TG-Focus

English   简体中文

Table of Contents

• About
• A Simple Demo
• Supported Platforms
• Supported Languages
• How To Use
  • Hardware Requirement
  • Deployment
    • Using Prebuilt Binaries
    • Using Docker Image
• Filtering Rules
  • Filter Examples
• Q & A
  • Can I trust tg-focus?
  • What makes tg-focus a gentle User Bot?
• Development
• Contributing
• License

About

tg-focus is a comprehensive Telegram message filter that helps you focus. A gentle User Bot for message forwarding.

Several forms of text messages are supported:

1. Chat Message(Raw/Reply/Forward/Quoted)
2. Media Caption
3. Emoji
4. Channel Update

By customizing Focus Filters, tg-focus can forward any specific message from any chat to a dedicated chat named TG-FOCUS, that way users have full control of the messages they see.

Telegram is famous for its openness. While users are free to make a speech in any public chat, they should be equally free to choose what they see. Telegram's official client apps already have excellent support for the former need, tg-focus and other counterparts are trying to fill the blank in the latter one.

Notable features:

• Reliable & Consistent: Built directly on top of official C++ interfaces(TDLib). All communications towards Telegram servers are completely handled by TDLib.

• Easy to customize: Users can customize Focus Filter any time they like. The filter accepts both simple text and regular expressions. The configuration file is in human-readable TOML format.

• Ready to deploy: Releases are made in three forms: source, Docker image and prebuilt binary.

A Simple Demo

Below is a simple 9-step demonstration of a common use case: forwarding the messages that contains the keywords we want, ignoring the message that sent from the users we don't want.

Back to start

Back to start

Back to start

Back to start

Back to start

Back to start

Back to start

Supported Platforms

Operating System	Support(x86_64)
Windows	❌
Windows with WSL 1	✅
Windows with WSL 2	✅
AlmaLinux 9	✅
AlmaLinux 8	✅
Amazon Linux 2023	✅
Amazon Linux 2	❌
Amazon Linux 1	❌
Arch Linux	✅
Debian 12	✅
Debian 11	✅
Debian 10	✅
Fedora 39	✅
Fedora 38	✅
Fedora 37	✅
openSUSE Leap 15.5	✅
openSUSE Leap 15.4	✅
openSUSE Tumbleweed	✅
Oracle Linux 9	✅
Oracle Linux 8	✅
Oracle Linux 7	❌
Rocky Linux 9	✅
Rocky Linux 8	✅
Slackware 15.0	✅
Slackware 14.2	❌
Solus	✅
Ubuntu 23.10	✅
Ubuntu 23.04	✅
Ubuntu 22.04	✅
Ubuntu 20.04	✅
Void Linux	✅

Note: this is not an exhaustive list, missing platforms can be added to the list by opening an issue or a pull request.

Supported Languages

Language	Status
English (Hong Kong)	✅
English (United States)	✅
English (Zimbabwe)	✅
Chinese (Mainland China)	✅
Chinese (Hong Kong)	✅
Others	Add Your Language

Note: More languages can be supported! Follow the internationalization instructions(docs/i18n.org) to add your language.

How To Use

Before anything make sure you have:

1. A Telegram account
2. A Telegram API ID and the corresponding API HASH, they can be obtained at https://my.telegram.org.

Hardware Requirement

There is no rigorous restriction on hardware, here is a memory usage overview in 7 days:

day1	day2	day3	day4	day5	day6	day7
76MB	84MB	90MB	96MB	103MB	108MB	115MB

(NOTE: this table is just for reference, it is recorded by an account with about 30 public chats, most of which have over 1000 members and 500 messages produced per day)

Deployment

There are two different way to deploy:

Using Prebuilt Binaries

Step 1:

Download the binaries at GitHub releases page. Please check whether your platform is supported or not.

Step 2:

Login your Telegram account first, it needs API ID, API HASH, your phone number (in international format, cannot contain spaces, e.g., input +1234567 if your country code is +1 and phone number is 234567) and a login code, the first two can be obtained in https://my.telegram.org:

/path/to/tgf-conf auth

If you've logged in successfully, the last three lines will be:

[tgf INFO] Log in successfully!
[tgf INFO] Closing...
[tgf INFO] Closed

Step 3: (Optional)

Using different language other than default(English) by:

/path/to/tgf-conf lang XX-YY

Note: XX is the language code, YY is the region code, e.g. "en-US" represents "English (United States)", "en-GB" represents "English (United Kingdom)", etc. All valid codes can be found at http://www.lingoes.net/en/translator/langcode.htm

Step 4:

Launch tg-focus daemon:

# run in foreground
/path/to/tgf-focusd

# or run in background
nohup /path/to/tgf-focusd &

Step 5:

If logined successfullly, a new chat named TG-FOCUS will be created, the subsequent messages will be filtered and forwarded to this chat.

Step 6:

Note that by default, tg-focus will forward all messages you receive. You can change the Focus Filters(the filter configuration) any time you like, to custom the messages you'd like to receive on that chat. See examples.

/path/to/tgf-conf filters

(NOTE: This will open the GNU nano editor for configuration modification. After modification, press Ctrl-O, Enter, Ctrl-X to finish the work. As long as the configuration is valid, the last line of output shall be "Saving filters...")

Step 7:

Done.

Using Docker Image

Make sure have Docker or Podman installed on your machine:

Step 1:

Pull the image and run it in the background, assuming the newly created container's name is CONTAINER-NAME:

docker pull docker.io/micl2e2/tg-focus:latest
docker run -d docker.io/micl2e2/tg-focus

Step 2:

Login your telegram account first, here you should provide API ID, API HASH, your phone number(in international format), and a login code, the first two can be obtained in https://my.telegram.org:

docker exec -it CONTAINER-NAME tgf-conf auth

If you've logged in successfully, the last three lines will be:

[tgf INFO] Log in successfully!
[tgf INFO] Closing...
[tgf INFO] Closed

Step 3: (Optional)

Using different language other than default(English) by:

docker exec -it CONTAINER-NAME tgf-conf lang XX-YY

Note: XX is the language code, YY is the region code, e.g. "en-US" represents "English (United States)", "en-GB" represents "English (United Kingdom)", etc. All valid codes can be found at http://www.lingoes.net/en/translator/langcode.htm

Step 4:

If logined successfullly, a new chat named TG-FOCUS will be created, the subsequent messages will be filtered and forwarded to this chat.

Step 5:

Note that by default, tg-focus will forward all messages you receive. You can change the Focus Filters(the filter configuration) any time you like, to custom the messages you'd like to receive on that chat. See examples.

docker exec -it CONTAINER-NAME tgf-conf filters

(NOTE: This will open the embeded GNU nano editor for configuration modification. After modification, press Ctrl-O, Enter, Ctrl-X to finish the work. If the modified one is valid, the last line of output shall be "Saving filters...")

Step 6:

Done.

Step 7: (Optional)

Use following command if you want to destroy the instance and its Telegram-related credentials and resource:

docker rm --force CONTAINER-NAME

Filtering Rules

A Focus Filter is a filter used by tg-focus to match against the message's text content(text, emoji, or media caption). They are:

• title
• keywords
• no-keywords
• senders
• no-senders
• rej-senders

Users can add as many filters as they like, they accept simple text or regular expression.

When a message comes, they are tried one by one, if there is any filter than can match the message and not skip the message , it will be forwarded.

Currently Focus Filter can match:

• Chat title(title)
• Keywords(keywords)
• Senders(senders)

can skip:

• Keywords(no-keywords)
• Senders(no-senders)

can reject:

• Senders(rej-senders)

A message is forwarded if and only if:

• it is matched by a Focus Filter, and
• it is not skipped by that Focus Filter, and
• it is not rejected by any previous Focus Filter.

(Note that the first three rules can be regarded as whitelist, the second two rules can be regarded as weak blacklist, the third one can be regarded as strong blacklist)

Filter Examples

Say we want to forward all messages in a chat named "👍AnAwesomeChat👍":

works, the chat title is fully matched:

[[focus-filter]]
title = "👍AnAwesomeChat👍"

works, the chat title is partly matched:

[[focus-filter]]
title = "AnAwesomeChat"

works, the chat title is partly matched:

[[focus-filter]]
title = "Awesome"

works, the chat title is partly matched:

[[focus-filter]]
title = "👍"

works, the regular expression matches the chat title:

[[focus-filter]]
title = ".*Awesome.*"

Say we want to forward the messages that contain "football", but not "basketball", from a chat named "Sports User Group". We can write our filters as

[[focus-filter]]
title = "Sports User Group"
keywords = ["football", "soccer"]
no-keywords = ["basketball"]

Say we want the messages containing "football", or any other messages except the ones containing "basketball", and the ones sent by the user whose full name is "Basketball Lover", or the ones sent by the user whose user ID is "alice_love_basketball", from a chat named "SportsUserGroup". We can write the filter as:

# 1st
[[focus-filter]]
title = "Sports User Group"
keywords = ["football", "soccer"]

# 2nd
[[focus-filter]]
title = "Sports User Group"
no-keywords = ["basketball"]
no-senders = ["Basketball Love", "@alice_love_basketball"]

So these message will be forwarded:

• Alice: I love football! (matching the 1st)
• Bob: I play soccer. (matching the 1st)
• Alice: I am good at swimming. (matching the 2nd)
• Bob: Sometime I also play badminton. (matching the 2nd)
• Alice: I play football and basketball. (matching the 1st, skip the 2nd)

these will NOT be forwarded:

• Alice: I play badminton and basketball. (matching the 2nd's "no-keywords" means a reject)
• Basketball Lover: Hellop everyone! (a reject by the 2nd)
• Alice @alice_love_basketball: Hi, I am Alice, I love sports. (a reject by the 2nd)

Q&A

Can I trust tg-focus?

Absolutely yes!

First of all, consider that tg-focus is a GPL-licensed free software.

tg-focus inherits the openness of TDLib, its source as well as its dependencies' source are completely open. Additionally, all its binaries and OCI images will not be built by any individual, but by Github-hosted machines. All build details can be found in a dedicated repository tg-focus-cicd.

Telegram bot is useful for some specific tasks, but not sufficient to be a comprehensive message filter. Therefore, during the authorization process, tg-focus needs to log in user's account, which in turn needs a phone number and login code, just as any functional Telegram client would require. It only saves API ID/HASH on the user's machine because these two are required each time client initialization. But under no circumstances would tg-focus save user's phone number or any other credentials.

What makes tg-focus a gentle User Bot?

TDLib states that it is a library for building Telegram clients. By embedding TDLib, tg-focus, like any other User Bot, is able to use all TDLib's public APIs. While it opens the possibility for many powerful features, it also means User Bots have far more privileges than a general-purpose Bot. This can be a double-edged sword.

There are already some examples where some of the good-purpose User Bots disappoint their users because of their misuse of Telegram APIs, such as repeatedly sending random messages to users' random chats, or other unmanaged behaviors. This is probably one of the reasons why Telegram User Bots have a bad reputation.

However, tg-focus will not follow their paths, we have one strict principle: all message collected by tg-focus will only be forwarded to a newly created chat.

With this principle, it is firmly guaranteed that no messages generated by tg-focus will be ever sent to the wrong chat. Additionally, there is no reason for tg-focus users to worry about whether any existing or new features will be harmful or not. By the time of updates, tg-focus will remain gentle.

Development

See docs/development.org.

Contributing

tg-focus is still far from perfect, any form of contribution is welcomed!

• Ask question (open an issue with label )
• Bug Report (open an issue with label )
• Feature Request (open an issue with label )
• Documentation Improvement
• Code Contribution (tg-focus follows GCC code style)

License

tg-focus is licensed under GNU General Public License Version 3.0.