-
Notifications
You must be signed in to change notification settings - Fork 0
Code snippets
This page can be read on its own to find the code snippet you need right now.
It is also a follow-up to the page Introduction to the API. If you come from there, you can leave your command line open and just try out a few of these snippets.
- Pure API
- General code snippets
- Advanced snippets
- What to read next?
Table of contents generated with markdown-toc
To fetch messages sent to your Bot, you can use the getUpdates API method.
Note: You don't have to use get_updates
if you are writing your bot with the telegram.ext
submodule, since telegram.ext.Updater
takes care of fetching all updates for you. Read more about that here.
>>> updates = bot.get_updates()
>>> print([u.message.text for u in updates])
>>> updates = bot.get_updates()
>>> print([u.message.photo for u in updates if u.message.photo])
You'll always need the chat_id
>>> chat_id = bot.get_updates()[-1].message.chat_id
These snippets usually apply to both ways of fetching updates. If you're using telegram.ext
, you can get the chat_id
in your handler callback with update.message.chat_id
.
Note: In general, you can send messages to users by passing their user id as the chat_id
.
If the bot has a chat with the user, it will send the message to that chat.
>>> bot.send_message(chat_id=chat_id, text="I'm sorry Dave I'm afraid I can't do that.")
Note: send_message
method (as any of send_*
methods of Bot
class) returns the instance of Message
class, so it can be used in code later.
This is a shortcut to bot.send_message
with sane defaults. Read more about it in the docs.
>>> update.message.reply_text("I'm sorry Dave I'm afraid I can't do that.")
Note: There are equivalents of this method for replying with photos, audio etc., and similar shortcuts exist throughout the library. Related PRs: #362, #420, #423
ᵀᴱᴸᴱᴳᴿᴬᴹ Use this to tell the user that something is happening on the bot's side:
>>> bot.send_chat_action(chat_id=chat_id, action=telegram.ChatAction.TYPING)
Alternatively, if you have several commands and don't want to repeat the above code snippet inside all commands, you can copy the snippet below and just decorate the callback functions with @send_typing_action
.
from functools import wraps
def send_typing_action(func):
"""Sends typing action while processing func command."""
@wraps(func)
def command_func(*args, **kwargs):
bot, update = args
bot.send_chat_action(chat_id=update.effective_message.chat_id, action=telegram.ChatAction.TYPING)
return func(bot, update, **kwargs)
return command_func
@send_typing_action
def my_handler(bot, update):
pass # Will send 'typing' action while processing the request.
>>> location_keyboard = telegram.KeyboardButton(text="send_location", request_location=True)
>>> contact_keyboard = telegram.KeyboardButton(text="send_contact", request_contact=True)
>>> custom_keyboard = [[ location_keyboard, contact_keyboard ]]
>>> reply_markup = telegram.ReplyKeyboardMarkup(custom_keyboard)
>>> bot.send_message(chat_id=chat_id,
... text="Would you mind sharing your location and contact with me?",
... reply_markup=reply_markup)
>>> bot.send_message(chat_id=chat_id,
... text="*bold* _italic_ `fixed width font` [link](http://google.com).",
... parse_mode=telegram.ParseMode.MARKDOWN)
>>> bot.send_message(chat_id=chat_id,
... text='<b>bold</b> <i>italic</i> <a href="http://google.com">link</a>.',
... parse_mode=telegram.ParseMode.HTML)
ᵀᴱᴸᴱᴳᴿᴬᴹ
To use MessageEntity, extract the entities from a Message object using parse_entities
.
Note: This method should always be used instead of the entities
attribute, since it calculates the correct substring from the message text based on UTF-16 codepoints - that is, it extracts the correct string even on when working with weird characters such as Emojis.
>>> entities = message.parse_entities()
There are many more API methods. To read the full API documentation, visit the Telegram API documentation or the library documentation of telegram.Bot
>>> bot.send_photo(chat_id=chat_id, photo=open('tests/test.png', 'rb'))
>>> bot.send_voice(chat_id=chat_id, voice=open('tests/telegram.ogg', 'rb'))
>>> bot.send_photo(chat_id=chat_id, photo='https://telegram.org/img/t_logo.png')
>>> bot.send_animation(chat_id, animation, duration=None, width=None, height=None, thumb=None, caption=None, parse_mode=None, disable_notification=False, reply_to_message_id=None, reply_markup=None, timeout=20, **kwargs)
See the online documentation
>>> bot.send_audio(chat_id=chat_id, audio=open('tests/test.mp3', 'rb'))
>>> bot.send_document(chat_id=chat_id, document=open('tests/test.zip', 'rb'))
In this example, image
is a PIL (or Pillow) Image
object, but it works the same with all media types.
>>> from io import BytesIO
>>> bio = BytesIO()
>>> bio.name = 'image.jpeg'
>>> image.save(bio, 'JPEG')
>>> bio.seek(0)
>>> bot.send_photo(chat_id, photo=bio)
Where photos
is a list of PhotoSize
objects and desired_size
is a tuple containing the desired size.
>>> def get_closest(photos, desired_size):
>>> def diff(p): return p.width - desired_size[0], p.height - desired_size[1]
>>> def norm(t): return abs(t[0] + t[1] * 1j)
>>> return min(photos, key=lambda p: norm(diff(p)))
>>> file_id = message.voice.file_id
>>> newFile = bot.get_file(file_id)
>>> newFile.download('voice.ogg')
Note: For downloading photos, keep in mind that update.message.photo
is an array of different photo sizes. Use update.message.photo[-1]
to get the biggest size.
>>> custom_keyboard = [['top-left', 'top-right'],
... ['bottom-left', 'bottom-right']]
>>> reply_markup = telegram.ReplyKeyboardMarkup(custom_keyboard)
>>> bot.send_message(chat_id=chat_id,
... text="Custom Keyboard Test",
... reply_markup=reply_markup)
See also: Build a menu with Buttons
>>> reply_markup = telegram.ReplyKeyboardRemove()
>>> bot.send_message(chat_id=chat_id, text="I'm back.", reply_markup=reply_markup)
The Unicode flag emoji for any country can by definition be calculated from the countries 2 letter country code. The following snippet only works in Python 3.
>>> OFFSET = 127462 - ord('A')
>>>
>>> def flag(code):
... code = code.upper()
... return chr(ord(code[0]) + OFFSET) + chr(ord(code[1]) + OFFSET)
...
>>> flag('de')
'🇩🇪'
>>> flag('us')
'🇺🇸'
>>> flag('ru')
'🇷🇺'
>>>
class NewMember(BaseFilter):
def filter(self, message):
if not message.new_chat_members:
return False
return True
def add_group(bot, update):
for members in update.message.new_chat_members:
bot.send_message(update.message.chat_id, text="{username} add group".format(username=members.username))
add_group_handle = MessageHandler(callback=add_group, filters=NewMember())
This decorator allows you to restrict the access of a handler to only the user_ids
specified in LIST_OF_ADMINS
.
from functools import wraps
LIST_OF_ADMINS = [12345678, 87654321]
def restricted(func):
@wraps(func)
def wrapped(bot, update, *args, **kwargs):
user_id = update.effective_user.id
if user_id not in LIST_OF_ADMINS:
print("Unauthorized access denied for {}.".format(user_id))
return
return func(bot, update, *args, **kwargs)
return wrapped
Add a @restricted
decorator on top of your handler declaration:
@restricted
def my_handler(bot, update):
pass # only accessible if `user_id` is in `LIST_OF_ADMINS`.
This parametrized decorator allows you to signal different actions depending on the type of response of your bot. This way users will have similar feedback from your bot as they would from a real human.
from functools import wraps
from telegram import ChatAction
def send_action(action):
"""Sends `action` while processing func command."""
def decorator(func):
@wraps(func)
def command_func(*args, **kwargs):
bot, update = args
bot.send_chat_action(chat_id=update.effective_message.chat_id, action=action)
return func(bot, update, **kwargs)
return command_func
return decorator
You can decorate handler callbacks directly with @send_action(ChatAction.<Action>)
or create aliases and decorate with them (more readable) .
send_typing_action = send_action(ChatAction.TYPING)
send_upload_video_action = send_action(ChatAction.UPLOAD_VIDEO)
send_upload_photo_action = send_action(ChatAction.UPLOAD_PHOTO)
With the above aliases, the following decorators are equivalent
@send_typing_action
def my_handler(bot, update):
pass # user will see 'typing' while your bot is handling the request.
@send_action(ChatAction.TYPING)
def my_handler(bot, update):
pass # user will see 'typing' while your bot is handling the request.
All possible actions are documented here.
Often times you will find yourself in need for a menu with dynamic content. Use the following build_menu
method to create a button layout with n_cols
columns out of a list of buttons
.
def build_menu(buttons,
n_cols,
header_buttons=None,
footer_buttons=None):
menu = [buttons[i:i + n_cols] for i in range(0, len(buttons), n_cols)]
if header_buttons:
menu.insert(0, header_buttons)
if footer_buttons:
menu.append(footer_buttons)
return menu
You can use the header_buttons
and footer_buttons
lists to put buttons in the first or last row respectively.
Replace the ...
in below snippet by an appropriate argument, as indicated in the InlineKeyboardButton documentation. If you want to use KeyboardButtons
, use ReplyKeyboardMarkup
instead of InlineKeyboardMarkup
.
button_list = [
InlineKeyboardButton("col1", callback_data=...),
InlineKeyboardButton("col2", callback_data=...),
InlineKeyboardButton("row 2", callback_data=...)
]
reply_markup = InlineKeyboardMarkup(util.build_menu(button_list, n_cols=2))
bot.send_message(..., "A two-column menu", reply_markup=reply_markup)
Or, if you need a dynamic version, use list comprehension to generate your button_list
dynamically from a list of strings:
some_strings = ["col1", "col2", "row2"]
button_list = [[KeyboardButton(s)] for s in some_strings]
This is especially useful if put inside a helper method like get_data_buttons
to work on dynamic data and updating the menu according to user input.
To handle the callback_data
, you need to set a CallbackQueryHandler
.
If you want to limit certain bot functions to group administrators, you have to test if a user is an administrator in the group in question. This however requires an extra API request, which is why it can make sense to cache this information for a certain time, especially if your bot is very busy.
This snippet requires this timeout-based cache decorator. (gist mirror)
Save the decorator to a new file named mwt.py
and add this line to your imports:
from mwt import MWT
Then, add the following decorated function to your script. You can change the timeout as required.
@MWT(timeout=60*60)
def get_admin_ids(bot, chat_id):
"""Returns a list of admin IDs for a given chat. Results are cached for 1 hour."""
return [admin.user.id for admin in bot.get_chat_administrators(chat_id)]
You can then use the function like this:
if update.message.from_user.id in get_admin_ids(bot, update.message.chat_id):
# admin only
Note: Private chats and groups with all_members_are_administrator
flag, are not covered by this snippet. Make sure you handle them.
The following example allows you to restart the bot from within a handler. It goes without saying that you should protect this method from access by unauthorized users, which is why we are using a Filters.user
filter. If you want multiple users to have access the restart command, you can pass a list of usernames as well. You can also filter by user IDs which is arguably a bit safer since they can't change. See the docs for more information.
This example is using closures so it has access to the updater
variable. Alternatively, you could make it global.
import os
import sys
from threading import Thread
# Other code
def main():
updater = Updater("TOKEN")
dp = updater.dispatcher
# Add your other handlers here...
def stop_and_restart():
"""Gracefully stop the Updater and replace the current process with a new one"""
updater.stop()
os.execl(sys.executable, sys.executable, *sys.argv)
def restart(bot, update):
update.message.reply_text('Bot is restarting...')
Thread(target=stop_and_restart).start()
# ...or here...
dp.add_handler(CommandHandler('r', restart, filters=Filters.user(username='@jh0ker')))
# ...or here, depending on your preference :)
updater.start_polling()
updater.idle()
if __name__ == '__main__':
main()
The following code allows you to store ConversationHandler States and UserData and reloading them when you restart the bot. Store procedure is executed every 60 seconds; to change this value, you can modify the time.sleep(60)
instruction.
You should declare the two methods at the end of the main method to use python closure for accessing ConversationHandler and UserData.
import time, threading, pickle
def main():
def loadData():
try:
f = open('backup/conversations', 'rb')
conv_handler.conversations = pickle.load(f)
f.close()
f = open('backup/userdata', 'rb')
dp.user_data = pickle.load(f)
f.close()
except FileNotFoundError:
utils.logging.error("Data file not found")
except:
utils.logging.error(sys.exc_info()[0])
def saveData():
while True:
time.sleep(60)
# Before pickling
resolved = dict()
for k, v in conv_handler.conversations.items():
if isinstance(v, tuple) and len(v) is 2 and isinstance(v[1], Promise):
try:
new_state = v[1].result() # Result of async function
except:
new_state = v[0] # In case async function raised an error, fallback to old state
resolved[k] = new_state
else:
resolved[k] = v
try:
f = open('backup/conversations', 'wb+')
pickle.dump(resolved, f)
f.close()
f = open('backup/userdata', 'wb+')
pickle.dump(dp.user_data, f)
f.close()
except:
utils.logging.error(sys.exc_info()[0])
def main():
...
loadData()
threading.Thread(target=saveData).start()
The following snippet pickles the jobs in the job queue periodically and on bot shutdown and unpickles and queues them again on startup. Since pickle
doesn't support threading primitives, they are converted.
Note: Race condition for asynchronous jobs that use job.job_queue
, job.removed
, job.schedule_removal
or job.enabled
while the job is being pickled.
import pickle
from threading import Event
from time import time
from datetime import timedelta
JOBS_PICKLE = 'job_tuples.pickle'
def load_jobs(jq):
now = time()
with open(JOBS_PICKLE, 'rb') as fp:
while True:
try:
next_t, job = pickle.load(fp)
except EOFError:
break # Loaded all job tuples
# Create threading primitives
enabled = job._enabled
removed = job._remove
job._enabled = Event()
job._remove = Event()
if enabled:
job._enabled.set()
if removed:
job._remove.set()
next_t -= now # Convert from absolute to relative time
jq._put(job, next_t)
def save_jobs(jq):
if jq:
job_tuples = jq._queue.queue
else:
job_tuples = []
with open(JOBS_PICKLE, 'wb') as fp:
for next_t, job in job_tuples:
# Back up objects
_job_queue = job._job_queue
_remove = job._remove
_enabled = job._enabled
# Replace un-pickleable threading primitives
job._job_queue = None # Will be reset in jq.put
job._remove = job.removed # Convert to boolean
job._enabled = job.enabled # Convert to boolean
# Pickle the job
pickle.dump((next_t, job), fp)
# Restore objects
job._job_queue = _job_queue
job._remove = _remove
job._enabled = _enabled
def save_jobs_job(bot, job):
save_jobs(job.job_queue)
def main():
# updater = Updater(..)
job_queue = updater.job_queue
# Periodically save jobs
job_queue.run_repeating(save_jobs_job, timedelta(minutes=1))
try:
load_jobs(job_queue)
except FileNotFoundError:
# First run
pass
# updater.start_[polling|webhook]()
# updater.idle()
# Run again after bot has been properly shut down
save_jobs(job_queue)
if __name__ == '__main__':
main()
If you haven't read the tutorial "Extensions – Your first Bot" yet, you might want to do it now.
- Wiki of
python-telegram-bot
© Copyright 2015-2025 – Licensed by Creative Commons
- Architecture Overview
- Builder Pattern for
Application
- Types of Handlers
- Working with Files and Media
- Exceptions, Warnings and Logging
- Concurrency in PTB
- Advanced Filters
- Storing data
- Making your bot persistent
- Adding Defaults
- Job Queue
- Arbitrary
callback_data
- Avoiding flood limits
- Webhooks
- Bot API Forward Compatiblity
- Frequently requested design patterns
- Code snippets
- Performance Optimizations
- Telegram Passport
- Bots built with PTB
- Automated Bot Tests