Telegram: 1. Basic

The following tutorial shows how to run a regular Chatsky script in Telegram. It asks users for the ‘/start’ command and then loops in one place.

Here, LongpollingInterface class and python-telegram-bot library are used for accessing telegram API in polling mode.

Telegram API token is required to access telegram API.

# installing dependencies
%pip install -q chatsky[telegram]

Note: you may need to restart the kernel to use updated packages.

import os

from chatsky import (
    RESPONSE,
    TRANSITIONS,
    Pipeline,
    Transition as Tr,
    conditions as cnd,
    destinations as dst,
)
from chatsky.messengers.telegram import LongpollingInterface
from chatsky.utils.testing.common import is_interactive_mode

In order to integrate your script with Telegram, you need an instance of the _AbstractTelegramInterface class. Two of its child subclasses that can be instantiated are LongpollingInterface and WebhookInterface. The latter requires a webserver, so here we use long polling interface.

_AbstractTelegramInterface encapsulates the bot logic. The only required argument for a bot to run is a token. Some other parameters (such as host, port, interval, etc.) can be passed as keyword arguments (for their specs see documentations of the child subclasses.

Either of the two interfaces connect the bot to Telegram. They can be passed directly to a Chatsky Pipeline instance.

Note

You can also import LongpollingInterface under the alias of TelegramInterface from chatsky.messengers:

from chatsky.messengers import TelegramInterface

script = {
    "greeting_flow": {
        "start_node": {
            TRANSITIONS: [
                Tr(dst="greeting_node", cnd=cnd.ExactMatch("/start"))
            ],
        },
        "greeting_node": {
            RESPONSE: "Hi",
            TRANSITIONS: [Tr(dst=dst.Current())],
        },
        "fallback_node": {
            RESPONSE: "Please, repeat the request",
            TRANSITIONS: [
                Tr(dst="greeting_node", cnd=cnd.ExactMatch("/start"))
            ],
        },
    }
}

interface = LongpollingInterface(token=os.environ["TG_BOT_TOKEN"])

pipeline = Pipeline(
    script=script,
    start_label=("greeting_flow", "start_node"),
    fallback_label=("greeting_flow", "fallback_node"),
    messenger_interface=interface,
    # The interface can be passed as a pipeline argument.
)


if __name__ == "__main__":
    if is_interactive_mode():
        # prevent run during doc building
        pipeline.run()