telegram-stats-bot 0.1.0

A logging and statistics bot for Telegram based on python-telegram-bot.

telegram-stats-bot

Telegram-stats-bot is a simple bot that lives in your Telegram group, logging messages to a Postgresql database and serving statistical tables and plots to users as Telegram messages.

Table of contents

• Introduction

• Basic Requirements

• Installation

• Setup

• Fetching Stats

• counts

  • hours

  • days

  • week

  • history

  • corr

  • delta

• License

Introduction

This software is intended to be run on a server, handling updates for a bot user with a single bot per channel (multi-channel support could be added at some point if there is interest), using the excellent Python-telegram-bot library.

The bot is still in active development but at the moment, it features:

• Message logging to Postgresql database with optional JSON file backup

• Statistics output for users in group as Telegram messages, with optional filtering by date or limiting to the querying user. Some statistics are more useful than others, but they are mainly intended to be fun for users to play with.

  • Tables:

    • Most active users

    • A user’s message time correlation with other users

    • A user’s median message time difference with other users

  • Plots:

    • Message activity by hour of day

    • Message activity by day of week

    • Message activity over the week by hour and day

    • Message activity history

Basic Requirements

• Python 3.7+

• A Telegram bot token with privacy mode disabled (needed to log messages)

  • See here for details

• Postgresql (Tested with 12.3, but there shouldn’t be anything that won’t work with 9.4 or up)

  • This can be on a different system than telegram-stats-bot and requires either table creation permissions on a database or database can be pre-initialized following the setup in db.py

Installation

The easiest way to install or upgrade is with pip:

$ pip install telegram-stats-bot --upgrade

Or you can install the latest git version:

$ pip install git+git://github.com/mkdryden/telegram-stats-bot.git

Installing directly with setup.py should also work.

Setup

Once installed, you can run the bot by calling the main module with a few required arguments:

$ python -m telegram_stats_bot.main BOT_TOKEN CHAT_ID POSTGRESQL_URL

• BOT_TOKEN: Your bot’s token e.g., 110201543:AAHdqTcvCH1vGWJxfSeofSAs0K5PALDsaw

• CHAT_ID: The chat id to monitor (will be a large integer, possibly negative, if unknown, set to 0 and see below)

• POSTGRESQL_URL: Connection information in the form: postgresql://USERNAME:PASSWORD@ADDRESS/DB_NAME

  • if DB_NAME exists, there must not be tables called messages_utc, user_events, or user_names with incorrect columns

On startup, the bot will attempt to create the database and tables, if they do not already exist. If you do not know the chat’s id and have set it to 0 as mentioned above, you can send the /chatid command inside the group, and the bot will reply with it, then restart the bot with the id. If you have forgotten to disable privacy mode, an error will be logged in the terminal.

The bot will now log all messages in the group, but will only respond to users who have sent a message that has been logged previously (and this list is only updated once an hour, so if you’re impatient, you can restart the bot after you’ve sent a message to trigger the update). You can see if messages are being logged correctly by reviewing the terminal output. You should see a line like 2020-06-04 02:08:39,212 - __main__ - INFO - 8, whenever a message is logged.

Fetching Stats

To fetch stats, simply message the bot, either inside the group being logged, or in a direct message, using the /stats command. /stats with no arguments prints the table of most active users and other statistics are available through various subcommands. All commands are documented and the built in help can be displayed with /stats -h or stats <subcommand> -h.

Most commands have optional arguments that change the behaviour of the output. Nearly all have:

• -start and -end followed by a timestamp (e.g., 2019, 2019-01, 2019-01-01, “2019-01-01 14:21”) specify the range of data to fetch, otherwise all available data will be used. Either or both options can be given.

• -me calculates statistics for the user sending the command, rather than all chat users.

Sample outputs of each available subcommand follow.

counts

/stats counts returns a list of the most active users in the group.

              User  Total Messages  Percent
        @ACoolUser           42150      7.0
        @NumberOne           37370      6.2
 @WinstonChurchill           32668      5.4
          @AAAAAAA           32134      5.4
@WhereAreMyManners           30481      5.1
@TheWorstOfTheBest           28705      4.8

hours

/stats hours returns a plot of message frequency for the hours of the day.

days

/stats days returns a plot of message frequency for the days of the week.

week

/stats week returns a plot of total messages over the data period by day of week and hour of day.

history

/stats history returns a plot of messages versus date.

corr

/stats corr returns a list of users with the highest and lowest message time correlations with the requesting user.

User Correlations for @TheManWhoWasThursday
HIGHEST CORRELATION:
@MyGoodFriend         0.335
@Rawr                 0.302
@MangesUnePoutine     0.284
@GreenBlood           0.251
@TooMuchVacuum        0.235

LOWEST CORRELATION:
@Shiny                0.146
@BlueDog              0.142
@CoolCat              0.122
@EatMe                0.116
@JustPassingBy        0.106

delta

/stats delta returns a list of users with the shortest differences in message times with the requesting user.

Median message delays for @KingLeer and:
@PolyamorousPasta     00:03:23
@AggressiveArgon      00:04:43
@AdjectiveNoun        00:08:27
@SuperSalad           00:09:05
@ABoredProgrammer     00:09:06

The Future

Telegram-stats-bot is a work in progress. New stats will be added, but no guarantees that the database structure will stay constant if Telegram’s message structure changes or I need to change something to make a new statistic work.

License

Telegram-stats-bot is free software: You can redistribute it and/or modify it under the terms of the GNU General Public License v3.0 or later. Derivative works must also be redistributed under the GPL v3 or later.