

The intention of the library is to have an existing MTProto library existing with hardly any dependencies (indeed, wherever Python is available, you can run this library).

Being written in Python means that performance will be nowhere close to other implementations written in, for instance, Java, C++, Rust, or pretty much any other compiled language. However, the library turns out to actually be pretty decent for common operations such as sending messages, receiving updates, or other scripting. Uploading files may be notably slower, but if you would like to contribute, pull requests are appreciated!

If libssl is available on your system, the library will make use of it to speed up some critical parts such as encrypting and decrypting the messages. Files will notably be sent and downloaded faster.

The main focus is to keep everything clean and simple, for everyone to understand how working with MTProto and Telegram works. Don’t be afraid to read the source, the code won’t bite you! It may prove useful when using the library on your own use cases.

Test Servers

To run Telethon on a test server, use the following code:

client = TelegramClient(None, api_id, api_hash)
client.session.set_dc(dc_id, '149.154.167.40', 80)

You can check your 'test ip' on https://my.telegram.org.

You should set None session so to ensure you’re generating a new authorization key for it (it would fail if you used a session where you had previously connected to another data center).

Note that port 443 might not work, so you can try with 80 instead.

Once you’re connected, you’ll likely be asked to either sign in or sign up. Remember anyone can access the phone you choose, so don’t store sensitive data here.

Valid phone numbers are 99966XYYYY, where X is the dc_id and YYYY is any number you want, for example, 1234 in dc_id = 2 would be 9996621234. The code sent by Telegram will be dc_id repeated five times, in this case, 22222 so we can hardcode that:

client = TelegramClient(None, api_id, api_hash)
client.session.set_dc(2, '149.154.167.40', 80)
client.start(
    phone='9996621234', code_callback=lambda: '22222'
)

Note that Telegram has changed the length of login codes multiple times in the past, so if dc_id repeated five times does not work, try repeating it six times.

Project Structure

Main interface

The library itself is under the telethon/ directory. The __init__.py file there exposes the main TelegramClient, a class that servers as a nice interface with the most commonly used methods on Telegram such as sending messages, retrieving the message history, handling updates, etc.

The TelegramClient inherits from several mixing Method classes, since there are so many methods that having them in a single file would make maintenance painful (it was three thousand lines before this separation happened!). It’s a “god object”, but there is only a way to interact with Telegram really.

The TelegramBaseClient is an ABC which will support all of these mixins so they can work together nicely. It doesn’t even know how to invoke things because they need to be resolved with user information first (to work with input entities comfortably).

The client makes use of the network/mtprotosender.py. The MTProtoSender is responsible for connecting, reconnecting, packing, unpacking, sending and receiving items from the network. Basically, the low-level communication with Telegram, and handling MTProto-related functions and types such as BadSalt.

The sender makes use of a Connection class which knows the format in which outgoing messages should be sent (how to encode their length and their body, if they’re further encrypted).

Auto-generated code

The files under telethon_generator/ are used to generate the code that gets placed under telethon/tl/. The parsers take in files in a specific format (such as .tl for objects and .json for errors) and spit out the generated classes which represent, as Python classes, the request and types defined in the .tl file. It also constructs an index so that they can be imported easily.

Custom documentation can also be generated to easily navigate through the vast amount of items offered by the API.

If you clone the repository, you will have to run python setup.py gen in order to generate the code. Installing the library runs the generator too, but the mentioned command will just generate code.

Coding Style

Basically, make it readable, while keeping the style similar to the code of whatever file you’re working on.

Also note that not everyone has 4K screens for their primary monitors, so please try to stick to the 80-columns limit. This makes it easy to git diff changes from a terminal before committing changes. If the line has to be long, please don’t exceed 120 characters.

For the commit messages, please make them explanatory. Not only they’re helpful to troubleshoot when certain issues could have been introduced, but they’re also used to construct the change log once a new version is ready.

If you don’t know enough Python, I strongly recommend reading Dive Into Python 3, available online for free. For instance, remember to do if x is None or if x is not None instead if x == None!

Tests

Telethon uses Pytest, for testing, Tox for environment setup, and pytest-asyncio and pytest-cov for asyncio and coverage integration.

While reading the full documentation for these is probably a good idea, there is a lot to read, so a brief summary of these tools is provided below for convienience.

Brief Introduction to Pytest

Pytest is a tool for discovering and running python tests, as well as allowing modular reuse of test setup code using fixtures.

Most Pytest tests will look something like this:

from module import my_thing, my_other_thing

def test_my_thing(fixture):
    assert my_thing(fixture) == 42

@pytest.mark.asyncio
async def test_my_thing(event_loop):
    assert await my_other_thing(loop=event_loop) == 42

Note here:

  1. The test imports one specific function. The role of unit tests is to test that the implementation of some unit, like a function or class, works. It’s role is not so much to test that components interact well with each other. I/O, such as connecting to remote servers, should be avoided. This helps with quickly identifying the source of an error, finding silent breakage, and makes it easier to cover all possible code paths.

    System or integration tests can also be useful, but are currently out of scope of Telethon’s automated testing.

  2. A function test_my_thing is declared. Pytest searches for files starting with test_, classes starting with Test and executes any functions or methods starting with test_ it finds.

  3. The function is declared with a parameter fixture. Fixtures are used to request things required to run the test, such as temporary directories, free TCP ports, Connections, etc. Fixtures are declared by simply adding the fixture name as parameter. A full list of available fixtures can be found with the pytest --fixtures command.

  4. The test uses a simple assert to test some condition is valid. Pytest uses some magic to ensure that the errors from this are readable and easy to debug.

  5. The pytest.mark.asyncio fixture is provided by pytest-asyncio. It starts a loop and executes a test function as coroutine. This should be used for testing asyncio code. It also declares the event_loop fixture, which will request an asyncio event loop.

Brief Introduction to Tox

Tox is a tool for automated setup of virtual environments for testing. While the tests can be run directly by just running pytest, this only tests one specific python version in your existing environment, which will not catch e.g. undeclared dependencies, or version incompatabilities.

Tox environments are declared in the tox.ini file. The default environments, declared at the top, can be simply run with tox. The option tox -e py36,flake can be used to request specific environments to be run.

Brief Introduction to Pytest-cov

Coverage is a useful metric for testing. It measures the lines of code and branches that are exercised by the tests. The higher the coverage, the more likely it is that any coding errors will be caught by the tests.

A brief coverage report can be generated with the --cov option to tox, which will be passed on to pytest. Additionally, the very useful HTML report can be generated with --cov --cov-report=html, which contains a browsable copy of the source code, annotated with coverage information for each line.

Understanding the Type Language

Telegram’s Type Language (also known as TL, found on .tl files) is a concise way to define what other programming languages commonly call classes or structs.

Every definition is written as follows for a Telegram object is defined as follows:

name#id argument_name:argument_type = CommonType

This means that in a single line you know what the TLObject name is. You know it’s unique ID, and you know what arguments it has. It really isn’t that hard to write a generator for generating code to any platform!

The generated code should also be able to encode the TLObject (let this be a request or a type) into bytes, so they can be sent over the network. This isn’t a big deal either, because you know how the TLObject’s are made, and how the types should be serialized.

You can either write your own code generator, or use the one this library provides, but please be kind and keep some special mention to this project for helping you out.

This is only a introduction. The TL language is not that easy. But it’s not that hard either. You’re free to sniff the telethon_generator/ files and learn how to parse other more complex lines, such as flags (to indicate things that may or may not be written at all) and vector’s.

Tips for Porting the Project

If you’re going to use the code on this repository to guide you, please be kind and don’t forget to mention it helped you!

You should start by reading the source code on the first release of the project, and start creating a MTProtoSender. Once this is made, you should write by hand the code to authenticate on the Telegram’s server, which are some steps required to get the key required to talk to them. Save it somewhere! Then, simply mimic, or reinvent other parts of the code, and it will be ready to go within a few days.

Good luck!

Telegram API in Other Languages

Telethon was made for Python, and it has inspired other libraries such as gramjs (JavaScript) and grammers (Rust). But there is a lot more beyond those, made independently by different developers.

If you’re looking for something like Telethon but in a different programming language, head over to Telegram API in Other Languages in the official wiki for a (mostly) up-to-date list.

Changelog (Version History)

This page lists all the available versions of the library, in chronological order. You should read this when upgrading the library to know where your code can break, and where it can take advantage of new goodies!

List of All Versions

Layer bump and small changes (v1.30)

Scheme layer used: 162

Some of the bug fixes were already present in patch versions of v1.29, but the new layer necessitated a minor bump.

Enhancements
  • Removed client-side checks for editing messages. This only affects Message.edit, as client.edit_message already had no checks.
  • Library should not understand more server-side errors during update handling which should reduce crashes.
  • Client-side image compression should behave better now.
Bug fixes
  • Some updates such as UpdateChatParticipant were being missed due to the order in which Telegram sent them. The library now more carefully checks for the sequence and pts contained in them to avoid dropping them.
  • Fixed is_inline check for KeyboardButtonWebView.
  • Fixed some issues getting entity from cache by ID.
  • reply_to should now work when sending albums.

More bug fixing (v1.29)

Scheme layer used: 160

This layer introduces the necessary raw API methods to work with stories.

The library is aiming to be “feature-frozen” for as long as v1 is active, so friendly client methods are not implemented, but example code to use stories can be found in the GitHub wiki of the project.

Enhancements
  • Removed client-side checks for methods dealing with chat permissions. In particular, this means you can now ban channels.
  • Improved some error messages and added new classes for more RPC errors.
  • The client-side check for valid usernames has been loosened, so that very short premium usernames are no longer considered invalid.
Bug fixes
  • Attempting to download a thumbnail from documnets without one would fail, rather than do nothing (since nothing can be downloaded if there is no thumb).
  • More errors are caught in the update handling loop.
  • HTML .text should now “unparse” any message contents correctly.
  • Fixed some problems related to logging.
  • comment_to should now work as expected with albums.
  • asyncio.CancelledError should now correctly propagate from the update loop.
  • Removed some absolute imports in favour of relative imports.
  • UserUpdate.last_seen should now behave correctly.
  • Fixed a rare ValueError during connect if the session cache was bad.

New Layer and housekeeping (v1.28)

Scheme layer used: 155

Plenty of stale issues closed, as well as improvements for some others.

Additions
  • New entity_cache_limit parameter in the TelegramClient constructor. This should help a bit in keeping memory usage in check.
Enhancements
  • progress_callback is now called when dealing with albums. See the documentation on client.send_file() for details.
  • Update state and entities are now periodically saved, so that the information isn’t lost in the case of crash or unexpected script terminations. You should still be calling disconnect or using the context-manager, though.
  • The client should no longer unnecessarily call get_me every time it’s started.
Bug fixes
  • Messages obtained via raw API could not be used in forward_messages.
  • force_sms and sign_up have been deprecated. See issue 4050 for details. It is no longer possible for third-party applications, such as those made with Telethon, to use those features.
  • events.ChatAction should now work in more cases in groups with hidden members.
  • Errors that occur at the connection level should now be properly propagated, so that you can actually have a chance to handle them.
  • Update handling should be more resilient.
  • PhoneCodeExpiredError will correctly clear the stored hash if it occurs in sign_in.
  • In patch v1.28.2, InputBotInlineMessageID64 can now be used to edit inline messages.

New Layer and some Bug fixes (v1.27)

Scheme layer used: 152
Bug fixes
  • When the account is logged-out, the library should now correctly propagate an error through run_until_disconnected to let you handle it.
  • The library no longer uses asyncio.get_event_loop() in newer Python versions, which should get rid of some deprecation warnings.
  • It could happen that bots would receive messages sent by themselves, very often right after they deleted a message. This should happen far less often now (but might still happen with unlucky timings).
  • Maximum photo size for automatic image resizing is now larger.
  • The initial request is now correctly wrapped in invokeWithoutUpdates when updates are disabled after constructing the client instance.
  • Using a pathlib.Path to download contacts and web documents should now work correctly.

New Layer and some Bug fixes (v1.26)

Scheme layer used: 149

This new layer includes things such as emoji status, more admin log events, forum topics and message reactions, among other things. You can access these using raw API. It also contains a few bug fixes.

These were fixed in the v1.25 series:

  • client.edit_admin did not work on small group chats.
  • client.get_messages could stop early in some channels.
  • client.download_profile_photo now should work even if User.min.
  • client.disconnect should no longer hang when being called from within an event handlers.
  • client.get_dialogs now initializes the update state for channels.
  • The message sender should not need to be fetched in more cases.
  • Lowered the severity of some log messages to be less spammy.

These are new to v1.26.0:

  • Layer update.
  • New documented RPC errors.
  • Sometimes the first message update to a channel could be missed if said message was read immediately.
  • client.get_dialogs would fail when the total count evenly divided the chunk size of 100.
  • client.get_messages could get stuck during a global search.
  • Potentially fixed some issues when sending certain videos.
  • Update handling should be more resilient.
  • The client should handle having its auth key destroyed more gracefully.
  • Fixed some issues when logging certain messages.

Bug fixes (v1.25.1)

This version should fix some of the problems that came with the revamped update handling.

  • Some inline URLs were not parsing correctly with markdown.
  • events.Raw was handling UpdateShort which it shouldn’t do.
  • events.Album should now work again.
  • CancelledError was being incorrectly logged as a fatal error.
  • Some fixes to update handling primarly aimed for bot accounts.
  • Update handling now can deal with more errors without crashing.
  • Unhandled errors from update handling will now be propagated through client.run_until_disconnected.
  • Invite links with + are now recognized.
  • Added new known RPC errors.
  • telethon.types could not be used as a module.
  • 0-length message entities are now stripped to avoid errors.
  • client.send_message was not returning a message with reply_to in some cases.
  • aggressive in client.iter_participants now does nothing (it did not really work anymore anyway, and this should prevent other errors).
  • client.iter_participants was failing in some groups.
  • Text with HTML URLs could sometimes fail to parse.
  • Added a hard timeout during disconnect in order to prevent the program from freezing.

Please be sure to report issues with update handling if you still encounter some errors!

Update handling overhaul (v1.25)

Scheme layer used: 144

I had plans to release v2 way earlier, but my motivation drained off, so that didn’t happen. The reason for another v1 release is that there was a clear need to fix some things regarding update handling (which were present in v2). I did not want to make this release. But with the release date for v2 still being unclear, I find it necessary to release another v1 version. I apologize for the delay (I should’ve done this a lot sooner but didn’t because in my head I would’ve pushed through and finished v2, but I underestimated how much work that was and I probably experienced burn-out).

I still don’t intend to make new additions to the v1 series (beyond updating the Telegram layer being used). I still have plans to finish v2 some day. But in the meantime, new features, such as reactions, will have to be used through raw API.

This update also backports the update overhaul from v2. If you experience issues with updates, please report them on the GitHub page for the project. However, this new update handling should be more reliable, and catch_up should actually work properly.

Breaking Changes
  • In order for catch_up to work (new flag in the TelegramClient constructor), sessions need to impleemnt the new get_update_states. Third-party session storages won’t have this implemented by the time this version released, so catch_up may not work with those.

Rushed release to fix login (v1.24)

Scheme layer used: 133

This is a rushed release. It contains a layer recent enough to not fail with UPDATE_APP_TO_LOGIN, but still not the latest, to avoid breaking more than necessary.

Breaking Changes
  • The biggest change is user identifiers (and chat identifiers, and others) now use up to 64 bits, rather than 32. If you were storing them in some storage with fixed size, you may need to update (such as database tables storing only integers).

There have been other changes which I currently don’t have the time to document. You can refer to the following link to see them early: https://github.com/LonamiWebs/Telethon/compare/v1.23.0…v1.24.0

New schema and bug fixes (v1.23)

Scheme layer used: 130

View new and changed raw API methods.

Enhancements
  • client.pin_message() can now pin on a single side in PMs.
  • Iterating participants should now be less expensive floodwait-wise.
Bug fixes
  • The QR login URL was being encoded incorrectly.
  • force_document was being ignored in inline queries for document.
  • manage_call permission was accidentally set to True by default.

New schema and bug fixes (v1.22)

Scheme layer used: 129

View new and changed raw API methods.

Enhancements
  • You can now specify a message in client.get_stats().
  • Metadata extraction from audio files with hachoir now recognises “artist”.
  • Get default chat permissions by not supplying a user to client.get_permissions().
  • You may now use thumb when editing messages.
Bug fixes
  • Fixes regarding bot markup in messages.
  • Gracefully handle ChannelForbidden in get_sender.

And from v1.21.1:

  • file.width and .height was not working correctly in photos.
  • Raw API was mis-interpreting False values on boolean flag parameters.

New schema and QoL improvements (v1.21)

Scheme layer used: 125

View new and changed raw API methods.

Not many changes in this release, mostly the layer change. Lately quite a few people have been reporting TypeNotFoundError, which occurs when the server sends types that it shouldn’t. This can happen when Telegram decides to add a new, incomplete layer, and then they change the layer without bumping the layer number (so some constructor IDs no longer match and the error occurs). This layer change should fix it.

Additions
  • Message.click() now supports a password parameter, needed when doing things like changing the owner of a bot via @BotFather.
Enhancements
  • tgcrypto will now be used for encryption when installed.
Bug fixes
  • Message.edit wasn’t working in your own chat on events other than NewMessage.
  • client.delete_dialog() was not working on chats.
  • events.UserUpdate should now handle channels’ typing status.
  • InputNotifyPeer auto-cast should now work on other TLObject.
  • For some objects, False was not correctly serialized.

New schema and QoL improvements (v1.20)

Scheme layer used: 124

View new and changed raw API methods.

A bit late to the party, but Telethon now offers a convenient way to comment on channel posts. It works very similar to reply_to:

client.send_message(channel, 'Great update!', comment_to=1134)

This code will leave a comment to the channel post with ID 1134 in channel.

In addition, the library now logs warning or error messages to stderr by default! You no longer should be left wondering “why isn’t my event handler working” if you forgot to configure logging. It took so long for this change to arrive because nobody noticed that Telethon was using a logging.NullHandler when it really shouldn’t have.

If you want the old behaviour of no messages being logged, you can configure logging to CRITICAL severity:

import logging
logging.basicConfig(level=logging.CRITICAL)

This is not considered a breaking change because stderr should only be used for logging purposes, not to emit information others may consume (use stdout for that).

Additions
Enhancements
  • utils.resolve_invite_link handles the newer link format.
  • Downloading files now retries once on TimeoutError, which has been happening recently. It is not guaranteed to work, but it should help.
  • Sending albums of photo URLs is now supported.
  • EXIF metadata is respected when automatically resizing photos, so the orientation information should no longer be lost.
  • Downloading a thumbnail by index should now use the correct size ordering.
Bug fixes
  • Fixed a KeyError on certain cases with Conversation.
  • Thumbnails should properly render on more clients. Installing hachoir may help.
  • Message search was broken when using a certain combination of parameters.
  • utils.resolve_id was misbehaving with some identifiers.
  • Fix TypeNotFoundError was not being propagated, causing deadlocks.
  • Invoking multiple requests at once with ordered=True was deadlocking.

New raw API call methods (v1.19)

Scheme layer used: 122

Telegram has had group calls for some weeks now. This new version contains the raw API methods needed to initiate and manage these group calls, however, the library will likely not offer ways to stream audio directly.

Telethon’s focus is being an asyncio-based, pure-Python implementation to interact with Telegram’s API. Streaming audio is beyond the current scope of the project and would be a big undertaking.

However, that doesn’t mean calls are not possible with Telethon. If you want to help design a Python library to perform audio calls, which can then be used with Telethon (so you can use Telethon + that new library to perform calls with Telethon), please refer to @pytgcallschat and join the relevant chat to discuss and help with the implementation!

The above message was also posted in the official Telegram group, if you wish to discuss it further.

With that out of the way, let’s list the additions and bug fixes in this release:

Additions
Enhancements
  • Updated documentation and list of known RPC errors.
  • The library now treats a lack of ping responses as a network error.
  • client.kick_participant() now returns the service message about the user being kicked, so you can delete it.
Bug fixes
  • When editing inline messages, the text parameter is preferred if provided.
  • Additional senders are unconditionally disconnected when disconnecting the main client, which should reduce the amount of asyncio warnings.
  • Automatic reconnection with no retries was failing.
  • PhotoPathSize is now ignored when determining a download size, since this “size” is not a JPEG thumbnail unlike the rest.
  • events.ChatAction should misbehave less.

New layer and QoL improvements (v1.18)

Scheme layer used: 120

Mostly fixes, and added some new things that can be done in this new layer.

For proxy users, a pull request was merged that will use the python-socks library when available for proxy support. This library natively supports asyncio, so it should work better than the old pysocks. pysocks will still be used if the new library is not available, and both will be handled transparently by Telethon so you don’t need to worry about it.

Additions
  • New client.set_proxy() method which lets you change the proxy without recreating the client. You will need to reconnect for it to take effect, but you won’t need to recreate the client. This is also an external contribution.
  • New method to unpin messages client.unpin_message().
Enhancements
  • Empty peers are excluded from the list of dialogs.
  • If the python-socks library is installed (new optional requirement), it will be used instead of pysocks for proxy support. This should fix some issues with proxy timeouts, because the new library natively supports asyncio.
  • client.send_file() will now group any media type, instead of sending non-image documents separatedly. This lets you create music albums, for example.
  • You can now search messages with a from_user that’s not a user. This is a Telegram feature, we know the name isn’t great, but backwards-compatibility has to be kept.
Bug fixes
  • Fixes related to conversation timeouts.
  • Large dates (over year 2038) now wrap around a 32-bit integer, which is the only way we can represent them to Telegram. Even if “wrong”, it makes things not crash, and it’s the best we can do with 32-bit dates.
  • The library was accidentally using a deprecated argument in one of its friendly methods, producing a warning.
  • Improvements to the way marked IDs are parsed.
  • SlowModeWaitError floods are no longer cached.
  • Getting the buttons for a message could fail sometimes.
  • Getting the display name for “forbidden” chats now works.
  • Better handling of errors in some internal methods.

Channel comments and Anonymous Admins (v1.17)

Scheme layer used: 119

New minor version, new layer change! This time is a good one to remind every consumer of Python libraries that you should always specify fixed versions of your dependencies! If you’re using a requirements.txt file and you want to stick with the old version (or any version) for the time being, you can use the following syntax:

telethon~=1.16.0

This will install any version compatible with the written version (so, any in the 1.16 series). Patch releases will never break your code (and if they do, it’s a bug). You can also use that syntax in pip install. Your code can’t know what new versions will look like, so saying it will work with all versions is a lie and will cause issues.

The reason to bring this up is that Telegram has changed things again, and with the introduction of anonymous administrators and channel comments, the sender of a message may not be a User! To accomodate for this, the field is now a Peer and not int. As a reminder, it’s always a good idea to use Telethon’s friendly methods and custom properties, which have a higher stability guarantee than accessing raw API fields.

Even if you don’t update, your code will still need to account for the fact that the sender of a message might be one of the accounts Telegram introduced to preserve backwards compatibility, because this is a server-side change, so it’s better to update and not lag behind. As it’s mostly just a single person driving the project on their free time, bug-fixes are not backported.

This version also updates the format of SQLite sessions (the default), so after upgrading and using an old session, the session will be updated, which means trying to use it back in older versions of the library won’t work.

For backwards-compatibility sake, the library has introduced the properties Message.reply_to_msg_id and Message.to_id that behave like they did before (Telegram has renamed and changed how these fields work).

Breaking Changes
  • Message.from_id is now a Peer, not int! If you want the marked sender ID (much like old behaviour), replace all uses of .from_id with .sender_id. This will mostly work, but of course in old and new versions you have to account for the fact that this sender may no longer be a user.
  • You can no longer assign to Message.reply_to_msg_id and Message.to_id because these are now properties that offer a “view” to the real value from a different field.
  • Answering inline queries with a photo or document will now send the photo or document used in the resulting message by default. Not sending the media was technically a bug, but some people may be relying on this old behaviour. You can use the old behaviour with include_media=False.
Additions
  • New raise_last_call_error parameter in the client constructor to raise the same error produced by the last failing call, rather than a generic ValueError.
  • New formatting_entities parameter in client.send_message(), and client.send_file() to bypass the parse mode and manually specify the formatting entities.
  • New client.get_permissions() method to query a participant’s permissions in a group or channel. This request is slightly expensive in small group chats because it has to fetch the entire chat to check just a user, so use of a cache is advised.
  • Message.click() now works on normal polls!
  • New local_addr parameter in the client constructor to use a specific local network address when connecting to Telegram.
  • client.inline_query() now lets you specify the chat where the query is being made from, which some bots need to provide certain functionality.
  • You can now get comments in a channel post with the reply_to parameter in client.iter_messages(). Comments are messages that “reply to” a specific channel message, hence the name (which is consistent with how Telegram’s API calls it).
Enhancements
  • Updated documentation and list of known errors.
  • If hachoir is available, the file metadata can now be extracted from streams and in-memory bytes.
  • The default parameters used to initialize a connection now match the format of those used by Telegram Desktop.
  • Specifying 0 retries will no longer cause the library to attempt to reconnect.
  • The library should now be able to reliably download very large files.
  • Global search should work more reliably now.
  • Old usernames are evicted from cache, so getting entities by cached username should now be more reliable.
  • Slightly less noisy logs.
  • Stability regarding transport-level errors (transport flood, authorization key not found) should be improved. In particular, you should no longer be getting unnecessarily logged out.
  • Reconnection should no longer occur if the client gets logged out (for example, another client revokes the session).
Bug fixes
  • In some cases, there were issues when using events.Album together with events.Raw.
  • For some channels, one of their channel photos would not show up in client.iter_profile_photos().
  • In some cases, a request that failed to be sent would be forgotten, causing the original caller to be “locked” forever for a response that would never arrive. Failing requests should now consistently be automatically re-sent.
  • The library should more reliably handle certain updates with “empty” data.
  • Sending documents in inline queries should now work fine.
  • Manually using client.sign_up should now work correctly, instead of claiming “code invalid”.

Special mention to some of the other changes in the 1.16.x series:

  • The thumb for download_media now supports both str and VideoSize.
  • Thumbnails are sorted, so -1 is always the largest.

Bug Fixes (v1.16.1)

The last release added support to force_file on any media, including things that were not possible before like .webp files. However, the force_document toggle commonly used for photos was applied “twice” (one told the library to send it as a document, and then to send that document as file), which prevented Telegram for analyzing the images. Long story short, sending files to the stickers bot stopped working, but that’s been fixed now, and sending photos as documents include the size attribute again as long as Telegram adds it.

Enhancements
  • When trying to client.start() to another account if you were previously logged in, the library will now warn you because this is probably not intended. To avoid the warning, make sure you’re logging in to the right account or logout from the other first.
  • Sending a copy of messages with polls will now work when possible.
  • The library now automatically retries on inter-dc call errors (which occur when Telegram has internal issues).
Bug Fixes
  • The aforementioned issue with force_document.
  • Square brackets removed from IPv6 addresses. This may fix IPv6 support.

Channel Statistics (v1.16)

Scheme layer used: 116

The newest Telegram update has a new method to also retrieve megagroup statistics, which can now be used with client.get_stats(). This way you’ll be able to access the raw data about your channel or megagroup statistics.

The maximum file size limit has also been increased to 2GB on the server, so you can send even larger files.

Breaking Changes
  • Besides the obvious layer change, the loop argument is now ignored. It has been deprecated since Python 3.8 and will be removed in Python 3.10, and also caused some annoying warning messages when using certain parts of the library. If you were (incorrectly) relying on using a different loop from the one that was set, things may break.
Enhancements
  • client.upload_file() now works better when streaming files (anything that has a .read()), instead of reading it all into memory when possible.

QR login (v1.15)

Published at 2020/07/04

Scheme layer used: 114

The library now has a friendly method to perform QR-login, as detailed in https://core.telegram.org/api/qr-login. It won’t generate QR images, but it provides a way for you to easily do so with any other library of your choice.

Additions
Enhancements
Bug fixes
  • No longer crash on updates received prior to logging in.
  • Server-side changes caused clicking on inline buttons to trigger a different error, which is now handled correctly.

Minor quality of life improvements (v1.14)

Published at 2020/05/26

Scheme layer used: 113

Some nice things that were missing, along with the usual bug-fixes.

Additions
  • New Message.dice property.
  • The func= parameter of events can now be an async function.
Bug fixes
  • Fixed client.action() having an alias wrong.
  • Fixed incorrect formatting of some errors.
  • Probably more reliable detection of pin events in small groups.
  • Fixed send methods on client.conversation() were not honoring cancellation.
  • Flood waits of zero seconds are handled better.
  • Getting the pinned message in a chat was failing.
  • Fixed the return value when forwarding messages if some were missing and also the return value of albums.
Enhancements
  • .tgs files are now recognised as animated stickers.
  • The service message produced by Message.pin() is now returned.
  • Sending a file with client.send_file() now works fine when you pass an existing dice media (e.g. sending a message copy).
  • client.edit_permissions() now has the embed_links parameter which was missing.

Bug Fixes (v1.13)

Published at 2020/04/25

Scheme layer used: 112

Bug fixes and layer bump.

Bug fixes
  • Passing None as the entity to client.delete_messages() would fail.
  • When downloading a thumbnail, the name inferred was wrong.

Bug Fixes (v1.12)

Published at 2020/04/20

Scheme layer used: 111

Once again nothing major, but a few bug fixes and primarily the new layer deserves a new minor release.

Bug fixes

These were already included in the v1.11.3 patch:

  • libssl check was failing on macOS.
  • Getting input users would sometimes fail on events.ChatAction.

These bug fixes are available in this release and beyond:

  • Avoid another occurrence of MemoryError.
  • Sending large files in albums would fail because it tried to cache them.
  • The thumb was being ignored when sending files from InputFile.
  • Fixed editing inline messages from callback queries in some cases.
  • Proxy connection is now blocking which should help avoid some errors.

Bug Fixes (v1.11)

Published at 2020/02/20

Scheme layer used: 110

It has been a while since the last release, and a few bug fixes have been made since then. This release includes them and updates the scheme layer.

Note that most of the bug-fixes are available in the v1.10.10 patch.

Bug fixes
  • Fix MemoryError when casting certain media.
  • Fix client.get_entity() on small group chats.
  • client.delete_dialog() now handles deactivated chats more gracefully.
  • Sending a message with file= would ignore some of the parameters.
  • Errors are now un-pickle-able once again.
  • Fixed some issues regarding markdown and HTML (un)parsing.

The following are also present in v1.10.10:

Enhancements
  • Videos can now be included when sending albums.
  • Getting updates after reconnect should be more reliable.
  • Updated documentation and added more examples.
  • More security checks during the generation of the authorization key.

The following are also present in v1.10.10:

  • URLs like t.me/@username are now valid.
  • Auto-sleep now works for slow-mode too.
  • Improved some error messages.
  • Some internal improvements and updating.
  • client.pin_message() now also works with message objects.
  • Asynchronous file descriptors are now allowed during download and upload.

Scheduled Messages (v1.10)

Published at 2019/09/08

Scheme layer used: 105

You can now schedule messages to be sent (or edited, or forwarded…) at a later time, which can also work as reminders for yourself when used in your own chat!

from datetime import timedelta

# Remind yourself to walk the dog in 10 minutes (after you play with Telethon's update)
await client.send_message('me', 'Walk the dog',
                          schedule=timedelta(minutes=10))

# Remind your friend tomorrow to update Telethon
await client.send_message(friend, 'Update Telethon!',
                          schedule=timedelta(days=1))
Additions
Bug fixes
  • Fix calling flush on file objects which lack this attribute.
  • Fix CallbackQuery pattern.
  • Fix client.action() not returning itself when used in a context manager (so the as would be None).
  • Fix sending InputKeyboardButtonUrlAuth as inline buttons.
  • Fix client.edit_permissions() defaults.
  • Fix Forward had its client as None.
  • Fix (de)serialization of negative timestamps (caused by the information in some sites with instant view, where the date could be very old).
  • Fix HTML un-parsing.
  • Fix to/from_id in private messages when using multiple clients.
  • Stop disconnecting from None (incorrect logging).
  • Fix double-read on double-connect.
  • Fix client.get_messages() when being passed more than 100 IDs.
  • Fix Message.document for documents coming from web-pages.
Enhancements
  • Some documentation improvements, including the TL reference.
  • Documentation now avoids telethon.sync, which should hopefully be less confusing.
  • Better error messages for flood wait.
  • You can now client.get_drafts() for a single entity (which means you can now get a single draft from a single chat).
  • New-style file IDs now work with Telethon.
  • The progress_callback for client.upload_file() can now be an async def.

Animated Stickers (v1.9)

Published at 2019/07/06

Scheme layer used: 103

With the layer 103, Telethon is now able to send and receive animated stickers! These use the 'application/x-tgsticker' mime-type and for now, you can access its raw data, which is a gzipped JSON.

Additions
Bug fixes

  • Fixed downloading contacts.

  • Fixed client.iter_dialogs() missing some under certain circumstances.

  • Fixed incredibly slow imports under some systems due to expensive path resolution when searching for libssl.

  • Fixed captions when sending albums.

  • Fixed invalid states in Conversation.

  • Fixes to some methods in utils regarding extensions.

  • Fixed memory cycle in Forward which let you do things like the following:

    original_fwd = message.forward.original_fwd.original_fwd.original_fwd.original_fwd.original_fwd.original_fwd

    Hopefully you didn’t rely on that in your code.

  • Fixed File.ext not working on unknown mime-types, despite the file name having the extension.

  • Fixed ids=..., reverse=True in client.iter_messages().

  • Fixed Draft not being aware of the entity.

  • Added missing re-exports in telethon.sync.

Enhancements

Documentation Overhaul (v1.8)

Published at 2019/05/30

Scheme layer used: 100

The documentation has been completely reworked from the ground up, with awesome new quick references such as Client Reference to help you quickly find what you need!

Raw methods also warn you when a friendly variant is available, so that you don’t accidentally make your life harder than it has to be.

In addition, all methods in the client now are fully annotated with type hints! More work needs to be done, but this should already help a lot when using Telethon from any IDEs.

You may have noticed that the patch versions between v1.7.2 to v1.7.7 have not been documented. This is because patch versions should only contain bug fixes, no new features or breaking changes. This hasn’t been the case in the past, but from now on, the library will try to adhere more strictly to the Semantic Versioning principles.

If you ever want to look at those bug fixes, please use the appropriated git command, such as git shortlog v1.7.1...v1.7.4, but in general, they probably just fixed your issue.

With that out of the way, let’s look at the full change set:

Breaking Changes
  • The layer changed, so take note if you use the raw API, as it’s usual.
  • The way photos are downloaded changed during the layer update of the previous version, and fixing that bug as a breaking change in itself. client.download_media() now offers a different way to deal with thumbnails.
Additions
Bug fixes
  • Correctly catch and raise all RPC errors.
  • Downloading stripped photos wouldn’t work correctly.
  • Under some systems, libssl would fail to load earlier than expected, causing the library to fail when being imported.
  • conv.get_response() after ID 0 wasn’t allowed when it should.
  • InlineBuilder only worked with local files, but files from anywhere are supported.
  • Accessing the text property from a raw-API call to fetch Message would fail (any any other property that needed the client).
  • Database is now upgraded if the version was lower, not different. From now on, this should help with upgrades and downgrades slightly.
  • Fixed saving pts and session-related stuff.
  • Disconnection should not raise any errors.
  • Invite links of the form tg://join?invite= now work.
  • client.iter_participants(search=...) now works on private chats again.
  • Iterating over messages in reverse with a date as offset wouldn’t work.
  • The conversation would behave weirdly when a timeout occurred.
Enhancements
  • telethon now re-export all the goodies that you commonly need when using the library, so e.g. from telethon import Button will now work.
  • telethon.sync now re-exports everything from telethon, so that you can trivially import from just one place everything that you need.
  • More attempts at reducing CPU usage after automatically fetching missing entities on events. This isn’t a big deal, even if it sounds like one.
  • Hexadecimal invite links are now supported. You didn’t need them, but they will now work.
Internal Changes
  • Deterministic code generation. This is good for diff.
  • On Python 3.7 and above, we properly close the connection.
  • A lot of micro-optimization.
  • Fixes to bugs introduced while making this release.
  • Custom commands on setup.py are nicer to use.

Fix-up for Photo Downloads (v1.7.1)

Published at 2019/04/24

Telegram changed the way thumbnails (which includes photos) are downloaded, so you can no longer use a PhotoSize alone to download a particular thumbnail size (this is a breaking change).

Instead, you will have to specify the new thumb parameter in client.download_media() to download a particular thumbnail size. This addition enables you to easily download thumbnails from documents, something you couldn’t do easily before.

Easier Events (v1.7)

Published at 2019/04/22

Scheme layer used: 98

If you have been using Telethon for a while, you probably know how annoying the “Could not find the input entity for…” error can be. In this new version, the library will try harder to find the input entity for you!

That is, instead of doing:

@client.on(events.NewMessage)
async def handler(event):
    await client.download_profile_photo(await event.get_input_sender())
    # ...... needs await, it's a method ^^^^^                       ^^

You can now do:

@client.on(events.NewMessage)
async def handler(event):
    await client.download_profile_photo(event.input_sender)
    # ...... no await, it's a property! ^
    # It's also 12 characters shorter :)

And even the following will hopefully work:

@client.on(events.NewMessage)
async def handler(event):
    await client.download_profile_photo(event.sender_id)

A lot of people use IDs thinking this is the right way of doing it. Ideally, you would always use input_*, not sender or sender_id (and the same applies to chats). But, with this change, IDs will work just the same as input_* inside events.

This feature still needs some more testing, so please do open an issue if you find strange behaviour.

Breaking Changes
  • The layer changed, and a lot of things did too. If you are using raw API, you should be careful with this. In addition, some attributes weren’t of type datetime when they should be, which has been fixed.
  • Due to the layer change, you can no longer download photos with just their PhotoSize. Version 1.7.1 introduces a new way to download thumbnails to work around this issue.
  • client.disconnect() is now asynchronous again. This means you need to await it. You don’t need to worry about this if you were using with client or client.run_until_disconnected. This should prevent the “pending task was destroyed” errors.
Additions

  • New in-memory cache for input entities. This should mean a lot less of disk look-ups.

  • New client.action method to easily indicate that you are doing some chat action:

    async with client.action(chat, 'typing'):
    await asyncio.sleep(2)  # type for 2 seconds
    await client.send_message(chat, 'Hello world! I type slow ^^')

    You can also easily use this for sending files, playing games, etc.

New bugs
  • Downloading photos is broken. This is fixed in v1.7.1.
Bug fixes
  • Fix sending photos from streams/bytes.
  • Fix unhandled error when sending requests that were too big.
  • Fix edits that arrive too early on conversations.
  • Fix client.edit_message() when trying to edit a file.
  • Fix method calls on the objects returned by client.iter_dialogs().
  • Attempt at fixing client.iter_dialogs() missing many dialogs.
  • offset_date in client.iter_messages() was being ignored in some cases. This has been worked around.
  • Fix callback_query.edit().
  • Fix CallbackQuery(func=...) was being ignored.
  • Fix UserUpdate not working for “typing” (and uploading file, etc.) status.
  • Fix library was not expecting IOError from PySocks.
  • Fix library was raising a generic ConnectionError and not the one that actually occurred.
  • Fix the blacklist_chats parameter in MessageRead not working as intended.
  • Fix client.download_media(contact).
  • Fix mime type when sending mp3 files.
  • Fix forcibly getting the sender or chat from events would not always return all their information.
  • Fix sending albums with client.send_file() was not returning the sent messages.
  • Fix forwarding albums with client.forward_messages().
  • Some fixes regarding filtering updates from chats.
  • Attempt at preventing duplicated updates.
  • Prevent double auto-reconnect.
Enhancements
  • Some improvements related to proxy connections.
  • Several updates and improvements to the documentation, such as optional dependencies now being properly listed.
  • You can now forward messages from different chats directly with client.forward_messages.

Tidying up Internals (v1.6)

Published at 2019/02/27

Scheme layer used: 95

First things first, sorry for updating the layer in the previous patch version. That should only be done between major versions ideally, but due to how Telegram works, it’s done between minor versions. However raw API has and will always be considered “unsafe”, this meaning that you should always use the convenience client methods instead. These methods don’t cover the full API yet, so pull requests are welcome.

Breaking Changes

  • The layer update, of course. This didn’t really need a mention here.

  • You can no longer pass a batch_size when iterating over messages. No other method exposed this parameter, and it was only meant for testing purposes. Instead, it’s now a private constant.

  • client.iter_* methods no longer have a _total parameter which was supposed to be private anyway. Instead, they return a new generator object which has a .total attribute:

    it = client.iter_messages(chat)
for i, message in enumerate(it, start=1):
    percentage = i / it.total
    print('{:.2%} {}'.format(percentage, message.text))
Additions

  • You can now pass phone and phone_code_hash in client.sign_up, although you probably don’t need that.

  • Thanks to the overhaul of all client.iter_* methods, you can now do:

    for message in reversed(client.iter_messages('me')):
    print(message.text)
Bug fixes
Enhancements
  • Cleared up some parts of the documentation.
  • Improved some auto-casts to make life easier.
  • Improved image detection. Now you can easily send bytes and streams of images as photos, unless you force document.
  • Sending images as photos that are too large will now be resized before uploading, reducing the time it takes to upload them and also avoiding errors when the image was too large (as long as pillow is installed). The images will remain unchanged if you send it as a document.
  • Treat errors.RpcMcgetFailError as a temporary server error to automatically retry shortly. This works around most issues.
Internal changes
  • New common way to deal with retries (retry_range).
  • Cleaned up the takeout client.
  • Completely overhauled asynchronous generators.

Layer Update (v1.5.5)

Published at 2019/01/14

Scheme layer used: 93

There isn’t an entry for v1.5.4 because it contained only one hot-fix regarding loggers. This update is slightly bigger so it deserves mention.

Additions
Bug fixes
  • Dealing with mimetypes should cause less issues in systems like Windows.
  • Potentially fix alternative session storages that had issues with dates.
Enhancements
  • Saner timeout defaults for conversations.
  • Path-like files are now supported for thumbnails.
  • Added new hot-keys to the online documentation at https://tl.telethon.dev/ such as / to search. Press ? to view them all.

Bug Fixes (v1.5.3)

Published at 2019/01/14

Several bug fixes and some quality of life enhancements.

Breaking Changes
  • message.edit now respects the previous message buttons or link preview being hidden. If you want to toggle them you need to explicitly set them. This is generally the desired behaviour, but may cause some bots to have buttons when they shouldn’t.
Additions
  • You can now “hide_via” when clicking on results from client.inline_query to @bing and @gif.
  • You can now further configure the logger Telethon uses to suit your needs.
Bug fixes
  • Fixes for ReadTheDocs to correctly build the documentation.
  • Fix UserEmpty not being expected when getting the input variant.
  • The message object returned when sending a message with buttons wouldn’t always contain the ReplyMarkup.
  • Setting email when configuring 2FA wasn’t properly supported.
  • utils.resolve_bot_file_id now works again for photos.
Enhancements

Takeout Sessions (v1.5.2)

Published at 2019/01/05

You can now easily start takeout sessions (also known as data export sessions) through client.takeout(). Some of the requests will have lower flood limits when done through the takeout session.

Bug fixes
  • The new AdminLogEvent had a bug that made it unusable.
  • client.iter_dialogs() will now locally check for the offset date, since Telegram ignores it.
  • Answering inline queries with media no works properly. You can now use the library to create inline bots and send stickers through them!

object.to_json() (v1.5.1)

Published at 2019/01/03

The library already had a way to easily convert the objects the API returned into dictionaries through object.to_dict(), but some of the fields are dates or bytes which JSON can’t serialize directly.

For convenience, a new object.to_json() has been added which will by default format both of those problematic types into something sensible.

Bug fixes
  • client.is_connected() would be wrong when the initial connection failed.
  • Fixed UnicodeDecodeError when accessing the text of messages with malformed offsets in their entities.
  • Fixed client.get_input_entity() for integer IDs that the client has not seen before.
Enhancements
  • You can now configure the reply markup when using Button as a bot.
  • More properties for Message to make accessing media convenient.
  • Downloading to file=bytes will now return a bytes object with the downloaded media.

Polls with the Latest Layer (v1.5)

Published at 2018/12/25

Scheme layer used: 91

This version doesn’t really bring many new features, but rather focuses on updating the code base to support the latest available Telegram layer, 91. This layer brings polls, and you can create and manage them through Telethon!

Breaking Changes
  • The layer change from 82 to 91 changed a lot of things in the raw API, so be aware that if you rely on raw API calls, you may need to update your code, in particular if you work with files. They have a new file_reference parameter that you must provide.
Bug fixes
  • Markdown and HTML parsing now behave correctly with leading whitespace.
  • HTTP connection should now work correctly again.
  • Using caption=None would raise an error instead of setting no caption.
  • KeyError is now handled properly when forwarding messages.
  • button.click() now works as expected for KeyboardButtonGame.
Enhancements
  • Some improvements to the search in the full API and generated examples.
  • Using entities with access_hash = 0 will now work in more cases.
Internal changes
  • Some changes to the documentation and code generation.
  • 2FA code was updated to work under the latest layer.

Error Descriptions in CSV files (v1.4.3)

Published at 2018/12/04

While this may seem like a minor thing, it’s a big usability improvement.

Anyone who wants to update the documentation for known errors, or whether some methods can be used as a bot, user or both, can now be easily edited. Everyone is encouraged to help document this better!

Bug fixes
  • TimeoutError was not handled during automatic reconnects.
  • Getting messages by ID using InputMessageReplyTo could fail.
  • Fixed message.get_reply_message as a bot when a user replied to a different bot.
  • Accessing some document properties in a Message would fail.
Enhancements
  • Accessing events.ChatAction properties such as input users may now work in more cases.
Internal changes
  • Error descriptions and information about methods is now loaded from a CSV file instead of being part of several messy JSON files.

Bug Fixes (v1.4.2)

Published at 2018/11/24

This version also includes the v1.4.1 hot-fix, which was a single quick fix and didn’t really deserve an entry in the changelog.

Bug fixes
  • Authorization key wouldn’t be saved correctly, requiring re-login.
  • Conversations with custom events failed to be cancelled.
  • Fixed telethon.sync when using other threads.
  • Fix markdown/HTML parser from failing with leading/trailing whitespace.
  • Fix accessing chat_action_event.input_user property.
  • Potentially improved handling unexpected disconnections.
Enhancements
  • Better default behaviour for client.send_read_acknowledge.
  • Clarified some points in the documentation.
  • Clearer errors for utils.get_peer*.

Connection Overhaul (v1.4)

Published at 2018/11/03

Yet again, a lot of work has been put into reworking the low level connection classes. This means asyncio.open_connection is now used correctly and the errors it can produce are handled properly. The separation between packing, encrypting and network is now abstracted away properly, so reasoning about the code is easier, making it more maintainable.

As a user, you shouldn’t worry about this, other than being aware that quite a few changes were made in the insides of the library and you should report any issues that you encounter with this version if any.

Breaking Changes
  • The threaded version of the library will no longer be maintained, primarily because it never was properly maintained anyway. If you have old code, stick with old versions of the library, such as 0.19.1.6.
  • Timeouts no longer accept timedelta. Simply use seconds.
  • The callback parameter from telethon.tl.custom.button.Button.inline() was removed, since it had always been a bad idea. Adding the callback there meant a lot of extra work for every message sent, and only registering it after the first message was sent! Instead, use telethon.events.callbackquery.CallbackQuery.
Additions
Bug fixes
Enhancements
  • Support next_offset in inline_query.answer().
  • Support <a href="tg://user?id=123"> mentions in HTML parse mode.
  • New auto-casts for InputDocument and InputChatPhoto.
  • Conversations are now exclusive per-chat by default.
  • The request that caused a RPC error is now shown in the error message.
  • New full API examples in the generated documentation.
  • Fixed some broken links in the documentation.
  • client.disconnect() is now synchronous, but you can still await it for consistency or compatibility.

Event Templates (v1.3)

Published at 2018/09/22

If you have worked with Flask templates, you will love this update, since it gives you the same features but even more conveniently:

# handlers/welcome.py
from telethon import events

@events.register(events.NewMessage('(?i)hello'))
async def handler(event):
    client = event.client
    await event.respond('Hi!')
    await client.send_message('me', 'Sent hello to someone')

This will register the handler callback to handle new message events. Note that you didn’t add this to any client yet, and this is the key point: you don’t need a client to define handlers! You can add it later:

# main.py
from telethon import TelegramClient
import handlers.welcome

with TelegramClient(...) as client:
    # This line adds the handler we defined before for new messages
    client.add_event_handler(handlers.welcome.handler)
    client.run_until_disconnected()

This should help you to split your big code base into a more modular design.

Breaking Changes
  • .sender is the .chat when the message is sent in a broadcast channel. This makes sense, because the sender of the message was the channel itself, but you now must take into consideration that it may be either a User or Channel instead of being None.
Additions
  • New MultiError class when invoking many requests at once through client([requests]).
  • New custom func= on all events. These will receive the entire event, and a good usage example is func=lambda e: e.is_private.
  • New .web_preview field on messages. The .photo and .document will also return the media in the web preview if any, for convenience.
  • Callback queries now have a .chat in most circumstances.
Bug fixes
  • Running code with python3 -O would remove critical code from asserts.
  • Fix some rare ghost disconnections after reconnecting.
  • Fix strange behavior for send_message(chat, Message, reply_to=foo).
  • The loop= argument was being pretty much ignored.
  • Fix MemorySession file caching.
  • The logic for getting entities from their username is now correct.
  • Fixes for sending stickers from .webp files in Windows, again.
  • Fix disconnection without being logged in.
  • Retrieving media from messages would fail.
  • Getting some messages by ID on private chats.
Enhancements
  • iter_participants will now use its search= as a symbol set when aggressive=True, so you can do client.get_participants(group, aggressive=True, search='абвгдеёжзийклмнопрст').
  • The StringSession supports custom encoding.
  • Callbacks for telethon.client.auth.AuthMethods.start can be async.
Internal changes
  • Cherry-picked a commit to use asyncio.open_connection in the lowest level of the library. Do open issues if this causes trouble, but it should otherwise improve performance and reliability.
  • Building and resolving events overhaul.

Conversations, String Sessions and More (v1.2)

Published at 2018/08/14

This is a big release! Quite a few things have been added to the library, such as the new Conversation. This makes it trivial to get tokens from @BotFather:

from telethon.tl import types

with client.conversation('BotFather') as conv:
    conv.send_message('/mybots')
    message = conv.get_response()
    message.click(0)
    message = conv.get_edit()
    message.click(0)
    message = conv.get_edit()
    for _, token in message.get_entities_text(types.MessageEntityCode):
        print(token)

In addition to that, you can now easily load and export session files without creating any on-disk file thanks to the StringSession:

from telethon.sessions import StringSession
string = StringSession.save(client.session)

Check out Session Files for more details.

For those who aren’t able to install cryptg, the support for libssl has been added back. While interfacing libssl is not as fast, the speed when downloading and sending files should really be noticeably faster.

While those are the biggest things, there are still more things to be excited about.

Additions
Bug fixes
  • UTC timezone for telethon.events.userupdate.UserUpdate.
  • Bug with certain input parameters when iterating messages.
  • RPC errors without parent requests caused a crash, and better logging.
  • incoming = outgoing = True was not working properly.
  • Getting a message’s ID was not working.
  • File attributes not being inferred for open()’ed files.
  • Use MemorySession if sqlite3 is not installed by default.
  • Self-user would not be saved to the session file after signing in.
  • client.catch_up() seems to be functional again.
Enhancements
  • Updated documentation.
  • Invite links will now use cache, so using them as entities is cheaper.
  • You can reuse message buttons to send new messages with those buttons.
  • .to_dict() will now work even on invalid TLObject’s.

Better Custom Message (v1.1.1)

Published at 2018/07/23

The custom.Message class has been rewritten in a cleaner way and overall feels less hacky in the library. This should perform better than the previous way in which it was patched.

The release is primarily intended to test this big change, but also fixes Python 3.5.2 compatibility which was broken due to a trailing comma.

Bug fixes
  • Using functools.partial on event handlers broke updates if they had uncaught exceptions.
  • A bug under some session files where the sender would export authorization for the same data center, which is unsupported.
  • Some logical bugs in the custom message class.

Bot Friendly (v1.1)

Published at 2018/07/21

Two new event handlers to ease creating normal bots with the library, namely events.InlineQuery and events.CallbackQuery for handling @InlineBot queries or reacting to a button click. For this second option, there is an even better way:

from telethon.tl.custom import Button

async def callback(event):
    await event.edit('Thank you!')

bot.send_message(chat, 'Hello!',
                 buttons=Button.inline('Click me', callback))

You can directly pass the callback when creating the button.

This is fine for small bots but it will add the callback every time you send a message, so you probably should do this instead once you are done testing:

markup = bot.build_reply_markup(Button.inline('Click me', callback))
bot.send_message(chat, 'Hello!', buttons=markup)

And yes, you can create more complex button layouts with lists:

from telethon import events

global phone = ''

@bot.on(events.CallbackQuery)
async def handler(event):
    global phone
    if event.data == b'<':
        phone = phone[:-1]
    else:
        phone += event.data.decode('utf-8')

    await event.answer('Phone is now {}'.format(phone))

markup = bot.build_reply_markup([
    [Button.inline('1'), Button.inline('2'), Button.inline('3')],
    [Button.inline('4'), Button.inline('5'), Button.inline('6')],
    [Button.inline('7'), Button.inline('8'), Button.inline('9')],
    [Button.inline('+'), Button.inline('0'), Button.inline('<')],
])
bot.send_message(chat, 'Enter a phone', buttons=markup)

(Yes, there are better ways to do this). Now for the rest of things:

Additions
  • New custom.Button class to help you create inline (or normal) reply keyboards. You must sign in as a bot to use the buttons= parameters.
  • New events usable if you sign in as a bot: events.InlineQuery and events.CallbackQuery.
  • New silent parameter when sending messages, usable in broadcast channels.
  • Documentation now has an entire section dedicate to how to use the client’s friendly methods at (removed broken link).
Bug fixes
  • Empty except are no longer used which means sending a keyboard interrupt should now work properly.
  • The pts of incoming updates could be None.
  • UTC timezone information is properly set for read datetime.
  • Some infinite recursion bugs in the custom message class.
  • Updates was being dispatched to raw handlers when it shouldn’t.
  • Using proxies and HTTPS connection mode may now work properly.
  • Less flood waits when downloading media from different data centers, and the library will now detect them even before sending requests.
Enhancements
  • Interactive sign in now supports signing in with a bot token.
  • timedelta is now supported where a date is expected, which means you can e.g. ban someone for timedelta(minutes=5).
  • Events are only built once and reused many times, which should save quite a few CPU cycles if you have a lot of the same type.
  • You can now click inline buttons directly if you know their data.
Internal changes
  • When downloading media, the right sender is directly used without previously triggering migrate errors.
  • Code reusing for getting the chat and the sender, which easily enables this feature for new types.

New HTTP(S) Connection Mode (v1.0.4)

Published at 2018/07/09

This release implements the HTTP connection mode to the library, which means certain proxies that only allow HTTP connections should now work properly. You can use it doing the following, like any other mode:

from telethon import TelegramClient, sync
from telethon.network import ConnectionHttp

client = TelegramClient(..., connection=ConnectionHttp)
with client:
    client.send_message('me', 'Hi!')
Additions
Bug fixes
  • If several TLMessage in a MessageContainer exceeds 1MB, it will no longer be automatically turned into one. This basically means that e.g. uploading 10 file parts at once will work properly again.
  • Documentation fixes and some missing await.
  • Revert named argument for client.forward_messages
Enhancements
Internal changes
  • Outgoing TLMessage are now pre-packed so if there’s an error when serializing the raw requests, the library will no longer swallow it. This also means re-sending packets doesn’t need to re-pack their bytes.

Iterate Messages in Reverse (v1.0.3)

Published at 2018/07/04

Scheme layer used: 82

Mostly bug fixes, but now there is a new parameter on client.iter_messages to support reversing the order in which messages are returned.

Additions
  • The mentioned reverse parameter when iterating over messages.
  • A new sequential_updates parameter when creating the client for updates to be processed sequentially. This is useful when you need to make sure that all updates are processed in order, such as a script that only forwards incoming messages somewhere else.
Bug fixes
  • Count was always None for message.button_count.
  • Some fixes when disconnecting upon dropping the client.
  • Support for Python 3.4 in the sync version, and fix media download.
  • Some issues with events when accessing the input chat or their media.
  • Hachoir wouldn’t automatically close the file after reading its metadata.
  • Signing in required a named code= parameter, but usage without a name was really widespread so it has been reverted.

Bug Fixes (v1.0.2)

Published at 2018/06/28

Updated some asserts and parallel downloads, as well as some fixes for sync.

Bug Fixes (v1.0.1)

Published at 2018/06/27

And as usual, every major release has a few bugs that make the library unusable! This quick update should fix those, namely:

Bug fixes
  • client.start() was completely broken due to a last-time change requiring named arguments everywhere.
  • Since the rewrite, if your system clock was wrong, the connection would get stuck in an infinite “bad message” loop of responses from Telegram.
  • Accessing the buttons of a custom message wouldn’t work in channels, which lead to fix a completely different bug regarding starting bots.
  • Disconnecting could complain if the magic telethon.sync was imported.
  • Successful automatic reconnections now ask Telegram to send updates to us once again as soon as the library is ready to listen for them.

Synchronous magic (v1.0)

Published at 2018/06/27

Important

If you come from Telethon pre-1.0 you really want to read Compatibility and Convenience to port your scripts to the new version.

The library has been around for well over a year. A lot of improvements have been made, a lot of user complaints have been fixed, and a lot of user desires have been implemented. It’s time to consider the public API as stable, and remove some of the old methods that were around until now for compatibility reasons. But there’s one more surprise!

There is a new magic telethon.sync module to let you use all the methods in the TelegramClient (and the types returned from its functions) in a synchronous way, while using asyncio behind the scenes! This means you’re now able to do both of the following:

import asyncio

async def main():
  await client.send_message('me', 'Hello!')

asyncio.run(main())

# ...can be rewritten as:

from telethon import sync
client.send_message('me', 'Hello!')

Both ways can coexist (you need to await if the loop is running).

You can also use the magic sync module in your own classes, and call sync.syncify(cls) to convert all their async def into magic variants.

Breaking Changes
  • message.get_fwd_sender is now in message.forward.
  • client.idle is now client.run_until_disconnected()
  • client.add_update_handler is now client.add_event_handler
  • client.remove_update_handler is now client.remove_event_handler
  • client.list_update_handlers is now client.list_event_handlers
  • client.get_message_history is now client.get_messages
  • client.send_voice_note is now client.send_file with is_voice=True.
  • client.invoke() is now client(...).
  • report_errors has been removed since it’s currently not used, and flood_sleep_threshold is now part of the client.
  • The update_workers and spawn_read_thread arguments are gone. Simply remove them from your code when you create the client.
  • Methods with a lot of arguments can no longer be used without specifying their argument. Instead you need to use named arguments. This improves readability and not needing to learn the order of the arguments, which can also change.
Additions

  • client.send_file now accepts external http:// and https:// URLs.

  • You can use the TelegramClient inside of with blocks, which will client.start() and disconnect() the client for you:

    from telethon import TelegramClient, sync

with TelegramClient(name, api_id, api_hash) as client:
    client.send_message('me', 'Hello!')

    Convenience at its maximum! You can even chain the .start() method since it returns the instance of the client:

    with TelegramClient(name, api_id, api_hash).start(bot_token=token) as bot:
    bot.send_message(chat, 'Hello!')
Bug fixes
  • There were some @property async def left, and some await property.
  • “User joined” event was being treated as “User was invited”.
  • SQLite’s cursor should not be closed properly after usage.
  • await the updates task upon disconnection.
  • Some bug in Python 3.5.2’s asyncio causing 100% CPU load if you forgot to call client.disconnect(). The method is called for you on object destruction, but you still should disconnect manually or use a with block.
  • Some fixes regarding disconnecting on client deletion and properly saving the authorization key.
  • Passing a class to message.get_entities_text now works properly.
  • Iterating messages from a specific user in private messages now works.
Enhancements

Core Rewrite in asyncio (v1.0-rc1)

Published at 2018/06/24

Scheme layer used: 81

This version is a major overhaul of the library internals. The core has been rewritten, cleaned up and refactored to fix some oddities that have been growing inside the library.

This means that the code is easier to understand and reason about, including the code flow such as conditions, exceptions, where to reconnect, how the library should behave, and separating different retry types such as disconnections or call fails, but it also means that some things will necessarily break in this version.

All requests that touch the network are now methods and need to have their await (or be ran until their completion).

Also, the library finally has the simple logo it deserved: a carefully hand-written .svg file representing a T following Python’s colours.

Breaking Changes
  • If you relied on internals like the MtProtoSender and the TelegramBareClient, both are gone. They are now MTProtoSender and TelegramBaseClient and they behave differently.
  • Underscores have been renamed from filenames. This means telethon.errors.rpc_error_list won’t work, but you should have been using telethon.errors all this time instead.
  • client.connect no longer returns True on success. Instead, you should except the possible ConnectionError and act accordingly. This makes it easier to not ignore the error.
  • You can no longer set retries=n when calling a request manually. The limit works differently now, and it’s done on a per-client basis.
  • Accessing .sender, .chat and similar may not work in events anymore, since previously they could access the network. The new rule is that properties are not allowed to make API calls. You should use .get_sender(), .get_chat() instead while using events. You can safely access properties if you get messages through client.get_messages() or other methods in the client.
  • The above point means reply_message is now .get_reply_message(), and fwd_from_entity is now get_fwd_sender(). Also forward was gone in the previous version, and you should be using fwd_from instead.
Additions
  • Telegram’s Terms Of Service are now accepted when creating a new account. This can possibly help avoid bans. This has no effect for accounts that were created before.
  • The method reference now shows which methods can be used if you sign in with a bot_token.
  • There’s a new client.disconnected future which you can wait on. When a disconnection occurs, you will now, instead letting it happen in the background.
  • More configurable retries parameters, such as auto-reconnection, retries when connecting, and retries when sending a request.
  • You can filter events.NewMessage by sender ID, and also whether they are forwards or not.
  • New ignore_migrated parameter for client.iter_dialogs.
Bug fixes
  • Several fixes to telethon.events.newmessage.NewMessage.
  • Removed named length argument in to_bytes for PyPy.
  • Raw events failed due to not having ._set_client.
  • message.get_entities_text properly supports filtering, even if there are no message entities.
  • message.click works better.
  • The server started sending DraftMessageEmpty which the library didn’t handle correctly when getting dialogs.
  • The “correct” chat is now always returned from returned messages.
  • to_id was not validated when retrieving messages by their IDs.
  • '__' is no longer considered valid in usernames.
  • The fd is removed from the reader upon closing the socket. This should be noticeable in Windows.
  • MessageEmpty is now handled when searching messages.
  • Fixed a rare infinite loop bug in client.iter_dialogs for some people.
  • Fixed TypeError when there is no .sender.
Enhancements
  • You can now delete over 100 messages at once with client.delete_messages.
  • Signing in now accounts for AuthRestartError itself, and also handles PasswordHashInvalidError.
  • __all__ is now defined, so from telethon import * imports sane defaults (client, events and utils). This is however discouraged and should be used only in quick scripts.
  • pathlib.Path is now supported for downloading and uploading media.
  • Messages you send to yourself are now considered outgoing, unless they are forwarded.
  • The documentation has been updated with a brand new asyncio crash course to encourage you use it. You can still use the threaded version if you want though.
  • .name property is now properly supported when sending and downloading files.
  • Custom parse_mode, which can now be set per-client, support MessageEntityMentionName so you can return those now.
  • The session file is saved less often, which could result in a noticeable speed-up when working with a lot of incoming updates.
Internal changes
  • The flow for sending a request is as follows: the TelegramClient creates a MTProtoSender with a Connection, and the sender starts send and receive loops. Sending a request means enqueueing it in the sender, which will eventually pack and encrypt it with its ConnectionState instead of using the entire Session instance. When the data is packed, it will be sent over the Connection and ultimately over the TcpClient.
  • Reconnection occurs at the MTProtoSender level, and receiving responses follows a similar process, but now asyncio.Future is used for the results which are no longer part of all TLObject, instead are part of the TLMessage which simplifies things.
  • Objects can no longer be content_related and instead subclass TLRequest, making the separation of concerns easier.
  • The TelegramClient has been split into several mixin classes to avoid having a 3,000-lines-long file with all the methods.
  • More special cases in the MTProtoSender have been cleaned up, and also some attributes from the Session which didn’t really belong there since they weren’t being saved.
  • The telethon_generator/ can now convert .tl files into .json, mostly as a proof of concept, but it might be useful for other people.

Custom Message class (v0.19.1)

Published at 2018/06/03

Scheme layer used: 80

This update brings a new telethon.tl.custom.message.Message object!

All the methods in the telethon.telegram_client.TelegramClient that used to return a Message will now return this object instead, which means you can do things like the following:

msg = client.send_message(chat, 'Hello!')
msg.edit('Hello there!')
msg.reply('Good day!')
print(msg.sender)

Refer to its documentation to see all you can do, again, click telethon.tl.custom.message.Message to go to its page.

Breaking Changes
Additions

  • You can get messages by their ID with telethon.telegram_client.TelegramClient.get_messages’s ids parameter:

    message = client.get_messages(chats, ids=123)  # Single message
message_list = client.get_messages(chats, ids=[777, 778])  # Multiple

  • More convenience properties for telethon.tl.custom.dialog.Dialog.

  • New default telethon.telegram_client.TelegramClient.parse_mode.

  • You can edit the media of messages that already have some media.

  • New dark theme in the online tl reference, check it out at https://tl.telethon.dev/.

Bug fixes
  • Some IDs start with 1000 and these would be wrongly treated as channels.
  • Some short usernames like @vote were being ignored.
  • telethon.telegram_client.TelegramClient.iter_messages’s from_user was failing if no filter had been set.
  • telethon.telegram_client.TelegramClient.iter_messages’s min_id/max_id was being ignored by Telegram. This is now worked around.
  • telethon.telegram_client.TelegramClient.catch_up would fail with empty states.
  • telethon.events.newmessage.NewMessage supports incoming=False to indicate outgoing=True.
Enhancements

  • You can now send multiple requests at once while preserving the order:

    from telethon.tl.functions.messages import SendMessageRequest
client([SendMessageRequest(chat, 'Hello 1!'),
        SendMessageRequest(chat, 'Hello 2!')], ordered=True)
Internal changes
  • without rowid is not used in SQLite anymore.
  • Unboxed serialization would fail.
  • Different default limit for iter_messages and get_messages.
  • Some clean-up in the telethon_generator/ package.

Catching up on Updates (v0.19)

Published at 2018/05/07

Scheme layer used: 76

This update prepares the library for catching up with updates with the new telethon.telegram_client.TelegramClient.catch_up method. This feature needs more testing, but for now it will let you “catch up” on some old updates that occurred while the library was offline, and brings some new features and bug fixes.

Additions
  • Add search, filter and from_user parameters to telethon.telegram_client.TelegramClient.iter_messages.
  • telethon.telegram_client.TelegramClient.download_file now supports a None path to return the file in memory and return its bytes.
  • Events now have a .original_update field.
Bug fixes
  • Fixed a race condition when receiving items from the network.
  • A disconnection is made when “retries reached 0”. This hasn’t been tested but it might fix the bug.
  • reply_to would not override Message object’s reply value.
  • Add missing caption when sending Message with media.
Enhancements
  • Retry automatically on RpcCallFailError. This error happened a lot when iterating over many messages, and retrying often fixes it.
  • Faster telethon.telegram_client.TelegramClient.iter_messages by sleeping only as much as needed.
  • telethon.telegram_client.TelegramClient.edit_message now supports omitting the entity if you pass a Message.
  • telethon.events.raw.Raw can now be filtered by type.
Internal changes
  • The library now distinguishes between MTProto and API schemas.
  • State is now persisted to the session file.
  • Connection won’t retry forever.
  • Fixed some errors and cleaned up the generation of code.
  • Fixed typos and enhanced some documentation in general.
  • Add auto-cast for InputMessage and InputLocation.

Pickle-able objects (v0.18.3)

Published at 2018/04/15

Now you can use Python’s pickle module to serialize RPCError and any other TLObject thanks to @vegeta1k95! A fix that was fairly simple, but still might be useful for many people.

As a side note, the documentation at https://tl.telethon.dev now lists known RPCError for all requests, so you know what to expect. This required a major rewrite, but it was well worth it!

Breaking changes
  • telethon.telegram_client.TelegramClient.forward_messages now returns a single item instead of a list if the input was also a single item.
Additions
Bug fixes
  • Possibly fix some bug regarding lost GzipPacked requests.
  • The library now uses the “real” layer 75, hopefully.
  • Fixed .entities name collision on updates by making it private.
  • AUTH_KEY_DUPLICATED is handled automatically on connection.
  • Markdown parser’s offset uses match.start() to allow custom regex.
  • Some filter types (as a type) were not supported by telethon.telegram_client.TelegramClient.iter_participants.
  • telethon.telegram_client.TelegramClient.remove_event_handler works.
  • telethon.telegram_client.TelegramClient.start works on all terminals.
  • InputPeerSelf case was missing from telethon.telegram_client.TelegramClient.get_input_entity.
Enhancements
  • The parse_mode for messages now accepts a callable.
  • telethon.telegram_client.TelegramClient.download_media accepts web previews.
  • telethon.tl.custom.dialog.Dialog instances can now be casted into InputPeer.
  • Better logging when reading packages “breaks”.
  • Better and more powerful setup.py gen command.
Internal changes
  • The library won’t call .get_dialogs() on entity not found. Instead, it will raise ValueError() so you can properly except it.
  • Several new examples and updated documentation.
  • py:obj is the default Sphinx’s role which simplifies .rst files.
  • setup.py now makes use of python_requires.
  • Events now live in separate files.
  • Other minor changes.

Several bug fixes (v0.18.2)

Published at 2018/03/27

Just a few bug fixes before they become too many.

Additions
  • Getting an entity by its positive ID should be enough, regardless of their type (whether it’s an User, a Chat or a Channel). Although wrapping them inside a Peer is still recommended, it’s not necessary.
  • New client.edit_2fa function to change your Two Factor Authentication settings.
  • .stringify() and string representation for custom Dialog/Draft.
Bug fixes
  • Some bug regarding .get_input_peer.
  • events.ChatAction wasn’t picking up all the pins.
  • force_document=True was being ignored for albums.
  • Now you’re able to send Photo and Document as files.
  • Wrong access to a member on chat forbidden error for .get_participants. An empty list is returned instead.
  • me/self check for .get[_input]_entity has been moved up so if someone has “me” or “self” as their name they won’t be retrieved.

Iterator methods (v0.18.1)

Published at 2018/03/17

All the .get_ methods in the TelegramClient now have a .iter_ counterpart, so you can do operations while retrieving items from them. For instance, you can client.iter_dialogs() and break once you find what you’re looking for instead fetching them all at once.

Another big thing, you can get entities by just their positive ID. This may cause some collisions (although it’s very unlikely), and you can (should) still be explicit about the type you want. However, it’s a lot more convenient and less confusing.

Breaking changes
  • The library only offers the default SQLiteSession again. See Session Files for more on how to use a different storage from now on.
Additions
  • Events now override __str__ and implement .stringify(), just like every other TLObject does.
  • events.ChatAction now has respond(), reply() and delete() for the message that triggered it.
  • client.iter_participants() (and its client.get_participants() counterpart) now expose the filter argument, and the returned users also expose the .participant they are.
  • You can now use client.remove_event_handler() and client.list_event_handlers() similar how you could with normal updates.
  • New properties on events.NewMessage, like .video_note and .gif to access only specific types of documents.
  • The Draft class now exposes .text and .raw_text, as well as a new Draft.send() to send it.
Bug fixes
  • MessageEdited was ignoring NewMessage constructor arguments.
  • Fixes for Event.delete_messages which wouldn’t handle MessageService.
  • Bot API style IDs not working on client.get_input_entity().
  • client.download_media() didn’t support PhotoSize.
Enhancements
  • Less RPC are made when accessing the .sender and .chat of some events (mostly those that occur in a channel).
  • You can send albums larger than 10 items (they will be sliced for you), as well as mixing normal files with photos.
  • TLObject now have Python type hints.
Internal changes
  • Several documentation corrections.
  • client.get_dialogs() is only called once again when an entity is not found to avoid flood waits.

Sessions overhaul (v0.18)

Published at 2018/03/04

Scheme layer used: 75

The Session’s have been revisited thanks to the work of @tulir and they now use an ABC so you can easily implement your own!

The default will still be a SQLiteSession, but you might want to use the new AlchemySessionContainer if you need. Refer to the section of the documentation on Session Files for more.

Breaking changes
  • events.MessageChanged doesn’t exist anymore. Use the new events.MessageEdited and events.MessageDeleted instead.
Additions
  • The mentioned addition of new session types.
  • You can omit the event type on client.add_event_handler to use Raw.
  • You can raise StopPropagation of events if you added several of them.
  • .get_participants() can now get up to 90,000 members from groups with 100,000 if when aggressive=True, “bypassing” Telegram’s limit.
  • You now can access NewMessage.Event.pattern_match.
  • Multiple captions are now supported when sending albums.
  • client.send_message() has an optional file= parameter, so you can do events.reply(file='/path/to/photo.jpg') and similar.
  • Added .input_ versions to events.ChatAction.
  • You can now access the public .client property on events.
  • New client.forward_messages, with its own wrapper on events, called event.forward_to(...).
Bug fixes
  • Silly bug regarding client.get_me(input_peer=True).
  • client.send_voice_note() was missing some parameters.
  • client.send_file() plays better with streams now.
  • Incoming messages from bots weren’t working with whitelists.
  • Markdown’s URL regex was not accepting newlines.
  • Better attempt at joining background update threads.
  • Use the right peer type when a marked integer ID is provided.
Internal changes
  • Resolving events.Raw is now a no-op.
  • Logging calls in the TcpClient to spot errors.
  • events resolution is postponed until you are successfully connected, so you can attach them before starting the client.
  • When an entity is not found, it is searched in all dialogs. This might not always be desirable but it’s more comfortable for legitimate uses.
  • Some non-persisting properties from the Session have been moved out.

Further easing library usage (v0.17.4)

Published at 2018/02/24

Some new things and patches that already deserved their own release.

Additions
  • New pattern argument to NewMessage to easily filter messages.
  • New .get_participants() convenience method to get members from chats.
  • .send_message() now accepts a Message as the message parameter.
  • You can now .get_entity() through exact name match instead username.
  • Raise ProxyConnectionError instead looping forever so you can except it on your own code and behave accordingly.
Bug fixes
  • .parse_username would fail with www. or a trailing slash.
  • events.MessageChanged would fail with UpdateDeleteMessages.
  • You can now send b'byte strings' directly as files again.
  • .send_file() was not respecting the original captions when passing another message (or media) as the file.
  • Downloading media from a different data center would always log a warning for the first time.
Internal changes
  • Use req_pq_multi instead req_pq when generating auth_key.
  • You can use .get_me(input_peer=True) if all you need is your self ID.
  • New addition to the interactive client example to show peer information.
  • Avoid special casing InputPeerSelf on some NewMessage events, so you can always safely rely on .sender to get the right ID.

New small convenience functions (v0.17.3)

Published at 2018/02/18

More bug fixes and a few others addition to make events easier to use.

Additions
  • Use hachoir to extract video and audio metadata before upload.
  • New .add_event_handler, .add_update_handler now deprecated.
Bug fixes
  • bot_token wouldn’t work on .start(), and changes to password (now it will ask you for it if you don’t provide it, as docstring hinted).
  • .edit_message() was ignoring the formatting (e.g. markdown).
  • Added missing case to the NewMessage event for normal groups.
  • Accessing the .text of the NewMessage event was failing due to a bug with the markdown unparser.
Internal changes

New small convenience functions (v0.17.2)

Published at 2018/02/15

Primarily bug fixing and a few welcomed additions.

Additions
  • New convenience .edit_message() method on the TelegramClient.
  • New .edit() and .delete() shorthands on the NewMessage event.
  • Default to markdown parsing when sending and editing messages.
  • Support for inline mentions when sending and editing messages. They work like inline urls (e.g. [text](@username)) and also support the Bot-API style (see here).
Bug fixes
  • Periodically send GetStateRequest automatically to keep the server sending updates even if you’re not invoking any request yourself.
  • HTML parsing was failing due to not handling surrogates properly.
  • .sign_up was not accepting int codes.
  • Whitelisting more than one chat on events wasn’t working.
  • Video files are sent as a video by default unless force_document.
Internal changes
  • More logging calls to help spot some bugs in the future.
  • Some more logic to retrieve input entities on events.
  • Clarified a few parts of the documentation.

Updates as Events (v0.17.1)

Published at 2018/02/09

Of course there was more work to be done regarding updates, and it’s here! The library comes with a new events module (which you will often import as from telethon import TelegramClient, events). This are pretty much all the additions that come with this version change, but they are a nice addition. Refer to (removed broken link) to get started with events.

Trust the Server with Updates (v0.17)

Published at 2018/02/03

The library trusts the server with updates again. The library will not check for duplicates anymore, and when the server kicks us, it will run GetStateRequest so the server starts sending updates again (something it wouldn’t do unless you invoked something, it seems). But this update also brings a few more changes!

Additions
  • TLObject’s override __eq__ and __ne__, so you can compare them.
  • Added some missing cases on .get_input_entity() and peer functions.
  • obj.to_dict() now has a '_' key with the type used.
  • .start() can also sign up now.
  • More parameters for .get_message_history().
  • Updated list of RPC errors.
  • HTML parsing thanks to @tulir! It can be used similar to markdown: client.send_message(..., parse_mode='html').
Enhancements
  • client.send_file() now accepts Message’s and MessageMedia’s as the file parameter.
  • Some documentation updates and fixed to clarify certain things.
  • New exact match feature on https://tl.telethon.dev.
  • Return as early as possible from .get_input_entity() and similar, to avoid penalizing you for doing this right.
Bug fixes
  • .download_media() wouldn’t accept a Document as parameter.
  • The SQLite is now closed properly on disconnection.
  • IPv6 addresses shouldn’t use square braces.
  • Fix regarding .log_out().
  • The time offset wasn’t being used (so having wrong system time would cause the library not to work at all).

New .resolve() method (v0.16.2)

Published at 2018/01/19

The TLObject’s (instances returned by the API and Request’s) have now acquired a new .resolve() method. While this should be used by the library alone (when invoking a request), it means that you can now use Peer types or even usernames where a InputPeer is required. The object now has access to the client, so that it can fetch the right type if needed, or access the session database. Furthermore, you can reuse requests that need “autocast” (e.g. you put User but InputPeer was needed), since .resolve() is called when invoking. Before, it was only done on object construction.

Additions
  • Album support. Just pass a list, tuple or any iterable to .send_file().
Enhancements
  • .start() asks for your phone only if required.
  • Better file cache. All files under 10MB, once uploaded, should never be needed to be re-uploaded again, as the sent media is cached to the session.
Bug fixes
  • setup.py now calls gen_tl when installing the library if needed.
Internal changes
  • The mentioned .resolve() to perform “autocast”, more powerful.
  • Upload and download methods are no longer part of TelegramBareClient.
  • Reuse .on_response(), .__str__ and .stringify(). Only override .on_response() if necessary (small amount of cases).
  • Reduced “autocast” overhead as much as possible. You shouldn’t be penalized if you’ve provided the right type.

MtProto 2.0 (v0.16.1)

Published at 2018/01/11

Scheme layer used: 74

The library is now using MtProto 2.0! This shouldn’t really affect you as an end user, but at least it means the library will be ready by the time MtProto 1.0 is deprecated.

Additions
  • New .start() method, to make the library avoid boilerplate code.
  • .send_file accepts a new optional thumbnail parameter, and returns the Message with the sent file.
Bug fixes
  • The library uses again only a single connection. Less updates are be dropped now, and the performance is even better than using temporary connections.
  • without rowid will only be used on the *.session if supported.
  • Phone code hash is associated with phone, so you can change your mind when calling .sign_in().
Internal changes
  • File cache now relies on the hash of the file uploaded instead its path, and is now persistent in the *.session file. Report any bugs on this!
  • Clearer error when invoking without being connected.
  • Markdown parser doesn’t work on bytes anymore (which makes it cleaner).

Sessions as sqlite databases (v0.16)

Published at 2017/12/28

In the beginning, session files used to be pickle. This proved to be bad as soon as one wanted to add more fields. For this reason, they were migrated to use JSON instead. But this proved to be bad as soon as one wanted to save things like entities (usernames, their ID and hash), so now it properly uses sqlite3, which has been well tested, to save the session files! Calling .get_input_entity using a username no longer will need to fetch it first, so it’s really 0 calls again. Calling .get_entity will always fetch the most up to date version.

Furthermore, nearly everything has been documented, thus preparing the library for Read the Docs (although there are a few things missing I’d like to polish first), and the logging are now better placed.

Breaking changes
  • .get_dialogs() now returns a single list instead a tuple consisting of a custom class that should make everything easier to work with.
  • .get_message_history() also returns a single list instead a tuple, with the Message instances modified to make them more convenient.

Both lists have a .total attribute so you can still know how many dialogs/messages are in total.

Additions
  • The mentioned use of sqlite3 for the session file.
  • .get_entity() now supports lists too, and it will make as little API calls as possible if you feed it InputPeer types. Usernames will always be resolved, since they may have changed.
  • .set_proxy() method, to avoid having to create a new TelegramClient.
  • More date types supported to represent a date parameter.
Bug fixes
  • Empty strings weren’t working when they were a flag parameter (e.g., setting no last name).
  • Fix invalid assertion regarding flag parameters as well.
  • Avoid joining the background thread on disconnect, as it would be None due to a race condition.
  • Correctly handle None dates when downloading media.
  • .download_profile_photo was failing for some channels.
  • .download_media wasn’t handling Photo.
Internal changes
  • date was being serialized as local date, but that was wrong.
  • date was being represented as a float instead of an int.
  • .tl parser wasn’t stripping inline comments.
  • Removed some redundant checks on update_state.py.
  • Use a synchronized queue instead a hand crafted version.
  • Use signed integers consistently (e.g. salt).
  • Always read the corresponding TLObject from API responses, except for some special cases still.
  • A few more except low level to correctly wrap errors.
  • More accurate exception types.
  • invokeWithLayer(initConnection(X)) now wraps every first request after .connect().

As always, report if you have issues with some of the changes!

IPv6 support (v0.15.5)

Published at 2017/11/16

Scheme layer used: 73

It’s here, it has come! The library now supports IPv6! Just pass use_ipv6=True when creating a TelegramClient. Note that I could not test this feature because my machine doesn’t have IPv6 setup. If you know IPv6 works in your machine but the library doesn’t, please refer to #425.

Additions
  • IPv6 support.
  • New method to extract the text surrounded by MessageEntity’s, in the extensions.markdown module.
Enhancements
  • Markdown parsing is Done Right.
  • Reconnection on failed invoke. Should avoid “number of retries reached 0” (#270).
  • Some missing autocast to Input* types.
  • The library uses the NullHandler for logging as it should have always done.
  • TcpClient.is_connected() is now more reliable.
Bug fixes
  • Getting an entity using their phone wasn’t actually working.
  • Full entities aren’t saved unless they have an access_hash, to avoid some None errors.
  • .get_message_history was failing when retrieving items that had messages forwarded from a channel.

General enhancements (v0.15.4)

Published at 2017/11/04

Scheme layer used: 72

This update brings a few general enhancements that are enough to deserve a new release, with a new feature: beta markdown-like parsing for .send_message()!

Additions
  • .send_message() supports parse_mode='md' for Markdown! It works in a similar fashion to the official clients (defaults to double underscore/asterisk, like **this**). Please report any issues with emojies or enhancements for the parser!
  • New .idle() method so your main thread can do useful job (listen for updates).
  • Add missing .to_dict(), __str__ and .stringify() for TLMessage and MessageContainer.
Bug fixes
  • The list of known peers could end “corrupted” and have users with access_hash=None, resulting in struct error for it not being an integer. You shouldn’t encounter this issue anymore.
  • The warning for “added update handler but no workers set” wasn’t actually working.
  • .get_input_peer was ignoring a case for InputPeerSelf.
  • There used to be an exception when logging exceptions (whoops) on update handlers.
  • “Downloading contacts” would produce strange output if they had semicolons (;) in their name.
  • Fix some cyclic imports and installing dependencies from the git repository.
  • Code generation was using f-strings, which are only supported on Python ≥3.6.
Internal changes
  • The auth_key generation has been moved from .connect() to .invoke(). There were some issues were .connect() failed and the auth_key was None so this will ensure to have a valid auth_key when needed, even if BrokenAuthKeyError is raised.
  • Support for higher limits on .get_history() and .get_dialogs().
  • Much faster integer factorization when generating the required auth_key. Thanks @delivrance for making me notice this, and for the pull request.

Bug fixes with updates (v0.15.3)

Published at 2017/10/20

Hopefully a very ungrateful bug has been removed. When you used to invoke some request through update handlers, it could potentially enter an infinite loop. This has been mitigated and it’s now safe to invoke things again! A lot of updates were being dropped (all those gzipped), and this has been fixed too.

More bug fixes include a correct parsing of certain TLObjects thanks to @stek29, and some wrong calls that would cause the library to crash thanks to @andr-04, and the ReadThread not re-starting if you were already authorized.

Internally, the .to_bytes() function has been replaced with __bytes__ so now you can do bytes(tlobject).

Bug fixes and new small features (v0.15.2)

Published at 2017/10/14

This release primarly focuses on a few bug fixes and enhancements. Although more stuff may have broken along the way.

Enhancements
  • You will be warned if you call .add_update_handler with no update_workers.
  • New customizable threshold value on the session to determine when to automatically sleep on flood waits. See client.session.flood_sleep_threshold.
  • New .get_drafts() method with a custom Draft class by @JosXa.
  • Join all threads when calling .disconnect(), to assert no dangling thread is left alive.
  • Larger chunk when downloading files should result in faster downloads.
  • You can use a callable key for the EntityDatabase, so it can be any filter you need.
Bug fixes
  • .get_input_entity was failing for IDs and other cases, also making more requests than it should.
  • Use basename instead abspath when sending a file. You can now also override the attributes.
  • EntityDatabase.__delitem__ wasn’t working.
  • .send_message() was failing with channels.
  • .get_dialogs(limit=None) should now return all the dialogs correctly.
  • Temporary fix for abusive duplicated updates.
Internal changes
  • MsgsAck is now sent in a container rather than its own request.
  • .get_input_photo is now used in the generated code.
  • .process_entities was being called from more places than only __call__.
  • MtProtoSender now relies more on the generated code to read responses.

Custom Entity Database (v0.15.1)

Published at 2017/10/05

The main feature of this release is that Telethon now has a custom database for all the entities you encounter, instead depending on @lru_cache on the .get_entity() method.

The EntityDatabase will, by default, cache all the users, chats and channels you find in memory for as long as the program is running. The session will, by default, save all key-value pairs of the entity identifiers and their hashes (since Telegram may send an ID that it thinks you already know about, we need to save this information).

You can prevent the EntityDatabase from saving users by setting client.session.entities.enabled = False, and prevent the Session from saving input entities at all by setting client.session.save_entities = False. You can also clear the cache for a certain user through client.session.entities.clear_cache(entity=None), which will clear all if no entity is given.

Additions
  • New method to .delete_messages().
  • New ChannelPrivateError class.
Enhancements
  • .sign_in accepts phones as integers.
  • Changing the IP to which you connect to is as simple as client.session.server_address = 'ip', since now the server address is always queried from the session.
Bug fixes
  • .get_dialogs() doesn’t fail on Windows anymore, and returns the right amount of dialogs.
  • GeneralProxyError should be passed to the main thread again, so that you can handle it.

Updates Overhaul Update (v0.15)

Published at 2017/10/01

After hundreds of lines changed on a major refactor, it’s finally here. It’s the Updates Overhaul Update; let’s get right into it!

Breaking changes
  • .create_new_connection() is gone for good. No need to deal with this manually since new connections are now handled on demand by the library itself.
Enhancements
  • You can invoke requests from update handlers. And any other thread. A new temporary will be made, so that you can be sending even several requests at the same time!
  • Several worker threads for your updates! By default, None will spawn. I recommend you to work with update_workers=4 to get started, these will be polling constantly for updates.
  • You can also change the number of workers at any given time.
  • The library can now run in a single thread again, if you don’t need to spawn any at all. Simply set spawn_read_thread=False when creating the TelegramClient!
  • You can specify limit=None on .get_dialogs() to get all of them[1].
  • Updates are expanded, so you don’t need to check if the update has .updates or an inner .update anymore.
  • All InputPeer entities are saved in the session file, but you can disable this by setting save_entities=False.
  • New .get_input_entity method, which makes use of the above feature. You should use this when a request needs a InputPeer, rather than the whole entity (although both work).
  • Assert that either all or None dependent-flag parameters are set before sending the request.
  • Phone numbers can have dashes, spaces, or parenthesis. They’ll be removed before making the request.
  • You can override the phone and its hash on .sign_in(), if you’re creating a new TelegramClient on two different places.
Bug fixes
  • .log_out() was consuming all retries. It should work just fine now.
  • The session would fail to load if the auth_key had been removed manually.
  • Updates.check_error was popping wrong side, although it’s been completely removed.
  • ServerError’s will be ignored, and the request will immediately be retried.
  • Cross-thread safety when saving the session file.
  • Some things changed on a matter of when to reconnect, so please report any bugs!
Internal changes
  • TelegramClient is now only an abstraction over the TelegramBareClient, which can only do basic things, such as invoking requests, working with files, etc. If you don’t need any of the abstractions the TelegramClient, you can now use the TelegramBareClient in a much more comfortable way.
  • MtProtoSender is not thread-safe, but it doesn’t need to be since a new connection will be spawned when needed.
  • New connections used to be cached and then reused. Now only their sessions are saved, as temporary connections are spawned only when needed.
  • Added more RPC errors to the list.

[1]: Broken due to a condition which should had been the opposite (sigh), fixed 4 commits ahead on https://github.com/LonamiWebs/Telethon/commit/62ea77cbeac7c42bfac85aa8766a1b5b35e3a76c.

That’s pretty much it, although there’s more work to be done to make the overall experience of working with updates even better. Stay tuned!

Serialization bug fixes (v0.14.2)

Published at 2017/09/29

Bug fixes
  • Important, related to the serialization. Every object or request that had to serialize a True/False type was always being serialized as false!
  • Another bug that didn’t allow you to leave as None flag parameters that needed a list has been fixed.
Internal changes
  • Other internal changes include a somewhat more readable .to_bytes() function and pre-computing the flag instead using bit shifting. The TLObject.constructor_id has been renamed to TLObject.CONSTRUCTOR_ID, and .subclass_of_id is also uppercase now.

Farewell, BinaryWriter (v0.14.1)

Published at 2017/09/28

Version v0.14 had started working on the new .to_bytes() method to dump the BinaryWriter and its usage on the .on_send() when serializing TLObjects, and this release finally removes it. The speed up when serializing things to bytes should now be over twice as fast wherever it’s needed.

Bug fixes
  • This version is again compatible with Python 3.x versions below 3.5 (there was a method call that was Python 3.5 and above).
Internal changes
  • Using proper classes (including the generated code) for generating authorization keys and to write out TLMessage’s.

Several requests at once and upload compression (v0.14)

Published at 2017/09/27

New major release, since I’ve decided that these two features are big enough:

Additions
  • Requests larger than 512 bytes will be compressed through gzip, and if the result is smaller, this will be uploaded instead.
  • You can now send multiple requests at once, they’re simply *var_args on the .invoke(). Note that the server doesn’t guarantee the order in which they’ll be executed!

Internally, another important change. The .on_send function on the TLObjects is gone, and now there’s a new .to_bytes(). From my tests, this has always been over twice as fast serializing objects, although more replacements need to be done, so please report any issues.

Enhancements
  • Implemented .get_input_media helper methods. Now you can even use another message as input media!
Bug fixes
  • Downloading media from CDNs wasn’t working (wrong access to a parameter).
  • Correct type hinting.
  • Added a tiny sleep when trying to perform automatic reconnection.
  • Error reporting is done in the background, and has a shorter timeout.
  • setup.py used to fail with wrongly generated code.

Quick fix-up (v0.13.6)

Published at 2017/09/23

Before getting any further, here’s a quick fix-up with things that should have been on v0.13.5 but were missed. Specifically, the timeout when receiving a request will now work properly.

Some other additions are a tiny fix when handling updates, which was ignoring some of them, nicer __str__ and .stringify() methods for the TLObject’s, and not stopping the ReadThread if you try invoking something there (now it simply returns None).

Attempts at more stability (v0.13.5)

Published at 2017/09/23

Yet another update to fix some bugs and increase the stability of the library, or, at least, that was the attempt!

This release should really improve the experience with the background thread that the library starts to read things from the network as soon as it can, but I can’t spot every use case, so please report any bug (and as always, minimal reproducible use cases will help a lot).

Bug fixes
  • setup.py was failing on Python < 3.5 due to some imports.
  • Duplicated updates should now be ignored.
  • .send_message would crash in some cases, due to having a typo using the wrong object.
  • "socket is None" when calling .connect() should not happen anymore.
  • BrokenPipeError was still being raised due to an incorrect order on the try/except block.
Enhancements
  • Type hinting for all the generated Request’s and TLObjects! IDEs like PyCharm will benefit from this.
  • ProxyConnectionError should properly be passed to the main thread for you to handle.
  • The background thread will only be started after you’re authorized on Telegram (i.e. logged in), and several other attempts at polishing the experience with this thread.
  • The Connection instance is only created once now, and reused later.
  • Calling .connect() should have a better behavior now (like actually trying to connect even if we seemingly were connected already).
  • .reconnect() behavior has been changed to also be more consistent by making the assumption that we’ll only reconnect if the server has disconnected us, and is now private.
Internal changes
  • TLObject.__repr__ doesn’t show the original TL definition anymore, it was a lot of clutter. If you have any complaints open an issue and we can discuss it.
  • Internally, the '+' from the phone number is now stripped, since it shouldn’t be included.
  • Spotted a new place where BrokenAuthKeyError would be raised, and it now is raised there.

More bug fixes and enhancements (v0.13.4)

Published at 2017/09/18

Additions
  • TelegramClient now exposes a .is_connected() method.
  • Initial authorization on a new data center will retry up to 5 times by default.
  • Errors that couldn’t be handled on the background thread will be raised on the next call to .invoke() or updates.poll().
Bug fixes
  • Now you should be able to sign in even if you have process_updates=True and no previous session.
  • Some errors and methods are documented a bit clearer.
  • .send_message() could randomly fail, as the returned type was not expected.
  • TimeoutError is now ignored, since the request will be retried up to 5 times by default.
  • “-404” errors (BrokenAuthKeyError’s) are now detected when first connecting to a new data center.
  • BufferError is handled more gracefully, in the same way as InvalidCheckSumError’s.
  • Attempt at fixing some “NoneType has no attribute…” errors (with the .sender).
Internal changes
  • Calling GetConfigRequest is now made less often.
  • The initial_query parameter from .connect() is gone, as it’s not needed anymore.
  • Renamed all_tlobjects.layer to all_tlobjects.LAYER (since it’s a constant).
  • The message from BufferError is now more useful.

Bug fixes and enhancements (v0.13.3)

Published at 2017/09/14

Bug fixes
  • Reconnection used to fail because it tried invoking things from the ReadThread.
  • Inferring random ids for ForwardMessagesRequest wasn’t working.
  • Downloading media from CDNs failed due to having forgotten to remove a single line.
  • TcpClient.close() now has a ``threading.Lock``, so NoneType has no close() should not happen.
  • New workaround for msg seqno too low/high. Also, both Session.id/seq are not saved anymore.
Enhancements
  • Request will be retried up to 5 times by default rather than failing on the first attempt.
  • InvalidChecksumError’s are now ignored by the library.
  • TelegramClient.get_entity() is now public, and uses the @lru_cache() decorator.
  • New method to ``.send_voice_note()``’s.
  • Methods to send message and media now support a ``reply_to`` parameter.
  • .send_message() now returns the full message which was just sent.

New way to work with updates (v0.13.2)

Published at 2017/09/08

This update brings a new way to work with updates, and it’s begging for your feedback, or better names or ways to do what you can do now.

Please refer to the wiki/Usage Modes for an in-depth description on how to work with updates now. Notice that you cannot invoke requests from within handlers anymore, only the v.0.13.1 patch allowed you to do so.

Bug fixes
  • Periodic pings are back.
  • The username regex mentioned on UsernameInvalidError was invalid, but it has now been fixed.
  • Sending a message to a phone number was failing because the type used for a request had changed on layer 71.
  • CDN downloads weren’t working properly, and now a few patches have been applied to ensure more reliability, although I couldn’t personally test this, so again, report any feedback.

Invoke other requests from within update callbacks (v0.13.1)

Published at 2017/09/04

Warning

This update brings some big changes to the update system, so please read it if you work with them!

A silly “bug” which hadn’t been spotted has now been fixed. Now you can invoke other requests from within your update callbacks. However this is not advised. You should post these updates to some other thread, and let that thread do the job instead. Invoking a request from within a callback will mean that, while this request is being invoked, no other things will be read.

Internally, the generated code now resides under a lot less files, simply for the sake of avoiding so many unnecessary files. The generated code is not meant to be read by anyone, simply to do its job.

Unused attributes have been removed from the TLObject class too, and .sign_up() returns the user that just logged in in a similar way to .sign_in() now.

Connection modes (v0.13)

Published at 2017/09/04

Scheme layer used: 71

The purpose of this release is to denote a big change, now you can connect to Telegram through different **connection modes**. Also, a second thread will always be started when you connect a TelegramClient, despite whether you’ll be handling updates or ignoring them, whose sole purpose is to constantly read from the network.

The reason for this change is as simple as “reading and writing shouldn’t be related”. Even when you’re simply ignoring updates, this way, once you send a request you will only need to read the result for the request. Whatever Telegram sent before has already been read and outside the buffer.

Additions
  • The mentioned different connection modes, and a new thread.
  • You can modify the Session attributes through the TelegramClient constructor (using **kwargs).
  • RPCError’s now belong to some request you’ve made, which makes more sense.
  • get_input_* now handles None (default) parameters more gracefully (it used to crash).
Enhancements
  • The low-level socket doesn’t use a handcrafted timeout anymore, which should benefit by avoiding the arbitrary sleep(0.1) that there used to be.
  • TelegramClient.sign_in will call .send_code_request if no code was provided.
Deprecation
  • .sign_up does not take a phone argument anymore. Change this or you will be using phone as code, and it will fail! The definition looks like def sign_up(self, code, first_name, last_name='').
  • The old JsonSession finally replaces the original Session (which used pickle). If you were overriding any of these, you should only worry about overriding Session now.

Added verification for CDN file (v0.12.2)

Published at 2017/08/28

Since the Content Distributed Network (CDN) is not handled by Telegram itself, the owners may tamper these files. Telegram sends their sha256 sum for clients to implement this additional verification step, which now the library has. If any CDN has altered the file you’re trying to download, CdnFileTamperedError will be raised to let you know.

Besides this. TLObject.stringify() was showing bytes as lists (now fixed) and RPC errors are reported by default:

In an attempt to help everyone who works with the Telegram API, Telethon will by default report all Remote Procedure Call errors to PWRTelegram, a public database anyone can query, made by Daniil. All the information sent is a GET request with the error code, error message and method used.

Note

If you still would like to opt out, simply set client.session.report_errors = False to disable this feature. However Daniil would really thank you if you helped him (and everyone) by keeping it on!

CDN support (v0.12.1)

Published at 2017/08/24

The biggest news for this update are that downloading media from CDN’s (you’ll often encounter this when working with popular channels) now works.

Bug fixes
  • The method used to download documents crashed because two lines were swapped.
  • Determining the right path when downloading any file was very weird, now it’s been enhanced.
  • The .sign_in() method didn’t support integer values for the code! Now it does again.

Some important internal changes are that the old way to deal with RSA public keys now uses a different module instead the old strange hand-crafted version.

Hope the new, super simple README.rst encourages people to use Telethon and make it better with either suggestions, or pull request. Pull requests are super appreciated, but showing some support by leaving a star also feels nice ⭐️.

Newbie friendly update (v0.12)

Published at 2017/08/22

Scheme layer used: 70

This update is overall an attempt to make Telethon a bit more user friendly, along with some other stability enhancements, although it brings quite a few changes.

Breaking changes
  • The TelegramClient methods .send_photo_file(), .send_document_file() and .send_media_file() are now a single method called .send_file(). It’s also important to note that the order of the parameters has been swapped: first to who you want to send it, then the file itself.
  • The same applies to .download_msg_media(), which has been renamed to .download_media(). The method now supports a Message itself too, rather than only Message.media. The specialized .download_photo(), .download_document() and .download_contact() still exist, but are private.
Additions
  • Updated to layer 70!
  • Both downloading and uploading now support stream-like objects.
  • A lot faster initial connection if sympy is installed (can be installed through pip).
  • libssl will also be used if available on your system (likely on Linux based systems). This speed boost should also apply to uploading and downloading files.
  • You can use a phone number or an username for methods like .send_message(), .send_file(), and all the other quick-access methods provided by the TelegramClient.
Bug fixes
  • Crashing when migrating to a new layer and receiving old updates should not happen now.
  • InputPeerChannel is now casted to InputChannel automtically too.
  • .get_new_msg_id() should now be thread-safe. No promises.
  • Logging out on macOS caused a crash, which should be gone now.
  • More checks to ensure that the connection is flagged correctly as either connected or not.

Note

Downloading files from CDN’s will not work yet (something new that comes with layer 70).

That’s it, any new idea or suggestion about how to make the project even more friendly is highly appreciated.

Note

Did you know that you can pretty print any result Telegram returns (called TLObject’s) by using their .stringify() function? Great for debugging!

get_input_* now works with vectors (v0.11.5)

Published at 2017/07/11

Quick fix-up of a bug which hadn’t been encountered until now. Auto-cast by using get_input_* now works.

get_input_* everywhere (v0.11.4)

Published at 2017/07/10

For some reason, Telegram doesn’t have enough with the InputPeer. There also exist InputChannel and InputUser! You don’t have to worry about those anymore, it’s handled internally now.

Besides this, every Telegram object now features a new default .__str__ look, and also a .stringify() method to pretty format them, if you ever need to inspect them.

The library now uses the DEBUG level everywhere, so no more warnings or information messages if you had logging enabled.

The no_webpage parameter from .send_message has been renamed to link_preview for clarity, so now it does the opposite (but has a clearer intention).

Quick .send_message() fix (v0.11.3)

Published at 2017/07/05

A very quick follow-up release to fix a tiny bug with .send_message(), no new features.

Callable TelegramClient (v0.11.2)

Published at 2017/07/04

Scheme layer used: 68

There is a new preferred way to invoke requests, which you’re encouraged to use:

# New!
result = client(SomeRequest())

# Old.
result = client.invoke(SomeRequest())

Existing code will continue working, since the old .invoke() has not been deprecated.

When you .create_new_connection(), it will also handle FileMigrateError’s for you, so you don’t need to worry about those anymore.

Bugs fixes
  • Fixed some errors when installing Telethon via pip (for those using either source distributions or a Python version ≤ 3.5).
  • ConnectionResetError didn’t flag sockets as closed, but now it does.

On a more technical side, msg_id’s are now more accurate.

Improvements to the updates (v0.11.1)

Published at 2017/06/24

Receiving new updates shouldn’t miss any anymore, also, periodic pings are back again so it should work on the long run.

On a different order of things, .connect() also features a timeout. Notice that the timeout= is not passed as a parameter anymore, and is instead specified when creating the TelegramClient.

Bug fixes
  • Fixed some name class when a request had a .msg_id parameter.
  • The correct amount of random bytes is now used in DH request
  • Fixed CONNECTION_APP_VERSION_EMPTY when using temporary sessions.
  • Avoid connecting if already connected.

Support for parallel connections (v0.11)

Published at 2017/06/16

This update brings a lot of changes, so it would be nice if you could read the whole change log!

Breaking changes
  • Every Telegram error has now its own class, so it’s easier to fine-tune your except’s.
  • Markdown parsing is not part of Telethon itself anymore, although there are plans to support it again through a some external module.
  • The .list_sessions() has been moved to the Session class instead.
  • The InteractiveTelegramClient is not shipped with pip anymore.
Additions
  • A new, more lightweight class has been added. The TelegramBareClient is now the base of the normal TelegramClient, and has the most basic features.
  • New method to .create_new_connection(), which can be ran in parallel with the original connection. This will return the previously mentioned TelegramBareClient already connected.
  • Any file object can now be used to download a file (for instance, a BytesIO() instead a file name).
  • Vales like random_id are now automatically inferred, so you can save yourself from the hassle of writing generate_random_long() everywhere. Same applies to .get_input_peer(), unless you really need the extra performance provided by skipping one if if called manually.
  • Every type now features a new .to_dict() method.
Bug fixes
  • Received errors are acknowledged to the server, so they don’t happen over and over.
  • Downloading media on different data centers is now up to x2 faster, since there used to be an InvalidDCError for each file part tried to be downloaded.
  • Lost messages are now properly skipped.
  • New way to handle the result of requests. The old ValueError “The previously sent request must be resent. However, no request was previously sent (possibly called from a different thread).” should not happen anymore.
Internal changes
  • Some fixes to the JsonSession.
  • Fixed possibly crashes if trying to .invoke() a Request while .reconnect() was being called on the UpdatesThread.
  • Some improvements on the TcpClient, such as not switching between blocking and non-blocking sockets.
  • The code now uses ASCII characters only.
  • Some enhancements to .find_user_or_chat() and .get_input_peer().

JSON session file (v0.10.1)

Published at 2017/06/07

This version is primarily for people to migrate their .session files, which are pickled, to the new JSON format. Although slightly slower, and a bit more vulnerable since it’s plain text, it’s a lot more resistant to upgrades.

Warning

You must upgrade to this version before any higher one if you’ve used Telethon ≤ v0.10. If you happen to upgrade to an higher version, that’s okay, but you will have to manually delete the *.session file, and logout from that session from an official client.

Additions
  • New .get_me() function to get the current user.
  • .is_user_authorized() is now more reliable.
  • New nice button to copy the from telethon.tl.xxx.yyy import Yyy on the online documentation.
  • More error codes added to the errors file.
Enhancements
  • Everything on the documentation is now, theoretically, sorted alphabetically.
  • No second thread is spawned unless one or more update handlers are added.

Full support for different DCs and ++stable (v0.10)

Published at 2017/06/03

Working with different data centers finally works! On a different order of things, reconnection is now performed automatically every time Telegram decides to kick us off their servers, so now Telethon can really run forever and ever! In theory.

Enhancements
  • Documentation improvements, such as showing the return type.
  • The msg_id too low/high error should happen less often, if any.
  • Sleeping on the main thread is not done anymore. You will have to except FloodWaitError’s.
  • You can now specify your own application version, device model, system version and language code.
  • Code is now more pythonic (such as making some members private), and other internal improvements (which affect the updates thread), such as using logger instead a bare print() too.

This brings Telethon a whole step closer to v1.0, though more things should preferably be changed.

Stability improvements (v0.9.1)

Published at 2017/05/23

Telethon used to crash a lot when logging in for the very first time. The reason for this was that the reconnection (or dead connections) were not handled properly. Now they are, so you should be able to login directly, without needing to delete the *.session file anymore. Notice that downloading from a different DC is still a WIP.

Enhancements
  • Updates thread is only started after a successful login.
  • Files meant to be ran by the user now use shebangs and proper permissions.
  • In-code documentation now shows the returning type.
  • Relative import is now used everywhere, so you can rename telethon to anything else.
  • Dead connections are now detected instead entering an infinite loop.
  • Sockets can now be closed (and re-opened) properly.
  • Telegram decided to update the layer 66 without increasing the number. This has been fixed and now we’re up-to-date again.

General improvements (v0.9)

Published at 2017/05/19

Scheme layer used: 66
Additions
  • The documentation, available online here, has a new search bar.
  • Better cross-thread safety by using threading.Event.
  • More improvements for running Telethon during a long period of time.
Bug fixes
  • Avoid a certain crash on login (occurred if an unexpected object ID was received).
  • Avoid crashing with certain invalid UTF-8 strings.
  • Avoid crashing on certain terminals by using known ASCII characters where possible.
  • The UpdatesThread is now a daemon, and should cause less issues.
  • Temporary sessions didn’t actually work (with session=None).
Internal changes
  • .get_dialogs(count= was renamed to .get_dialogs(limit=.

Bot login and proxy support (v0.8)

Published at 2017/04/14

Additions
  • Bot login, thanks to @JuanPotato for hinting me about how to do it.
  • Proxy support, thanks to @exzhawk for implementing it.
  • Logging support, used by passing --telethon-log=DEBUG (or INFO) as a command line argument.
Bug fixes
  • Connection fixes, such as avoiding connection until .connect() is explicitly invoked.
  • Uploading big files now works correctly.
  • Fix uploading big files.
  • Some fixes on the updates thread, such as correctly sleeping when required.

Long-run bug fix (v0.7.1)

Published at 2017/02/19

If you’re one of those who runs Telethon for a long time (more than 30 minutes), this update by @strayge will be great for you. It sends periodic pings to the Telegram servers so you don’t get disconnected and you can still send and receive updates!

Two factor authentication (v0.7)

Published at 2017/01/31

Scheme layer used: 62

If you’re one of those who love security the most, these are good news. You can now use two factor authentication with Telethon too! As internal changes, the coding style has been improved, and you can easily use custom session objects, and various little bugs have been fixed.

Updated pip version (v0.6)

Published at 2016/11/13

Scheme layer used: 57

This release has no new major features. However, it contains some small changes that make using Telethon a little bit easier. Now those who have installed Telethon via pip can also take advantage of changes, such as less bugs, creating empty instances of TLObjects, specifying a timeout and more!

Ready, pip, go! (v0.5)

Published at 2016/09/18

Telethon is now available as a `Python package <https://pypi.python.org/pypi?name=Telethon>`__! Those are really exciting news (except, sadly, the project structure had to change a lot to be able to do that; but hopefully it won’t need to change much more, any more!)

Not only that, but more improvements have also been made: you’re now able to both sign up and logout, watch a pretty “Uploading/Downloading… x%” progress, and other minor changes which make using Telethon easier.

Made InteractiveTelegramClient cool (v0.4)

Published at 2016/09/12

Yes, really cool! I promise. Even though this is meant to be a library, that doesn’t mean it can’t have a good interactive client for you to try the library out. This is why now you can do many, many things with the InteractiveTelegramClient:

  • List dialogs (chats) and pick any you wish.
  • Send any message you like, text, photos or even documents.
  • List the latest messages in the chat.
  • Download any message’s media (photos, documents or even contacts!).
  • Receive message updates as you talk (i.e., someone sent you a message).

It actually is a usable-enough client for your day by day. You could even add libnotify and pop, you’re done! A great cli-client with desktop notifications.

Also, being able to download and upload media implies that you can do the same with the library itself. Did I need to mention that? Oh, and now, with even less bugs! I hope.

Media revolution and improvements to update handling! (v0.3)

Published at 2016/09/11

Telegram is more than an application to send and receive messages. You can also send and receive media. Now, this implementation also gives you the power to upload and download media from any message that contains it! Nothing can now stop you from filling up all your disk space with all the photos! If you want to, of course.

Handle updates in their own thread! (v0.2)

Published at 2016/09/10

This version handles updates in a different thread (if you wish to do so). This means that both the low level TcpClient and the not-so-low-level MtProtoSender are now multi-thread safe, so you can use them with more than a single thread without worrying!

This also implies that you won’t need to send a request to receive an update (is someone typing? did they send me a message? has someone gone offline?). They will all be received instantly.

Some other cool examples of things that you can do: when someone tells you “Hello”, you can automatically reply with another “Hello” without even needing to type it by yourself :)

However, be careful with spamming!! Do not use the program for that!

First working alpha version! (v0.1)

Published at 2016/09/06

Scheme layer used: 55

There probably are some bugs left, which haven’t yet been found. However, the majority of code works and the application is already usable! Not only that, but also uses the latest scheme as of now and handles way better the errors. This tag is being used to mark this release as stable enough.

Wall of Shame

This project has an issues section for you to file issues whenever you encounter any when working with the library. Said section is not for issues on your program but rather issues with Telethon itself.

If you have not made the effort to 1. read through the docs and 2. look for the method you need, you will end up on the Wall of Shame, i.e. all issues labeled “RTFM”:

rtfm Literally “Read The F–king Manual”; a term showing the frustration of being bothered with questions so trivial that the asker could have quickly figured out the answer on their own with minimal effort, usually by reading readily-available documents. People who say”RTFM!” might be considered rude, but the true rude ones are the annoying people who take absolutely no self-responibility and expect to have all the answers handed to them personally.

“Damn, that’s the twelveth time that somebody posted this question to the messageboard today! RTFM, already!”

by Bill M. July 27, 2004

If you have indeed read the docs, and have tried looking for the method, and yet you didn’t find what you need, that’s fine. Telegram’s API can have some obscure names at times, and for this reason, there is a “question” label with questions that are okay to ask. Just state what you’ve tried so that we know you’ve made an effort, or you’ll go to the Wall of Shame.

Of course, if the issue you’re going to open is not even a question but a real issue with the library (thankfully, most of the issues have been that!), you won’t end up here. Don’t worry.

Current winner

The current winner is issue 213:

Issue:

https://user-images.githubusercontent.com/6297805/29822978-9a9a6ef0-8ccd-11e7-9ec5-934ea0f57681.jpg
alt:Winner issue

Winner issue

Answer:

https://user-images.githubusercontent.com/6297805/29822983-9d523402-8ccd-11e7-9fb1-5783740ee366.jpg
alt:Winner issue answer

Winner issue answer

Compatibility and Convenience

Telethon is an asyncio library. Compatibility is an important concern, and while it can’t always be kept and mistakes happens, the Changelog (Version History) is there to tell you when these important changes happen.

Compatibility

Some decisions when developing will inevitable be proven wrong in the future. One of these decisions was using threads. Now that Python 3.4 is reaching EOL and using asyncio is usable as of Python 3.5 it makes sense for a library like Telethon to make a good use of it.

If you have old code, just use old versions of the library! There is nothing wrong with that other than not getting new updates or fixes, but using a fixed version with pip install telethon==0.19.1.6 is easy enough to do.

You might want to consider using Virtual Environments in your projects.

There’s no point in maintaining a synchronous version because the whole point is that people don’t have time to upgrade, and there has been several changes and clean-ups. Using an older version is the right way to go.

Sometimes, other small decisions are made. These all will be reflected in the Changelog (Version History) which you should read when upgrading.

If you want to jump the asyncio boat, here are some of the things you will need to start migrating really old code:

# 1. Import the client from telethon.sync
from telethon.sync import TelegramClient

# 2. Change this monster...
try:
    assert client.connect()
    if not client.is_user_authorized():
        client.send_code_request(phone_number)
        me = client.sign_in(phone_number, input('Enter code: '))

    ...  # REST OF YOUR CODE
finally:
    client.disconnect()

# ...for this:
with client:
    ...  # REST OF YOUR CODE

# 3. client.idle() no longer exists.
# Change this...
client.idle()
# ...to this:
client.run_until_disconnected()

# 4. client.add_update_handler no longer exists.
# Change this...
client.add_update_handler(handler)
# ...to this:
client.add_event_handler(handler)

In addition, all the update handlers must be async def, and you need to await method calls that rely on network requests, such as getting the chat or sender. If you don’t use updates, you’re done!

Convenience

Note

The entire documentation assumes you have done one of the following:

from telethon import TelegramClient, sync
# or
from telethon.sync import TelegramClient

This makes the examples shorter and easier to think about.

For quick scripts that don’t need updates, it’s a lot more convenient to forget about asyncio and just work with sequential code. This can prove to be a powerful hybrid for running under the Python REPL too.

from telethon.sync import TelegramClient
#            ^~~~~ note this part; it will manage the asyncio loop for you

with TelegramClient(...) as client:
    print(client.get_me().username)
    #     ^ notice the lack of await, or loop.run_until_complete().
    #       Since there is no loop running, this is done behind the scenes.
    #
    message = client.send_message('me', 'Hi!')
    import time
    time.sleep(5)
    message.delete()

    # You can also have an hybrid between a synchronous
    # part and asynchronous event handlers.
    #
    from telethon import events
    @client.on(events.NewMessage(pattern='(?i)hi|hello'))
    async def handler(event):
        await event.reply('hey')

    client.run_until_disconnected()

Some methods, such as with, start, disconnect and run_until_disconnected work both in synchronous and asynchronous contexts by default for convenience, and to avoid the little overhead it has when using methods like sending a message, getting messages, etc. This keeps the best of both worlds as a sane default.

Note

As a rule of thumb, if you’re inside an async def and you need the client, you need to await calls to the API. If you call other functions that also need API calls, make them async def and await them too. Otherwise, there is no need to do so with this mode.

Speed

When you’re ready to micro-optimize your application, or if you simply don’t need to call any non-basic methods from a synchronous context, just get rid of telethon.sync and work inside an async def:

import asyncio
from telethon import TelegramClient, events

async def main():
    async with TelegramClient(...) as client:
        print((await client.get_me()).username)
        #     ^_____________________^ notice these parenthesis
        #     You want to ``await`` the call, not the username.
        #
        message = await client.send_message('me', 'Hi!')
        await asyncio.sleep(5)
        await message.delete()

        @client.on(events.NewMessage(pattern='(?i)hi|hello'))
        async def handler(event):
            await event.reply('hey')

        await client.run_until_disconnected()

asyncio.run(main())

The telethon.sync magic module essentially wraps every method behind:

asyncio.run(main())

With some other tricks, so that you don’t have to write it yourself every time. That’s the overhead you pay if you import it, and what you save if you don’t.

Learning

You know the library uses asyncio everywhere, and you want to learn how to do things right. Even though asyncio is its own topic, the documentation wants you to learn how to use Telethon correctly, and for that, you need to use asyncio correctly too. For this reason, there is a section called Mastering asyncio that will introduce you to the asyncio world, with links to more resources for learning how to use it. Feel free to check that section out once you have read the rest.

TelegramClient

The TelegramClient aggregates several mixin classes to provide all the common functionality in a nice, Pythonic interface. Each mixin has its own methods, which you all can use.

In short, to create a client you must run:

from telethon import TelegramClient

client = TelegramClient(name, api_id, api_hash)

async def main():
    # Now you can use all client methods listed below, like for example...
    await client.send_message('me', 'Hello to myself!')

with client:
    client.loop.run_until_complete(main())

You don’t need to import these AuthMethods, MessageMethods, etc. Together they are the TelegramClient and you can access all of their methods.

See Client Reference for a short summary.

class telethon.client.telegramclient.TelegramClient(session: typing.Union[str, Session], api_id: int, api_hash: str, *, connection: typing.Type[Connection] = <class 'telethon.network.connection.tcpfull.ConnectionTcpFull'>, use_ipv6: bool = False, proxy: Union[tuple, dict] = None, local_addr: Union[str, tuple] = None, timeout: int = 10, request_retries: int = 5, connection_retries: int = 5, retry_delay: int = 1, auto_reconnect: bool = True, sequential_updates: bool = False, flood_sleep_threshold: int = 60, raise_last_call_error: bool = False, device_model: str = None, system_version: str = None, app_version: str = None, lang_code: str = 'en', system_lang_code: str = 'en', loop: asyncio.events.AbstractEventLoop = None, base_logger: Union[str, logging.Logger] = None, receive_updates: bool = True, catch_up: bool = False, entity_cache_limit: int = 5000)

Bases: telethon.client.account.AccountMethods, telethon.client.auth.AuthMethods, telethon.client.downloads.DownloadMethods, telethon.client.dialogs.DialogMethods, telethon.client.chats.ChatMethods, telethon.client.bots.BotMethods, telethon.client.messages.MessageMethods, telethon.client.uploads.UploadMethods, telethon.client.buttons.ButtonMethods, telethon.client.updates.UpdateMethods, telethon.client.messageparse.MessageParseMethods, telethon.client.users.UserMethods, telethon.client.telegrambaseclient.TelegramBaseClient

class telethon.client.telegrambaseclient.TelegramBaseClient(session: typing.Union[str, Session], api_id: int, api_hash: str, *, connection: typing.Type[Connection] = <class 'telethon.network.connection.tcpfull.ConnectionTcpFull'>, use_ipv6: bool = False, proxy: Union[tuple, dict] = None, local_addr: Union[str, tuple] = None, timeout: int = 10, request_retries: int = 5, connection_retries: int = 5, retry_delay: int = 1, auto_reconnect: bool = True, sequential_updates: bool = False, flood_sleep_threshold: int = 60, raise_last_call_error: bool = False, device_model: str = None, system_version: str = None, app_version: str = None, lang_code: str = 'en', system_lang_code: str = 'en', loop: asyncio.events.AbstractEventLoop = None, base_logger: Union[str, logging.Logger] = None, receive_updates: bool = True, catch_up: bool = False, entity_cache_limit: int = 5000)

Bases: abc.ABC

This is the abstract base class for the client. It defines some basic stuff like connecting, switching data center, etc, and leaves the __call__ unimplemented.

Arguments
session (str | telethon.sessions.abstract.Session, None):

The file name of the session file to be used if a string is given (it may be a full path), or the Session instance to be used otherwise. If it’s None, the session will not be saved, and you should call log_out() when you’re done.

Note that if you pass a string it will be a file in the current working directory, although you can also pass absolute paths.

The session file contains enough information for you to login without re-sending the code, so if you have to enter the code more than once, maybe you’re changing the working directory, renaming or removing the file, or using random names.

api_id (int | str):
The API ID you obtained from https://my.telegram.org.
api_hash (str):
The API hash you obtained from https://my.telegram.org.
connection (telethon.network.connection.common.Connection, optional):

The connection instance to be used when creating a new connection to the servers. It must be a type.

Defaults to telethon.network.connection.tcpfull.ConnectionTcpFull.

use_ipv6 (bool, optional):
Whether to connect to the servers through IPv6 or not. By default this is False as IPv6 support is not too widespread yet.
proxy (tuple | list | dict, optional):
An iterable consisting of the proxy info. If connection is one of MTProxy, then it should contain MTProxy credentials: ('hostname', port, 'secret'). Otherwise, it’s meant to store function parameters for PySocks, like (type, 'hostname', port). See https://github.com/Anorov/PySocks#usage-1 for more.
local_addr (str | tuple, optional):
Local host address (and port, optionally) used to bind the socket to locally. You only need to use this if you have multiple network cards and want to use a specific one.
timeout (int | float, optional):
The timeout in seconds to be used when connecting. This is not the timeout to be used when await’ing for invoked requests, and you should use asyncio.wait or asyncio.wait_for for that.
request_retries (int | None, optional):

How many times a request should be retried. Request are retried when Telegram is having internal issues (due to either errors.ServerError or errors.RpcCallFailError), when there is a errors.FloodWaitError less than flood_sleep_threshold, or when there’s a migrate error.

May take a negative or None value for infinite retries, but this is not recommended, since some requests can always trigger a call fail (such as searching for messages).

connection_retries (int | None, optional):
How many times the reconnection should retry, either on the initial connection or when Telegram disconnects us. May be set to a negative or None value for infinite retries, but this is not recommended, since the program can get stuck in an infinite loop.
retry_delay (int | float, optional):
The delay in seconds to sleep between automatic reconnections.
auto_reconnect (bool, optional):
Whether reconnection should be retried connection_retries times automatically if Telegram disconnects us or not.
sequential_updates (bool, optional):

By default every incoming update will create a new task, so you can handle several updates in parallel. Some scripts need the order in which updates are processed to be sequential, and this setting allows them to do so.

If set to True, incoming updates will be put in a queue and processed sequentially. This means your event handlers should not perform long-running operations since new updates are put inside of an unbounded queue.

flood_sleep_threshold (int | float, optional):
The threshold below which the library should automatically sleep on flood wait and slow mode wait errors (inclusive). For instance, if a FloodWaitError for 17s occurs and flood_sleep_threshold is 20s, the library will sleep automatically. If the error was for 21s, it would raise FloodWaitError instead. Values larger than a day (like float('inf')) will be changed to a day.
raise_last_call_error (bool, optional):
When API calls fail in a way that causes Telethon to retry automatically, should the RPC error of the last attempt be raised instead of a generic ValueError. This is mostly useful for detecting when Telegram has internal issues.
device_model (str, optional):
“Device model” to be sent when creating the initial connection. Defaults to ‘PC (n)bit’ derived from platform.uname().machine, or its direct value if unknown.
system_version (str, optional):
“System version” to be sent when creating the initial connection. Defaults to platform.uname().release stripped of everything ahead of -.
app_version (str, optional):
“App version” to be sent when creating the initial connection. Defaults to telethon.version.__version__.
lang_code (str, optional):
“Language code” to be sent when creating the initial connection. Defaults to 'en'.
system_lang_code (str, optional):
“System lang code” to be sent when creating the initial connection. Defaults to lang_code.
loop (asyncio.AbstractEventLoop, optional):
Asyncio event loop to use. Defaults to asyncio.get_running_loop(). This argument is ignored.
base_logger (str | logging.Logger, optional):
Base logger name or instance to use. If a str is given, it’ll be passed to logging.getLogger(). If a logging.Logger is given, it’ll be used directly. If something else or nothing is given, the default logger will be used.
receive_updates (bool, optional):

Whether the client will receive updates or not. By default, updates will be received from Telegram as they occur.

Turning this off means that Telegram will not send updates at all so event handlers, conversations, and QR login will not work. However, certain scripts don’t need updates, so this will reduce the amount of bandwidth used.

entity_cache_limit (int, optional):

How many users, chats and channels to keep in the in-memory cache at most. This limit is checked against when processing updates.

When this limit is reached or exceeded, all entities that are not required for update handling will be flushed to the session file.

Note that this implies that there is a lower bound to the amount of entities that must be kept in memory.

Setting this limit too low will cause the library to attempt to flush entities to the session file even if no entities can be removed from the in-memory cache, which will degrade performance.

__call__(request, ordered=False)

Invokes (sends) one or more MTProtoRequests and returns (receives) their result.

Args:
request (TLObject | list):
The request or requests to be invoked.
ordered (bool, optional):
Whether the requests (if more than one was given) should be executed sequentially on the server. They run in arbitrary order by default.
flood_sleep_threshold (int | None, optional):
The flood sleep threshold to use for this request. This overrides the default value stored in client.flood_sleep_threshold
Returns:
The result of the request (often a TLObject) or a list of results if more than one request was given.
__version__ = '1.30.3'
__weakref__

list of weak references to the object (if defined)

connect() → None

Connects to Telegram.

Note

Connect means connect and nothing else, and only one low-level request is made to notify Telegram about which layer we will be using.

Before Telegram sends you updates, you need to make a high-level request, like client.get_me(), as described in https://core.telegram.org/api/updates.

Example
try:
    await client.connect()
except OSError:
    print('Failed to connect')
disconnect()

Disconnects from Telegram.

If the event loop is already running, this method returns a coroutine that you should await on your own code; otherwise the loop is ran until said coroutine completes.

Event handlers which are currently running will be cancelled before this function returns (in order to properly clean-up their tasks). In particular, this means that using disconnect in a handler will cause code after the disconnect to never run. If this is needed, consider spawning a separate task to do the remaining work.

Example
# You don't need to use this if you used "with client"
await client.disconnect()
disconnected

Property with a Future that resolves upon disconnection.

Example
# Wait for a disconnection to occur
try:
    await client.disconnected
except OSError:
    print('Error on disconnect')
flood_sleep_threshold
is_connected() → bool

Returns True if the user has connected.

This method is not asynchronous (don’t use await on it).

Example
while client.is_connected():
    await asyncio.sleep(1)
loop

Property with the asyncio event loop used by this client.

Example
# Download media in the background
task = client.loop.create_task(message.download_media())

# Do some work
...

# Join the task (wait for it to complete)
await task
set_proxy(proxy: Union[tuple, dict])

Changes the proxy which will be used on next (re)connection.

Method has no immediate effects if the client is currently connected.

The new proxy will take it’s effect on the next reconnection attempt:
  • on a call await client.connect() (after complete disconnect)
  • on auto-reconnect attempt (e.g, after previous connection was lost)
class telethon.client.account.AccountMethods

Bases: object

__weakref__

list of weak references to the object (if defined)

end_takeout(success: bool) → bool

Finishes the current takeout session.

Arguments
success (bool):
Whether the takeout completed successfully or not.
Returns
True if the operation was successful, False otherwise.
Example
await client.end_takeout(success=False)
takeout(finalize: bool = True, *, contacts: bool = None, users: bool = None, chats: bool = None, megagroups: bool = None, channels: bool = None, files: bool = None, max_file_size: bool = None) → TelegramClient

Returns a TelegramClient which calls methods behind a takeout session.

It does so by creating a proxy object over the current client through which making requests will use InvokeWithTakeoutRequest to wrap them. In other words, returns the current client modified so that requests are done as a takeout:

Some of the calls made through the takeout session will have lower flood limits. This is useful if you want to export the data from conversations or mass-download media, since the rate limits will be lower. Only some requests will be affected, and you will need to adjust the wait_time of methods like client.iter_messages.

By default, all parameters are None, and you need to enable those you plan to use by setting them to either True or False.

You should except errors.TakeoutInitDelayError as e, since this exception will raise depending on the condition of the session. You can then access e.seconds to know how long you should wait for before calling the method again.

There’s also a success property available in the takeout proxy object, so from the with body you can set the boolean result that will be sent back to Telegram. But if it’s left None as by default, then the action is based on the finalize parameter. If it’s True then the takeout will be finished, and if no exception occurred during it, then True will be considered as a result. Otherwise, the takeout will not be finished and its ID will be preserved for future usage as client.session.takeout_id.

Arguments
finalize (bool):
Whether the takeout session should be finalized upon exit or not.
contacts (bool):
Set to True if you plan on downloading contacts.
users (bool):
Set to True if you plan on downloading information from users and their private conversations with you.
chats (bool):
Set to True if you plan on downloading information from small group chats, such as messages and media.
megagroups (bool):
Set to True if you plan on downloading information from megagroups (channels), such as messages and media.
channels (bool):
Set to True if you plan on downloading information from broadcast channels, such as messages and media.
files (bool):
Set to True if you plan on downloading media and you don’t only wish to export messages.
max_file_size (int):
The maximum file size, in bytes, that you plan to download for each message with media.
Example
from telethon import errors

try:
    async with client.takeout() as takeout:
        await client.get_messages('me')  # normal call
        await takeout.get_messages('me')  # wrapped through takeout (less limits)

        async for message in takeout.iter_messages(chat, wait_time=0):
            ...  # Do something with the message

except errors.TakeoutInitDelayError as e:
    print('Must wait', e.seconds, 'before takeout')
class telethon.client.auth.AuthMethods

Bases: object

__aenter__()
__aexit__(*args)
__enter__()

Helps to cut boilerplate on async context managers that offer synchronous variants.

__exit__(*args)
__weakref__

list of weak references to the object (if defined)

edit_2fa(current_password: str = None, new_password: str = None, *, hint: str = '', email: str = None, email_code_callback: Callable[[int], str] = None) → bool

Changes the 2FA settings of the logged in user.

Review carefully the parameter explanations before using this method.

Note that this method may be incredibly slow depending on the prime numbers that must be used during the process to make sure that everything is safe.

Has no effect if both current and new password are omitted.

Arguments
current_password (str, optional):
The current password, to authorize changing to new_password. Must be set if changing existing 2FA settings. Must not be set if 2FA is currently disabled. Passing this by itself will remove 2FA (if correct).
new_password (str, optional):
The password to set as 2FA. If 2FA was already enabled, current_password must be set. Leaving this blank or None will remove the password.
hint (str, optional):
Hint to be displayed by Telegram when it asks for 2FA. Leaving unspecified is highly discouraged. Has no effect if new_password is not set.
email (str, optional):
Recovery and verification email. If present, you must also set email_code_callback, else it raises ValueError.
email_code_callback (callable, optional):

If an email is provided, a callback that returns the code sent to it must also be set. This callback may be asynchronous. It should return a string with the code. The length of the code will be passed to the callback as an input parameter.

If the callback returns an invalid code, it will raise CodeInvalidError.

Returns
True if successful, False otherwise.
Example
# Setting a password for your account which didn't have
await client.edit_2fa(new_password='I_<3_Telethon')

# Removing the password
await client.edit_2fa(current_password='I_<3_Telethon')
log_out() → bool

Logs out Telegram and deletes the current *.session file.

The client is unusable after logging out and a new instance should be created.

Returns
True if the operation was successful.
Example
# Note: you will need to login again!
await client.log_out()
qr_login(ignored_ids: List[int] = None) → telethon.tl.custom.qrlogin.QRLogin

Initiates the QR login procedure.

Note that you must be connected before invoking this, as with any other request.

It is up to the caller to decide how to present the code to the user, whether it’s the URL, using the token bytes directly, or generating a QR code and displaying it by other means.

See the documentation for QRLogin to see how to proceed after this.

Arguments
ignored_ids (List[int]):
List of already logged-in user IDs, to prevent logging in twice with the same user.
Returns
An instance of QRLogin.
Example
def display_url_as_qr(url):
    pass  # do whatever to show url as a qr to the user

qr_login = await client.qr_login()
display_url_as_qr(qr_login.url)

# Important! You need to wait for the login to complete!
await qr_login.wait()

# If you have 2FA enabled, `wait` will raise `telethon.errors.SessionPasswordNeededError`.
# You should except that error and call `sign_in` with the password if this happens.
send_code_request(phone: str, *, force_sms: bool = False, _retry_count: int = 0) → types.auth.SentCode

Sends the Telegram code needed to login to the given phone number.

Arguments
phone (str | int):
The phone to which the code will be sent.
force_sms (bool, optional):
Whether to force sending as SMS. This has been deprecated. See issue #4050 for context.
Returns
An instance of SentCode.
Example
phone = '+34 123 123 123'
sent = await client.send_code_request(phone)
print(sent)
sign_in(phone: str = None, code: Union[str, int] = None, *, password: str = None, bot_token: str = None, phone_code_hash: str = None) → typing.Union[types.User, types.auth.SentCode]

Logs in to Telegram to an existing user or bot account.

You should only use this if you are not authorized yet.

This method will send the code if it’s not provided.

Note

In most cases, you should simply use start() and not this method.

Arguments
phone (str | int):
The phone to send the code to if no code was provided, or to override the phone that was previously used with these requests.
code (str | int):
The code that Telegram sent. Note that if you have sent this code through the application itself it will immediately expire. If you want to send the code, obfuscate it somehow. If you’re not doing any of this you can ignore this note.
password (str):
2FA password, should be used if a previous call raised SessionPasswordNeededError.
bot_token (str):
Used to sign in as a bot. Not all requests will be available. This should be the hash the @BotFather gave you.
phone_code_hash (str, optional):
The hash returned by send_code_request. This can be left as None to use the last hash known for the phone to be used.
Returns
The signed in user, or the information about send_code_request().
Example
phone = '+34 123 123 123'
await client.sign_in(phone)  # send code

code = input('enter code: ')
await client.sign_in(phone, code)
sign_up(code: Union[str, int], first_name: str, last_name: str = '', *, phone: str = None, phone_code_hash: str = None) → types.User

This method can no longer be used, and will immediately raise a ValueError. See issue #4050 for context.

start(phone: Callable[[], str] = <function AuthMethods.<lambda>>, password: Callable[[], str] = <function AuthMethods.<lambda>>, *, bot_token: str = None, force_sms: bool = False, code_callback: Callable[[], Union[str, int]] = None, first_name: str = 'New User', last_name: str = '', max_attempts: int = 3) → TelegramClient

Starts the client (connects and logs in if necessary).

By default, this method will be interactive (asking for user input if needed), and will handle 2FA if enabled too.

If the event loop is already running, this method returns a coroutine that you should await on your own code; otherwise the loop is ran until said coroutine completes.

Arguments
phone (str | int | callable):
The phone (or callable without arguments to get it) to which the code will be sent. If a bot-token-like string is given, it will be used as such instead. The argument may be a coroutine.
password (str, callable, optional):
The password for 2 Factor Authentication (2FA). This is only required if it is enabled in your account. The argument may be a coroutine.
bot_token (str):
Bot Token obtained by @BotFather to log in as a bot. Cannot be specified with phone (only one of either allowed).
force_sms (bool, optional):
Whether to force sending the code request as SMS. This only makes sense when signing in with a phone.
code_callback (callable, optional):
A callable that will be used to retrieve the Telegram login code. Defaults to input(). The argument may be a coroutine.
first_name (str, optional):
The first name to be used if signing up. This has no effect if the account already exists and you sign in.
last_name (str, optional):
Similar to the first name, but for the last. Optional.
max_attempts (int, optional):
How many times the code/password callback should be retried or switching between signing in and signing up.
Returns
This TelegramClient, so initialization can be chained with .start().
Example
client = TelegramClient('anon', api_id, api_hash)

# Starting as a bot account
await client.start(bot_token=bot_token)

# Starting as a user account
await client.start(phone)
# Please enter the code you received: 12345
# Please enter your password: *******
# (You are now logged in)

# Starting using a context manager (this calls start()):
with client:
    pass
class telethon.client.bots.BotMethods

Bases: object

__weakref__

list of weak references to the object (if defined)

inline_query(bot: hints.EntityLike, query: str, *, entity: hints.EntityLike = None, offset: str = None, geo_point: types.GeoPoint = None) → telethon.tl.custom.inlineresults.InlineResults

Makes an inline query to the specified bot (@vote New Poll).

Arguments
bot (entity):
The bot entity to which the inline query should be made.
query (str):
The query that should be made to the bot.
entity (entity, optional):

The entity where the inline query is being made from. Certain bots use this to display different results depending on where it’s used, such as private chats, groups or channels.

If specified, it will also be the default entity where the message will be sent after clicked. Otherwise, the “empty peer” will be used, which some bots may not handle correctly.

offset (str, optional):
The string offset to use for the bot.
geo_point (GeoPoint, optional)
The geo point location information to send to the bot for localised results. Available under some bots.
Returns
A list of custom.InlineResult.
Example
# Make an inline query to @like
results = await client.inline_query('like', 'Do you like Telethon?')

# Send the first result to some chat
message = await results[0].click('TelethonOffTopic')
class telethon.client.buttons.ButtonMethods

Bases: object

__weakref__

list of weak references to the object (if defined)

static build_reply_markup(buttons: Union[telethon.tl.types.ReplyKeyboardHide, telethon.tl.types.ReplyKeyboardForceReply, telethon.tl.types.ReplyKeyboardMarkup, telethon.tl.types.ReplyInlineMarkup, telethon.tl.types.KeyboardButton, telethon.tl.types.KeyboardButtonUrl, telethon.tl.types.KeyboardButtonCallback, telethon.tl.types.KeyboardButtonRequestPhone, telethon.tl.types.KeyboardButtonRequestGeoLocation, telethon.tl.types.KeyboardButtonSwitchInline, telethon.tl.types.KeyboardButtonGame, telethon.tl.types.KeyboardButtonBuy, telethon.tl.types.KeyboardButtonUrlAuth, telethon.tl.types.InputKeyboardButtonUrlAuth, telethon.tl.types.KeyboardButtonRequestPoll, telethon.tl.types.InputKeyboardButtonUserProfile, telethon.tl.types.KeyboardButtonUserProfile, telethon.tl.types.KeyboardButtonWebView, telethon.tl.types.KeyboardButtonSimpleWebView, telethon.tl.types.KeyboardButtonRequestPeer, telethon.tl.custom.button.Button, Sequence[Union[telethon.tl.types.KeyboardButton, telethon.tl.types.KeyboardButtonUrl, telethon.tl.types.KeyboardButtonCallback, telethon.tl.types.KeyboardButtonRequestPhone, telethon.tl.types.KeyboardButtonRequestGeoLocation, telethon.tl.types.KeyboardButtonSwitchInline, telethon.tl.types.KeyboardButtonGame, telethon.tl.types.KeyboardButtonBuy, telethon.tl.types.KeyboardButtonUrlAuth, telethon.tl.types.InputKeyboardButtonUrlAuth, telethon.tl.types.KeyboardButtonRequestPoll, telethon.tl.types.InputKeyboardButtonUserProfile, telethon.tl.types.KeyboardButtonUserProfile, telethon.tl.types.KeyboardButtonWebView, telethon.tl.types.KeyboardButtonSimpleWebView, telethon.tl.types.KeyboardButtonRequestPeer, telethon.tl.custom.button.Button]], Sequence[Sequence[Union[telethon.tl.types.KeyboardButton, telethon.tl.types.KeyboardButtonUrl, telethon.tl.types.KeyboardButtonCallback, telethon.tl.types.KeyboardButtonRequestPhone, telethon.tl.types.KeyboardButtonRequestGeoLocation, telethon.tl.types.KeyboardButtonSwitchInline, telethon.tl.types.KeyboardButtonGame, telethon.tl.types.KeyboardButtonBuy, telethon.tl.types.KeyboardButtonUrlAuth, telethon.tl.types.InputKeyboardButtonUrlAuth, telethon.tl.types.KeyboardButtonRequestPoll, telethon.tl.types.InputKeyboardButtonUserProfile, telethon.tl.types.KeyboardButtonUserProfile, telethon.tl.types.KeyboardButtonWebView, telethon.tl.types.KeyboardButtonSimpleWebView, telethon.tl.types.KeyboardButtonRequestPeer, telethon.tl.custom.button.Button]]], None], inline_only: bool = False) → Union[telethon.tl.types.ReplyKeyboardHide, telethon.tl.types.ReplyKeyboardForceReply, telethon.tl.types.ReplyKeyboardMarkup, telethon.tl.types.ReplyInlineMarkup, None]

Builds a ReplyInlineMarkup or ReplyKeyboardMarkup for the given buttons.

Does nothing if either no buttons are provided or the provided argument is already a reply markup.

You should consider using this method if you are going to reuse the markup very often. Otherwise, it is not necessary.

This method is not asynchronous (don’t use await on it).

Arguments
buttons (hints.MarkupLike):
The button, list of buttons, array of buttons or markup to convert into a markup.
inline_only (bool, optional):
Whether the buttons must be inline buttons only or not.
Example
from telethon import Button

markup = client.build_reply_markup(Button.inline('hi'))
# later
await client.send_message(chat, 'click me', buttons=markup)
class telethon.client.chats.ChatMethods

Bases: object

__weakref__

list of weak references to the object (if defined)

action(entity: hints.EntityLike, action: typing.Union[str, types.TypeSendMessageAction], *, delay: float = 4, auto_cancel: bool = True) → typing.Union[_ChatAction, typing.Coroutine]

Returns a context-manager object to represent a “chat action”.

Chat actions indicate things like “user is typing”, “user is uploading a photo”, etc.

If the action is 'cancel', you should just await the result, since it makes no sense to use a context-manager for it.

See the example below for intended usage.

Arguments
entity (entity):
The entity where the action should be showed in.
action (str | SendMessageAction):

The action to show. You can either pass a instance of SendMessageAction or better, a string used while:

  • 'typing': typing a text message.
  • 'contact': choosing a contact.
  • 'game': playing a game.
  • 'location': choosing a geo location.
  • 'sticker': choosing a sticker.
  • 'record-audio': recording a voice note. You may use 'record-voice' as alias.
  • 'record-round': recording a round video.
  • 'record-video': recording a normal video.
  • 'audio': sending an audio file (voice note or song). You may use 'voice' and 'song' as aliases.
  • 'round': uploading a round video.
  • 'video': uploading a video file.
  • 'photo': uploading a photo.
  • 'document': uploading a document file. You may use 'file' as alias.
  • 'cancel': cancel any pending action in this chat.

Invalid strings will raise a ValueError.

delay (int | float):
The delay, in seconds, to wait between sending actions. For example, if the delay is 5 and it takes 7 seconds to do something, three requests will be made at 0s, 5s, and 7s to cancel the action.
auto_cancel (bool):
Whether the action should be cancelled once the context manager exists or not. The default is True, since you don’t want progress to be shown when it has already completed.
Returns
Either a context-manager object or a coroutine.
Example
# Type for 2 seconds, then send a message
async with client.action(chat, 'typing'):
    await asyncio.sleep(2)
    await client.send_message(chat, 'Hello world! I type slow ^^')

# Cancel any previous action
await client.action(chat, 'cancel')

# Upload a document, showing its progress (most clients ignore this)
async with client.action(chat, 'document') as action:
    await client.send_file(chat, zip_file, progress_callback=action.progress)
edit_admin(entity: hints.EntityLike, user: hints.EntityLike, *, change_info: bool = None, post_messages: bool = None, edit_messages: bool = None, delete_messages: bool = None, ban_users: bool = None, invite_users: bool = None, pin_messages: bool = None, add_admins: bool = None, manage_call: bool = None, anonymous: bool = None, is_admin: bool = None, title: str = None) → telethon.tl.types.Updates

Edits admin permissions for someone in a chat.

Raises an error if a wrong combination of rights are given (e.g. you don’t have enough permissions to grant one).

Unless otherwise stated, permissions will work in channels and megagroups.

Arguments
entity (entity):
The channel, megagroup or chat where the promotion should happen.
user (entity):
The user to be promoted.
change_info (bool, optional):
Whether the user will be able to change info.
post_messages (bool, optional):
Whether the user will be able to post in the channel. This will only work in broadcast channels.
edit_messages (bool, optional):
Whether the user will be able to edit messages in the channel. This will only work in broadcast channels.
delete_messages (bool, optional):
Whether the user will be able to delete messages.
ban_users (bool, optional):
Whether the user will be able to ban users.
invite_users (bool, optional):
Whether the user will be able to invite users. Needs some testing.
pin_messages (bool, optional):
Whether the user will be able to pin messages.
add_admins (bool, optional):
Whether the user will be able to add admins.
manage_call (bool, optional):
Whether the user will be able to manage group calls.
anonymous (bool, optional):

Whether the user will remain anonymous when sending messages. The sender of the anonymous messages becomes the group itself.

Note

Users may be able to identify the anonymous admin by its custom title, so additional care is needed when using both anonymous and custom titles. For example, if multiple anonymous admins share the same title, users won’t be able to distinguish them.

is_admin (bool, optional):

Whether the user will be an admin in the chat. This will only work in small group chats. Whether the user will be an admin in the chat. This is the only permission available in small group chats, and when used in megagroups, all non-explicitly set permissions will have this value.

Essentially, only passing is_admin=True will grant all permissions, but you can still disable those you need.

title (str, optional):

The custom title (also known as “rank”) to show for this admin. This text will be shown instead of the “admin” badge. This will only work in channels and megagroups.

When left unspecified or empty, the default localized “admin” badge will be shown.

Returns
The resulting Updates object.
Example
# Allowing `user` to pin messages in `chat`
await client.edit_admin(chat, user, pin_messages=True)

# Granting all permissions except for `add_admins`
await client.edit_admin(chat, user, is_admin=True, add_admins=False)
edit_permissions(entity: hints.EntityLike, user: typing.Optional[hints.EntityLike] = None, until_date: hints.DateLike = None, *, view_messages: bool = True, send_messages: bool = True, send_media: bool = True, send_stickers: bool = True, send_gifs: bool = True, send_games: bool = True, send_inline: bool = True, embed_link_previews: bool = True, send_polls: bool = True, change_info: bool = True, invite_users: bool = True, pin_messages: bool = True) → telethon.tl.types.Updates

Edits user restrictions in a chat.

Set an argument to False to apply a restriction (i.e. remove the permission), or omit them to use the default True (i.e. don’t apply a restriction).

Raises an error if a wrong combination of rights are given (e.g. you don’t have enough permissions to revoke one).

By default, each boolean argument is True, meaning that it is true that the user has access to the default permission and may be able to make use of it.

If you set an argument to False, then a restriction is applied regardless of the default permissions.

It is important to note that True does not mean grant, only “don’t restrict”, and this is where the default permissions come in. A user may have not been revoked the pin_messages permission (it is True) but they won’t be able to use it if the default permissions don’t allow it either.

Arguments
entity (entity):
The channel or megagroup where the restriction should happen.
user (entity, optional):
If specified, the permission will be changed for the specific user. If left as None, the default chat permissions will be updated.
until_date (DateLike, optional):

When the user will be unbanned.

If the due date or duration is longer than 366 days or shorter than 30 seconds, the ban will be forever. Defaults to 0 (ban forever).

view_messages (bool, optional):
Whether the user is able to view messages or not. Forbidding someone from viewing messages equals to banning them. This will only work if user is set.
send_messages (bool, optional):
Whether the user is able to send messages or not.
send_media (bool, optional):
Whether the user is able to send media or not.
send_stickers (bool, optional):
Whether the user is able to send stickers or not.
send_gifs (bool, optional):
Whether the user is able to send animated gifs or not.
send_games (bool, optional):
Whether the user is able to send games or not.
send_inline (bool, optional):
Whether the user is able to use inline bots or not.
embed_link_previews (bool, optional):
Whether the user is able to enable the link preview in the messages they send. Note that the user will still be able to send messages with links if this permission is removed, but these links won’t display a link preview.
send_polls (bool, optional):
Whether the user is able to send polls or not.
change_info (bool, optional):
Whether the user is able to change info or not.
invite_users (bool, optional):
Whether the user is able to invite other users or not.
pin_messages (bool, optional):
Whether the user is able to pin messages or not.
Returns
The resulting Updates object.
Example
from datetime import timedelta

# Banning `user` from `chat` for 1 minute
await client.edit_permissions(chat, user, timedelta(minutes=1),
                              view_messages=False)

# Banning `user` from `chat` forever
await client.edit_permissions(chat, user, view_messages=False)

# Kicking someone (ban + un-ban)
await client.edit_permissions(chat, user, view_messages=False)
await client.edit_permissions(chat, user)
get_admin_log(entity: hints.EntityLike, limit: float = None, *, max_id: int = 0, min_id: int = 0, search: str = None, admins: hints.EntitiesLike = None, join: bool = None, leave: bool = None, invite: bool = None, restrict: bool = None, unrestrict: bool = None, ban: bool = None, unban: bool = None, promote: bool = None, demote: bool = None, info: bool = None, settings: bool = None, pinned: bool = None, edit: bool = None, delete: bool = None, group_call: bool = None) → telethon.client.chats._AdminLogIter

Same as iter_admin_log(), but returns a list instead.

Example
# Get a list of deleted message events which said "heck"
events = await client.get_admin_log(channel, search='heck', delete=True)

# Print the old message before it was deleted
print(events[0].old)
get_participants(entity: hints.EntityLike, limit: float = None, *, search: str = '', filter: types.TypeChannelParticipantsFilter = None, aggressive: bool = False) → telethon.client.chats._ParticipantsIter

Same as iter_participants(), but returns a TotalList instead.

Example
users = await client.get_participants(chat)
print(users[0].first_name)

for user in users:
    if user.username is not None:
        print(user.username)
get_permissions(entity: hints.EntityLike, user: hints.EntityLike = None) → typing.Optional[custom.ParticipantPermissions]

Fetches the permissions of a user in a specific chat or channel or get Default Restricted Rights of Chat or Channel.

Note

This request has to fetch the entire chat for small group chats, which can get somewhat expensive, so use of a cache is advised.

Arguments
entity (entity):
The channel or chat the user is participant of.
user (entity, optional):
Target user.
Returns
A ParticipantPermissions instance. Refer to its documentation to see what properties are available.
Example
permissions = await client.get_permissions(chat, user)
if permissions.is_admin:
    # do something

# Get Banned Permissions of Chat
await client.get_permissions(chat)
get_profile_photos(entity: hints.EntityLike, limit: int = None, *, offset: int = 0, max_id: int = 0) → telethon.client.chats._ProfilePhotoIter

Same as iter_profile_photos(), but returns a TotalList instead.

Example
# Get the photos of a channel
photos = await client.get_profile_photos(channel)

# Download the oldest photo
await client.download_media(photos[-1])
get_stats(entity: hints.EntityLike, message: typing.Union[int, types.Message] = None)

Retrieves statistics from the given megagroup or broadcast channel.

Note that some restrictions apply before being able to fetch statistics, in particular the channel must have enough members (for megagroups, this requires at least 500 members).

Arguments
entity (entity):
The channel from which to get statistics.
message (int | Message, optional):
The message ID from which to get statistics, if your goal is to obtain the statistics of a single message.
Raises

If the given entity is not a channel (broadcast or megagroup), a TypeError is raised.

If there are not enough members (poorly named) errors such as telethon.errors.ChatAdminRequiredError will appear.

Returns
If both entity and message were provided, returns MessageStats. Otherwise, either BroadcastStats or MegagroupStats, depending on whether the input belonged to a broadcast channel or megagroup.
Example
# Some megagroup or channel username or ID to fetch
channel = -100123
stats = await client.get_stats(channel)
print('Stats from', stats.period.min_date, 'to', stats.period.max_date, ':')
print(stats.stringify())
iter_admin_log(entity: hints.EntityLike, limit: float = None, *, max_id: int = 0, min_id: int = 0, search: str = None, admins: hints.EntitiesLike = None, join: bool = None, leave: bool = None, invite: bool = None, restrict: bool = None, unrestrict: bool = None, ban: bool = None, unban: bool = None, promote: bool = None, demote: bool = None, info: bool = None, settings: bool = None, pinned: bool = None, edit: bool = None, delete: bool = None, group_call: bool = None) → telethon.client.chats._AdminLogIter

Iterator over the admin log for the specified channel.

The default order is from the most recent event to to the oldest.

Note that you must be an administrator of it to use this method.

If none of the filters are present (i.e. they all are None), all event types will be returned. If at least one of them is True, only those that are true will be returned.

Arguments
entity (entity):
The channel entity from which to get its admin log.
limit (int | None, optional):

Number of events to be retrieved.

The limit may also be None, which would eventually return the whole history.

max_id (int):
All the events with a higher (newer) ID or equal to this will be excluded.
min_id (int):
All the events with a lower (older) ID or equal to this will be excluded.
search (str):
The string to be used as a search query.
admins (entity | list):
If present, the events will be filtered by these admins (or single admin) and only those caused by them will be returned.
join (bool):
If True, events for when a user joined will be returned.
leave (bool):
If True, events for when a user leaves will be returned.
invite (bool):
If True, events for when a user joins through an invite link will be returned.
restrict (bool):
If True, events with partial restrictions will be returned. This is what the API calls “ban”.
unrestrict (bool):
If True, events removing restrictions will be returned. This is what the API calls “unban”.
ban (bool):
If True, events applying or removing all restrictions will be returned. This is what the API calls “kick” (restricting all permissions removed is a ban, which kicks the user).
unban (bool):
If True, events removing all restrictions will be returned. This is what the API calls “unkick”.
promote (bool):
If True, events with admin promotions will be returned.
demote (bool):
If True, events with admin demotions will be returned.
info (bool):
If True, events changing the group info will be returned.
settings (bool):
If True, events changing the group settings will be returned.
pinned (bool):
If True, events of new pinned messages will be returned.
edit (bool):
If True, events of message edits will be returned.
delete (bool):
If True, events of message deletions will be returned.
group_call (bool):
If True, events related to group calls will be returned.
Yields
Instances of AdminLogEvent.
Example
async for event in client.iter_admin_log(channel):
    if event.changed_title:
        print('The title changed from', event.old, 'to', event.new)
iter_participants(entity: hints.EntityLike, limit: float = None, *, search: str = '', filter: types.TypeChannelParticipantsFilter = None, aggressive: bool = False) → telethon.client.chats._ParticipantsIter

Iterator over the participants belonging to the specified chat.

The order is unspecified.

Arguments
entity (entity):
The entity from which to retrieve the participants list.
limit (int):
Limits amount of participants fetched.
search (str, optional):
Look for participants with this string in name/username.
filter (ChannelParticipantsFilter, optional):

The filter to be used, if you want e.g. only admins Note that you might not have permissions for some filter. This has no effect for normal chats or users.

Note

The filter ChannelParticipantsBanned will return restricted users. If you want banned users you should use ChannelParticipantsKicked instead.

aggressive (bool, optional):

Does nothing. This is kept for backwards-compatibility.

There have been several changes to Telegram’s API that limits the amount of members that can be retrieved, and this was a hack that no longer works.

Yields
The User objects returned by GetParticipantsRequest with an additional .participant attribute which is the matched ChannelParticipant type for channels/megagroups or ChatParticipants for normal chats.
Example
# Show all user IDs in a chat
async for user in client.iter_participants(chat):
    print(user.id)

# Search by name
async for user in client.iter_participants(chat, search='name'):
    print(user.username)

# Filter by admins
from telethon.tl.types import ChannelParticipantsAdmins
async for user in client.iter_participants(chat, filter=ChannelParticipantsAdmins):
    print(user.first_name)
iter_profile_photos(entity: hints.EntityLike, limit: int = None, *, offset: int = 0, max_id: int = 0) → telethon.client.chats._ProfilePhotoIter

Iterator over a user’s profile photos or a chat’s photos.

The order is from the most recent photo to the oldest.

Arguments
entity (entity):
The entity from which to get the profile or chat photos.
limit (int | None, optional):

Number of photos to be retrieved.

The limit may also be None, which would eventually all the photos that are still available.

offset (int):
How many photos should be skipped before returning the first one.
max_id (int):
The maximum ID allowed when fetching photos.
Yields
Instances of Photo.
Example
# Download all the profile photos of some user
async for photo in client.iter_profile_photos(user):
    await client.download_media(photo)
kick_participant(entity: hints.EntityLike, user: typing.Optional[hints.EntityLike])

Kicks a user from a chat.

Kicking yourself ('me') will result in leaving the chat.

Note

Attempting to kick someone who was banned will remove their restrictions (and thus unbanning them), since kicking is just ban + unban.

Arguments
entity (entity):
The channel or chat where the user should be kicked from.
user (entity, optional):
The user to kick.
Returns
Returns the service Message produced about a user being kicked, if any.
Example
# Kick some user from some chat, and deleting the service message
msg = await client.kick_participant(chat, user)
await msg.delete()

# Leaving chat
await client.kick_participant(chat, 'me')
class telethon.client.dialogs.DialogMethods

Bases: object

__weakref__

list of weak references to the object (if defined)

conversation(entity: hints.EntityLike, *, timeout: float = 60, total_timeout: float = None, max_messages: int = 100, exclusive: bool = True, replies_are_responses: bool = True) → telethon.tl.custom.conversation.Conversation

Creates a Conversation with the given entity.

Note

This Conversation API has certain shortcomings, such as lacking persistence, poor interaction with other event handlers, and overcomplicated usage for anything beyond the simplest case.

If you plan to interact with a bot without handlers, this works fine, but when running a bot yourself, you may instead prefer to follow the advice from https://stackoverflow.com/a/62246569/.

This is not the same as just sending a message to create a “dialog” with them, but rather a way to easily send messages and await for responses or other reactions. Refer to its documentation for more.

Arguments
entity (entity):
The entity with which a new conversation should be opened.
timeout (int | float, optional):
The default timeout (in seconds) per action to be used. You may also override this timeout on a per-method basis. By default each action can take up to 60 seconds (the value of this timeout).
total_timeout (int | float, optional):
The total timeout (in seconds) to use for the whole conversation. This takes priority over per-action timeouts. After these many seconds pass, subsequent actions will result in asyncio.TimeoutError.
max_messages (int, optional):
The maximum amount of messages this conversation will remember. After these many messages arrive in the specified chat, subsequent actions will result in ValueError.
exclusive (bool, optional):

By default, conversations are exclusive within a single chat. That means that while a conversation is open in a chat, you can’t open another one in the same chat, unless you disable this flag.

If you try opening an exclusive conversation for a chat where it’s already open, it will raise AlreadyInConversationError.

replies_are_responses (bool, optional):

Whether replies should be treated as responses or not.

If the setting is enabled, calls to conv.get_response and a subsequent call to conv.get_reply will return different messages, otherwise they may return the same message.

Consider the following scenario with one outgoing message, 1, and two incoming messages, the second one replying:

                    Hello! <1
2> (reply to 1) Hi!
3> (reply to 1) How are you?

And the following code:

async with client.conversation(chat) as conv:
    msg1 = await conv.send_message('Hello!')
    msg2 = await conv.get_response()
    msg3 = await conv.get_reply()

With the setting enabled, msg2 will be 'Hi!' and msg3 be 'How are you?' since replies are also responses, and a response was already returned.

With the setting disabled, both msg2 and msg3 will be 'Hi!' since one is a response and also a reply.

Returns
A Conversation.
Example
# <you> denotes outgoing messages you sent
# <usr> denotes incoming response messages
with bot.conversation(chat) as conv:
    # <you> Hi!
    conv.send_message('Hi!')

    # <usr> Hello!
    hello = conv.get_response()

    # <you> Please tell me your name
    conv.send_message('Please tell me your name')

    # <usr> ?
    name = conv.get_response().raw_text

    while not any(x.isalpha() for x in name):
        # <you> Your name didn't have any letters! Try again
        conv.send_message("Your name didn't have any letters! Try again")

        # <usr> Human
        name = conv.get_response().raw_text

    # <you> Thanks Human!
    conv.send_message('Thanks {}!'.format(name))
delete_dialog(entity: hints.EntityLike, *, revoke: bool = False)

Deletes a dialog (leaves a chat or channel).

This method can be used as a user and as a bot. However, bots will only be able to use it to leave groups and channels (trying to delete a private conversation will do nothing).

See also Dialog.delete().

Arguments
entity (entities):
The entity of the dialog to delete. If it’s a chat or channel, you will leave it. Note that the chat itself is not deleted, only the dialog, because you left it.
revoke (bool, optional):

On private chats, you may revoke the messages from the other peer too. By default, it’s False. Set it to True to delete the history for both.

This makes no difference for bot accounts, who can only leave groups and channels.

Returns
The Updates object that the request produces, or nothing for private conversations.
Example
# Deleting the first dialog
dialogs = await client.get_dialogs(5)
await client.delete_dialog(dialogs[0])

# Leaving a channel by username
await client.delete_dialog('username')
edit_folder(entity: hints.EntitiesLike = None, folder: Union[int, Sequence[int]] = None, *, unpack=None) → telethon.tl.types.Updates

Edits the folder used by one or more dialogs to archive them.

Arguments
entity (entities):
The entity or list of entities to move to the desired archive folder.
folder (int):

The folder to which the dialog should be archived to.

If you want to “archive” a dialog, use folder=1.

If you want to “un-archive” it, use folder=0.

You may also pass a list with the same length as entities if you want to control where each entity will go.

unpack (int, optional):

If you want to unpack an archived folder, set this parameter to the folder number that you want to delete.

When you unpack a folder, all the dialogs inside are moved to the folder number 0.

You can only use this parameter if the other two are not set.

Returns
The Updates object that the request produces.
Example
# Archiving the first 5 dialogs
dialogs = await client.get_dialogs(5)
await client.edit_folder(dialogs, 1)

# Un-archiving the third dialog (archiving to folder 0)
await client.edit_folder(dialog[2], 0)

# Moving the first dialog to folder 0 and the second to 1
dialogs = await client.get_dialogs(2)
await client.edit_folder(dialogs, [0, 1])

# Un-archiving all dialogs
await client.edit_folder(unpack=1)
get_dialogs(limit: float = None, *, offset_date: hints.DateLike = None, offset_id: int = 0, offset_peer: hints.EntityLike = <telethon.tl.types.InputPeerEmpty object>, ignore_pinned: bool = False, ignore_migrated: bool = False, folder: int = None, archived: bool = None) → telethon.client.dialogs._DialogsIter

Same as iter_dialogs(), but returns a TotalList instead.

Example
# Get all open conversation, print the title of the first
dialogs = await client.get_dialogs()
first = dialogs[0]
print(first.title)

# Use the dialog somewhere else
await client.send_message(first, 'hi')

# Getting only non-archived dialogs (both equivalent)
non_archived = await client.get_dialogs(folder=0)
non_archived = await client.get_dialogs(archived=False)

# Getting only archived dialogs (both equivalent)
archived = await client.get_dialogs(folder=1)
archived = await client.get_dialogs(archived=True)
get_drafts(entity: hints.EntitiesLike = None) → hints.TotalList

Same as iter_drafts(), but returns a list instead.

Example
# Get drafts, print the text of the first
drafts = await client.get_drafts()
print(drafts[0].text)

# Get the draft in your chat
draft = await client.get_drafts('me')
print(drafts.text)
iter_dialogs(limit: float = None, *, offset_date: hints.DateLike = None, offset_id: int = 0, offset_peer: hints.EntityLike = <telethon.tl.types.InputPeerEmpty object>, ignore_pinned: bool = False, ignore_migrated: bool = False, folder: int = None, archived: bool = None) → telethon.client.dialogs._DialogsIter

Iterator over the dialogs (open conversations/subscribed channels).

The order is the same as the one seen in official applications (first pinned, them from those with the most recent message to those with the oldest message).

Arguments
limit (int | None):
How many dialogs to be retrieved as maximum. Can be set to None to retrieve all dialogs. Note that this may take whole minutes if you have hundreds of dialogs, as Telegram will tell the library to slow down through a FloodWaitError.
offset_date (datetime, optional):
The offset date to be used.
offset_id (int, optional):
The message ID to be used as an offset.
offset_peer (InputPeer, optional):
The peer to be used as an offset.
ignore_pinned (bool, optional):
Whether pinned dialogs should be ignored or not. When set to True, these won’t be yielded at all.
ignore_migrated (bool, optional):
Whether Chat that have migrated_to a Channel should be included or not. By default all the chats in your dialogs are returned, but setting this to True will ignore (i.e. skip) them in the same way official applications do.
folder (int, optional):

The folder from which the dialogs should be retrieved.

If left unspecified, all dialogs (including those from folders) will be returned.

If set to 0, all dialogs that don’t belong to any folder will be returned.

If set to a folder number like 1, only those from said folder will be returned.

By default Telegram assigns the folder ID 1 to archived chats, so you should use that if you need to fetch the archived dialogs.

archived (bool, optional):
Alias for folder. If unspecified, all will be returned, False implies folder=0 and True implies folder=1.
Yields
Instances of Dialog.
Example
# Print all dialog IDs and the title, nicely formatted
async for dialog in client.iter_dialogs():
    print('{:>14}: {}'.format(dialog.id, dialog.title))
iter_drafts(entity: hints.EntitiesLike = None) → telethon.client.dialogs._DraftsIter

Iterator over draft messages.

The order is unspecified.

Arguments
entity (hints.EntitiesLike, optional):
The entity or entities for which to fetch the draft messages. If left unspecified, all draft messages will be returned.
Yields
Instances of Draft.
Example
# Clear all drafts
async for draft in client.get_drafts():
    await draft.delete()

# Getting the drafts with 'bot1' and 'bot2'
async for draft in client.iter_drafts(['bot1', 'bot2']):
    print(draft.text)
class telethon.client.downloads.DownloadMethods

Bases: object

__weakref__

list of weak references to the object (if defined)

download_file(input_location: hints.FileLike, file: hints.OutFileLike = None, *, part_size_kb: float = None, file_size: int = None, progress_callback: hints.ProgressCallback = None, dc_id: int = None, key: bytes = None, iv: bytes = None) → Optional[bytes]

Low-level method to download files from their input location.

Note

Generally, you should instead use download_media. This method is intended to be a bit more low-level.

Arguments
input_location (InputFileLocation):
The file location from which the file will be downloaded. See telethon.utils.get_input_location source for a complete list of supported types.
file (str | file, optional):

The output file path, directory, or stream-like object. If the path exists and is a file, it will be overwritten.

If the file path is None or bytes, then the result will be saved in memory and returned as bytes.

part_size_kb (int, optional):
Chunk size when downloading files. The larger, the less requests will be made (up to 512KB maximum).
file_size (int, optional):
The file size that is about to be downloaded, if known. Only used if progress_callback is specified.
progress_callback (callable, optional):
A callback function accepting two parameters: (downloaded bytes, total). Note that the total is the provided file_size.
dc_id (int, optional):
The data center the library should connect to in order to download the file. You shouldn’t worry about this.
key (‘bytes’, optional):
In case of an encrypted upload (secret chats) a key is supplied
iv (‘bytes’, optional):
In case of an encrypted upload (secret chats) an iv is supplied
Example
# Download a file and print its header
data = await client.download_file(input_file, bytes)
print(data[:16])
download_media(message: hints.MessageLike, file: hints.FileLike = None, *, thumb: typing.Union[int, types.TypePhotoSize] = None, progress_callback: hints.ProgressCallback = None) → Union[str, bytes, None]

Downloads the given media from a message object.

Note that if the download is too slow, you should consider installing cryptg (through pip install cryptg) so that decrypting the received data is done in C instead of Python (much faster).

See also Message.download_media().

Arguments
message (Message | Media):
The media or message containing the media that will be downloaded.
file (str | file, optional):
The output file path, directory, or stream-like object. If the path exists and is a file, it will be overwritten. If file is the type bytes, it will be downloaded in-memory and returned as a bytestring (i.e. file=bytes, without parentheses or quotes).
progress_callback (callable, optional):
A callback function accepting two parameters: (received bytes, total).
thumb (int | PhotoSize, optional):

Which thumbnail size from the document or photo to download, instead of downloading the document or photo itself.

If it’s specified but the file does not have a thumbnail, this method will return None.

The parameter should be an integer index between 0 and len(sizes). 0 will download the smallest thumbnail, and len(sizes) - 1 will download the largest thumbnail. You can also use negative indices, which work the same as they do in Python’s list.

You can also pass the PhotoSize instance to use. Alternatively, the thumb size type str may be used.

In short, use thumb=0 if you want the smallest thumbnail and thumb=-1 if you want the largest thumbnail.

Note

The largest thumbnail may be a video instead of a photo, as they are available since layer 116 and are bigger than any of the photos.

Returns
None if no media was provided, or if it was Empty. On success the file path is returned since it may differ from the one given.
Example
path = await client.download_media(message)
await client.download_media(message, filename)
# or
path = await message.download_media()
await message.download_media(filename)

# Downloading to memory
blob = await client.download_media(message, bytes)

# Printing download progress
def callback(current, total):
    print('Downloaded', current, 'out of', total,
          'bytes: {:.2%}'.format(current / total))

await client.download_media(message, progress_callback=callback)
download_profile_photo(entity: hints.EntityLike, file: hints.FileLike = None, *, download_big: bool = True) → Optional[str]

Downloads the profile photo from the given user, chat or channel.

Arguments
entity (entity):

From who the photo will be downloaded.

Note

This method expects the full entity (which has the data to download the photo), not an input variant.

It’s possible that sometimes you can’t fetch the entity from its input (since you can get errors like ChannelPrivateError) but you already have it through another call, like getting a forwarded message from it.

file (str | file, optional):
The output file path, directory, or stream-like object. If the path exists and is a file, it will be overwritten. If file is the type bytes, it will be downloaded in-memory and returned as a bytestring (i.e. file=bytes, without parentheses or quotes).
download_big (bool, optional):
Whether to use the big version of the available photos.
Returns
None if no photo was provided, or if it was Empty. On success the file path is returned since it may differ from the one given.
Example
# Download your own profile photo
path = await client.download_profile_photo('me')
print(path)
iter_download(file: hints.FileLike, *, offset: int = 0, stride: int = None, limit: int = None, chunk_size: int = None, request_size: int = 524288, file_size: int = None, dc_id: int = None)

Iterates over a file download, yielding chunks of the file.

This method can be used to stream files in a more convenient way, since it offers more control (pausing, resuming, etc.)

Note

Using a value for offset or stride which is not a multiple of the minimum allowed request_size, or if chunk_size is different from request_size, the library will need to do a bit more work to fetch the data in the way you intend it to.

You normally shouldn’t worry about this.

Arguments
file (hints.FileLike):
The file of which contents you want to iterate over.
offset (int, optional):
The offset in bytes into the file from where the download should start. For example, if a file is 1024KB long and you just want the last 512KB, you would use offset=512 * 1024.
stride (int, optional):

The stride of each chunk (how much the offset should advance between reading each chunk). This parameter should only be used for more advanced use cases.

It must be bigger than or equal to the chunk_size.

limit (int, optional):
The limit for how many chunks will be yielded at most.
chunk_size (int, optional):
The maximum size of the chunks that will be yielded. Note that the last chunk may be less than this value. By default, it equals to request_size.
request_size (int, optional):

How many bytes will be requested to Telegram when more data is required. By default, as many bytes as possible are requested. If you would like to request data in smaller sizes, adjust this parameter.

Note that values outside the valid range will be clamped, and the final value will also be a multiple of the minimum allowed size.

file_size (int, optional):
If the file size is known beforehand, you should set this parameter to said value. Depending on the type of the input file passed, this may be set automatically.
dc_id (int, optional):
The data center the library should connect to in order to download the file. You shouldn’t worry about this.

Yields

bytes objects representing the chunks of the file if the right conditions are met, or memoryview objects instead.
Example
# Streaming `media` to an output file
# After the iteration ends, the sender is cleaned up
with open('photo.jpg', 'wb') as fd:
    async for chunk in client.iter_download(media):
        fd.write(chunk)

# Fetching only the header of a file (32 bytes)
# You should manually close the iterator in this case.
#
# "stream" is a common name for asynchronous generators,
# and iter_download will yield `bytes` (chunks of the file).
stream = client.iter_download(media, request_size=32)
header = await stream.__anext__()  # "manual" version of `async for`
await stream.close()
assert len(header) == 32
class telethon.client.messageparse.MessageParseMethods

Bases: object

__weakref__

list of weak references to the object (if defined)

parse_mode

This property is the default parse mode used when sending messages. Defaults to telethon.extensions.markdown. It will always be either None or an object with parse and unparse methods.

When setting a different value it should be one of:

  • Object with parse and unparse methods.
  • A callable to act as the parse method.
  • A str indicating the parse_mode. For Markdown 'md' or 'markdown' may be used. For HTML, 'htm' or 'html' may be used.

The parse method should be a function accepting a single parameter, the text to parse, and returning a tuple consisting of (parsed message str, [MessageEntity instances]).

The unparse method should be the inverse of parse such that assert text == unparse(*parse(text)).

See MessageEntity for allowed message entities.

Example
# Disabling default formatting
client.parse_mode = None

# Enabling HTML as the default format
client.parse_mode = 'html'
class telethon.client.messages.MessageMethods

Bases: object

__weakref__

list of weak references to the object (if defined)

delete_messages(entity: hints.EntityLike, message_ids: typing.Union[hints.MessageIDLike, typing.Sequence[hints.MessageIDLike]], *, revoke: bool = True) → typing.Sequence[types.messages.AffectedMessages]

Deletes the given messages, optionally “for everyone”.

See also Message.delete().

Warning

This method does not validate that the message IDs belong to the chat that you passed! It’s possible for the method to delete messages from different private chats and small group chats at once, so make sure to pass the right IDs.

Arguments
entity (entity):
From who the message will be deleted. This can actually be None for normal chats, but must be present for channels and megagroups.
message_ids (list | int | Message):
The IDs (or ID) or messages to be deleted.
revoke (bool, optional):

Whether the message should be deleted for everyone or not. By default it has the opposite behaviour of official clients, and it will delete the message for everyone.

Since 24 March 2019, you can also revoke messages of any age (i.e. messages sent long in the past) the other person sent in private conversations (and of course your messages too).

Disabling this has no effect on channels or megagroups, since it will unconditionally delete the message for everyone.

Returns
A list of AffectedMessages, each item being the result for the delete calls of the messages in chunks of 100 each.
Example
await client.delete_messages(chat, messages)
edit_message(entity: typing.Union[hints.EntityLike, types.Message], message: hints.MessageLike = None, text: str = None, *, parse_mode: str = (), attributes: typing.Sequence[types.TypeDocumentAttribute] = None, formatting_entities: Optional[List[Union[telethon.tl.types.MessageEntityUnknown, telethon.tl.types.MessageEntityMention, telethon.tl.types.MessageEntityHashtag, telethon.tl.types.MessageEntityBotCommand, telethon.tl.types.MessageEntityUrl, telethon.tl.types.MessageEntityEmail, telethon.tl.types.MessageEntityBold, telethon.tl.types.MessageEntityItalic, telethon.tl.types.MessageEntityCode, telethon.tl.types.MessageEntityPre, telethon.tl.types.MessageEntityTextUrl, telethon.tl.types.MessageEntityMentionName, telethon.tl.types.InputMessageEntityMentionName, telethon.tl.types.MessageEntityPhone, telethon.tl.types.MessageEntityCashtag, telethon.tl.types.MessageEntityUnderline, telethon.tl.types.MessageEntityStrike, telethon.tl.types.MessageEntityBlockquote, telethon.tl.types.MessageEntityBankCard, telethon.tl.types.MessageEntitySpoiler, telethon.tl.types.MessageEntityCustomEmoji]]] = None, link_preview: bool = True, file: hints.FileLike = None, thumb: hints.FileLike = None, force_document: bool = False, buttons: Optional[hints.MarkupLike] = None, supports_streaming: bool = False, schedule: hints.DateLike = None) → types.Message

Edits the given message to change its text or media.

See also Message.edit().

Arguments
entity (entity | Message):

From which chat to edit the message. This can also be the message to be edited, and the entity will be inferred from it, so the next parameter will be assumed to be the message text.

You may also pass a InputBotInlineMessageID or InputBotInlineMessageID64, which is the only way to edit messages that were sent after the user selects an inline query result.

message (int | Message | str):
The ID of the message (or Message itself) to be edited. If the entity was a Message, then this message will be treated as the new text.
text (str, optional):
The new text of the message. Does nothing if the entity was a Message.
parse_mode (object, optional):
See the TelegramClient.parse_mode property for allowed values. Markdown parsing will be used by default.
attributes (list, optional):
Optional attributes that override the inferred ones, like DocumentAttributeFilename and so on.
formatting_entities (list, optional):
A list of message formatting entities. When provided, the parse_mode is ignored.
link_preview (bool, optional):
Should the link preview be shown?
file (str | bytes | file | media, optional):
The file object that should replace the existing media in the message.
thumb (str | bytes | file, optional):
Optional JPEG thumbnail (for documents). Telegram will ignore this parameter unless you pass a .jpg file! The file must also be small in dimensions and in disk size. Successful thumbnails were files below 20kB and 320x320px. Width/height and dimensions/size ratios may be important. For Telegram to accept a thumbnail, you must provide the dimensions of the underlying media through attributes= with DocumentAttributesVideo or by installing the optional hachoir dependency.
force_document (bool, optional):
Whether to send the given file as a document or not.
buttons (list, custom.Button, KeyboardButton):
The matrix (list of lists), row list or button to be shown after sending the message. This parameter will only work if you have signed in as a bot. You can also pass your own ReplyMarkup here.
supports_streaming (bool, optional):
Whether the sent video supports streaming or not. Note that Telegram only recognizes as streamable some formats like MP4, and others like AVI or MKV will not work. You should convert these to MP4 before sending if you want them to be streamable. Unsupported formats will result in VideoContentTypeError.
schedule (hints.DateLike, optional):

If set, the message won’t be edited immediately, and instead it will be scheduled to be automatically edited at a later time.

Note that this parameter will have no effect if you are trying to edit a message that was sent via inline bots.

Returns
The edited Message, unless entity was a InputBotInlineMessageID or InputBotInlineMessageID64 in which case this method returns a boolean.
Raises

MessageAuthorRequiredError if you’re not the author of the message but tried editing it anyway.

MessageNotModifiedError if the contents of the message were not modified at all.

MessageIdInvalidError if the ID of the message is invalid (the ID itself may be correct, but the message with that ID cannot be edited). For example, when trying to edit messages with a reply markup (or clear markup) this error will be raised.

Example
message = await client.send_message(chat, 'hello')

await client.edit_message(chat, message, 'hello!')
# or
await client.edit_message(chat, message.id, 'hello!!')
# or
await client.edit_message(message, 'hello!!!')
forward_messages(entity: hints.EntityLike, messages: typing.Union[hints.MessageIDLike, typing.Sequence[hints.MessageIDLike]], from_peer: hints.EntityLike = None, *, background: bool = None, with_my_score: bool = None, silent: bool = None, as_album: bool = None, schedule: hints.DateLike = None) → typing.Sequence[types.Message]

Forwards the given messages to the specified entity.

If you want to “forward” a message without the forward header (the “forwarded from” text), you should use send_message with the original message instead. This will send a copy of it.

See also Message.forward_to().

Arguments
entity (entity):
To which entity the message(s) will be forwarded.
messages (list | int | Message):
The message(s) to forward, or their integer IDs.
from_peer (entity):
If the given messages are integer IDs and not instances of the Message class, this must be specified in order for the forward to work. This parameter indicates the entity from which the messages should be forwarded.
silent (bool, optional):
Whether the message should notify people with sound or not. Defaults to False (send with a notification sound unless the person has the chat muted). Set it to True to alter this behaviour.
background (bool, optional):
Whether the message should be forwarded in background.
with_my_score (bool, optional):
Whether forwarded should contain your game score.
as_album (bool, optional):
This flag no longer has any effect.
schedule (hints.DateLike, optional):
If set, the message(s) won’t forward immediately, and instead they will be scheduled to be automatically sent at a later time.
Returns

The list of forwarded Message, or a single one if a list wasn’t provided as input.

Note that if all messages are invalid (i.e. deleted) the call will fail with MessageIdInvalidError. If only some are invalid, the list will have None instead of those messages.

Example
# a single one
await client.forward_messages(chat, message)
# or
await client.forward_messages(chat, message_id, from_chat)
# or
await message.forward_to(chat)

# multiple
await client.forward_messages(chat, messages)
# or
await client.forward_messages(chat, message_ids, from_chat)

# Forwarding as a copy
await client.send_message(chat, message)
get_messages(entity: hints.EntityLike, limit: float = None, *, offset_date: hints.DateLike = None, offset_id: int = 0, max_id: int = 0, min_id: int = 0, add_offset: int = 0, search: str = None, filter: typing.Union[types.TypeMessagesFilter, typing.Type[types.TypeMessagesFilter]] = None, from_user: hints.EntityLike = None, wait_time: float = None, ids: typing.Union[int, typing.Sequence[int]] = None, reverse: bool = False, reply_to: int = None, scheduled: bool = False) → typing.Union[_MessagesIter, _IDsIter]

Same as iter_messages(), but returns a TotalList instead.

If the limit is not set, it will be 1 by default unless both min_id and max_id are set (as named arguments), in which case the entire range will be returned.

This is so because any integer limit would be rather arbitrary and it’s common to only want to fetch one message, but if a range is specified it makes sense that it should return the entirety of it.

If ids is present in the named arguments and is not a list, a single Message will be returned for convenience instead of a list.

Example
# Get 0 photos and print the total to show how many photos there are
from telethon.tl.types import InputMessagesFilterPhotos
photos = await client.get_messages(chat, 0, filter=InputMessagesFilterPhotos)
print(photos.total)

# Get all the photos
photos = await client.get_messages(chat, None, filter=InputMessagesFilterPhotos)

# Get messages by ID:
message_1337 = await client.get_messages(chat, ids=1337)
iter_messages(entity: hints.EntityLike, limit: float = None, *, offset_date: hints.DateLike = None, offset_id: int = 0, max_id: int = 0, min_id: int = 0, add_offset: int = 0, search: str = None, filter: typing.Union[types.TypeMessagesFilter, typing.Type[types.TypeMessagesFilter]] = None, from_user: hints.EntityLike = None, wait_time: float = None, ids: typing.Union[int, typing.Sequence[int]] = None, reverse: bool = False, reply_to: int = None, scheduled: bool = False) → typing.Union[_MessagesIter, _IDsIter]

Iterator over the messages for the given chat.

The default order is from newest to oldest, but this behaviour can be changed with the reverse parameter.

If either search, filter or from_user are provided, messages.Search will be used instead of messages.getHistory.

Note

Telegram’s flood wait limit for GetHistoryRequest seems to be around 30 seconds per 10 requests, therefore a sleep of 1 second is the default for this limit (or above).

Arguments
entity (entity):

The entity from whom to retrieve the message history.

It may be None to perform a global search, or to get messages by their ID from no particular chat. Note that some of the offsets will not work if this is the case.

Note that if you want to perform a global search, you must set a non-empty search string, a filter. or from_user.

limit (int | None, optional):

Number of messages to be retrieved. Due to limitations with the API retrieving more than 3000 messages will take longer than half a minute (or even more based on previous calls).

The limit may also be None, which would eventually return the whole history.

offset_date (datetime):
Offset date (messages previous to this date will be retrieved). Exclusive.
offset_id (int):
Offset message ID (only messages previous to the given ID will be retrieved). Exclusive.
max_id (int):
All the messages with a higher (newer) ID or equal to this will be excluded.
min_id (int):
All the messages with a lower (older) ID or equal to this will be excluded.
add_offset (int):
Additional message offset (all of the specified offsets + this offset = older messages).
search (str):
The string to be used as a search query.
filter (MessagesFilter | type):
The filter to use when returning messages. For instance, InputMessagesFilterPhotos would yield only messages containing photos.
from_user (entity):
Only messages from this entity will be returned.
wait_time (int):

Wait time (in seconds) between different GetHistoryRequest. Use this parameter to avoid hitting the FloodWaitError as needed. If left to None, it will default to 1 second only if the limit is higher than 3000.

If the ids parameter is used, this time will default to 10 seconds only if the amount of IDs is higher than 300.

ids (int, list):

A single integer ID (or several IDs) for the message that should be returned. This parameter takes precedence over the rest (which will be ignored if this is set). This can for instance be used to get the message with ID 123 from a channel. Note that if the message doesn’t exist, None will appear in its place, so that zipping the list of IDs with the messages can match one-to-one.

Note

At the time of writing, Telegram will not return MessageEmpty for InputMessageReplyTo IDs that failed (i.e. the message is not replying to any, or is replying to a deleted message). This means that it is not possible to match messages one-by-one, so be careful if you use non-integers in this parameter.

reverse (bool, optional):

If set to True, the messages will be returned in reverse order (from oldest to newest, instead of the default newest to oldest). This also means that the meaning of offset_id and offset_date parameters is reversed, although they will still be exclusive. min_id becomes equivalent to offset_id instead of being max_id as well since messages are returned in ascending order.

You cannot use this if both entity and ids are None.

reply_to (int, optional):

If set to a message ID, the messages that reply to this ID will be returned. This feature is also known as comments in posts of broadcast channels, or viewing threads in groups.

This feature can only be used in broadcast channels and their linked megagroups. Using it in a chat or private conversation will result in telethon.errors.PeerIdInvalidError to occur.

When using this parameter, the filter and search parameters have no effect, since Telegram’s API doesn’t support searching messages in replies.

Note

This feature is used to get replies to a message in the discussion group. If the same broadcast channel sends a message and replies to it itself, that reply will not be included in the results.

scheduled (bool, optional):
If set to True, messages which are scheduled will be returned. All other parameter will be ignored for this, except entity.
Yields
Instances of Message.
Example
# From most-recent to oldest
async for message in client.iter_messages(chat):
    print(message.id, message.text)

# From oldest to most-recent
async for message in client.iter_messages(chat, reverse=True):
    print(message.id, message.text)

# Filter by sender
async for message in client.iter_messages(chat, from_user='me'):
    print(message.text)

# Server-side search with fuzzy text
async for message in client.iter_messages(chat, search='hello'):
    print(message.id)

# Filter by message type:
from telethon.tl.types import InputMessagesFilterPhotos
async for message in client.iter_messages(chat, filter=InputMessagesFilterPhotos):
    print(message.photo)

# Getting comments from a post in a channel:
async for message in client.iter_messages(channel, reply_to=123):
    print(message.chat.title, message.text)
pin_message(entity: hints.EntityLike, message: typing.Optional[hints.MessageIDLike], *, notify: bool = False, pm_oneside: bool = False)

Pins a message in a chat.

The default behaviour is to not notify members, unlike the official applications.

See also Message.pin().

Arguments
entity (entity):
The chat where the message should be pinned.
message (int | Message):
The message or the message ID to pin. If it’s None, all messages will be unpinned instead.
notify (bool, optional):
Whether the pin should notify people or not.
pm_oneside (bool, optional):
Whether the message should be pinned for everyone or not. By default it has the opposite behaviour of official clients, and it will pin the message for both sides, in private chats.
Example
# Send and pin a message to annoy everyone
message = await client.send_message(chat, 'Pinotifying is fun!')
await client.pin_message(chat, message, notify=True)
send_message(entity: hints.EntityLike, message: hints.MessageLike = '', *, reply_to: typing.Union[int, types.Message] = None, attributes: typing.Sequence[types.TypeDocumentAttribute] = None, parse_mode: Optional[str] = (), formatting_entities: Optional[List[Union[telethon.tl.types.MessageEntityUnknown, telethon.tl.types.MessageEntityMention, telethon.tl.types.MessageEntityHashtag, telethon.tl.types.MessageEntityBotCommand, telethon.tl.types.MessageEntityUrl, telethon.tl.types.MessageEntityEmail, telethon.tl.types.MessageEntityBold, telethon.tl.types.MessageEntityItalic, telethon.tl.types.MessageEntityCode, telethon.tl.types.MessageEntityPre, telethon.tl.types.MessageEntityTextUrl, telethon.tl.types.MessageEntityMentionName, telethon.tl.types.InputMessageEntityMentionName, telethon.tl.types.MessageEntityPhone, telethon.tl.types.MessageEntityCashtag, telethon.tl.types.MessageEntityUnderline, telethon.tl.types.MessageEntityStrike, telethon.tl.types.MessageEntityBlockquote, telethon.tl.types.MessageEntityBankCard, telethon.tl.types.MessageEntitySpoiler, telethon.tl.types.MessageEntityCustomEmoji]]] = None, link_preview: bool = True, file: typing.Union[hints.FileLike, typing.Sequence[hints.FileLike]] = None, thumb: hints.FileLike = None, force_document: bool = False, clear_draft: bool = False, buttons: Optional[hints.MarkupLike] = None, silent: bool = None, background: bool = None, supports_streaming: bool = False, schedule: hints.DateLike = None, comment_to: typing.Union[int, types.Message] = None, nosound_video: bool = None) → types.Message

Sends a message to the specified user, chat or channel.

The default parse mode is the same as the official applications (a custom flavour of markdown). **bold**, `code` or __italic__ are available. In addition you can send [links](https://example.com) and [mentions](@username) (or using IDs like in the Bot API: [mention](tg://user?id=123456789)) and pre blocks with three backticks.

Sending a /start command with a parameter (like ?start=data) is also done through this method. Simply send '/start data' to the bot.

See also Message.respond() and Message.reply().

Arguments
entity (entity):
To who will it be sent.
message (str | Message):

The message to be sent, or another message object to resend.

The maximum length for a message is 35,000 bytes or 4,096 characters. Longer messages will not be sliced automatically, and you should slice them manually if the text to send is longer than said length.

reply_to (int | Message, optional):
Whether to reply to a message or not. If an integer is provided, it should be the ID of the message that it should reply to.
attributes (list, optional):
Optional attributes that override the inferred ones, like DocumentAttributeFilename and so on.
parse_mode (object, optional):
See the TelegramClient.parse_mode property for allowed values. Markdown parsing will be used by default.
formatting_entities (list, optional):
A list of message formatting entities. When provided, the parse_mode is ignored.
link_preview (bool, optional):
Should the link preview be shown?
file (file, optional):
Sends a message with a file attached (e.g. a photo, video, audio or document). The message may be empty.
thumb (str | bytes | file, optional):
Optional JPEG thumbnail (for documents). Telegram will ignore this parameter unless you pass a .jpg file! The file must also be small in dimensions and in disk size. Successful thumbnails were files below 20kB and 320x320px. Width/height and dimensions/size ratios may be important. For Telegram to accept a thumbnail, you must provide the dimensions of the underlying media through attributes= with DocumentAttributesVideo or by installing the optional hachoir dependency.
force_document (bool, optional):
Whether to send the given file as a document or not.
clear_draft (bool, optional):
Whether the existing draft should be cleared or not.
buttons (list, custom.Button, KeyboardButton):

The matrix (list of lists), row list or button to be shown after sending the message. This parameter will only work if you have signed in as a bot. You can also pass your own ReplyMarkup here.

All the following limits apply together:

  • There can be 100 buttons at most (any more are ignored).
  • There can be 8 buttons per row at most (more are ignored).
  • The maximum callback data per button is 64 bytes.
  • The maximum data that can be embedded in total is just over 4KB, shared between inline callback data and text.
silent (bool, optional):
Whether the message should notify people in a broadcast channel or not. Defaults to False, which means it will notify them. Set it to True to alter this behaviour.
background (bool, optional):
Whether the message should be send in background.
supports_streaming (bool, optional):
Whether the sent video supports streaming or not. Note that Telegram only recognizes as streamable some formats like MP4, and others like AVI or MKV will not work. You should convert these to MP4 before sending if you want them to be streamable. Unsupported formats will result in VideoContentTypeError.
schedule (hints.DateLike, optional):
If set, the message won’t send immediately, and instead it will be scheduled to be automatically sent at a later time.
comment_to (int | Message, optional):

Similar to reply_to, but replies in the linked group of a broadcast channel instead (effectively leaving a “comment to” the specified message).

This parameter takes precedence over reply_to. If there is no linked chat, telethon.errors.sgIdInvalidError is raised.

nosound_video (bool, optional):
Only applicable when sending a video file without an audio track. If set to True, the video will be displayed in Telegram as a video. If set to False, Telegram will attempt to display the video as an animated gif. (It may still display as a video due to other factors.) The value is ignored if set on non-video files. This is set to True for albums, as gifs cannot be sent in albums.
Returns
The sent custom.Message.
Example
# Markdown is the default
await client.send_message('me', 'Hello **world**!')

# Default to another parse mode
client.parse_mode = 'html'

await client.send_message('me', 'Some <b>bold</b> and <i>italic</i> text')
await client.send_message('me', 'An <a href="https://example.com">URL</a>')
# code and pre tags also work, but those break the documentation :)
await client.send_message('me', '<a href="tg://user?id=me">Mentions</a>')

# Explicit parse mode
# No parse mode by default
client.parse_mode = None

# ...but here I want markdown
await client.send_message('me', 'Hello, **world**!', parse_mode='md')

# ...and here I need HTML
await client.send_message('me', 'Hello, <i>world</i>!', parse_mode='html')

# If you logged in as a bot account, you can send buttons
from telethon import events, Button

@client.on(events.CallbackQuery)
async def callback(event):
    await event.edit('Thank you for clicking {}!'.format(event.data))

# Single inline button
await client.send_message(chat, 'A single button, with "clk1" as data',
                          buttons=Button.inline('Click me', b'clk1'))

# Matrix of inline buttons
await client.send_message(chat, 'Pick one from this grid', buttons=[
    [Button.inline('Left'), Button.inline('Right')],
    [Button.url('Check this site!', 'https://example.com')]
])

# Reply keyboard
await client.send_message(chat, 'Welcome', buttons=[
    Button.text('Thanks!', resize=True, single_use=True),
    Button.request_phone('Send phone'),
    Button.request_location('Send location')
])

# Forcing replies or clearing buttons.
await client.send_message(chat, 'Reply to me', buttons=Button.force_reply())
await client.send_message(chat, 'Bye Keyboard!', buttons=Button.clear())

# Scheduling a message to be sent after 5 minutes
from datetime import timedelta
await client.send_message(chat, 'Hi, future!', schedule=timedelta(minutes=5))
send_read_acknowledge(entity: hints.EntityLike, message: typing.Union[hints.MessageIDLike, typing.Sequence[hints.MessageIDLike]] = None, *, max_id: int = None, clear_mentions: bool = False, clear_reactions: bool = False) → bool

Marks messages as read and optionally clears mentions.

This effectively marks a message as read (or more than one) in the given conversation.

If neither message nor maximum ID are provided, all messages will be marked as read by assuming that max_id = 0.

If a message or maximum ID is provided, all the messages up to and including such ID will be marked as read (for all messages whose ID ≤ max_id).

See also Message.mark_read().

Arguments
entity (entity):
The chat where these messages are located.
message (list | Message):
Either a list of messages or a single message.
max_id (int):
Until which message should the read acknowledge be sent for. This has priority over the message parameter.
clear_mentions (bool):

Whether the mention badge should be cleared (so that there are no more mentions) or not for the given entity.

If no message is provided, this will be the only action taken.

clear_reactions (bool):

Whether the reactions badge should be cleared (so that there are no more reaction notifications) or not for the given entity.

If no message is provided, this will be the only action taken.

Example
# using a Message object
await client.send_read_acknowledge(chat, message)
# ...or using the int ID of a Message
await client.send_read_acknowledge(chat, message_id)
# ...or passing a list of messages to mark as read
await client.send_read_acknowledge(chat, messages)
unpin_message(entity: hints.EntityLike, message: typing.Optional[hints.MessageIDLike] = None, *, notify: bool = False)

Unpins a message in a chat.

If no message ID is specified, all pinned messages will be unpinned.

See also Message.unpin().

Arguments
entity (entity):
The chat where the message should be pinned.
message (int | Message):
The message or the message ID to unpin. If it’s None, all messages will be unpinned instead.
Example
# Unpin all messages from a chat
await client.unpin_message(chat)
class telethon.client.updates.EventBuilderDict(client: TelegramClient, update, others)

Bases: object

Helper “dictionary” to return events from types and cache them.

__getitem__(builder)
__weakref__

list of weak references to the object (if defined)

class telethon.client.updates.UpdateMethods

Bases: object

__weakref__

list of weak references to the object (if defined)

add_event_handler(callback: Callable[[Any], Any], event: telethon.events.common.EventBuilder = None)

Registers a new event handler callback.

The callback will be called when the specified event occurs.

Arguments
callback (callable):

The callable function accepting one parameter to be used.

Note that if you have used telethon.events.register in the callback, event will be ignored, and instead the events you previously registered will be used.

event (_EventBuilder | type, optional):

The event builder class or instance to be used, for instance events.NewMessage.

If left unspecified, telethon.events.raw.Raw (the Update objects with no further processing) will be passed instead.

Example
from telethon import TelegramClient, events
client = TelegramClient(...)

async def handler(event):
    ...

client.add_event_handler(handler, events.NewMessage)
catch_up()

“Catches up” on the missed updates while the client was offline. You should call this method after registering the event handlers so that the updates it loads can by processed by your script.

This can also be used to forcibly fetch new updates if there are any.

Example
await client.catch_up()
list_event_handlers() → typing.Sequence[typing.Tuple[Callback, EventBuilder]]

Lists all registered event handlers.

Returns
A list of pairs consisting of (callback, event).
Example
@client.on(events.NewMessage(pattern='hello'))
async def on_greeting(event):
    '''Greets someone'''
    await event.reply('Hi')

for callback, event in client.list_event_handlers():
    print(id(callback), type(event))
on(event: telethon.events.common.EventBuilder)

Decorator used to add_event_handler more conveniently.

Arguments
event (_EventBuilder | type):
The event builder class or instance to be used, for instance events.NewMessage.
Example
from telethon import TelegramClient, events
client = TelegramClient(...)

# Here we use client.on
@client.on(events.NewMessage)
async def handler(event):
    ...
remove_event_handler(callback: Callable[[Any], Any], event: telethon.events.common.EventBuilder = None) → int

Inverse operation of add_event_handler().

If no event is given, all events for this callback are removed. Returns how many callbacks were removed.

Example
@client.on(events.Raw)
@client.on(events.NewMessage)
async def handler(event):
    ...

# Removes only the "Raw" handling
# "handler" will still receive "events.NewMessage"
client.remove_event_handler(handler, events.Raw)

# "handler" will stop receiving anything
client.remove_event_handler(handler)
run_until_disconnected()

Runs the event loop until the library is disconnected.

It also notifies Telegram that we want to receive updates as described in https://core.telegram.org/api/updates. If an unexpected error occurs during update handling, the client will disconnect and said error will be raised.

Manual disconnections can be made by calling disconnect() or sending a KeyboardInterrupt (e.g. by pressing Ctrl+C on the console window running the script).

If a disconnection error occurs (i.e. the library fails to reconnect automatically), said error will be raised through here, so you have a chance to except it on your own code.

If the loop is already running, this method returns a coroutine that you should await on your own code.

Note

If you want to handle KeyboardInterrupt in your code, simply run the event loop in your code too in any way, such as loop.run_forever() or await client.disconnected (e.g. loop.run_until_complete(client.disconnected)).

Example
# Blocks the current task here until a disconnection occurs.
#
# You will still receive updates, since this prevents the
# script from exiting.
await client.run_until_disconnected()
set_receive_updates(receive_updates)

Change the value of receive_updates.

This is an async method, because in order for Telegram to start sending updates again, a request must be made.

class telethon.client.uploads.UploadMethods

Bases: object

__weakref__

list of weak references to the object (if defined)

send_file(entity: hints.EntityLike, file: typing.Union[hints.FileLike, typing.Sequence[hints.FileLike]], *, caption: Union[str, Sequence[str]] = None, force_document: bool = False, file_size: int = None, clear_draft: bool = False, progress_callback: hints.ProgressCallback = None, reply_to: hints.MessageIDLike = None, attributes: typing.Sequence[types.TypeDocumentAttribute] = None, thumb: hints.FileLike = None, allow_cache: bool = True, parse_mode: str = (), formatting_entities: Optional[List[Union[telethon.tl.types.MessageEntityUnknown, telethon.tl.types.MessageEntityMention, telethon.tl.types.MessageEntityHashtag, telethon.tl.types.MessageEntityBotCommand, telethon.tl.types.MessageEntityUrl, telethon.tl.types.MessageEntityEmail, telethon.tl.types.MessageEntityBold, telethon.tl.types.MessageEntityItalic, telethon.tl.types.MessageEntityCode, telethon.tl.types.MessageEntityPre, telethon.tl.types.MessageEntityTextUrl, telethon.tl.types.MessageEntityMentionName, telethon.tl.types.InputMessageEntityMentionName, telethon.tl.types.MessageEntityPhone, telethon.tl.types.MessageEntityCashtag, telethon.tl.types.MessageEntityUnderline, telethon.tl.types.MessageEntityStrike, telethon.tl.types.MessageEntityBlockquote, telethon.tl.types.MessageEntityBankCard, telethon.tl.types.MessageEntitySpoiler, telethon.tl.types.MessageEntityCustomEmoji]]] = None, voice_note: bool = False, video_note: bool = False, buttons: Optional[hints.MarkupLike] = None, silent: bool = None, background: bool = None, supports_streaming: bool = False, schedule: hints.DateLike = None, comment_to: typing.Union[int, types.Message] = None, ttl: int = None, nosound_video: bool = None, **kwargs) → types.Message

Sends message with the given file to the specified entity.

Note

If the hachoir3 package (hachoir module) is installed, it will be used to determine metadata from audio and video files.

If the pillow package is installed and you are sending a photo, it will be resized to fit within the maximum dimensions allowed by Telegram to avoid errors.PhotoInvalidDimensionsError. This cannot be done if you are sending InputFile, however.

Arguments
entity (entity):
Who will receive the file.
file (str | bytes | file | media):

The file to send, which can be one of:

  • A local file path to an in-disk file. The file name will be the path’s base name.
  • A bytes byte array with the file’s data to send (for example, by using text.encode('utf-8')). A default file name will be used.
  • A bytes io.IOBase stream over the file to send (for example, by using open(file, 'rb')). Its .name property will be used for the file name, or a default if it doesn’t have one.
  • An external URL to a file over the internet. This will send the file as “external” media, and Telegram is the one that will fetch the media and send it.
  • A Bot API-like file_id. You can convert previously sent media to file IDs for later reusing with telethon.utils.pack_bot_file_id.
  • A handle to an existing file (for example, if you sent a message with media before, you can use its message.media as a file here).
  • A handle to an uploaded file (from upload_file).
  • A InputMedia instance. For example, if you want to send a dice use InputMediaDice, or if you want to send a contact use InputMediaContact.

To send an album, you should provide a list in this parameter.

If a list or similar is provided, the files in it will be sent as an album in the order in which they appear, sliced in chunks of 10 if more than 10 are given.

caption (str, optional):
Optional caption for the sent media message. When sending an album, the caption may be a list of strings, which will be assigned to the files pairwise.
force_document (bool, optional):
If left to False and the file is a path that ends with the extension of an image file or a video file, it will be sent as such. Otherwise always as a document.
file_size (int, optional):

The size of the file to be uploaded if it needs to be uploaded, which will be determined automatically if not specified.

If the file size can’t be determined beforehand, the entire file will be read in-memory to find out how large it is.

clear_draft (bool, optional):
Whether the existing draft should be cleared or not.
progress_callback (callable, optional):
A callback function accepting two parameters: (sent bytes, total).
reply_to (int | Message):
Same as reply_to from send_message.
attributes (list, optional):
Optional attributes that override the inferred ones, like DocumentAttributeFilename and so on.
thumb (str | bytes | file, optional):

Optional JPEG thumbnail (for documents). Telegram will ignore this parameter unless you pass a .jpg file!

The file must also be small in dimensions and in disk size. Successful thumbnails were files below 20kB and 320x320px. Width/height and dimensions/size ratios may be important. For Telegram to accept a thumbnail, you must provide the dimensions of the underlying media through attributes= with DocumentAttributesVideo or by installing the optional hachoir dependency.

allow_cache (bool, optional):
This parameter currently does nothing, but is kept for backward-compatibility (and it may get its use back in the future).
parse_mode (object, optional):
See the TelegramClient.parse_mode property for allowed values. Markdown parsing will be used by default.
formatting_entities (list, optional):
A list of message formatting entities. When provided, the parse_mode is ignored.
voice_note (bool, optional):
If True the audio will be sent as a voice note.
video_note (bool, optional):
If True the video will be sent as a video note, also known as a round video message.
buttons (list, custom.Button, KeyboardButton):
The matrix (list of lists), row list or button to be shown after sending the message. This parameter will only work if you have signed in as a bot. You can also pass your own ReplyMarkup here.
silent (bool, optional):
Whether the message should notify people with sound or not. Defaults to False (send with a notification sound unless the person has the chat muted). Set it to True to alter this behaviour.
background (bool, optional):
Whether the message should be send in background.
supports_streaming (bool, optional):
Whether the sent video supports streaming or not. Note that Telegram only recognizes as streamable some formats like MP4, and others like AVI or MKV will not work. You should convert these to MP4 before sending if you want them to be streamable. Unsupported formats will result in VideoContentTypeError.
schedule (hints.DateLike, optional):
If set, the file won’t send immediately, and instead it will be scheduled to be automatically sent at a later time.
comment_to (int | Message, optional):

Similar to reply_to, but replies in the linked group of a broadcast channel instead (effectively leaving a “comment to” the specified message).

This parameter takes precedence over reply_to. If there is no linked chat, telethon.errors.sgIdInvalidError is raised.

ttl (int. optional):

The Time-To-Live of the file (also known as “self-destruct timer” or “self-destructing media”). If set, files can only be viewed for a short period of time before they disappear from the message history automatically.

The value must be at least 1 second, and at most 60 seconds, otherwise Telegram will ignore this parameter.

Not all types of media can be used with this parameter, such as text documents, which will fail with TtlMediaInvalidError.

nosound_video (bool, optional):
Only applicable when sending a video file without an audio track. If set to True, the video will be displayed in Telegram as a video. If set to False, Telegram will attempt to display the video as an animated gif. (It may still display as a video due to other factors.) The value is ignored if set on non-video files. This is set to True for albums, as gifs cannot be sent in albums.
Returns
The Message (or messages) containing the sent file, or messages if a list of them was passed.
Example
# Normal files like photos
await client.send_file(chat, '/my/photos/me.jpg', caption="It's me!")
# or
await client.send_message(chat, "It's me!", file='/my/photos/me.jpg')

# Voice notes or round videos
await client.send_file(chat, '/my/songs/song.mp3', voice_note=True)
await client.send_file(chat, '/my/videos/video.mp4', video_note=True)

# Custom thumbnails
await client.send_file(chat, '/my/documents/doc.txt', thumb='photo.jpg')

# Only documents
await client.send_file(chat, '/my/photos/photo.png', force_document=True)

# Albums
await client.send_file(chat, [
    '/my/photos/holiday1.jpg',
    '/my/photos/holiday2.jpg',
    '/my/drawings/portrait.png'
])

# Printing upload progress
def callback(current, total):
    print('Uploaded', current, 'out of', total,
          'bytes: {:.2%}'.format(current / total))

await client.send_file(chat, file, progress_callback=callback)

# Dices, including dart and other future emoji
from telethon.tl import types
await client.send_file(chat, types.InputMediaDice(''))
await client.send_file(chat, types.InputMediaDice('🎯'))

# Contacts
await client.send_file(chat, types.InputMediaContact(
    phone_number='+34 123 456 789',
    first_name='Example',
    last_name='',
    vcard=''
))
upload_file(file: hints.FileLike, *, part_size_kb: float = None, file_size: int = None, file_name: str = None, use_cache: type = None, key: bytes = None, iv: bytes = None, progress_callback: hints.ProgressCallback = None) → types.TypeInputFile

Uploads a file to Telegram’s servers, without sending it.

Note

Generally, you want to use send_file instead.

This method returns a handle (an instance of InputFile or InputFileBig, as required) which can be later used before it expires (they are usable during less than a day).

Uploading a file will simply return a “handle” to the file stored remotely in the Telegram servers, which can be later used on. This will not upload the file to your own chat or any chat at all.

Arguments
file (str | bytes | file):
The path of the file, byte array, or stream that will be sent. Note that if a byte array or a stream is given, a filename or its type won’t be inferred, and it will be sent as an “unnamed application/octet-stream”.
part_size_kb (int, optional):
Chunk size when uploading files. The larger, the less requests will be made (up to 512KB maximum).
file_size (int, optional):

The size of the file to be uploaded, which will be determined automatically if not specified.

If the file size can’t be determined beforehand, the entire file will be read in-memory to find out how large it is.

file_name (str, optional):
The file name which will be used on the resulting InputFile. If not specified, the name will be taken from the file and if this is not a str, it will be "unnamed".
use_cache (type, optional):
This parameter currently does nothing, but is kept for backward-compatibility (and it may get its use back in the future).
key (‘bytes’, optional):
In case of an encrypted upload (secret chats) a key is supplied
iv (‘bytes’, optional):
In case of an encrypted upload (secret chats) an iv is supplied
progress_callback (callable, optional):

A callback function accepting two parameters: (sent bytes, total).

When sending an album, the callback will receive a number between 0 and the amount of files as the “sent” parameter, and the amount of files as the “total”. Note that the first parameter will be a floating point number to indicate progress within a file (e.g. 2.5 means it has sent 50% of the third file, because it’s between 2 and 3).

Returns
InputFileBig if the file size is larger than 10MB, InputSizedFile (subclass of InputFile) otherwise.
Example
# Photos as photo and document
file = await client.upload_file('photo.jpg')
await client.send_file(chat, file)                       # sends as photo
await client.send_file(chat, file, force_document=True)  # sends as document

file.name = 'not a photo.jpg'
await client.send_file(chat, file, force_document=True)  # document, new name

# As song or as voice note
file = await client.upload_file('song.ogg')
await client.send_file(chat, file)                   # sends as song
await client.send_file(chat, file, voice_note=True)  # sends as voice note
class telethon.client.users.UserMethods

Bases: object

__call__(request, ordered=False, flood_sleep_threshold=None)

Call self as a function.

__weakref__

list of weak references to the object (if defined)

get_entity(entity: hints.EntitiesLike) → hints.Entity

Turns the given entity into a valid Telegram User, Chat or Channel. You can also pass a list or iterable of entities, and they will be efficiently fetched from the network.

Arguments
entity (str | int | Peer | InputPeer):

If a username is given, the username will be resolved making an API call every time. Resolving usernames is an expensive operation and will start hitting flood waits around 50 usernames in a short period of time.

If you want to get the entity for a cached username, you should first get_input_entity(username) which will use the cache), and then use get_entity with the result of the previous call.

Similar limits apply to invite links, and you should use their ID instead.

Using phone numbers (from people in your contact list), exact names, integer IDs or Peer rely on a get_input_entity first, which in turn needs the entity to be in cache, unless a InputPeer was passed.

Unsupported types will raise TypeError.

If the entity can’t be found, ValueError will be raised.

Returns
User, Chat or Channel corresponding to the input entity. A list will be returned if more than one was given.
Example
from telethon import utils

me = await client.get_entity('me')
print(utils.get_display_name(me))

chat = await client.get_input_entity('username')
async for message in client.iter_messages(chat):
    ...

# Note that you could have used the username directly, but it's
# good to use get_input_entity if you will reuse it a lot.
async for message in client.iter_messages('username'):
    ...

# Note that for this to work the phone number must be in your contacts
some_id = await client.get_peer_id('+34123456789')
get_input_entity(peer: hints.EntityLike) → types.TypeInputPeer

Turns the given entity into its input entity version.

Most requests use this kind of InputPeer, so this is the most suitable call to make for those cases. Generally you should let the library do its job and don’t worry about getting the input entity first, but if you’re going to use an entity often, consider making the call:

Arguments
entity (str | int | Peer | InputPeer):

If a username or invite link is given, the library will use the cache. This means that it’s possible to be using a username that changed or an old invite link (this only happens if an invite link for a small group chat is used after it was upgraded to a mega-group).

If the username or ID from the invite link is not found in the cache, it will be fetched. The same rules apply to phone numbers ('+34 123456789') from people in your contact list.

If an exact name is given, it must be in the cache too. This is not reliable as different people can share the same name and which entity is returned is arbitrary, and should be used only for quick tests.

If a positive integer ID is given, the entity will be searched in cached users, chats or channels, without making any call.

If a negative integer ID is given, the entity will be searched exactly as either a chat (prefixed with -) or as a channel (prefixed with -100).

If a Peer is given, it will be searched exactly in the cache as either a user, chat or channel.

If the given object can be turned into an input entity directly, said operation will be done.

Unsupported types will raise TypeError.

If the entity can’t be found, ValueError will be raised.

Returns

InputPeerUser, InputPeerChat or InputPeerChannel or InputPeerSelf if the parameter is 'me' or 'self'.

If you need to get the ID of yourself, you should use get_me with input_peer=True) instead.

Example
# If you're going to use "username" often in your code
# (make a lot of calls), consider getting its input entity
# once, and then using the "user" everywhere instead.
user = await client.get_input_entity('username')

# The same applies to IDs, chats or channels.
chat = await client.get_input_entity(-123456789)
get_me(input_peer: bool = False) → typing.Union[types.User, types.InputPeerUser]

Gets “me”, the current User who is logged in.

If the user has not logged in yet, this method returns None.

Arguments
input_peer (bool, optional):
Whether to return the InputPeerUser version or the normal User. This can be useful if you just need to know the ID of yourself.
Returns
Your own User.
Example
me = await client.get_me()
print(me.username)
get_peer_id(peer: hints.EntityLike, add_mark: bool = True) → int

Gets the ID for the given entity.

This method needs to be async because peer supports usernames, invite-links, phone numbers (from people in your contact list), etc.

If add_mark is False, then a positive ID will be returned instead. By default, bot-API style IDs (signed) are returned.

Example
print(await client.get_peer_id('me'))
is_bot() → bool

Return True if the signed-in user is a bot, False otherwise.

Example
if await client.is_bot():
    print('Beep')
else:
    print('Hello')
is_user_authorized() → bool

Returns True if the user is authorized (logged in).

Example
if not await client.is_user_authorized():
    await client.send_code_request(phone)
    code = input('enter code: ')
    await client.sign_in(phone, code)

Update Events

Every event (builder) subclasses common.EventBuilder, so all the methods in it can be used from any event builder/event instance.

class telethon.events.common.EventBuilder(chats=None, *, blacklist_chats=False, func=None)

Bases: abc.ABC

The common event builder, with builtin support to filter per chat.

Args:
chats (entity, optional):
May be one or more entities (username/peer/etc.), preferably IDs. By default, only matching chats will be handled.
blacklist_chats (bool, optional):
Whether to treat the chats as a blacklist instead of as a whitelist (default). This means that every chat will be handled except those specified in chats which will be ignored if blacklist_chats=True.
func (callable, optional):

A callable (async or not) function that should accept the event as input parameter, and return a value indicating whether the event should be dispatched or not (any truthy value will do, it does not need to be a bool). It works like a custom filter:

@client.on(events.NewMessage(func=lambda e: e.is_private))
async def handler(event):
    pass  # code here
__weakref__

list of weak references to the object (if defined)

classmethod build(update, others=None, self_id=None)

Builds an event for the given update if possible, or returns None.

others are the rest of updates that came in the same container as the current update.

self_id should be the current user’s ID, since it is required for some events which lack this information but still need it.

filter(event)

Returns a truthy value if the event passed the filter and should be used, or falsy otherwise. The return value may need to be awaited.

The events must have been resolved before this can be called.

resolve(client)

Helper method to allow event builders to be resolved before usage

class telethon.events.common.EventCommon(chat_peer=None, msg_id=None, broadcast=None)

Bases: telethon.tl.custom.chatgetter.ChatGetter, abc.ABC

Intermediate class with common things to all events.

Remember that this class implements ChatGetter which means you have access to all chat properties and methods.

In addition, you can access the original_update field which contains the original Update.

__str__()

Return str(self).

client

The telethon.TelegramClient that created this event.

stringify()
to_dict()
telethon.events.common.name_inner_event(cls)

Decorator to rename cls.Event ‘Event’ as ‘cls.Event’

class telethon.events.newmessage.NewMessage(chats=None, *, blacklist_chats=False, func=None, incoming=None, outgoing=None, from_users=None, forwards=None, pattern=None)

Bases: telethon.events.common.EventBuilder

Occurs whenever a new text message or a message with media arrives.

Args:
incoming (bool, optional):
If set to True, only incoming messages will be handled. Mutually exclusive with outgoing (can only set one of either).
outgoing (bool, optional):
If set to True, only outgoing messages will be handled. Mutually exclusive with incoming (can only set one of either).
from_users (entity, optional):
Unlike chats, this parameter filters the senders of the message. That is, only messages sent by these users will be handled. Use chats if you want private messages with this/these users. from_users lets you filter by messages sent by one or more users across the desired chats (doesn’t need a list).
forwards (bool, optional):
Whether forwarded messages should be handled or not. By default, both forwarded and normal messages are included. If it’s True only forwards will be handled. If it’s False only messages that are not forwards will be handled.
pattern (str, callable, Pattern, optional):
If set, only messages matching this pattern will be handled. You can specify a regex-like string which will be matched against the message, a callable function that returns True if a message is acceptable, or a compiled regex pattern.
Example
import asyncio
from telethon import events

@client.on(events.NewMessage(pattern='(?i)hello.+'))
async def handler(event):
    # Respond whenever someone says "Hello" and something else
    await event.reply('Hey!')

@client.on(events.NewMessage(outgoing=True, pattern='!ping'))
async def handler(event):
    # Say "!pong" whenever you send "!ping", then delete both messages
    m = await event.respond('!pong')
    await asyncio.sleep(5)
    await client.delete_messages(event.chat_id, [event.id, m.id])
class Event(message)

Bases: telethon.events.common.EventCommon

Represents the event of a new message. This event can be treated to all effects as a Message, so please refer to its documentation to know what you can do with this event.

Members:
message (Message):

This is the only difference with the received Message, and will return the telethon.tl.custom.message.Message itself, not the text.

See Message for the rest of available members and methods.

pattern_match (obj):

The resulting object from calling the passed pattern function. Here’s an example using a string (defaults to regex match):

>>> from telethon import TelegramClient, events
>>> client = TelegramClient(...)
>>>
>>> @client.on(events.NewMessage(pattern=r'hi (\w+)!'))
... async def handler(event):
...     # In this case, the result is a ``Match`` object
...     # since the `str` pattern was converted into
...     # the ``re.compile(pattern).match`` function.
...     print('Welcomed', event.pattern_match.group(1))
...
>>>
__getattr__(item)
__setattr__(name, value)

Implement setattr(self, name, value).

classmethod build(update, others=None, self_id=None)

Builds an event for the given update if possible, or returns None.

others are the rest of updates that came in the same container as the current update.

self_id should be the current user’s ID, since it is required for some events which lack this information but still need it.

filter(event)

Returns a truthy value if the event passed the filter and should be used, or falsy otherwise. The return value may need to be awaited.

The events must have been resolved before this can be called.

class telethon.events.chataction.ChatAction(chats=None, *, blacklist_chats=False, func=None)

Bases: telethon.events.common.EventBuilder

Occurs on certain chat actions:

  • Whenever a new chat is created.
  • Whenever a chat’s title or photo is changed or removed.
  • Whenever a new message is pinned.
  • Whenever a user scores in a game.
  • Whenever a user joins or is added to the group.
  • Whenever a user is removed or leaves a group if it has less than 50 members or the removed user was a bot.

Note that “chat” refers to “small group, megagroup and broadcast channel”, whereas “group” refers to “small group and megagroup” only.

Example
from telethon import events

@client.on(events.ChatAction)
async def handler(event):
    # Welcome every new user
    if event.user_joined:
        await event.reply('Welcome to the group!')
class Event(where, new_photo=None, added_by=None, kicked_by=None, created=None, users=None, new_title=None, pin_ids=None, pin=None, new_score=None)

Bases: telethon.events.common.EventCommon

Represents the event of a new chat action.

Members:
action_message (MessageAction):
The message invoked by this Chat Action.
new_pin (bool):
True if there is a new pin.
new_photo (bool):
True if there’s a new chat photo (or it was removed).
photo (Photo, optional):
The new photo (or None if it was removed).
user_added (bool):
True if the user was added by some other.
user_joined (bool):
True if the user joined on their own.
user_left (bool):
True if the user left on their own.
user_kicked (bool):
True if the user was kicked by some other.
created (bool, optional):
True if this chat was just created.
new_title (str, optional):
The new title string for the chat, if applicable.
new_score (str, optional):
The new score string for the game, if applicable.
unpin (bool):
True if the existing pin gets unpinned.
added_by

The user who added users, if applicable (None otherwise).

delete(*args, **kwargs)

Deletes the chat action message. You’re responsible for checking whether you have the permission to do so, or to except the error otherwise. Shorthand for telethon.client.messages.MessageMethods.delete_messages with entity and message_ids already set.

Does nothing if no message action triggered this event.

get_added_by()

Returns added_by but will make an API call if necessary.

get_input_user()

Returns input_user but will make an API call if necessary.

get_input_users()

Returns input_users but will make an API call if necessary.

get_kicked_by()

Returns kicked_by but will make an API call if necessary.

get_pinned_message()

If new_pin is True, this returns the Message object that was pinned.

get_pinned_messages()

If new_pin is True, this returns a list of Message objects that were pinned.

get_user()

Returns user but will make an API call if necessary.

get_users()

Returns users but will make an API call if necessary.

input_user

Input version of the self.user property.

input_users

Input version of the self.users property.

kicked_by

The user who kicked users, if applicable (None otherwise).

reply(*args, **kwargs)

Replies to the chat action message (as a reply). Shorthand for telethon.client.messages.MessageMethods.send_message with both entity and reply_to already set.

Has the same effect as respond if there is no message.

respond(*args, **kwargs)

Responds to the chat action message (not as a reply). Shorthand for telethon.client.messages.MessageMethods.send_message with entity already set.

user

The first user that takes part in this action. For example, who joined.

Might be None if the information can’t be retrieved or there is no user taking part.

user_id

Returns the marked signed ID of the first user, if any.

user_ids

Returns the marked signed ID of the users, if any.

users

A list of users that take part in this action. For example, who joined.

Might be empty if the information can’t be retrieved or there are no users taking part.

classmethod build(update, others=None, self_id=None)

Builds an event for the given update if possible, or returns None.

others are the rest of updates that came in the same container as the current update.

self_id should be the current user’s ID, since it is required for some events which lack this information but still need it.

class telethon.events.userupdate.UserUpdate(chats=None, *, blacklist_chats=False, func=None)

Bases: telethon.events.common.EventBuilder

Occurs whenever a user goes online, starts typing, etc.

Example
from telethon import events

@client.on(events.UserUpdate)
async def handler(event):
    # If someone is uploading, say something
    if event.uploading:
        await client.send_message(event.user_id, 'What are you sending?')
class Event(peer, *, status=None, chat_peer=None, typing=None)

Bases: telethon.events.common.EventCommon, telethon.tl.custom.sendergetter.SenderGetter

Represents the event of a user update such as gone online, started typing, etc.

Members:
status (UserStatus, optional):

The user status if the update is about going online or offline.

You should check this attribute first before checking any of the seen within properties, since they will all be None if the status is not set.

action (SendMessageAction, optional):

The “typing” action if any the user is performing if any.

You should check this attribute first before checking any of the typing properties, since they will all be None if the action is not set.

audio

True if what’s being recorded/uploaded is an audio.

cancel

True if the action was cancelling other actions.

contact

True if what’s being uploaded (selected) is a contact.

document

True if what’s being uploaded is document.

geo

True if what’s being uploaded is a geo.

get_input_user()

Alias for get_input_sender.

get_user()

Alias for get_sender.

input_user

Alias for input_sender.

last_seen

Exact datetime.datetime when the user was last seen if known.

online

True if the user is currently online,

photo

True if what’s being uploaded is a photo.

playing

True if the action is playing a game.

recently

True if the user was seen within a day.

recording

True if the action is recording something.

round

True if what’s being recorded/uploaded is a round video.

sticker

True if what’s being uploaded is a sticker.

typing

True if the action is typing a message.

until

The datetime.datetime until when the user should appear online.

uploading

True if the action is uploading something.

user

Alias for sender.

user_id

Alias for sender_id.

video

True if what’s being recorded/uploaded is an video.

within_months

True if the user was seen within 30 days.

within_weeks

True if the user was seen within 7 days.

classmethod build(update, others=None, self_id=None)

Builds an event for the given update if possible, or returns None.

others are the rest of updates that came in the same container as the current update.

self_id should be the current user’s ID, since it is required for some events which lack this information but still need it.

class telethon.events.messageedited.MessageEdited(chats=None, *, blacklist_chats=False, func=None, incoming=None, outgoing=None, from_users=None, forwards=None, pattern=None)

Bases: telethon.events.newmessage.NewMessage

Occurs whenever a message is edited. Just like NewMessage, you should treat this event as a Message.

Warning

On channels, Message.out will be True if you sent the message originally, not if you edited it! This can be dangerous if you run outgoing commands on edits.

Some examples follow:

  • You send a message “A”, out is True.
  • You edit “A” to “B”, out is True.
  • Someone else edits “B” to “C”, out is True (be careful!).
  • Someone sends “X”, out is False.
  • Someone edits “X” to “Y”, out is False.
  • You edit “Y” to “Z”, out is False.

Since there are useful cases where you need the right out value, the library cannot do anything automatically to help you. Instead, consider using from_users='me' (it won’t work in broadcast channels at all since the sender is the channel and not you).

Example
from telethon import events

@client.on(events.MessageEdited)
async def handler(event):
    # Log the date of new edits
    print('Message', event.id, 'changed at', event.date)
class Event(message)

Bases: telethon.events.newmessage.Event

classmethod build(update, others=None, self_id=None)

Builds an event for the given update if possible, or returns None.

others are the rest of updates that came in the same container as the current update.

self_id should be the current user’s ID, since it is required for some events which lack this information but still need it.

class telethon.events.messagedeleted.MessageDeleted(chats=None, *, blacklist_chats=False, func=None)

Bases: telethon.events.common.EventBuilder

Occurs whenever a message is deleted. Note that this event isn’t 100% reliable, since Telegram doesn’t always notify the clients that a message was deleted.

Important

Telegram does not send information about where a message was deleted if it occurs in private conversations with other users or in small group chats, because message IDs are unique and you can identify the chat with the message ID alone if you saved it previously.

Telethon does not save information of where messages occur, so it cannot know in which chat a message was deleted (this will only work in channels, where the channel ID is present).

This means that the chats= parameter will not work reliably, unless you intend on working with channels and super-groups only.

Example
from telethon import events

@client.on(events.MessageDeleted)
async def handler(event):
    # Log all deleted message IDs
    for msg_id in event.deleted_ids:
        print('Message', msg_id, 'was deleted in', event.chat_id)
class Event(deleted_ids, peer)

Bases: telethon.events.common.EventCommon

classmethod build(update, others=None, self_id=None)

Builds an event for the given update if possible, or returns None.

others are the rest of updates that came in the same container as the current update.

self_id should be the current user’s ID, since it is required for some events which lack this information but still need it.

class telethon.events.messageread.MessageRead(chats=None, *, blacklist_chats=False, func=None, inbox=False)

Bases: telethon.events.common.EventBuilder

Occurs whenever one or more messages are read in a chat.

Args:
inbox (bool, optional):
If this argument is True, then when you read someone else’s messages the event will be fired. By default (False) only when messages you sent are read by someone else will fire it.
Example
from telethon import events

@client.on(events.MessageRead)
async def handler(event):
    # Log when someone reads your messages
    print('Someone has read all your messages until', event.max_id)

@client.on(events.MessageRead(inbox=True))
async def handler(event):
    # Log when you read message in a chat (from your "inbox")
    print('You have read messages until', event.max_id)
class Event(peer=None, max_id=None, out=False, contents=False, message_ids=None)

Bases: telethon.events.common.EventCommon

Represents the event of one or more messages being read.

Members:
max_id (int):
Up to which message ID has been read. Every message with an ID equal or lower to it have been read.
outbox (bool):
True if someone else has read your messages.
contents (bool):
True if what was read were the contents of a message. This will be the case when e.g. you play a voice note. It may only be set on inbox events.
__contains__(message)

True if the message(s) are read message.

get_messages()

Returns the list of Message which contents’ were read.

Use is_read() if you need to check whether a message was read instead checking if it’s in here.

inbox

True if you have read someone else’s messages.

is_read(message)

Returns True if the given message (or its ID) has been read.

If a list-like argument is provided, this method will return a list of booleans indicating which messages have been read.

message_ids

The IDs of the messages which contents’ were read.

Use is_read() if you need to check whether a message was read instead checking if it’s in here.

classmethod build(update, others=None, self_id=None)

Builds an event for the given update if possible, or returns None.

others are the rest of updates that came in the same container as the current update.

self_id should be the current user’s ID, since it is required for some events which lack this information but still need it.

filter(event)

Returns a truthy value if the event passed the filter and should be used, or falsy otherwise. The return value may need to be awaited.

The events must have been resolved before this can be called.

class telethon.events.callbackquery.CallbackQuery(chats=None, *, blacklist_chats=False, func=None, data=None, pattern=None)

Bases: telethon.events.common.EventBuilder

Occurs whenever you sign in as a bot and a user clicks one of the inline buttons on your messages.

Note that the chats parameter will not work with normal IDs or peers if the clicked inline button comes from a “via bot” message. The chats parameter also supports checking against the chat_instance which should be used for inline callbacks.

Args:
data (bytes, str, callable, optional):
If set, the inline button payload data must match this data. A UTF-8 string can also be given, a regex or a callable. For instance, to check against 'data_1' and 'data_2' you can use re.compile(b'data_').
pattern (bytes, str, callable, Pattern, optional):
If set, only buttons with payload matching this pattern will be handled. You can specify a regex-like string which will be matched against the payload data, a callable function that returns True if a the payload data is acceptable, or a compiled regex pattern.
Example
from telethon import events, Button

# Handle all callback queries and check data inside the handler
@client.on(events.CallbackQuery)
async def handler(event):
    if event.data == b'yes':
        await event.answer('Correct answer!')

# Handle only callback queries with data being b'no'
@client.on(events.CallbackQuery(data=b'no'))
async def handler(event):
    # Pop-up message with alert
    await event.answer('Wrong answer!', alert=True)

# Send a message with buttons users can click
async def main():
    await client.send_message(user, 'Yes or no?', buttons=[
        Button.inline('Yes!', b'yes'),
        Button.inline('Nope', b'no')
    ])
class Event(query, peer, msg_id)

Bases: telethon.events.common.EventCommon, telethon.tl.custom.sendergetter.SenderGetter

Represents the event of a new callback query.

Members:
query (UpdateBotCallbackQuery):
The original UpdateBotCallbackQuery.
data_match (obj, optional):
The object returned by the data= parameter when creating the event builder, if any. Similar to pattern_match for the new message event.
pattern_match (obj, optional):
Alias for data_match.
answer(message=None, cache_time=0, *, url=None, alert=False)

Answers the callback query (and stops the loading circle).

Args:
message (str, optional):
The toast message to show feedback to the user.
cache_time (int, optional):
For how long this result should be cached on the user’s client. Defaults to 0 for no cache.
url (str, optional):
The URL to be opened in the user’s client. Note that the only valid URLs are those of games your bot has, or alternatively a ‘t.me/your_bot?start=xyz’ parameter.
alert (bool, optional):
Whether an alert (a pop-up dialog) should be used instead of showing a toast. Defaults to False.
chat_instance

Unique identifier for the chat where the callback occurred. Useful for high scores in games.

data

Returns the data payload from the original inline button.

delete(*args, **kwargs)

Deletes the message. Shorthand for telethon.client.messages.MessageMethods.delete_messages with entity and message_ids already set.

If you need to delete more than one message at once, don’t use this delete method. Use a telethon.client.telegramclient.TelegramClient instance directly.

This method also creates a task to answer the callback.

This method will likely fail if via_inline is True.

edit(*args, **kwargs)

Edits the message. Shorthand for telethon.client.messages.MessageMethods.edit_message with the entity set to the correct InputBotInlineMessageID or InputBotInlineMessageID64.

Returns True if the edit was successful.

This method also creates a task to answer the callback.

Note

This method won’t respect the previous message unlike Message.edit, since the message object is normally not present.

get_message()

Returns the message to which the clicked inline button belongs.

id

Returns the query ID. The user clicking the inline button is the one who generated this random ID.

message_id

Returns the message ID to which the clicked inline button belongs.

reply(*args, **kwargs)

Replies to the message (as a reply). Shorthand for telethon.client.messages.MessageMethods.send_message with both entity and reply_to already set.

This method also creates a task to answer the callback.

This method will likely fail if via_inline is True.

respond(*args, **kwargs)

Responds to the message (not as a reply). Shorthand for telethon.client.messages.MessageMethods.send_message with entity already set.

This method also creates a task to answer the callback.

This method will likely fail if via_inline is True.

via_inline

Whether this callback was generated from an inline button sent via an inline query or not. If the bot sent the message itself with buttons, and one of those is clicked, this will be False. If a user sent the message coming from an inline query to the bot, and one of those is clicked, this will be True.

If it’s True, it’s likely that the bot is not in the chat, so methods like respond or delete won’t work (but edit will always work).

classmethod build(update, others=None, self_id=None)

Builds an event for the given update if possible, or returns None.

others are the rest of updates that came in the same container as the current update.

self_id should be the current user’s ID, since it is required for some events which lack this information but still need it.

filter(event)

Returns a truthy value if the event passed the filter and should be used, or falsy otherwise. The return value may need to be awaited.

The events must have been resolved before this can be called.

class telethon.events.inlinequery.InlineQuery(users=None, *, blacklist_users=False, func=None, pattern=None)

Bases: telethon.events.common.EventBuilder

Occurs whenever you sign in as a bot and a user sends an inline query such as @bot query.

Args:
users (entity, optional):
May be one or more entities (username/peer/etc.), preferably IDs. By default, only inline queries from these users will be handled.
blacklist_users (bool, optional):
Whether to treat the users as a blacklist instead of as a whitelist (default). This means that every chat will be handled except those specified in users which will be ignored if blacklist_users=True.
pattern (str, callable, Pattern, optional):
If set, only queries matching this pattern will be handled. You can specify a regex-like string which will be matched against the message, a callable function that returns True if a message is acceptable, or a compiled regex pattern.
Example
from telethon import events

@client.on(events.InlineQuery)
async def handler(event):
    builder = event.builder

    # Two options (convert user text to UPPERCASE or lowercase)
    await event.answer([
        builder.article('UPPERCASE', text=event.text.upper()),
        builder.article('lowercase', text=event.text.lower()),
    ])
class Event(query)

Bases: telethon.events.common.EventCommon, telethon.tl.custom.sendergetter.SenderGetter

Represents the event of a new callback query.

Members:
query (UpdateBotInlineQuery):

The original UpdateBotInlineQuery.

Make sure to access the text property of the query if you want the text rather than the actual query object.

pattern_match (obj, optional):
The resulting object from calling the passed pattern function, which is re.compile(...).match by default.
answer(results=None, cache_time=0, *, gallery=False, next_offset=None, private=False, switch_pm=None, switch_pm_param='')

Answers the inline query with the given results.

See the documentation for builder to know what kind of answers can be given.

Args:
results (list, optional):

A list of InputBotInlineResult to use. You should use builder to create these:

builder = inline.builder
r1 = builder.article('Be nice', text='Have a nice day')
r2 = builder.article('Be bad', text="I don't like you")
await inline.answer([r1, r2])

You can send up to 50 results as documented in https://core.telegram.org/bots/api#answerinlinequery. Sending more will raise ResultsTooMuchError, and you should consider using next_offset to paginate them.

cache_time (int, optional):
For how long this result should be cached on the user’s client. Defaults to 0 for no cache.
gallery (bool, optional):
Whether the results should show as a gallery (grid) or not.
next_offset (str, optional):
The offset the client will send when the user scrolls the results and it repeats the request.
private (bool, optional):
Whether the results should be cached by Telegram (not private) or by the user’s client (private).
switch_pm (str, optional):
If set, this text will be shown in the results to allow the user to switch to private messages.
switch_pm_param (str, optional):
Optional parameter to start the bot with if switch_pm was used.

Example:

@bot.on(events.InlineQuery)
async def handler(event):
    builder = event.builder

    rev_text = event.text[::-1]
    await event.answer([
        builder.article('Reverse text', text=rev_text),
        builder.photo('/path/to/photo.jpg')
    ])
builder

Returns a new InlineBuilder instance.

geo

If the user location is requested when using inline mode and the user’s device is able to send it, this will return the GeoPoint with the position of the user.

id

Returns the unique identifier for the query ID.

offset

The string the user’s client used as an offset for the query. This will either be empty or equal to offsets passed to answer.

text

Returns the text the user used to make the inline query.

classmethod build(update, others=None, self_id=None)

Builds an event for the given update if possible, or returns None.

others are the rest of updates that came in the same container as the current update.

self_id should be the current user’s ID, since it is required for some events which lack this information but still need it.

filter(event)

Returns a truthy value if the event passed the filter and should be used, or falsy otherwise. The return value may need to be awaited.

The events must have been resolved before this can be called.

class telethon.events.album.Album(chats=None, *, blacklist_chats=False, func=None)

Bases: telethon.events.common.EventBuilder

Occurs whenever you receive an album. This event only exists to ease dealing with an unknown amount of messages that belong to the same album.

Example
from telethon import events

@client.on(events.Album)
async def handler(event):
    # Counting how many photos or videos the album has
    print('Got an album with', len(event), 'items')

    # Forwarding the album as a whole to some chat
    event.forward_to(chat)

    # Printing the caption
    print(event.text)

    # Replying to the fifth item in the album
    await event.messages[4].reply('Cool!')
class Event(messages)

Bases: telethon.events.common.EventCommon, telethon.tl.custom.sendergetter.SenderGetter

Represents the event of a new album.

Members:
messages (Sequence[Message]):
The list of messages belonging to the same album.
__getitem__(n)

Access the n’th message in the album.

Equivalent to event.messages[n].

__iter__()

Iterate over the messages in the album.

Equivalent to iter(self.messages).

__len__()

Return the amount of messages in the album.

Equivalent to len(self.messages).

delete(*args, **kwargs)

Deletes the entire album. You’re responsible for checking whether you have the permission to do so, or to except the error otherwise. Shorthand for telethon.client.messages.MessageMethods.delete_messages with entity and message_ids already set.

edit(*args, **kwargs)

Edits the first caption or the message, or the first messages’ caption if no caption is set, iff it’s outgoing. Shorthand for telethon.client.messages.MessageMethods.edit_message with both entity and message already set.

Returns None if the message was incoming, or the edited Message otherwise.

Note

This is different from client.edit_message and will respect the previous state of the message. For example, if the message didn’t have a link preview, the edit won’t add one by default, and you should force it by setting it to True if you want it.

This is generally the most desired and convenient behaviour, and will work for link previews and message buttons.

forward

The Forward information for the first message in the album if it was forwarded.

forward_to(*args, **kwargs)

Forwards the entire album. Shorthand for telethon.client.messages.MessageMethods.forward_messages with both messages and from_peer already set.

get_reply_message()

The Message that this album is replying to, or None.

The result will be cached after its first use.

grouped_id

The shared grouped_id between all the messages.

is_reply

True if the album is a reply to some other message.

Remember that you can access the ID of the message this one is replying to through reply_to_msg_id, and the Message object with get_reply_message().

mark_read()

Marks the entire album as read. Shorthand for client.send_read_acknowledge() with both entity and message already set.

pin(*, notify=False)

Pins the first photo in the album. Shorthand for telethon.client.messages.MessageMethods.pin_message with both entity and message already set.

raw_text

The raw message text of the first photo with a caption, ignoring any formatting.

reply(*args, **kwargs)

Replies to the first photo in the album (as a reply). Shorthand for telethon.client.messages.MessageMethods.send_message with both entity and reply_to already set.

respond(*args, **kwargs)

Responds to the album (not as a reply). Shorthand for telethon.client.messages.MessageMethods.send_message with entity already set.

text

The message text of the first photo with a caption, formatted using the client’s default parse mode.

classmethod build(update, others=None, self_id=None)

Builds an event for the given update if possible, or returns None.

others are the rest of updates that came in the same container as the current update.

self_id should be the current user’s ID, since it is required for some events which lack this information but still need it.

filter(event)

Returns a truthy value if the event passed the filter and should be used, or falsy otherwise. The return value may need to be awaited.

The events must have been resolved before this can be called.

class telethon.events.album.AlbumHack(client, event)

Bases: object

When receiving an album from a different data-center, they will come in separate Updates, so we need to temporarily remember them for a while and only after produce the event.

Of course events are not designed for this kind of wizardy, so this is a dirty hack that gets the job done.

When cleaning up the code base we may want to figure out a better way to do this, or just leave the album problem to the users; the update handling code is bad enough as it is.

__weakref__

list of weak references to the object (if defined)

deliver_event()
extend(messages)
class telethon.events.raw.Raw(types=None, *, func=None)

Bases: telethon.events.common.EventBuilder

Raw events are not actual events. Instead, they are the raw Update object that Telegram sends. You normally shouldn’t need these.

Args:
types (list | tuple | type, optional):
The type or types that the Update instance must be. Equivalent to if not isinstance(update, types): return.
Example
from telethon import events

@client.on(events.Raw)
async def handler(update):
    # Print all incoming updates
    print(update.stringify())
classmethod build(update, others=None, self_id=None)

Builds an event for the given update if possible, or returns None.

others are the rest of updates that came in the same container as the current update.

self_id should be the current user’s ID, since it is required for some events which lack this information but still need it.

filter(event)

Returns a truthy value if the event passed the filter and should be used, or falsy otherwise. The return value may need to be awaited.

The events must have been resolved before this can be called.

resolve(client)

Helper method to allow event builders to be resolved before usage

exception telethon.events.StopPropagation

Bases: Exception

If this exception is raised in any of the handlers for a given event, it will stop the execution of all other registered event handlers. It can be seen as the StopIteration in a for loop but for events.

Example usage:

>>> from telethon import TelegramClient, events
>>> client = TelegramClient(...)
>>>
>>> @client.on(events.NewMessage)
... async def delete(event):
...     await event.delete()
...     # No other event handler will have a chance to handle this event
...     raise StopPropagation
...
>>> @client.on(events.NewMessage)
... async def _(event):
...     # Will never be reached, because it is the second handler
...     pass
__weakref__

list of weak references to the object (if defined)

telethon.events.is_handler(callback)

Returns True if the given callback is an event handler (i.e. you used register on it).

telethon.events.list(callback)

Returns a list containing the registered event builders inside the specified callback handler.

telethon.events.register(event=None)

Decorator method to register event handlers. This is the client-less add_event_handler() variant.

Note that this method only registers callbacks as handlers, and does not attach them to any client. This is useful for external modules that don’t have access to the client, but still want to define themselves as a handler. Example:

>>> from telethon import events
>>> @events.register(events.NewMessage)
... async def handler(event):
...     ...
...
>>> # (somewhere else)
...
>>> from telethon import TelegramClient
>>> client = TelegramClient(...)
>>> client.add_event_handler(handler)

Remember that you can use this as a non-decorator through register(event)(callback).

Args:
event (_EventBuilder | type):
The event builder class or instance to be used, for instance events.NewMessage.
telethon.events.unregister(callback, event=None)

Inverse operation of register (though not a decorator). Client-less remove_event_handler variant. Note that this won’t remove handlers from the client, because it simply can’t, so you would generally use this before adding the handlers to the client.

This method is here for symmetry. You will rarely need to unregister events, since you can simply just not add them to any client.

If no event is given, all events for this callback are removed. Returns how many callbacks were removed.

Custom package

The telethon.tl.custom package contains custom classes that the library uses in order to make working with Telegram easier. Only those that you are supposed to use will be documented here. You can use undocumented ones at your own risk.

More often than not, you don’t need to import these (unless you want type hinting), nor do you need to manually create instances of these classes. They are returned by client methods.

AdminLogEvent

class telethon.tl.custom.adminlogevent.AdminLogEvent(original, entities)

Bases: object

Represents a more friendly interface for admin log events.

Members:
original (ChannelAdminLogEvent):
The original ChannelAdminLogEvent.
entities (dict):

A dictionary mapping user IDs to User.

When old and new are ChannelParticipant, you can use this dictionary to map the user_id, kicked_by, inviter_id and promoted_by IDs to their User.

user (User):
The user that caused this action (entities[original.user_id]).
input_user (InputPeerUser):
Input variant of user.
__str__()

Return str(self).

__weakref__

list of weak references to the object (if defined)

action

The original ChannelAdminLogEventAction.

changed_about

Whether the channel’s about was changed or not.

If True, old and new will be present as str.

changed_admin

Whether the permissions for an admin in this channel changed or not.

If True, old and new will be present as ChannelParticipant.

changed_call_settings

Whether the group call settings were changed or not.

If True, new will be True if new users are muted on join.

changed_default_banned_rights

Whether the default banned rights were changed or not.

If True, old and new will be present as ChatBannedRights.

changed_hide_history

Whether hiding the previous message history for new members in the channel was toggled or not.

If True, old and new will be present as bool.

changed_history_ttl

Whether the Time To Live of the message history has changed.

Messages sent after this change will have a ttl_period in seconds indicating how long they should live for before being auto-deleted.

If True, old will be the old TTL, and new the new TTL, in seconds.

changed_invites

Whether the invites in the channel were toggled or not.

If True, old and new will be present as bool.

changed_location

Whether the location setting of the channel has changed or not.

If True, old and new will be present as ChannelLocation.

changed_message

Whether a message in this channel was edited or not.

If True, old and new will be present as Message.

changed_photo

Whether the channel’s photo was changed or not.

If True, old and new will be present as Photo.

changed_pin

Whether a new message in this channel was pinned or not.

If True, new will be present as Message.

changed_restrictions

Whether a message in this channel was edited or not.

If True, old and new will be present as ChannelParticipant.

changed_signatures

Whether the message signatures in the channel were toggled or not.

If True, old and new will be present as bool.

changed_sticker_set

Whether the channel’s sticker set was changed or not.

If True, old and new will be present as InputStickerSet.

changed_title

Whether the channel’s title was changed or not.

If True, old and new will be present as str.

changed_user_volume

Whether a participant’s volume in a call has been changed.

If True, new will be the updated GroupCallParticipant.

changed_username

Whether the channel’s username was changed or not.

If True, old and new will be present as str.

date

The date when this event occurred.

deleted_exported_invite

Whether the exported chat invite has been deleted.

If True, old will be the deleted ExportedChatInvite.

deleted_message

Whether a message in this channel was deleted or not.

If True, old will be present as Message.

discarded_group_call

Whether a group call was started or not.

If True, old will be present as InputGroupCall.

edited_exported_invite

Whether the exported chat invite has been edited.

If True, old and new will be the old and new ExportedChatInvite, respectively.

id

The ID of this event.

joined

Whether user joined through the channel’s public username or not.

joined_by_invite

Whether a new participant has joined with the use of an invite link.

If True, old will be pre-existing (old) ExportedChatInvite used to join.

joined_invite

Whether a new user joined through an invite link to the channel or not.

If True, new will be present as ChannelParticipant.

left

Whether user left the channel or not.

new

The new value present in the event.

old

The old value from the event.

revoked_exported_invite

Whether the exported chat invite has been revoked.

If True, old will be the revoked ExportedChatInvite.

started_group_call

Whether a group call was started or not.

If True, new will be present as InputGroupCall.

stopped_poll

Whether a poll was stopped or not.

If True, new will be present as Message.

stringify()
user_id

The ID of the user that triggered this event.

user_muted

Whether a participant was muted in the ongoing group call or not.

If True, new will be present as GroupCallParticipant.

user_unmutted

Whether a participant was unmuted from the ongoing group call or not.

If True, new will be present as GroupCallParticipant.

Button

class telethon.tl.custom.button.Button(button, *, resize, single_use, selective)

Bases: object

Note

This class is used to define reply markups, e.g. when sending a message or replying to events. When you access Message.buttons they are actually MessageButton, so you might want to refer to that class instead.

Helper class to allow defining reply_markup when sending a message with inline or keyboard buttons.

You should make use of the defined class methods to create button instances instead making them yourself (i.e. don’t do Button(...) but instead use methods line Button.inline(...) etc.

You can use inline, switch_inline, url, auth, buy and game together to create inline buttons (under the message).

You can use text, request_location, request_phone and request_poll together to create a reply markup (replaces the user keyboard). You can also configure the aspect of the reply with these. The latest message with a reply markup will be the one shown to the user (messages contain the buttons, not the chat itself).

You cannot mix the two type of buttons together, and it will error if you try to do so.

The text for all buttons may be at most 142 characters. If more characters are given, Telegram will cut the text to 128 characters and add the ellipsis (…) character as the 129.

__weakref__

list of weak references to the object (if defined)

static auth(text, url=None, *, bot=None, write_access=False, fwd_text=None)

Creates a new inline button to authorize the user at the given URL.

You should set the url to be on the same domain as the one configured for the desired bot via @BotFather using the /setdomain command.

For more information about letting the user login via Telegram to a certain domain, see https://core.telegram.org/widgets/login.

If no url is specified, it will default to text.

Args:
bot (hints.EntityLike):

The bot that requires this authorization. By default, this is the bot that is currently logged in (itself), although you may pass a different input peer.

Note

For now, you cannot use ID or username for this argument. If you want to use a different bot than the one currently logged in, you must manually use client.get_input_entity().

write_access (bool):
Whether write access is required or not. This is False by default (read-only access).
fwd_text (str):
The new text to show in the button if the message is forwarded. By default, the button text will be the same.

When the user clicks this button, a confirmation box will be shown to the user asking whether they want to login to the specified domain.

static buy(text)

Creates a new inline button to buy a product.

This can only be used when sending files of type InputMediaInvoice, and must be the first button.

If the button is not specified, Telegram will automatically add the button to the message. See the Payments API documentation for more information.

static clear(selective=None)
Clears all keyboard buttons after sending a message with this markup. When used, no other button should be present or it will be ignored.

selective is as documented in text.

static force_reply(single_use=None, selective=None, placeholder=None)

Forces a reply to the message with this markup. If used, no other button should be present or it will be ignored.

single_use and selective are as documented in text.

Args:
placeholder (str):

text to show the user at typing place of message.

If the placeholder is too long, Telegram applications will crop the text (for example, to 64 characters and adding an ellipsis (…) character as the 65th).

static game(text)

Creates a new inline button to start playing a game.

This should be used when sending files of type InputMediaGame, and must be the first button.

See the Games documentation for more information on using games.

static inline(text, data=None)

Creates a new inline button with some payload data in it.

If data is omitted, the given text will be used as data. In any case data should be either bytes or str.

Note that the given data must be less or equal to 64 bytes. If more than 64 bytes are passed as data, ValueError is raised. If you need to store more than 64 bytes, consider saving the real data in a database and a reference to that data inside the button.

When the user clicks this button, events.CallbackQuery will trigger with the same data that the button contained, so that you can determine which button was pressed.

classmethod request_location(text, *, resize=None, single_use=None, selective=None)

Creates a new keyboard button to request the user’s location on click.

resize, single_use and selective are documented in text.

When the user clicks this button, a confirmation box will be shown to the user asking whether they want to share their location with the bot, and if confirmed a message with geo media will be sent.

classmethod request_phone(text, *, resize=None, single_use=None, selective=None)

Creates a new keyboard button to request the user’s phone on click.

resize, single_use and selective are documented in text.

When the user clicks this button, a confirmation box will be shown to the user asking whether they want to share their phone with the bot, and if confirmed a message with contact media will be sent.

classmethod request_poll(text, *, force_quiz=False, resize=None, single_use=None, selective=None)

Creates a new keyboard button to request the user to create a poll.

If force_quiz is False, the user will be allowed to choose whether they want their poll to be a quiz or not. Otherwise, the user will be forced to create a quiz when creating the poll.

If a poll is a quiz, there will be only one answer that is valid, and the votes cannot be retracted. Otherwise, users can vote and retract the vote, and the pol might be multiple choice.

resize, single_use and selective are documented in text.

When the user clicks this button, a screen letting the user create a poll will be shown, and if they do create one, the poll will be sent.

static switch_inline(text, query='', same_peer=False)

Creates a new inline button to switch to inline query.

If query is given, it will be the default text to be used when making the inline query.

If same_peer is True the inline query will directly be set under the currently opened chat. Otherwise, the user will have to select a different dialog to make the query.

When the user clicks this button, after a chat is selected, their input field will be filled with the username of your bot followed by the query text, ready to make inline queries.

classmethod text(text, *, resize=None, single_use=None, selective=None)

Creates a new keyboard button with the given text.

Args:
resize (bool):
If present, the entire keyboard will be reconfigured to be resized and be smaller if there are not many buttons.
single_use (bool):
If present, the entire keyboard will be reconfigured to be usable only once before it hides itself.
selective (bool):
If present, the entire keyboard will be reconfigured to be “selective”. The keyboard will be shown only to specific users. It will target users that are @mentioned in the text of the message or to the sender of the message you reply to.

When the user clicks this button, a text message with the same text as the button will be sent, and can be handled with events.NewMessage. You cannot distinguish between a button press and the user typing and sending exactly the same text on their own.

static url(text, url=None)

Creates a new inline button to open the desired URL on click.

If no url is given, the text will be used as said URL instead.

You cannot detect that the user clicked this button directly.

When the user clicks this button, a confirmation box will be shown to the user asking whether they want to open the displayed URL unless the domain is trusted, and once confirmed the URL will open in their device.

ChatGetter

class telethon.tl.custom.chatgetter.ChatGetter(chat_peer=None, *, input_chat=None, chat=None, broadcast=None)

Bases: abc.ABC

Helper base class that introduces the chat, input_chat and chat_id properties and get_chat and get_input_chat methods.

__weakref__

list of weak references to the object (if defined)

chat

Returns the User, Chat or Channel where this object belongs to. It may be None if Telegram didn’t send the chat.

If you only need the ID, use chat_id instead.

If you need to call a method which needs this chat, use input_chat instead.

If you’re using telethon.events, use get_chat() instead.

chat_id

Returns the marked chat integer ID. Note that this value will be different from peer_id for incoming private messages, since the chat to which the messages go is to your own person, but the chat itself is with the one who sent the message.

TL;DR; this gets the ID that you expect.

If there is a chat in the object, chat_id will always be set, which is why you should use it instead of chat.id.

get_chat()

Returns chat, but will make an API call to find the chat unless it’s already cached.

If you only need the ID, use chat_id instead.

If you need to call a method which needs this chat, use get_input_chat() instead.

get_input_chat()

Returns input_chat, but will make an API call to find the input chat unless it’s already cached.

input_chat

This InputPeer is the input version of the chat where the message was sent. Similarly to input_sender, this doesn’t have things like username or similar, but still useful in some cases.

Note that this might not be available if the library doesn’t have enough information available.

is_channel

True if the message was sent on a megagroup or channel.

is_group

True if the message was sent on a group or megagroup.

Returns None if there isn’t enough information (e.g. on events.MessageDeleted).

is_private

True if the message was sent as a private message.

Returns None if there isn’t enough information (e.g. on events.MessageDeleted).

Conversation

class telethon.tl.custom.conversation.Conversation(client, input_chat, *, timeout, total_timeout, max_messages, exclusive, replies_are_responses)

Bases: telethon.tl.custom.chatgetter.ChatGetter

Represents a conversation inside an specific chat.

A conversation keeps track of new messages since it was created until its exit and easily lets you query the current state.

If you need a conversation across two or more chats, you should use two conversations and synchronize them as you better see fit.

__aenter__()
__aexit__(exc_type, exc_val, exc_tb)
__enter__()

Helps to cut boilerplate on async context managers that offer synchronous variants.

__exit__(*args)
cancel()

Cancels the current conversation. Pending responses and subsequent calls to get a response will raise asyncio.CancelledError.

This method is synchronous and should not be awaited.

cancel_all()

Calls cancel on all conversations in this chat.

Note that you should await this method, since it’s meant to be used outside of a context manager, and it needs to resolve the chat.

get_edit(message=None, *, timeout=None)

Awaits for an edit after the last message to arrive. The arguments are the same as those for get_response.

get_reply(message=None, *, timeout=None)

Gets the next message that explicitly replies to a previous one.

get_response(message=None, *, timeout=None)

Gets the next message that responds to a previous one. This is the method you need most of the time, along with get_edit.

Args:
message (Message | int, optional):
The message (or the message ID) for which a response is expected. By default this is the last sent message.
timeout (int | float, optional):
If present, this timeout (in seconds) will override the per-action timeout defined for the conversation.
async with client.conversation(...) as conv:
    await conv.send_message('Hey, what is your name?')

    response = await conv.get_response()
    name = response.text

    await conv.send_message('Nice to meet you, {}!'.format(name))
mark_read(message=None)

Marks as read the latest received message if message is None. Otherwise, marks as read until the given message (or message ID).

This is equivalent to calling client.send_read_acknowledge.

send_file(*args, **kwargs)

Sends a file in the context of this conversation. Shorthand for telethon.client.uploads.UploadMethods.send_file with entity already set.

send_message(*args, **kwargs)

Sends a message in the context of this conversation. Shorthand for telethon.client.messages.MessageMethods.send_message with entity already set.

wait_event(event, *, timeout=None)

Waits for a custom event to occur. Timeouts still apply.

Note

Only use this if there isn’t another method available! For example, don’t use wait_event for new messages, since get_response already exists, etc.

Unless you’re certain that your code will run fast enough, generally you should get a “handle” of this special coroutine before acting. In this example you will see how to wait for a user to join a group with proper use of wait_event:

from telethon import TelegramClient, events

client = TelegramClient(...)
group_id = ...

async def main():
    # Could also get the user id from an event; this is just an example
    user_id = ...

    async with client.conversation(user_id) as conv:
        # Get a handle to the future event we'll wait for
        handle = conv.wait_event(events.ChatAction(
            group_id,
            func=lambda e: e.user_joined and e.user_id == user_id
        ))

        # Perform whatever action in between
        await conv.send_message('Please join this group before speaking to me!')

        # Wait for the event we registered above to fire
        event = await handle

        # Continue with the conversation
        await conv.send_message('Thanks!')

This way your event can be registered before acting, since the response may arrive before your event was registered. It depends on your use case since this also means the event can arrive before you send a previous action.

wait_read(message=None, *, timeout=None)

Awaits for the sent message to be marked as read. Note that receiving a response doesn’t imply the message was read, and this action will also trigger even without a response.

Dialog

class telethon.tl.custom.dialog.Dialog(client, dialog, entities, message)

Bases: object

Custom class that encapsulates a dialog (an open “conversation” with someone, a group or a channel) providing an abstraction to easily access the input version/normal entity/message etc. The library will return instances of this class when calling get_dialogs().

Args:
dialog (Dialog):
The original Dialog instance.
pinned (bool):
Whether this dialog is pinned to the top or not.
folder_id (folder_id):
The folder ID that this dialog belongs to.
archived (bool):
Whether this dialog is archived or not (folder_id is None).
message (Message):
The last message sent on this dialog. Note that this member will not be updated when new messages arrive, it’s only set on creation of the instance.
date (datetime):
The date of the last message sent on this dialog.
entity (entity):
The entity that belongs to this dialog (user, chat or channel).
input_entity (InputPeer):
Input version of the entity.
id (int):
The marked ID of the entity, which is guaranteed to be unique.
name (str):
Display name for this dialog. For chats and channels this is their title, and for users it’s “First-Name Last-Name”.
title (str):
Alias for name.
unread_count (int):
How many messages are currently unread in this dialog. Note that this value won’t update when new messages arrive.
unread_mentions_count (int):
How many mentions are currently unread in this dialog. Note that this value won’t update when new messages arrive.
draft (Draft):
The draft object in this dialog. It will not be None, so you can call draft.set_message(...).
is_user (bool):
True if the entity is a User.
is_group (bool):
True if the entity is a Chat or a Channel megagroup.
is_channel (bool):
True if the entity is a Channel.
__str__()

Return str(self).

__weakref__

list of weak references to the object (if defined)

archive(folder=1)

Archives (or un-archives) this dialog.

Args:
folder (int, optional):

The folder to which the dialog should be archived to.

If you want to “un-archive” it, use folder=0.

Returns:
The Updates object that the request produces.

Example:

# Archiving
dialog.archive()

# Un-archiving
dialog.archive(0)
delete(revoke=False)

Deletes the dialog from your dialog list. If you own the channel this won’t destroy it, only delete it from the list.

Shorthand for telethon.client.dialogs.DialogMethods.delete_dialog with entity already set.

send_message(*args, **kwargs)

Sends a message to this dialog. This is just a wrapper around client.send_message(dialog.input_entity, *args, **kwargs).

stringify()
to_dict()

Draft

class telethon.tl.custom.draft.Draft(client, entity, draft)

Bases: object

Custom class that encapsulates a draft on the Telegram servers, providing an abstraction to change the message conveniently. The library will return instances of this class when calling get_drafts().

Args:
date (datetime):
The date of the draft.
link_preview (bool):
Whether the link preview is enabled or not.
reply_to_msg_id (int):
The message ID that the draft will reply to.
__str__()

Return str(self).

__weakref__

list of weak references to the object (if defined)

delete()

Deletes this draft, and returns True on success.

entity

The entity that belongs to this dialog (user, chat or channel).

get_entity()

Returns entity but will make an API call if necessary.

get_input_entity()

Returns input_entity but will make an API call if necessary.

input_entity

Input version of the entity.

is_empty

Convenience bool to determine if the draft is empty or not.

raw_text

The raw (text without formatting) contained in the draft. It will be empty if there is no text (thus draft not set).

send(clear=True, parse_mode=())

Sends the contents of this draft to the dialog. This is just a wrapper around send_message(dialog.input_entity, *args, **kwargs).

set_message(text=None, reply_to=0, parse_mode=(), link_preview=None)

Changes the draft message on the Telegram servers. The changes are reflected in this object.

Parameters:
  • text (str) – New text of the draft. Preserved if left as None.
  • reply_to (int) – Message ID to reply to. Preserved if left as 0, erased if set to None.
  • link_preview (bool) – Whether to attach a web page preview. Preserved if left as None.
  • parse_mode (str) – The parse mode to be used for the text.
Return bool:

True on success.
stringify()
text

The markdown text contained in the draft. It will be empty if there is no text (and hence no draft is set).

to_dict()

File

class telethon.tl.custom.file.File(media)

Bases: object

Convenience class over media like photos or documents, which supports accessing the attributes in a more convenient way.

If any of the attributes are not present in the current media, the properties will be None.

The original media is available through the media attribute.

__weakref__

list of weak references to the object (if defined)

duration

The duration in seconds of the audio or video.

emoji

A string with all emoji that represent the current sticker.

ext

The extension from the mime type of this file.

If the mime type is unknown, the extension from the file name (if any) will be used.

height

The height in pixels of this media if it’s a photo or a video.

id

The old bot-API style file_id representing this file.

Warning

This feature has not been maintained for a long time and may not work. It will be removed in future versions.

Note

This file ID may not work under user accounts, but should still be usable by bot accounts.

You can, however, still use it to identify a file in for example a database.

mime_type

The mime-type of this file.

name

The file name of this document.

performer

The performer of the song.

size

The size in bytes of this file.

For photos, this is the heaviest thumbnail, as it often repressents the largest dimensions.

sticker_set

The InputStickerSet to which the sticker file belongs.

title

The title of the song.

width

The width in pixels of this media if it’s a photo or a video.

Forward

class telethon.tl.custom.forward.Forward(client, original, entities)

Bases: telethon.tl.custom.chatgetter.ChatGetter, telethon.tl.custom.sendergetter.SenderGetter

Custom class that encapsulates a MessageFwdHeader providing an abstraction to easily access information like the original sender.

Remember that this class implements ChatGetter and SenderGetter which means you have access to all their sender and chat properties and methods.

Attributes:

original_fwd (MessageFwdHeader):
The original MessageFwdHeader instance.
Any other attribute:
Attributes not described here are the same as those available in the original MessageFwdHeader.

InlineBuilder

class telethon.tl.custom.inlinebuilder.InlineBuilder(client)

Bases: object

Helper class to allow defining InlineQuery results.

Common arguments to all methods are explained here to avoid repetition:

text (str, optional):
If present, the user will send a text message with this text upon being clicked.
link_preview (bool, optional):
Whether to show a link preview in the sent text message or not.
geo (InputGeoPoint, GeoPoint, InputMediaVenue, MessageMediaVenue, optional):
If present, it may either be a geo point or a venue.
period (int, optional):
The period in seconds to be used for geo points.
contact (InputMediaContact, MessageMediaContact, optional):
If present, it must be the contact information to send.
game (bool, optional):
May be True to indicate that the game will be sent.
buttons (list, custom.Button, KeyboardButton, optional):
Same as buttons for client.send_message().
parse_mode (str, optional):
Same as parse_mode for client.send_message().
id (str, optional):

The string ID to use for this result. If not present, it will be the SHA256 hexadecimal digest of converting the created InputBotInlineResult with empty ID to bytes(), so that the ID will be deterministic for the same input.

Note

If two inputs are exactly the same, their IDs will be the same too. If you send two articles with the same ID, it will raise ResultIdDuplicateError. Consider giving them an explicit ID if you need to send two results that are the same.

__weakref__

list of weak references to the object (if defined)

article(title, description=None, *, url=None, thumb=None, content=None, id=None, text=None, parse_mode=(), link_preview=True, geo=None, period=60, contact=None, game=False, buttons=None)

Creates new inline result of article type.

Args:
title (str):
The title to be shown for this result.
description (str, optional):
Further explanation of what this result means.
url (str, optional):
The URL to be shown for this result.
thumb (InputWebDocument, optional):
The thumbnail to be shown for this result. For now it has to be a InputWebDocument if present.
content (InputWebDocument, optional):
The content to be shown for this result. For now it has to be a InputWebDocument if present.
Example:
results = [
    # Option with title and description sending a message.
    builder.article(
        title='First option',
        description='This is the first option',
        text='Text sent after clicking this option',
    ),
    # Option with title URL to be opened when clicked.
    builder.article(
        title='Second option',
        url='https://example.com',
        text='Text sent if the user clicks the option and not the URL',
    ),
    # Sending a message with buttons.
    # You can use a list or a list of lists to include more buttons.
    builder.article(
        title='Third option',
        text='Text sent with buttons below',
        buttons=Button.url('https://example.com'),
    ),
]
document(file, title=None, *, description=None, type=None, mime_type=None, attributes=None, force_document=False, voice_note=False, video_note=False, use_cache=True, id=None, text=None, parse_mode=(), link_preview=True, geo=None, period=60, contact=None, game=False, buttons=None, include_media=True)

Creates a new inline result of document type.

use_cache, mime_type, attributes, force_document, voice_note and video_note are described in client.send_file.

Args:
file (obj):
Same as file for client.send_file().
title (str, optional):
The title to be shown for this result.
description (str, optional):
Further explanation of what this result means.
type (str, optional):
The type of the document. May be one of: article, audio, contact, file, geo, gif, photo, sticker, venue, video, voice. It will be automatically set if mime_type is specified, and default to 'file' if no matching mime type is found. you may need to pass attributes in order to use type effectively.
attributes (list, optional):
Optional attributes that override the inferred ones, like DocumentAttributeFilename and so on.
include_media (bool, optional):
Whether the document file used to display the result should be included in the message itself or not. By default, the document is included, and the text parameter alters the caption.
Example:
results = [
    # Sending just the file when the user selects it.
    builder.document('/path/to/file.pdf'),

    # Including a caption with some in-memory file.
    file_bytesio = ...
    builder.document(
        file_bytesio,
        text='This will be the caption of the sent file',
    ),

    # Sending just the message without including the file.
    builder.document(
        photo,
        text='This will be a normal text message',
        include_media=False,
    ),
]
game(short_name, *, id=None, text=None, parse_mode=(), link_preview=True, geo=None, period=60, contact=None, game=False, buttons=None)

Creates a new inline result of game type.

Args:
short_name (str):
The short name of the game to use.
photo(file, *, id=None, include_media=True, text=None, parse_mode=(), link_preview=True, geo=None, period=60, contact=None, game=False, buttons=None)

Creates a new inline result of photo type.

Args:
include_media (bool, optional):
Whether the photo file used to display the result should be included in the message itself or not. By default, the photo is included, and the text parameter alters the caption.
file (obj, optional):
Same as file for client.send_file().
Example:
results = [
    # Sending just the photo when the user selects it.
    builder.photo('/path/to/photo.jpg'),

    # Including a caption with some in-memory photo.
    photo_bytesio = ...
    builder.photo(
        photo_bytesio,
        text='This will be the caption of the sent photo',
    ),

    # Sending just the message without including the photo.
    builder.photo(
        photo,
        text='This will be a normal text message',
        include_media=False,
    ),
]

InlineResult

class telethon.tl.custom.inlineresult.InlineResult(client, original, query_id=None, *, entity=None)

Bases: object

Custom class that encapsulates a bot inline result providing an abstraction to easily access some commonly needed features (such as clicking a result to select it).

Attributes:

result (BotInlineResult):
The original BotInlineResult object.
ARTICLE = 'article'
AUDIO = 'audio'
CONTACT = 'contact'
DOCUMENT = 'document'
GAME = 'game'
GIF = 'gif'
LOCATION = 'location'
PHOTO = 'photo'
VENUE = 'venue'
VIDEO = 'video'
VIDEO_GIF = 'mpeg4_gif'
__weakref__

list of weak references to the object (if defined)

click(entity=None, reply_to=None, comment_to=None, silent=False, clear_draft=False, hide_via=False, background=None)

Clicks this result and sends the associated message.

Args:
entity (entity):
The entity to which the message of this result should be sent.
reply_to (int | Message, optional):
If present, the sent message will reply to this ID or message.
comment_to (int | Message, optional):
Similar to reply_to, but replies in the linked group of a broadcast channel instead (effectively leaving a “comment to” the specified message).
silent (bool, optional):
Whether the message should notify people with sound or not. Defaults to False (send with a notification sound unless the person has the chat muted). Set it to True to alter this behaviour.
clear_draft (bool, optional):
Whether the draft should be removed after sending the message from this result or not. Defaults to False.
hide_via (bool, optional):
Whether the “via @bot” should be hidden or not. Only works with certain bots (like @bing or @gif).
background (bool, optional):
Whether the message should be send in background.
description

The description for this inline result. It may be None.

document

Returns either the WebDocument content for normal results or the Document for media results.

download_media(*args, **kwargs)

Downloads the media in this result (if there is a document, the document will be downloaded; otherwise, the photo will if present).

This is a wrapper around client.download_media.

message

The always-present BotInlineMessage that will be sent if click is called on this result.

photo

Returns either the WebDocument thumbnail for normal results or the Photo for media results.

title

The title for this inline result. It may be None.

type

The always-present type of this result. It will be one of: 'article', 'photo', 'gif', 'mpeg4_gif', 'video', 'audio', 'voice', 'document', 'location', 'venue', 'contact', 'game'.

You can access all of these constants through InlineResult, such as InlineResult.ARTICLE, InlineResult.VIDEO_GIF, etc.

url

The URL present in this inline results. If you want to “click” this URL to open it in your browser, you should use Python’s webbrowser.open(url) for such task.

InlineResults

class telethon.tl.custom.inlineresults.InlineResults(client, original, *, entity=None)

Bases: list

Custom class that encapsulates BotResults providing an abstraction to easily access some commonly needed features (such as clicking one of the results to select it)

Note that this is a list of InlineResult so you can iterate over it or use indices to access its elements. In addition, it has some attributes.

Attributes:
result (BotResults):
The original BotResults object.
query_id (int):
The random ID that identifies this query.
cache_time (int):
For how long the results should be considered valid. You can call results_valid at any moment to determine if the results are still valid or not.
users (User):
The users present in this inline query.
gallery (bool):
Whether these results should be presented in a grid (as a gallery of images) or not.
next_offset (str, optional):
The string to be used as an offset to get the next chunk of results, if any.
switch_pm (InlineBotSwitchPM, optional):
If presents, the results should show a button to switch to a private conversation with the bot using the text in this object.
__repr__()

Return repr(self).

__str__()

Return str(self).

__weakref__

list of weak references to the object (if defined)

results_valid()

Returns True if the cache time has not expired yet and the results can still be considered valid.

Message

class telethon.tl.custom.message.Message(id: int, peer_id: Union[telethon.tl.types.PeerUser, telethon.tl.types.PeerChat, telethon.tl.types.PeerChannel] = None, date: Optional[datetime.datetime] = None, out: Optional[bool] = None, mentioned: Optional[bool] = None, media_unread: Optional[bool] = None, silent: Optional[bool] = None, post: Optional[bool] = None, from_id: Union[telethon.tl.types.PeerUser, telethon.tl.types.PeerChat, telethon.tl.types.PeerChannel, None] = None, reply_to: Union[telethon.tl.types.MessageReplyHeader, telethon.tl.types.MessageReplyStoryHeader, None] = None, ttl_period: Optional[int] = None, message: Optional[str] = None, fwd_from: Optional[telethon.tl.types.MessageFwdHeader] = None, via_bot_id: Optional[int] = None, media: Union[telethon.tl.types.MessageMediaEmpty, telethon.tl.types.MessageMediaPhoto, telethon.tl.types.MessageMediaGeo, telethon.tl.types.MessageMediaContact, telethon.tl.types.MessageMediaUnsupported, telethon.tl.types.MessageMediaDocument, telethon.tl.types.MessageMediaWebPage, telethon.tl.types.MessageMediaVenue, telethon.tl.types.MessageMediaGame, telethon.tl.types.MessageMediaInvoice, telethon.tl.types.MessageMediaGeoLive, telethon.tl.types.MessageMediaPoll, telethon.tl.types.MessageMediaDice, telethon.tl.types.MessageMediaStory, None] = None, reply_markup: Union[telethon.tl.types.ReplyKeyboardHide, telethon.tl.types.ReplyKeyboardForceReply, telethon.tl.types.ReplyKeyboardMarkup, telethon.tl.types.ReplyInlineMarkup, None] = None, entities: Optional[List[Union[telethon.tl.types.MessageEntityUnknown, telethon.tl.types.MessageEntityMention, telethon.tl.types.MessageEntityHashtag, telethon.tl.types.MessageEntityBotCommand, telethon.tl.types.MessageEntityUrl, telethon.tl.types.MessageEntityEmail, telethon.tl.types.MessageEntityBold, telethon.tl.types.MessageEntityItalic, telethon.tl.types.MessageEntityCode, telethon.tl.types.MessageEntityPre, telethon.tl.types.MessageEntityTextUrl, telethon.tl.types.MessageEntityMentionName, telethon.tl.types.InputMessageEntityMentionName, telethon.tl.types.MessageEntityPhone, telethon.tl.types.MessageEntityCashtag, telethon.tl.types.MessageEntityUnderline, telethon.tl.types.MessageEntityStrike, telethon.tl.types.MessageEntityBlockquote, telethon.tl.types.MessageEntityBankCard, telethon.tl.types.MessageEntitySpoiler, telethon.tl.types.MessageEntityCustomEmoji]]] = None, views: Optional[int] = None, edit_date: Optional[datetime.datetime] = None, post_author: Optional[str] = None, grouped_id: Optional[int] = None, from_scheduled: Optional[bool] = None, legacy: Optional[bool] = None, edit_hide: Optional[bool] = None, pinned: Optional[bool] = None, noforwards: Optional[bool] = None, reactions: Optional[telethon.tl.types.MessageReactions] = None, restriction_reason: Optional[telethon.tl.types.RestrictionReason] = None, forwards: Optional[int] = None, replies: Optional[telethon.tl.types.MessageReplies] = None, action: Union[telethon.tl.types.MessageActionEmpty, telethon.tl.types.MessageActionChatCreate, telethon.tl.types.MessageActionChatEditTitle, telethon.tl.types.MessageActionChatEditPhoto, telethon.tl.types.MessageActionChatDeletePhoto, telethon.tl.types.MessageActionChatAddUser, telethon.tl.types.MessageActionChatDeleteUser, telethon.tl.types.MessageActionChatJoinedByLink, telethon.tl.types.MessageActionChannelCreate, telethon.tl.types.MessageActionChatMigrateTo, telethon.tl.types.MessageActionChannelMigrateFrom, telethon.tl.types.MessageActionPinMessage, telethon.tl.types.MessageActionHistoryClear, telethon.tl.types.MessageActionGameScore, telethon.tl.types.MessageActionPaymentSentMe, telethon.tl.types.MessageActionPaymentSent, telethon.tl.types.MessageActionPhoneCall, telethon.tl.types.MessageActionScreenshotTaken, telethon.tl.types.MessageActionCustomAction, telethon.tl.types.MessageActionBotAllowed, telethon.tl.types.MessageActionSecureValuesSentMe, telethon.tl.types.MessageActionSecureValuesSent, telethon.tl.types.MessageActionContactSignUp, telethon.tl.types.MessageActionGeoProximityReached, telethon.tl.types.MessageActionGroupCall, telethon.tl.types.MessageActionInviteToGroupCall, telethon.tl.types.MessageActionSetMessagesTTL, telethon.tl.types.MessageActionGroupCallScheduled, telethon.tl.types.MessageActionSetChatTheme, telethon.tl.types.MessageActionChatJoinedByRequest, telethon.tl.types.MessageActionWebViewDataSentMe, telethon.tl.types.MessageActionWebViewDataSent, telethon.tl.types.MessageActionGiftPremium, telethon.tl.types.MessageActionTopicCreate, telethon.tl.types.MessageActionTopicEdit, telethon.tl.types.MessageActionSuggestProfilePhoto, telethon.tl.types.MessageActionRequestedPeer, telethon.tl.types.MessageActionSetChatWallPaper, telethon.tl.types.MessageActionSetSameChatWallPaper, None] = None)

Bases: telethon.tl.custom.chatgetter.ChatGetter, telethon.tl.custom.sendergetter.SenderGetter, telethon.tl.tlobject.TLObject

This custom class aggregates both Message and MessageService to ease accessing their members.

Remember that this class implements ChatGetter and SenderGetter which means you have access to all their sender and chat properties and methods.

Members:
out (bool):

Whether the message is outgoing (i.e. you sent it from another session) or incoming (i.e. someone else sent it).

Note that messages in your own chat are always incoming, but this member will be True if you send a message to your own chat. Messages you forward to your chat are not considered outgoing, just like official clients display them.

mentioned (bool):
Whether you were mentioned in this message or not. Note that replies to your own messages also count as mentions.
media_unread (bool):
Whether you have read the media in this message or not, e.g. listened to the voice note media.
silent (bool):
Whether the message should notify people with sound or not. Previously used in channels, but since 9 August 2019, it can also be used in private chats.
post (bool):
Whether this message is a post in a broadcast channel or not.
from_scheduled (bool):
Whether this message was originated from a previously-scheduled message or not.
legacy (bool):
Whether this is a legacy message or not.
edit_hide (bool):
Whether the edited mark of this message is edited should be hidden (e.g. in GUI clients) or shown.
pinned (bool):
Whether this message is currently pinned or not.
noforwards (bool):
Whether this message can be forwarded or not.
id (int):
The ID of this message. This field is always present. Any other member is optional and may be None.
from_id (Peer):
The peer who sent this message, which is either PeerUser, PeerChat or PeerChannel. This value will be None for anonymous messages.
peer_id (Peer):
The peer to which this message was sent, which is either PeerUser, PeerChat or PeerChannel. This will always be present except for empty messages.
fwd_from (MessageFwdHeader):
The original forward header if this message is a forward. You should probably use the forward property instead.
via_bot_id (int):
The ID of the bot used to send this message through its inline mode (e.g. “via @like”).
reply_to (MessageReplyHeader):
The original reply header if this message is replying to another.
date (datetime):
The UTC+0 datetime object indicating when this message was sent. This will always be present except for empty messages.
message (str):
The string text of the message for Message instances, which will be None for other types of messages.
media (MessageMedia):

The media sent with this message if any (such as photos, videos, documents, gifs, stickers, etc.).

You may want to access the photo, document etc. properties instead.

If the media was not present or it was MessageMediaEmpty, this member will instead be None for convenience.

reply_markup (ReplyMarkup):
The reply markup for this message (which was sent either via a bot or by a bot). You probably want to access buttons instead.
entities (List[MessageEntity]):
The list of markup entities in this message, such as bold, italics, code, hyperlinks, etc.
views (int):
The number of views this message from a broadcast channel has. This is also present in forwards.
forwards (int):
The number of times this message has been forwarded.
replies (int):
The number of times another message has replied to this message.
edit_date (datetime):
The date when this message was last edited.
post_author (str):
The display name of the message sender to show in messages sent to broadcast channels.
grouped_id (int):
If this message belongs to a group of messages (photo albums or video albums), all of them will have the same value here.
reactions (MessageReactions)
Reactions to this message.
restriction_reason (List[RestrictionReason])
An optional list of reasons why this message was restricted. If the list is None, this message has not been restricted.
ttl_period (int):
The Time To Live period configured for this message. The message should be erased from wherever it’s stored (memory, a local database, etc.) when datetime.now() > message.date + timedelta(seconds=message.ttl_period).
action (MessageAction):
The message action object of the message for MessageService instances, which will be None for other types of messages.
action_entities

Returns a list of entities that took part in this action.

Possible cases for this are MessageActionChatAddUser, types.MessageActionChatCreate, MessageActionChatDeleteUser, MessageActionChatJoinedByLink MessageActionChatMigrateTo and MessageActionChannelMigrateFrom.

If the action is neither of those, the result will be None. If some entities could not be retrieved, the list may contain some None items in it.

audio

The Document media in this message, if it’s an audio file.

button_count

Returns the total button count (sum of all buttons rows).

buttons

Returns a list of lists of MessageButton, if any.

Otherwise, it returns None.

click(i=None, j=None, *, text=None, filter=None, data=None, share_phone=None, share_geo=None, password=None)

Calls SendVote with the specified poll option or button.click on the specified button.

Does nothing if the message is not a poll or has no buttons.

Args:
i (int | list):

Clicks the i’th button or poll option (starting from the index 0). For multiple-choice polls, a list with the indices should be used. Will raise IndexError if out of bounds. Example:

>>> message = ...  # get the message somehow
>>> # Clicking the 3rd button
>>> # [button1] [button2]
>>> # [     button3     ]
>>> # [button4] [button5]
>>> await message.click(2)  # index
j (int):

Clicks the button at position (i, j), these being the indices for the (row, column) respectively. Example:

>>> # Clicking the 2nd button on the 1st row.
>>> # [button1] [button2]
>>> # [     button3     ]
>>> # [button4] [button5]
>>> await message.click(0, 1)  # (row, column)

This is equivalent to message.buttons[0][1].click().

text (str | callable):

Clicks the first button or poll option with the text “text”. This may also be a callable, like a re.compile(...).match, and the text will be passed to it.

If you need to select multiple options in a poll, pass a list of indices to the i parameter.

filter (callable):

Clicks the first button or poll option for which the callable returns True. The callable should accept a single MessageButton or PollAnswer argument.

If you need to select multiple options in a poll, pass a list of indices to the i parameter.

data (bytes):
This argument overrides the rest and will not search any buttons. Instead, it will directly send the request to behave as if it clicked a button with said data. Note that if the message does not have this data, it will raise DataInvalidError.
share_phone (bool | str | tl:InputMediaContact):

When clicking on a keyboard button requesting a phone number (KeyboardButtonRequestPhone), this argument must be explicitly set to avoid accidentally sharing the number.

It can be True to automatically share the current user’s phone, a string to share a specific phone number, or a contact media to specify all details.

If the button is pressed without this, ValueError is raised.

share_geo (tuple | list | tl:InputMediaGeoPoint):

When clicking on a keyboard button requesting a geo location (KeyboardButtonRequestGeoLocation), this argument must be explicitly set to avoid accidentally sharing the location.

It must be a tuple of float as (longitude, latitude), or a InputGeoPoint instance to avoid accidentally using the wrong roder.

If the button is pressed without this, ValueError is raised.

password (str):
When clicking certain buttons (such as BotFather’s confirmation button to transfer ownership), if your account has 2FA enabled, you need to provide your account’s password. Otherwise, teltehon.errors.PasswordHashInvalidError is raised.

Example:

# Click the first button
await message.click(0)

# Click some row/column
await message.click(row, column)

# Click by text
await message.click(text='👍')

# Click by data
await message.click(data=b'payload')

# Click on a button requesting a phone
await message.click(0, share_phone=True)
client

Returns the TelegramClient that patched this message. This will only be present if you use the friendly methods, it won’t be there if you invoke raw API methods manually, in which case you should only access members, not properties.

contact

The MessageMediaContact in this message, if it’s a contact.

delete(*args, **kwargs)

Deletes the message. You’re responsible for checking whether you have the permission to do so, or to except the error otherwise. Shorthand for telethon.client.messages.MessageMethods.delete_messages with entity and message_ids already set.

If you need to delete more than one message at once, don’t use this delete method. Use a telethon.client.telegramclient.TelegramClient instance directly.

dice

The MessageMediaDice in this message, if it’s a dice roll.

document

The Document media in this message, if any.

download_media(*args, **kwargs)

Downloads the media contained in the message, if any. Shorthand for telethon.client.downloads.DownloadMethods.download_media with the message already set.

edit(*args, **kwargs)

Edits the message if it’s outgoing. Shorthand for telethon.client.messages.MessageMethods.edit_message with both entity and message already set.

Returns
The edited Message, unless entity was a InputBotInlineMessageID or InputBotInlineMessageID64 in which case this method returns a boolean.
Raises

MessageAuthorRequiredError if you’re not the author of the message but tried editing it anyway.

MessageNotModifiedError if the contents of the message were not modified at all.

MessageIdInvalidError if the ID of the message is invalid (the ID itself may be correct, but the message with that ID cannot be edited). For example, when trying to edit messages with a reply markup (or clear markup) this error will be raised.

Note

This is different from client.edit_message and will respect the previous state of the message. For example, if the message didn’t have a link preview, the edit won’t add one by default, and you should force it by setting it to True if you want it.

This is generally the most desired and convenient behaviour, and will work for link previews and message buttons.

file

Returns a File wrapping the photo or document in this message. If the media type is different (polls, games, none, etc.), this property will be None.

This instance lets you easily access other properties, such as file.id, file.name, etc., without having to manually inspect the document.attributes.

forward

The Forward information if this message is a forwarded message.

forward_to(*args, **kwargs)

Forwards the message. Shorthand for telethon.client.messages.MessageMethods.forward_messages with both messages and from_peer already set.

If you need to forward more than one message at once, don’t use this forward_to method. Use a telethon.client.telegramclient.TelegramClient instance directly.

game

The Game media in this message, if it’s a game.

geo

The GeoPoint media in this message, if it has a location.

get_buttons()

Returns buttons when that property fails (this is rarely needed).

get_entities_text(cls=None)

Returns a list of (markup entity, inner text) (like bold or italics).

The markup entity is a MessageEntity that represents bold, italics, etc., and the inner text is the str inside that markup entity.

For example:

print(repr(message.text))  # shows: 'Hello **world**!'

for ent, txt in message.get_entities_text():
    print(ent)  # shows: MessageEntityBold(offset=6, length=5)
    print(txt)  # shows: world
Args:
cls (type):

Returns entities matching this type only. For example, the following will print the text for all code entities:

>>> from telethon.tl.types import MessageEntityCode
>>>
>>> m = ...  # get the message
>>> for _, inner_text in m.get_entities_text(MessageEntityCode):
>>>     print(inner_text)
get_reply_message()

The Message that this message is replying to, or None.

The result will be cached after its first use.

gif

The Document media in this message, if it’s a “gif”.

“Gif” files by Telegram are normally .mp4 video files without sound, the so called “animated” media. However, it may be the actual gif format if the file is too large.

invoice

The MessageMediaInvoice in this message, if it’s an invoice.

is_reply

True if the message is a reply to some other message.

Remember that you can access the ID of the message this one is replying to through reply_to.reply_to_msg_id, and the Message object with get_reply_message().

mark_read()

Marks the message as read. Shorthand for client.send_read_acknowledge() with both entity and message already set.

photo

The Photo media in this message, if any.

This will also return the photo for MessageService if its action is MessageActionChatEditPhoto, or if the message has a web preview with a photo.

pin(*, notify=False, pm_oneside=False)

Pins the message. Shorthand for telethon.client.messages.MessageMethods.pin_message with both entity and message already set.

poll

The MessageMediaPoll in this message, if it’s a poll.

raw_text

The raw message text, ignoring any formatting. Will be None for MessageService.

Setting a value to this field will erase the entities, unlike changing the message member.

reply(*args, **kwargs)

Replies to the message (as a reply). Shorthand for telethon.client.messages.MessageMethods.send_message with both entity and reply_to already set.

reply_to_msg_id

Returns the message ID this message is replying to, if any. This is equivalent to accessing .reply_to.reply_to_msg_id.

respond(*args, **kwargs)

Responds to the message (not as a reply). Shorthand for telethon.client.messages.MessageMethods.send_message with entity already set.

sticker

The Document media in this message, if it’s a sticker.

text

The message text, formatted using the client’s default parse mode. Will be None for MessageService.

to_id

Returns the peer to which this message was sent to. This used to exist to infer the .peer_id.

unpin()

Unpins the message. Shorthand for telethon.client.messages.MessageMethods.unpin_message with both entity and message already set.

venue

The MessageMediaVenue in this message, if it’s a venue.

via_bot

The bot User if the message was sent via said bot.

This will only be present if via_bot_id is not None and the entity is known.

via_input_bot

Returns the input variant of via_bot.

video

The Document media in this message, if it’s a video.

video_note

The Document media in this message, if it’s a video note.

voice

The Document media in this message, if it’s a voice note.

web_preview

The WebPage media in this message, if any.

MessageButton

class telethon.tl.custom.messagebutton.MessageButton(client, original, chat, bot, msg_id)

Bases: object

Note

Message.buttons are instances of this type. If you want to define a reply markup for e.g. sending messages, refer to Button instead.

Custom class that encapsulates a message button providing an abstraction to easily access some commonly needed features (such as clicking the button itself).

Attributes:

button (KeyboardButton):
The original KeyboardButton object.
__weakref__

list of weak references to the object (if defined)

click(share_phone=None, share_geo=None, *, password=None)

Emulates the behaviour of clicking this button.

If it’s a normal KeyboardButton with text, a message will be sent, and the sent Message returned.

If it’s an inline KeyboardButtonCallback with text and data, it will be “clicked” and the BotCallbackAnswer returned.

If it’s an inline KeyboardButtonSwitchInline button, the StartBotRequest will be invoked and the resulting updates returned.

If it’s a KeyboardButtonUrl, the URL of the button will be passed to webbrowser.open and return True on success.

If it’s a KeyboardButtonRequestPhone, you must indicate that you want to share_phone=True in order to share it. Sharing it is not a default because it is a privacy concern and could happen accidentally.

You may also use share_phone=phone to share a specific number, in which case either str or InputMediaContact should be used.

If it’s a KeyboardButtonRequestGeoLocation, you must pass a tuple in share_geo=(longitude, latitude). Note that Telegram seems to have some heuristics to determine impossible locations, so changing this value a lot quickly may not work as expected. You may also pass a InputGeoPoint if you find the order confusing.

client

Returns the telethon.client.telegramclient.TelegramClient instance that created this instance.

data

The bytes data for KeyboardButtonCallback objects.

inline_query

The query str for KeyboardButtonSwitchInline objects.

text

The text string of the button.

url

The url str for KeyboardButtonUrl objects.

ParticipantPermissions

class telethon.tl.custom.participantpermissions.ParticipantPermissions(participant, chat: bool)

Bases: object

Participant permissions information.

The properties in this objects are boolean values indicating whether the user has the permission or not.

Example
permissions = ...

if permissions.is_banned:
    "this user is banned"
elif permissions.is_admin:
    "this user is an administrator"
__weakref__

list of weak references to the object (if defined)

add_admins

Whether the administrator can add new administrators with the same or less permissions than them.

anonymous

Whether the administrator will remain anonymous when sending messages.

ban_users

Whether the administrator can ban other users or not.

change_info

Whether the administrator can change the information about the chat, such as title or description.

delete_messages

Whether the administrator can delete messages from other participants.

edit_messages

Whether the administrator can edit messages.

has_default_permissions

Whether the user is a normal user of the chat (not administrator, but not banned either, and has no restrictions applied).

has_left

Whether the user left the chat.

invite_users

Whether the administrator can add new users to the chat.

is_admin

Whether the user is an administrator of the chat or not. The creator also counts as begin an administrator, since they have all permissions.

is_banned

Whether the user is banned in the chat.

is_creator

Whether the user is the creator of the chat or not.

manage_call

Whether the user will be able to manage group calls.

pin_messages

Whether the administrator can pin messages or not.

post_messages

Whether the administrator can post messages in the broadcast channel.

QRLogin

class telethon.tl.custom.qrlogin.QRLogin(client, ignored_ids)

Bases: object

QR login information.

Most of the time, you will present the url as a QR code to the user, and while it’s being shown, call wait.

__weakref__

list of weak references to the object (if defined)

expires

The datetime at which the QR code will expire.

If you want to try again, you will need to call recreate.

recreate()

Generates a new token and URL for a new QR code, useful if the code has expired before it was imported.

token

The binary data representing the token.

It can be used by a previously-authorized client in a call to auth.importLoginToken to log the client that originally requested the QR login.

url

The tg://login URI with the token. When opened by a Telegram application where the user is logged in, it will import the login token.

If you want to display a QR code to the user, this is the URL that should be launched when the QR code is scanned (the URL that should be contained in the QR code image you generate).

Whether you generate the QR code image or not is up to you, and the library can’t do this for you due to the vast ways of generating and displaying the QR code that exist.

The URL simply consists of token base64-encoded.

wait(timeout: float = None)

Waits for the token to be imported by a previously-authorized client, either by scanning the QR, launching the URL directly, or calling the import method.

This method must be called before the QR code is scanned, and must be executing while the QR code is being scanned. Otherwise, the login will not complete.

Will raise asyncio.TimeoutError if the login doesn’t complete on time.

Arguments
timeout (float):
The timeout, in seconds, to wait before giving up. By default the library will wait until the token expires, which is often what you want.
Returns
On success, an instance of User. On failure it will raise.

SenderGetter

class telethon.tl.custom.sendergetter.SenderGetter(sender_id=None, *, sender=None, input_sender=None)

Bases: abc.ABC

Helper base class that introduces the sender, input_sender and sender_id properties and get_sender and get_input_sender methods.

__weakref__

list of weak references to the object (if defined)

get_input_sender()

Returns input_sender, but will make an API call to find the input sender unless it’s already cached.

get_sender()

Returns sender, but will make an API call to find the sender unless it’s already cached.

If you only need the ID, use sender_id instead.

If you need to call a method which needs this sender, use get_input_sender() instead.

input_sender

This InputPeer is the input version of the user/channel who sent the message. Similarly to input_chat, this doesn’t have things like username or similar, but still useful in some cases.

Note that this might not be available if the library can’t find the input chat, or if the message a broadcast on a channel.

sender

Returns the User or Channel that sent this object. It may be None if Telegram didn’t send the sender.

If you only need the ID, use sender_id instead.

If you need to call a method which needs this chat, use input_sender instead.

If you’re using telethon.events, use get_sender() instead.

sender_id

Returns the marked sender integer ID, if present.

If there is a sender in the object, sender_id will always be set, which is why you should use it instead of sender.id.

Utilities

These are the utilities that the library has to offer.

Utilities for working with the Telegram API itself (such as handy methods to convert between an entity like a User, Chat, etc. into its Input version)

class telethon.utils.AsyncClassWrapper(wrapped)

Bases: object

__getattr__(item)
__weakref__

list of weak references to the object (if defined)

telethon.utils.chunks(iterable, size=100)

Turns the given iterable into chunks of the specified size, which is 100 by default since that’s what Telegram uses the most.

telethon.utils.decode_waveform(waveform)

Inverse operation of encode_waveform.

telethon.utils.encode_waveform(waveform)

Encodes the input bytes into a 5-bit byte-string to be used as a voice note’s waveform. See decode_waveform for the reverse operation.

Example
chat = ...
file = 'my.ogg'

# Send 'my.ogg' with a ascending-triangle waveform
await client.send_file(chat, file, attributes=[types.DocumentAttributeAudio(
    duration=7,
    voice=True,
    waveform=utils.encode_waveform(bytes(range(2 ** 5))  # 2**5 because 5-bit
)]

# Send 'my.ogg' with a square waveform
await client.send_file(chat, file, attributes=[types.DocumentAttributeAudio(
    duration=7,
    voice=True,
    waveform=utils.encode_waveform(bytes((31, 31, 15, 15, 15, 15, 31, 31)) * 4)
)]
telethon.utils.get_appropriated_part_size(file_size)

Gets the appropriated part size when uploading or downloading files, given an initial file size.

telethon.utils.get_attributes(file, *, attributes=None, mime_type=None, force_document=False, voice_note=False, video_note=False, supports_streaming=False, thumb=None)

Get a list of attributes for the given file and the mime type as a tuple ([attribute], mime_type).

telethon.utils.get_display_name(entity)

Gets the display name for the given User, Chat or Channel. Returns an empty string otherwise.

telethon.utils.get_extension(media)

Gets the corresponding extension for any Telegram media.

telethon.utils.get_inner_text(text, entities)

Gets the inner text that’s surrounded by the given entities. For instance: text = ‘hey!’, entity = MessageEntityBold(2, 2) -> ‘y!’.

Parameters:
  • text – the original text.
  • entities – the entity or entities that must be matched.
Returns:

a single result or a list of the text surrounded by the entities.
telethon.utils.get_input_channel(entity)

Similar to get_input_peer(), but for InputChannel’s alone.

Important

This method does not validate for invalid general-purpose access hashes, unlike get_input_peer. Consider using instead: get_input_channel(get_input_peer(channel)).

telethon.utils.get_input_chat_photo(photo)

Similar to get_input_peer(), but for chat photos

telethon.utils.get_input_dialog(dialog)

Similar to get_input_peer(), but for dialogs

telethon.utils.get_input_document(document)

Similar to get_input_peer(), but for documents

telethon.utils.get_input_geo(geo)

Similar to get_input_peer(), but for geo points

telethon.utils.get_input_group_call(call)

Similar to get_input_peer(), but for input calls.

telethon.utils.get_input_location(location)

Similar to get_input_peer(), but for input messages.

Note that this returns a tuple (dc_id, location), the dc_id being present if known.

telethon.utils.get_input_media(media, *, is_photo=False, attributes=None, force_document=False, voice_note=False, video_note=False, supports_streaming=False, ttl=None)

Similar to get_input_peer(), but for media.

If the media is InputFile and is_photo is known to be True, it will be treated as an InputMediaUploadedPhoto. Else, the rest of parameters will indicate how to treat it.

telethon.utils.get_input_message(message)

Similar to get_input_peer(), but for input messages.

telethon.utils.get_input_peer(entity, allow_self=True, check_hash=True)

Gets the input peer for the given “entity” (user, chat or channel).

A TypeError is raised if the given entity isn’t a supported type or if check_hash is True but the entity’s access_hash is None or the entity contains min information. In this case, the hash cannot be used for general purposes, and thus is not returned to avoid any issues which can derive from invalid access hashes.

Note that check_hash is ignored if an input peer is already passed since in that case we assume the user knows what they’re doing. This is key to getting entities by explicitly passing hash = 0.

telethon.utils.get_input_photo(photo)

Similar to get_input_peer(), but for photos

telethon.utils.get_input_user(entity)

Similar to get_input_peer(), but for InputUser’s alone.

Important

This method does not validate for invalid general-purpose access hashes, unlike get_input_peer. Consider using instead: get_input_channel(get_input_peer(channel)).

telethon.utils.get_message_id(message)

Similar to get_input_peer(), but for message IDs.

telethon.utils.get_peer(peer)
telethon.utils.get_peer_id(peer, add_mark=True)

Convert the given peer into its marked ID by default.

This “mark” comes from the “bot api” format, and with it the peer type can be identified back. User ID is left unmodified, chat ID is negated, and channel ID is “prefixed” with -100:

  • user_id
  • -chat_id
  • -100channel_id

The original ID and the peer type class can be returned with a call to resolve_id(marked_id)().

telethon.utils.is_audio(file)

Returns True if the file has an audio mime type.

telethon.utils.is_gif(file)

Returns True if the file extension looks like a gif file to Telegram.

telethon.utils.is_image(file)

Returns True if the file extension looks like an image file to Telegram.

telethon.utils.is_list_like(obj)

Returns True if the given object looks like a list.

Checking if hasattr(obj, '__iter__') and ignoring str/bytes is not enough. Things like open() are also iterable (and probably many other things), so just support the commonly known list-like objects.

telethon.utils.is_video(file)

Returns True if the file has a video mime type.

telethon.utils.pack_bot_file_id(file)

Inverse operation for resolve_bot_file_id.

The only parameters this method will accept are Document and Photo, and it will return a variable-length file_id string.

If an invalid parameter is given, it will return None.

telethon.utils.parse_phone(phone)

Parses the given phone, or returns None if it’s invalid.

telethon.utils.parse_username(username)

Parses the given username or channel access hash, given a string, username or URL. Returns a tuple consisting of both the stripped, lowercase username and whether it is a joinchat/ hash (in which case is not lowercase’d).

Returns (None, False) if the username or link is not valid.

telethon.utils.resolve_bot_file_id(file_id)

Given a Bot API-style file_id, returns the media it represents. If the file_id is not valid, None is returned instead.

Note that the file_id does not have information such as image dimensions or file size, so these will be zero if present.

For thumbnails, the photo ID and hash will always be zero.

telethon.utils.resolve_id(marked_id)

Given a marked ID, returns the original ID and its Peer type.

telethon.utils.resolve_inline_message_id(inline_msg_id)

Resolves an inline message ID. Returns a tuple of (message id, peer, dc id, access hash)

The peer may either be a PeerUser referencing the user who sent the message via the bot in a private conversation or small group chat, or a PeerChannel if the message was sent in a channel.

The access_hash does not have any use yet.

Resolves the given invite link. Returns a tuple of (link creator user id, global chat id, random int).

Note that for broadcast channels or with the newest link format, the link creator user ID will be zero to protect their identity. Normal chats and megagroup channels will have such ID.

Note that the chat ID may not be accurate for chats with a link that were upgraded to megagroup, since the link can remain the same, but the chat ID will be correct once a new link is generated.

telethon.utils.sanitize_parse_mode(mode)

Converts the given parse mode into an object with parse and unparse callable properties.

telethon.utils.split_text(text, entities, *, limit=4096, max_entities=100, split_at=('\\n', '\\s', '.'))

Split a message text and entities into multiple messages, each with their own set of entities. This allows sending a very large message as multiple messages while respecting the formatting.

Arguments
text (str):
The message text.
entities (List[MessageEntity])
The formatting entities.
limit (int):
The maximum message length of each individual message.
max_entities (int):
The maximum amount of entities that will be present in each individual message.
split_at (Tuplel[str]):

The list of regular expressions that will determine where to split the text. By default, a newline is searched. If no newline is present, a space is searched. If no space is found, the split will be made at any character.

The last expression should always match a character, or else the text will stop being splitted and the resulting text may be larger than the limit.

Yields
Pairs of (str, entities) with the split message.
Example
from telethon import utils
from telethon.extensions import markdown

very_long_markdown_text = "..."
text, entities = markdown.parse(very_long_markdown_text)

for text, entities in utils.split_text(text, entities):
    await client.send_message(chat, text, formatting_entities=entities)
telethon.utils.stripped_photo_to_jpg(stripped)

Adds the JPG header and footer to a stripped image.

Ported from https://github.com/telegramdesktop/tdesktop/blob/bec39d89e19670eb436dc794a8f20b657cb87c71/Telegram/SourceFiles/ui/image/image.cpp#L225

API Errors

These are the base errors that Telegram’s API may raise.

See RPC Errors for a more in-depth explanation on how to handle all known possible errors and learning to determine what a method may raise.

Errors not related to the Telegram API itself

exception telethon.errors.common.AlreadyInConversationError

Bases: Exception

Occurs when another exclusive conversation is opened in the same chat.

__weakref__

list of weak references to the object (if defined)

exception telethon.errors.common.AuthKeyNotFound

Bases: Exception

The server claims it doesn’t know about the authorization key (session file) currently being used. This might be because it either has never seen this authorization key, or it used to know about the authorization key but has forgotten it, either temporarily or permanently (possibly due to server errors).

If the issue persists, you may need to recreate the session file and login again. This is not done automatically because it is not possible to know if the issue is temporary or permanent.

__weakref__

list of weak references to the object (if defined)

exception telethon.errors.common.BadMessageError(request, code)

Bases: Exception

Occurs when handling a bad_message_notification.

ErrorMessages = {16: 'msg_id too low (most likely, client time is wrong it would be worthwhile to synchronize it using msg_id notifications and re-send the original message with the "correct" msg_id or wrap it in a container with a new msg_id if the original message had waited too long on the client to be transmitted).', 17: 'msg_id too high (similar to the previous case, the client time has to be synchronized, and the message re-sent with the correct msg_id).', 18: 'Incorrect two lower order msg_id bits (the server expects client message msg_id to be divisible by 4).', 19: 'Container msg_id is the same as msg_id of a previously received message (this must never happen).', 20: 'Message too old, and it cannot be verified whether the server has received a message with this msg_id or not.', 32: 'msg_seqno too low (the server has already received a message with a lower msg_id but with either a higher or an equal and odd seqno).', 33: 'msg_seqno too high (similarly, there is a message with a higher msg_id but with either a lower or an equal and odd seqno).', 34: 'An even msg_seqno expected (irrelevant message), but odd received.', 35: 'Odd msg_seqno expected (relevant message), but even received.', 48: 'Incorrect server salt (in this case, the bad_server_salt response is received with the correct salt, and the message is to be re-sent with it).', 64: 'Invalid container.'}
__weakref__

list of weak references to the object (if defined)

exception telethon.errors.common.CdnFileTamperedError

Bases: telethon.errors.common.SecurityError

Occurs when there’s a hash mismatch between the decrypted CDN file and its expected hash.

exception telethon.errors.common.InvalidBufferError(payload)

Bases: BufferError

Occurs when the buffer is invalid, and may contain an HTTP error code. For instance, 404 means “forgotten/broken authorization key”, while

__weakref__

list of weak references to the object (if defined)

exception telethon.errors.common.InvalidChecksumError(checksum, valid_checksum)

Bases: Exception

Occurs when using the TCP full mode and the checksum of a received packet doesn’t match the expected checksum.

__weakref__

list of weak references to the object (if defined)

exception telethon.errors.common.MultiError

Bases: Exception

Exception container for multiple TLRequest’s.

static __new__(cls, exceptions, result, requests)

Create and return a new object. See help(type) for accurate signature.

__weakref__

list of weak references to the object (if defined)

exception telethon.errors.common.ReadCancelledError

Bases: Exception

Occurs when a read operation was cancelled.

__weakref__

list of weak references to the object (if defined)

exception telethon.errors.common.SecurityError(*args)

Bases: Exception

Generic security error, mostly used when generating a new AuthKey.

__weakref__

list of weak references to the object (if defined)

exception telethon.errors.common.TypeNotFoundError(invalid_constructor_id, remaining)

Bases: Exception

Occurs when a type is not found, for example, when trying to read a TLObject with an invalid constructor code.

__weakref__

list of weak references to the object (if defined)

exception telethon.errors.rpcbaseerrors.AuthKeyError(request, message, code=None)

Bases: telethon.errors.rpcbaseerrors.RPCError

Errors related to invalid authorization key, like AUTH_KEY_DUPLICATED which can cause the connection to fail.

code = 406
message = 'AUTH_KEY'
exception telethon.errors.rpcbaseerrors.BadRequestError(request, message, code=None)

Bases: telethon.errors.rpcbaseerrors.RPCError

The query contains errors. In the event that a request was created using a form and contains user generated data, the user should be notified that the data must be corrected before the query is repeated.

code = 400
message = 'BAD_REQUEST'
telethon.errors.rpcbaseerrors.BotTimeout

alias of telethon.errors.rpcbaseerrors.TimedOutError

exception telethon.errors.rpcbaseerrors.FloodError(request, message, code=None)

Bases: telethon.errors.rpcbaseerrors.RPCError

The maximum allowed number of attempts to invoke the given method with the given input parameters has been exceeded. For example, in an attempt to request a large number of text messages (SMS) for the same phone number.

code = 420
message = 'FLOOD'
exception telethon.errors.rpcbaseerrors.ForbiddenError(request, message, code=None)

Bases: telethon.errors.rpcbaseerrors.RPCError

Privacy violation. For example, an attempt to write a message to someone who has blacklisted the current user.

code = 403
message = 'FORBIDDEN'
exception telethon.errors.rpcbaseerrors.InvalidDCError(request, message, code=None)

Bases: telethon.errors.rpcbaseerrors.RPCError

The request must be repeated, but directed to a different data center.

code = 303
message = 'ERROR_SEE_OTHER'
exception telethon.errors.rpcbaseerrors.NotFoundError(request, message, code=None)

Bases: telethon.errors.rpcbaseerrors.RPCError

An attempt to invoke a non-existent object, such as a method.

code = 404
message = 'NOT_FOUND'
exception telethon.errors.rpcbaseerrors.RPCError(request, message, code=None)

Bases: Exception

Base class for all Remote Procedure Call errors.

__reduce__()

Helper for pickle.

__weakref__

list of weak references to the object (if defined)

code = None
message = None
exception telethon.errors.rpcbaseerrors.ServerError(request, message, code=None)

Bases: telethon.errors.rpcbaseerrors.RPCError

An internal server error occurred while a request was being processed for example, there was a disruption while accessing a database or file storage.

code = 500
message = 'INTERNAL'
exception telethon.errors.rpcbaseerrors.TimedOutError(request, message, code=None)

Bases: telethon.errors.rpcbaseerrors.RPCError

Clicking the inline buttons of bots that never (or take to long to) call answerCallbackQuery will result in this “special” RPCError.

code = 503
message = 'Timeout'
exception telethon.errors.rpcbaseerrors.UnauthorizedError(request, message, code=None)

Bases: telethon.errors.rpcbaseerrors.RPCError

There was an unauthorized attempt to use functionality available only to authorized users.

code = 401
message = 'UNAUTHORIZED'

Sessions

These are the different built-in session storage that you may subclass.

class telethon.sessions.abstract.Session

Bases: abc.ABC

__weakref__

list of weak references to the object (if defined)

auth_key

Returns an AuthKey instance associated with the saved data center, or None if a new one should be generated.

cache_file(md5_digest, file_size, instance)

Caches the given file information persistently, so that it doesn’t need to be re-uploaded in case the file is used again.

The instance will be either an InputPhoto or InputDocument, both with an .id and .access_hash attributes.

clone(to_instance=None)

Creates a clone of this session file.

close()

Called on client disconnection. Should be used to free any used resources. Can be left empty if none.

dc_id

Returns the currently-used data center ID.

delete()

Called upon client.log_out(). Should delete the stored information from disk since it’s not valid anymore.

get_file(md5_digest, file_size, cls)

Returns an instance of cls if the md5_digest and file_size match an existing saved record. The class will either be an InputPhoto or InputDocument, both with two parameters id and access_hash in that order.

get_input_entity(key)

Turns the given key into an InputPeer (e.g. InputPeerUser). The library uses this method whenever an InputPeer is needed to suit several purposes (e.g. user only provided its ID or wishes to use a cached username to avoid extra RPC).

get_update_state(entity_id)

Returns the UpdateState associated with the given entity_id. If the entity_id is 0, it should return the UpdateState for no specific channel (the “general” state). If no state is known it should return None.

get_update_states()

Returns an iterable over all known pairs of (entity ID, update state).

classmethod list_sessions()

Lists available sessions. Not used by the library itself.

port

Returns the port to which the library should connect to.

process_entities(tlo)

Processes the input TLObject or list and saves whatever information is relevant (e.g., ID or access hash).

save()

Called whenever important properties change. It should make persist the relevant session information to disk.

server_address

Returns the server address where the library should connect to.

set_dc(dc_id, server_address, port)

Sets the information of the data center address and port that the library should connect to, as well as the data center ID, which is currently unused.

set_update_state(entity_id, state)

Sets the given UpdateState for the specified entity_id, which should be 0 if the UpdateState is the “general” state (and not for any specific channel).

takeout_id

Returns an ID of the takeout process initialized for this session, or None if there’s no were any unfinished takeout requests.

class telethon.sessions.memory.MemorySession

Bases: telethon.sessions.abstract.Session

auth_key

Returns an AuthKey instance associated with the saved data center, or None if a new one should be generated.

cache_file(md5_digest, file_size, instance)

Caches the given file information persistently, so that it doesn’t need to be re-uploaded in case the file is used again.

The instance will be either an InputPhoto or InputDocument, both with an .id and .access_hash attributes.

close()

Called on client disconnection. Should be used to free any used resources. Can be left empty if none.

dc_id

Returns the currently-used data center ID.

delete()

Called upon client.log_out(). Should delete the stored information from disk since it’s not valid anymore.

get_entity_rows_by_id(id, exact=True)
get_entity_rows_by_name(name)
get_entity_rows_by_phone(phone)
get_entity_rows_by_username(username)
get_file(md5_digest, file_size, cls)

Returns an instance of cls if the md5_digest and file_size match an existing saved record. The class will either be an InputPhoto or InputDocument, both with two parameters id and access_hash in that order.

get_input_entity(key)

Turns the given key into an InputPeer (e.g. InputPeerUser). The library uses this method whenever an InputPeer is needed to suit several purposes (e.g. user only provided its ID or wishes to use a cached username to avoid extra RPC).

get_update_state(entity_id)

Returns the UpdateState associated with the given entity_id. If the entity_id is 0, it should return the UpdateState for no specific channel (the “general” state). If no state is known it should return None.

get_update_states()

Returns an iterable over all known pairs of (entity ID, update state).

port

Returns the port to which the library should connect to.

process_entities(tlo)

Processes the input TLObject or list and saves whatever information is relevant (e.g., ID or access hash).

save()

Called whenever important properties change. It should make persist the relevant session information to disk.

server_address

Returns the server address where the library should connect to.

set_dc(dc_id, server_address, port)

Sets the information of the data center address and port that the library should connect to, as well as the data center ID, which is currently unused.

set_update_state(entity_id, state)

Sets the given UpdateState for the specified entity_id, which should be 0 if the UpdateState is the “general” state (and not for any specific channel).

takeout_id

Returns an ID of the takeout process initialized for this session, or None if there’s no were any unfinished takeout requests.

class telethon.sessions.sqlite.SQLiteSession(session_id=None)

Bases: telethon.sessions.memory.MemorySession

This session contains the required information to login into your Telegram account. NEVER give the saved session file to anyone, since they would gain instant access to all your messages and contacts.

If you think the session has been compromised, close all the sessions through an official Telegram client to revoke the authorization.

auth_key
cache_file(md5_digest, file_size, instance)

Caches the given file information persistently, so that it doesn’t need to be re-uploaded in case the file is used again.

The instance will be either an InputPhoto or InputDocument, both with an .id and .access_hash attributes.

clone(to_instance=None)

Creates a clone of this session file.

close()

Closes the connection unless we’re working in-memory

delete()

Deletes the current session file

get_entity_rows_by_id(id, exact=True)
get_entity_rows_by_name(name)
get_entity_rows_by_phone(phone)
get_entity_rows_by_username(username)
get_file(md5_digest, file_size, cls)

Returns an instance of cls if the md5_digest and file_size match an existing saved record. The class will either be an InputPhoto or InputDocument, both with two parameters id and access_hash in that order.

get_update_state(entity_id)

Returns the UpdateState associated with the given entity_id. If the entity_id is 0, it should return the UpdateState for no specific channel (the “general” state). If no state is known it should return None.

get_update_states()

Returns an iterable over all known pairs of (entity ID, update state).

classmethod list_sessions()

Lists all the sessions of the users who have ever connected using this client and never logged out

process_entities(tlo)

Processes all the found entities on the given TLObject, unless .save_entities is False.

save()

Saves the current session object as session_user_id.session

set_dc(dc_id, server_address, port)

Sets the information of the data center address and port that the library should connect to, as well as the data center ID, which is currently unused.

set_update_state(entity_id, state)

Sets the given UpdateState for the specified entity_id, which should be 0 if the UpdateState is the “general” state (and not for any specific channel).

takeout_id
class telethon.sessions.string.StringSession(string: str = None)

Bases: telethon.sessions.memory.MemorySession

This session file can be easily saved and loaded as a string. According to the initial design, it contains only the data that is necessary for successful connection and authentication, so takeout ID is not stored.

It is thought to be used where you don’t want to create any on-disk files but would still like to be able to save and load existing sessions by other means.

You can use custom encode and decode functions, if present:

  • encode definition must be def encode(value: bytes) -> str:.
  • decode definition must be def decode(value: str) -> bytes:.
static decode(x: str) → bytes
static encode(x: bytes) → str
save()

Called whenever important properties change. It should make persist the relevant session information to disk.

Connection Modes

The only part about network that you should worry about are the different connection modes, which are the following:

class telethon.network.connection.tcpfull.ConnectionTcpFull(ip, port, dc_id, *, loggers, proxy=None, local_addr=None)

Bases: telethon.network.connection.connection.Connection

Default Telegram mode. Sends 12 additional bytes and needs to calculate the CRC value of the packet itself.

packet_codec

alias of FullPacketCodec

class telethon.network.connection.tcpfull.FullPacketCodec(connection)

Bases: telethon.network.connection.connection.PacketCodec

encode_packet(data)

Encodes single packet and returns encoded bytes.

read_packet(reader)

Reads single packet from reader object that should have readexactly(n) method.

tag = None
class telethon.network.connection.tcpabridged.AbridgedPacketCodec(connection)

Bases: telethon.network.connection.connection.PacketCodec

encode_packet(data)

Encodes single packet and returns encoded bytes.

obfuscate_tag = b'\xef\xef\xef\xef'
read_packet(reader)

Reads single packet from reader object that should have readexactly(n) method.

tag = b'\xef'
class telethon.network.connection.tcpabridged.ConnectionTcpAbridged(ip, port, dc_id, *, loggers, proxy=None, local_addr=None)

Bases: telethon.network.connection.connection.Connection

This is the mode with the lowest overhead, as it will only require 1 byte if the packet length is less than 508 bytes (127 << 2, which is very common).

packet_codec

alias of AbridgedPacketCodec

class telethon.network.connection.tcpintermediate.ConnectionTcpIntermediate(ip, port, dc_id, *, loggers, proxy=None, local_addr=None)

Bases: telethon.network.connection.connection.Connection

Intermediate mode between ConnectionTcpFull and ConnectionTcpAbridged. Always sends 4 extra bytes for the packet length.

packet_codec

alias of IntermediatePacketCodec

class telethon.network.connection.tcpintermediate.IntermediatePacketCodec(connection)

Bases: telethon.network.connection.connection.PacketCodec

encode_packet(data)

Encodes single packet and returns encoded bytes.

obfuscate_tag = b'\xee\xee\xee\xee'
read_packet(reader)

Reads single packet from reader object that should have readexactly(n) method.

tag = b'\xee\xee\xee\xee'
class telethon.network.connection.tcpintermediate.RandomizedIntermediatePacketCodec(connection)

Bases: telethon.network.connection.tcpintermediate.IntermediatePacketCodec

Data packets are aligned to 4bytes. This codec adds random bytes of size from 0 to 3 bytes, which are ignored by decoder.

encode_packet(data)

Encodes single packet and returns encoded bytes.

obfuscate_tag = b'\xdd\xdd\xdd\xdd'
read_packet(reader)

Reads single packet from reader object that should have readexactly(n) method.

tag = None
class telethon.network.connection.tcpobfuscated.ConnectionTcpObfuscated(ip, port, dc_id, *, loggers, proxy=None, local_addr=None)

Bases: telethon.network.connection.connection.ObfuscatedConnection

Mode that Telegram defines as “obfuscated2”. Encodes the packet just like ConnectionTcpAbridged, but encrypts every message with a randomly generated key using the AES-CTR mode so the packets are harder to discern.

obfuscated_io

alias of ObfuscatedIO

packet_codec

alias of telethon.network.connection.tcpabridged.AbridgedPacketCodec

class telethon.network.connection.tcpobfuscated.ObfuscatedIO(connection)

Bases: object

__weakref__

list of weak references to the object (if defined)

header = None
static init_header(packet_codec)
readexactly(n)
write(data)
class telethon.network.connection.http.ConnectionHttp(ip, port, dc_id, *, loggers, proxy=None, local_addr=None)

Bases: telethon.network.connection.connection.Connection

connect(timeout=None, ssl=None)

Establishes a connection with the server.

packet_codec

alias of HttpPacketCodec

class telethon.network.connection.http.HttpPacketCodec(connection)

Bases: telethon.network.connection.connection.PacketCodec

encode_packet(data)

Encodes single packet and returns encoded bytes.

obfuscate_tag = None
read_packet(reader)

Reads single packet from reader object that should have readexactly(n) method.

tag = None

Helpers

Various helpers not related to the Telegram API itself

class telethon.helpers.TotalList(*args, **kwargs)

Bases: list

A list with an extra total property, which may not match its len since the total represents the total amount of items available somewhere else, not the items in this list.

Examples:

# Telethon returns these lists in some cases (for example,
# only when a chunk is returned, but the "total" count
# is available).
result = await client.get_messages(chat, limit=10)

print(result.total)  # large number
print(len(result))  # 10
print(result[0])  # latest message

for x in result:  # show the 10 messages
    print(x.text)
__repr__()

Return repr(self).

__str__()

Return str(self).

__weakref__

list of weak references to the object (if defined)

telethon.helpers.add_surrogate(text)
telethon.helpers.del_surrogate(text)
telethon.helpers.ensure_parent_dir_exists(file_path)

Ensures that the parent directory exists

telethon.helpers.generate_key_data_from_nonce(server_nonce, new_nonce)

Generates the key data corresponding to the given nonce

telethon.helpers.generate_random_long(signed=True)

Generates a random long integer (8 bytes), which is optionally signed

telethon.helpers.get_running_loop()
telethon.helpers.retry_range(retries, force_retry=True)

Generates an integer sequence starting from 1. If retries is not a zero or a positive integer value, the sequence will be infinite, otherwise it will end at retries + 1.

telethon.helpers.strip_text(text, entities)

Strips whitespace from the given surrogated text modifying the provided entities, also removing any empty (0-length) entities.

This assumes that the length of entities is greater or equal to 0, and that no entity is out of bounds.

telethon.helpers.within_surrogate(text, index, *, length=None)

True if index is within a surrogate (before and after it, not at!).