ChatDocs: A private knowledge base chatbot to chat with your documents

This is a RAG chatbot that uses LangChain and LangGraph to build a chatbot that can answer questions based on a given document.

Project Structure

The project is structured as follows:

chat-docs/
    ├── dockerfiles/
    │     ├── backend.Dockerfile
    │     └── frontend.Dockerfile
    ├── docs/
    │     ├── cover.jpg
    │     └── image.png
    ├── src/
    │    ├── __init__.py
    │    ├── backend/
    │    │     ├── __init__.py
    │    │     ├── ...
    │    │     └── ...
    │    ├── rag/
    │    │     ├── __init__.py
    │    │     ├── ...
    │    │     └── ...
    │    ├── ui/
    │    │     ├── __init__.py
    │    │     ├── ...
    │    │     └── ...
    │    └── __init__.py
    ├── .env
    ├── .gitignore
    ├── .dockerignore
    ├── docker-compose.yml
    ├── requirements.txt
    ├── README.md
    └── LICENSE

Setup

git clone https://github.com/sayedshaun/chat-docs.git
cd chat-docs

Setup Environment Variables

Create .env file and add the following environment variables: You can customize the environment variables as per your requirements.

OLLAMA_MODEL=llama3.2
POSTGRES_DB=chatdocs
CHROMADB_INDEX=chatdocs
POSTGRES_USER=user
POSTGRES_PASSWORD=password
CHROMADB_PRESET=./vectorstore
BACKEND_PORT=9010
FRONTEND_PORT=9011
CHROMADB_PORT=9012
POSTGRES_PORT=9013

All Valid Values for Environment Variables

Environment Variable	Valid Values	Example
GEMINI_API_KEY	str	API_KEY
GEMINI_MODEL	str	gemini-2.5-flash
GEMINI_EMBEDDING_MODEL	str	gemini-embedding-001
HUGGINGFACE_EMBEDDING_MODEL	str	all-MiniLM-L6-v2
OLLAMA_MODEL	str	llama3.2
CHUNK_SIZE	int	1000
CHUNK_OVERLAP	int	200
CHROMADB_INDEX	str	chatdocs
CHROMADB_PRESET	str	./vectorstore
POSTGRES_USER	str	user
POSTGRES_PASSWORD	str	password
POSTGRES_DB	str	chatdocs
BACKEND_PORT	int	9010
FRONTEND_PORT	int	9011
CHROMADB_PORT	int	9012
POSTGRES_PORT	int	9013

If you use ollama model, you need to set up ollama first.

Usage

Start Server

docker compose up --build

API

ChatDocs built with FastAPI, so you can view the API documentation with Swagger UI. Here is the reference of the available endpoints:

API Reference

Action	Endpoint
Ask Question	http://{HOST}:{BACKEND_PORT}/ask
Get stream response	http://{HOST}:{BACKEND_PORT}/ask_stream
Upload Files	http://{HOST}:{BACKEND_PORT}/upload_files
Update Database	http://{HOST}:{BACKEND_PORT}/update_database

Prebuilt UI

ChatDocs provides a prebuilt UI, you can access it at http://{HOST}:{FRONTEND_PORT}, e.g. http://localhost:9011.

Supported Document

File Type	Extensions	Description
PDF	.pdf	Portable Document Format
Word	.doc, .docx	Microsoft Word Documents
Text	.txt	Plain Text Files
Markdown	.md	Markdown Files
HTML	.html, .htm	HyperText Markup Language
PowerPoint	.ppt, .pptx	Microsoft PowerPoint Files
Excel	.xls, .xlsx	Microsoft Excel Files
XML	.xml	XML Data Files
Email	.eml	Email Message Files
JSON	.json	JavaScript Object Notation
CSV	.csv	Comma-Separated Values

Supported Language

Based on embedding model and document type. text pictures does not support yet but will be available soon.