🤖 Gemini FileSearch Chat

Chat with your documents using Google Gemini 2.5 FileSearch and FastAPI

📸 Screenshots

Upload Interface

Chat Interface

✨ Features

• 📄 Multi-format Support - Upload PDF, TXT, or MD files
• 💬 Bilingual Chat - Works with both Thai and English
• 🔍 RAG-Powered - Accurate answers using Google FileSearch
• 🎨 Beautiful UI - Modern, responsive design with gradient backgrounds
• 📊 Smart Formatting - Automatic table and markdown rendering
• 🚀 Fast & Lightweight - Built with FastAPI for high performance

🛠️ Tech Stack

• Backend: FastAPI + Uvicorn
• AI Model: Google Gemini 2.5 Flash
• RAG: Google FileSearch API
• Frontend: Vanilla JavaScript (No frameworks!)
• Styling: Custom CSS3 with responsive design

📋 Prerequisites

• Python 3.11 or higher
• Google API Key (Get one here)

🚀 Quick Start

1. Clone the repository

   git clone https://github.com/spped2000/gemini-filesearch-chat.git
   cd gemini-filesearch-chat

2. Create virtual environment

   python -m venv .venv
   source .venv/bin/activate  # On Windows: .venv\Scripts\activate

3. Install dependencies

   pip install fastapi uvicorn python-multipart aiofiles google-genai python-dotenv

4. Configure environment variables

   cp .env.example .env
   # Edit .env and add your Google API key

5. Run the application

   python app.py

6. Open your browser

   http://localhost:8000
   

How to Use

1. Upload a Document:

   • Drag and drop a file onto the upload area, or
   • Click "Or Browse Files" to select a file from your computer
   • Supported formats: PDF, TXT, MD

2. Chat with Your Document:

   • Once uploaded, you'll be taken to the chat interface
   • Type your question in the input field
   • Press Enter or click the send button
   • The AI will search through your document and provide relevant answers

3. Start Over:

   • Click the × button next to the document name to close and upload a new document

🔧 Technical Details

Backend

• Framework: FastAPI with async/await support
• Server: Uvicorn ASGI server
• File Handling: python-multipart + aiofiles for async file operations
• Environment: python-dotenv for configuration

AI & RAG

• Model: Google Gemini 2.5 Flash
• RAG System: Google FileSearch API
• Document Processing: Automatic file upload and indexing
• Search: Vector-based document retrieval

Frontend

• Vanilla JavaScript - No frameworks, pure JS
• Modern CSS3 - Gradient backgrounds, animations
• Responsive Design - Works on desktop and mobile
• Markdown Rendering - Smart formatting for tables and structured data

Features Implementation

• Drag & Drop: Native HTML5 drag-drop API
• File Upload: FormData with async fetch
• Chat Interface: Real-time message streaming
• Table Formatting: Markdown to HTML conversion
• Bilingual Support: Automatic Thai/English detection

API Endpoints

• GET / - Main application interface
• POST /upload - Upload a document and create a file search store
• POST /chat - Send a message and get AI response
• GET /stores - List all uploaded documents
• DELETE /store/{store_id} - Delete a document store

Project Structure

file_search_test/
├── app.py                 # FastAPI backend
├── test_filesearch.py     # Original test script
├── .env                   # Environment variables
├── static/
│   ├── index.html        # Main UI
│   ├── styles.css        # Styling
│   └── script.js         # Frontend logic
└── uploads/              # Uploaded files directory

Notes

• Files are automatically renamed to ASCII-safe names to avoid encoding issues
• Each document upload creates a unique file search store
• Chat history is maintained per session but not persisted
• The application runs on port 8000 by default