FbChatSummarizer

A chatbot that summarizes Messenger group conversations using AI and stores relevant data in a PostgreSQL database.

Features

• Fetches messages from Facebook Messenger groups using fbchat_muqit
• Summarizes conversations with OpenAI's API (also supports DeepSeek and Gemini)
• Stores user data in a PostgreSQL database (messages will be stored in the future)
• Currently only supports text messages (image support planned)

Requirements

• SQLAlchemy
• openai
• fbchat_muqit
• Note: The fbchat_muqit library has a minor issue where fetched messages are concatenated. Until it's addressed, you can use my fork with a quick fix.
• Python 3.x

Example usage

Start the bot by run the following command:

  python bot.py 

It will begin listening for incoming messages.

1. Add a group

You can add a group using the /addGroup command, in two ways:

Add a different group
If you want to add a group that is different from where you are sending the command, you need to provide it's thread_id. It can be found by looking at the URL in your browser when you're in a Messenger chat: https://www.messenger.com/t/6161463540609695

In this case, 6161463540609695 is the thread_id for the group.

The command would be: /addGroup "Group Name" 6161463540609695

Add the current group
If you want to add the group where the command is being sent, you only need to provide the group name. The bot will use the current groups's thread_id. /addGroup "Friends Chat"

2. Summarizing a conversation

To summarize the "Friends Chat" group from 12:00 to 14:00:

• First, use the /listGroups command to get the group index.
• Then, use the /recap command with the group index and time range: /recap 1 12:00-14:00

Installation

1. Clone the repository:

   git clone https://github.com/robertlippai/FbChatSummarizer.git
   cd FbChatSummarizer

2. Install dependencies:

    pip install -r requirements.txt

3. Set up the database:

• Ensure a PostgreSQL based database is installed and running properly.

4. Set up the environment variables:

   • Create a .env file in the root directory of the project.
   • Add the following variables to the .env file:

   OPENAI_API_KEY=your_openai_api_key
   DATABASE_URL=your_database_connection_url
   AI_ENDPOINT_URL="https://generativelanguage.googleapis.com/v1beta/"  # Default endpoint for Gemini
   AI_MODEL_NAME="models/gemini-2.0-flash-exp" 

   Some example endpoints:

   • OpenAI: "https://api.openai.com/v1"
   • For Gemini (Google DeepMind): "https://generativelanguage.googleapis.com/v1beta/"
   • For DeepSeek AI: "https://api.deepseek.com/v1"

5. Authenticate with Facebook:

   • Fill cookies.json with your extracted Facebook cookies.
   • You can obtain them using the C3C FbState Utility.
   • The file shuld be placed in the root directory of the project.

Future Plans

• Support for images with AI-generated descriptions
• Automatic Summarization: If a user misses a certain number of messages, the bot will automatically generate and send a recap of the missed conversation as soon as they open the group.
• Different types of summary length: short, detailed, or bullet-point.