NLSearch - Natural Language Code Search

A web application that converts natural language queries into Sourcegraph code search queries using Deep Search API.

Features

• 🔍 Natural language to search query translation
• 🌐 Clean, modern web interface
• ⚡ Real-time Deep Search API integration
• 🎯 Example queries to get started quickly
• 📊 Source attribution for search results

Prerequisites

• Go 1.23 or higher
• A Sourcegraph access token
• Access to a Sourcegraph instance (default: sourcegraph.com)

Quick Start

1. Clone or navigate to the project:

   cd nlsearch

2. Set up environment variables:

   cp .env.example .env
   # Edit .env and add your Sourcegraph token

3. Run the application:

   ./run.sh

4. Open your browser: Navigate to http://localhost:8080

Configuration

Configure the app using environment variables:

Variable	Description	Default
SOURCEGRAPH_TOKEN	Your Sourcegraph access token	Required
SOURCEGRAPH_URL	Sourcegraph instance URL	https://sourcegraph.com
PORT	Server port	8080

Getting a Sourcegraph Token

1. Go to your Sourcegraph instance (e.g., https://sourcegraph.com)
2. Navigate to Settings → Access tokens
3. Click "Generate new token"
4. Give it a name (e.g., "NLSearch App")
5. Copy the token and use it in your configuration

Usage

1. Type your natural language query in the search box
2. Click "Search" or press Enter
3. Wait for the Deep Search API to process your query
4. View the answer and sources

Example Queries

• "all repos which have python files"
• "find all TypeScript interfaces in the frontend"
• "show me error handling patterns in Go code"
• "repositories using React hooks"

Project Structure

nlsearch/
├── backend/
│   ├── main.go          # Go backend server with API endpoints
│   └── go.mod           # Go module definition
├── frontend/
│   └── index.html       # Web UI (HTML/CSS/JS)
├── .env.example         # Example environment variables
└── README.md            # This file

API Endpoints

POST /api/query

Submit a natural language query.

Request:

{
  "query": "all repos which have python files"
}

Response:

{
  "answer": "Here are the repositories...",
  "sources": [
    {
      "type": "Repository",
      "label": "github.com/example/repo"
    }
  ]
}

GET /health

Health check endpoint.

How It Works

1. User submits a natural language query via the web UI
2. Frontend sends the query to the backend API
3. Backend creates a Deep Search conversation with the Sourcegraph API
4. Backend polls for completion (up to 60 seconds)
5. Result is returned to the frontend and displayed

Development

Running in Development Mode

The backend serves both the API and the frontend static files. Any changes to the frontend HTML will be reflected immediately on refresh.

For backend changes, restart the Go server:

cd backend
go run main.go

Building for Production

cd backend
go build -o nlsearch-server main.go

Then run:

SOURCEGRAPH_TOKEN=your_token ./nlsearch-server

Troubleshooting

"SOURCEGRAPH_TOKEN environment variable is required"

• Make sure you've set the SOURCEGRAPH_TOKEN environment variable

"Failed to create conversation: unexpected status 401"

• Your access token is invalid or expired
• Generate a new token from your Sourcegraph instance

"timeout waiting for response"

• The Deep Search query is taking too long
• Try a simpler query
• The timeout is currently set to 60 seconds

"Network error"

• Check your internet connection
• Verify the SOURCEGRAPH_URL is correct
• Ensure the Sourcegraph instance is accessible

License

MIT License - feel free to use and modify as needed.

Contributing

Contributions are welcome! Feel free to submit issues or pull requests.