@textUtilsBot

A simple bot for Telegram, that has some useful handlers of inline queries to make text conversions.

The bot supports both ways of communication with Telegram: polling (intended for debugging) and web hooks (for production use).

Feel free to suggest additional functionality, new features, and configuration examples! Just create an issue or send me a pull request.

Requirements

To execute the tests:

How to run the bot in debug mode from PyCharm

To run the bot in debug mode from PyCharm, add a new Python configuration:

Script path: app/bot.py
Working directory: absolute path to the repo directory

Also, you must copy the examples/config.py file into the app/data directory. After that, set your TOKEN. That's all! Click the run button!

Don't forget to enable VPN if Telegram is banned in your country!

How to deploy

As of v2.0.0, there are two options to run the bot. The first one is to use a Docker container. To use this option, you need Docker and Docker Compose to be installed on your system. This approach allows you to don't care about the version of the Python interpreter installed on your machine.

The other way is to run the bot within a virtual environment. There is a special initialization script, that can help you on Linux. On Windows, you have to manually run venv, install all dependencies using pip and copy the config.py file from the examples/ directory into app/data/. Note, however, that there is only built-in support for Linux based servers for production use. But it's OK to utilize Windows machines for development and debugging.

Using Docker

Clone the repository.
Configure nginx or any other front-end web server (keep reading for more information).
Run the ./start-container.sh script.
Edit app/data/config.py according to your environment.
Run ./start-container.sh again.

Using venv

Clone the repository.
./init.sh
Configure nginx or any other front-end web server (keep reading for more information).
Edit app/data/config.py according to your environment.
Run ./start.sh using one of the following ways:
- directly (nohup ./start.sh &>/dev/null &);
- configure supervisord or systemd to do it for you (see exemplary configuration files);
- configure any other service manager on your choice (but you have to write configuration by yourself).

Common notes

I encourage you to use the application as a local server that receives requests from Telegram from an external web server. In such case, you can configure a TLS certificate for all at once. This is especially handy in the case if you're using Cloudflare services.

If you use nginx, look at the exemplary configuration file. In other cases, you have to write it by yourself (and send me a pull request, of course 😊).

Nevertheless, nobody forbids you to use the application as a front-end web server directly. However, to reach this, you have to make some changes in app/bot.py:

import ssl
...
SSL_CERT = "/your/path/to/ssl/certificate.pem"
SSL_PRIV = "/your/path/to/ssl/private.key"
...
context = ssl.SSLContext(ssl.PROTOCOL_TLSv1_2)
context.load_cert_chain(SSL_CERT, SSL_PRIV)
web.run_app(app, host='0.0.0.0', port=APP_PORT, ssl_context=context)

How to contribute a new feature

As of v1.1.0, it became much easier to add new text processors to the bot. If earlier you had to embed calls of new functions into the entangled and confusing inline_request_handler function itself, now this handler is implemented by using dynamic autoloading of modules from the strconv package. All you need is to add a new module there, implement the TextProcessor interface (strictly speaking in terms of Python, extend the ABC class) and append strings with a key hint_{classname_in_snake_case} to the localizations.ini file.

For more information, you may look at either real-world examples (banner, typographer) or the fictional ones, written exclusively for educational purposes, below.

Exemplary text processor

We're gonna create a very simple and stupid text processor. Let's, for instance, search input queries for the word anime and block other processors if it's found, suggesting only one message with text Baka-chan! ("baka" means a fool in Japanese), stylized by the bold font.

There are several ABC and mix-in classes in the txtproc.abc module. The most important is the TextProcessor class, of course. It defines the interface of all text processors and implements the name property, used by the localization system.

Text processors can be exclusive, reversible and universal.

The exclusivity means here that there is only a subset of processors which should handle the query. If such processors are found, the others will be discarded. It's mostly useful for decoders.

The reversibility means, obviously, that the transformed string may be transformed back and this reversed transformation has sense. Mostly useful for encoders.

Universal text processors are just the processors that can handle any text. That's it.

Implementation

Firstly, let's look at the TextInteface as a whole and implement the processor without other mix-in classes.

# app/strconv/baka.py

import re
# In fact, there is a shortcut for TextProcessor in the 'txtproc' module itself,
# but it's mostly intended for the loading system. Later you'll understand why
# you almost always should prefer the `abc` submodule.
from txtproc.abc import TextProcessor

class BakaDetector(TextProcessor):
    # Discard all non-exclusive processors if this one is matched.
    is_exclusive = True
    # By default, the text will be rendered by Telegram as-is.
    use_html = True

    # Let the system determine if the processor is able to handle a query.
    @classmethod
    def can_process(cls, query: str) -> bool:
        return re.search(r"\banime\b", query, re.IGNORECASE) is not None

    # Return the resulted message.
    def process(self, query: str) -> str:
        return "<b>Baka-chan!</b>"

    # Description of the message mustn't contain HTML tags since they will be rendered as-is.
    def get_description(self, query: str) -> str:
        return "Baka-chan!"

OK. Let's make the class a bit simpler by using the built-in mix-ins.

import re
from txtproc.abc import *

# Here we describe our text processor in one line.
class BakaDetector(Exclusive, HTML, TextProcessor):    
    @classmethod
    def can_process(cls, query: str) -> bool:
        return re.search(r"\banime\b", query, re.IGNORECASE) is not None

    def process(self, query: str) -> str:
        return "<b>Baka-chan!</b>"

    def get_description(self, query: str) -> str:
        return "Baka-chan!"

Note that you're, as a text processor developer, responsible for escaping of the input query! Use the strconv.util.escape_html function for that. For the same reason, you're not allowed to use Markdown, since Telegram flavored Markdown is much harder to escape properly.

Well, let's write another simple text processor. For example, let's convert strings to sequences of their numerical codes and vice versa.

# app/strconv/char2code.py

from typing import *
from txtproc.abc import *

# The 'Encoder' class is just a shortcut for "Reversible, TextProcessor". 
class CharEncoder(Universal, Encoder):
    def process(self, query: str) -> str:
        codes = [str(ord(c)) for c in query]
        return " ".join(codes)

# The 'Decoder' class is a shortcut for "Exclusive, TextProcessor". 
class CharDecoder(Decoder):
    @classmethod
    def can_process(cls, query: str) -> bool:
        codes = query.split()
        return all(cls._try_parse_int(code) for code in codes)

    def process(self, query: str) -> str:
        codes = query.split()
        chars = [chr(int(code)) for code in codes]
        return "".join(chars)

    @staticmethod
    def _try_parse_int(s: str) -> Optional[int]:
        try:
            return int(s)
        except ValueError:
            return None

Use the Encoder and Decoder classes for pairs of reversible text processors.

Titles and localization

Do you know what we forgot to do? Titles for our processors, that will be used by the localization system! Let's fix it. Go to the localizations.ini file and add the following lines:

# Under [DEFAULT] section
hint_baka_detector = Baka?
hint_char_encoder = Char codes
hint_char_decoder = Get the string back

...

# Under [ru] section
hint_baka_detector = Бака?
hint_char_encoder = Коды символов
hint_char_decoder = Вернуть строку обратно

If you want your processors to be described in the help, don't forget to add the following lines:

# Under [DEFAULT] section
help_baka_detector = You're a baka if you watch anime!
help_char_encoder = Converts strings to sequences of their numerical codes.

...

# Under [ru] section
help_baka_detector = Если ты смотришь аниме, то ты бака!
help_char_encoder = Превращает строки в последовательности их числовых кодов.

Much better! Actually, there is one more thing we need to do. Write tests for our code!

Testing

Code without tests is buggy and bad code.
— Mattias Petter Johansson

Let's create files test_baka.py and test_char2code.py files inside the tests/test_strconv/ directory.

# tests/test_strconv/test_baka.py

import pytest
from strconv.baka import BakaDetector

@pytest.fixture
def processor() -> BakaDetector:
    return BakaDetector()

def test_matcher(processor):
    assert not processor.can_process("hello world")
    assert processor.can_process("hello anime")
    assert processor.can_process("Hello Anime!")

def test_processor(processor):
    assert processor.process("hello anime") == "<b>Baka-chan!</b>"
    assert processor.get_description("hello anime") == "Baka-chan!"

    assert processor.process("Hello Anime!") == "<b>Baka-chan!</b>"
    assert processor.get_description("Hello Anime!") == "Baka-chan!"

# tests/test_strconv/test_baka.py

from strconv.char2code import CharEncoder, CharDecoder

def test_encoder():
    assert CharEncoder().process("hello world") == "104 101 108 108 111 32 119 111 114 108 100"

def test_decoder():
    decoder = CharDecoder()
    assert not decoder.can_process("hello world")
    assert decoder.can_process("104 101 108 108 111 32 119 111 114 108 100")
    assert decoder.process("104 101 108 108 111 32 119 111 114 108 100") == "hello world"

Run tests on Linux

PYTHONPATH=app pytest

Run tests on Windows

Command Prompt:

set PYTHONPATH=app && pytest

PowerShell:

$env:PYTHONPATH='app'; pytest

All tests must finish successfully.

What's next?

At this point, our work is done, and we're ready to commit these two features and make a pull request! Surely, we won't do it now, but if you develop a new awesome feature, feel free to send it to the upstream! All text processors have a chance to be included in the official hosted version of the bot. Especially if they're not universal and intended for some special cases.

kozalosev / textUtilsBot

readme