Mastering Telethon

You’ve come far! In this section you will learn best practices, as well as how to fix some silly (yet common) errors you may have found. Let’s start with a simple one.

Asyncio madness

We promise asyncio is worth learning. Take your time to learn it. It’s a powerful tool that enables you to use this powerful library. You need to be comfortable with it if you want to master Telethon.

AttributeError: 'coroutine' object has no attribute 'id'

You probably had a previous version, upgraded, and expected everything to work. Remember, just add this line:

import telethon.sync

If you’re inside an event handler you need to await everything that makes a network request. Getting users, sending messages, and nearly everything in the library needs access to the network, so they need to be awaited:

@client.on(events.NewMessage)
async def handler(event):
    print((await event.get_sender()).username)

You may want to read https://lonamiwebs.github.io/blog/asyncio/ to help you understand asyncio better. I’m open for feedback regarding that blog post

Entities

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 is an entire section on this at Users, Chats and Channels due to their importance.

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 short, unlike in most bot API libraries where you use the ID, you should not use the ID if you have the input entity. This is OK:

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

However, this is better:

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

Note that this also works for message instead of event. Telegram may not send the sender information, so if you want to be 99% confident that the above will work you should do this:

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

Methods are able to make network requests to get information that could be missing. Properties will never make a network request.

Of course, it is convenient to IDs or usernames for most purposes. It will be fast enough and caching with client.get_input_entity(...) will be a micro-optimization. However it’s worth knowing, and it will also let you know if the entity cannot be found beforehand.

Note

Sometimes Telegram doesn’t send the access hash inside entities, so using chat or sender may not work, but input_chat and input_sender while making requests definitely will since that’s what they exist for. If Telegram did not send information about the access hash, you will get something like “Invalid channel object” or “Invalid user object”.

Debugging

Please enable logging:

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

Change it for logging.DEBUG if you are asked for logs. It will save you a lot of headaches and time when you work with events. This is for errors.

Debugging is really important. Telegram’s API is really big and there is 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:

user = client.get_entity('Lonami')
print(user)

That will show a huge line similar to the following:

User(id=10885151, is_self=False, contact=False, mutual_contact=False, deleted=False, bot=False, bot_chat_history=False, bot_nochats=False, verified=False, restricted=False, min=False, bot_inline_geo=False, access_hash=123456789012345678, first_name='Lonami', last_name=None, username='Lonami', phone=None, photo=UserProfilePhoto(photo_id=123456789012345678, photo_small=FileLocation(dc_id=4, volume_id=1234567890, local_id=1234567890, secret=123456789012345678), photo_big=FileLocation(dc_id=4, volume_id=1234567890, local_id=1234567890, secret=123456789012345678)), status=UserStatusOffline(was_online=datetime.datetime(2018, 1, 2, 3, 4, 5, tzinfo=datetime.timezone.utc)), bot_info_version=None, restriction_reason=None, bot_inline_placeholder=None, lang_code=None)

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

username = user.username

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

print(user.stringify())

Will show a much better:

User(
    id=10885151,
    is_self=False,
    contact=False,
    mutual_contact=False,
    deleted=False,
    bot=False,
    bot_chat_history=False,
    bot_nochats=False,
    verified=False,
    restricted=False,
    min=False,
    bot_inline_geo=False,
    access_hash=123456789012345678,
    first_name='Lonami',
    last_name=None,
    username='Lonami',
    phone=None,
    photo=UserProfilePhoto(
        photo_id=123456789012345678,
        photo_small=FileLocation(
            dc_id=4,
            volume_id=123456789,
            local_id=123456789,
            secret=-123456789012345678
        ),
        photo_big=FileLocation(
            dc_id=4,
            volume_id=123456789,
            local_id=123456789,
            secret=123456789012345678
        )
    ),
    status=UserStatusOffline(
        was_online=datetime.datetime(2018, 1, 2, 3, 4, 5, tzinfo=datetime.timezone.utc)
    ),
    bot_info_version=None,
    restriction_reason=None,
    bot_inline_placeholder=None,
    lang_code=None
)

Now it’s easy to see how we could get, for example, the was_online time. It’s inside status:

online_at = user.status.was_online

You don’t need to print everything to see what all the possible values can be. You can just search in http://lonamiwebs.github.io/Telethon/.

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

from telethon import types

if isinstance(user.status, types.UserStatusOffline):
    print(user.status.was_online)

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.

Chat or User From Messages

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
message.get_chat()
# ...etc

SenderGetter is similar:

message.user_id
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.

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!).

However, some people have a lot of trouble with SQLite, especially in Windows:

...some lines of traceback
'insert or replace into entities values (?,?,?,?,?)', rows)
sqlite3.OperationalError: database is locked

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 two 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.

Final Words

Now you are aware of some common errors and use cases, this should help you master your Telethon skills to get the most out of the library. Have fun developing awesome things!