# Local API Docs Search

A CLI tool that indexes local API documentation (OpenAPI specs, code comments, README files) and enables natural language semantic search using local embedding models. Developers can query their project's API docs offline without external API calls.

## Features

- **Semantic Search**: Natural language queries using local embedding models
- **Offline Processing**: All processing happens locally - no external API calls
- **Multiple Formats**: Support for OpenAPI/Swagger specs, Markdown READMEs, and code comments
- **Interactive Mode**: Rich CLI interface with history and navigation
- **Configurable**: Easy configuration via YAML or environment variables

## Installation

```bash
# Clone the repository
git clone <repository-url>
cd local-api-docs-search

# Create virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install dependencies
pip install -e .

# Or install with dev dependencies
pip install -e ".[dev]"
```

## Quick Start

1. **Configure your environment**:
   ```bash
   cp .env.example .env
   # Edit .env to set your documentation path
   ```

2. **Index your documentation**:
   ```bash
   # Index all supported formats
   api-docs index ./docs

   # Index specific format
   api-docs index ./docs --type openapi
   api-docs index ./docs --type readme
   api-docs index ./docs --type code

   # Recursive indexing
   api-docs index ./docs --recursive
   ```

3. **Search your documentation**:
   ```bash
   # Simple search
   api-docs search "how to authenticate users"

   # With limit
   api-docs search "authentication" --limit 5

   # Filter by type
   api-docs search "endpoints" --type openapi
   ```

4. **Interactive mode**:
   ```bash
   api-docs interactive
   ```

## Configuration

### Environment Variables

| Variable | Default | Description |
|----------|---------|-------------|
| `API_DOCS_INDEX_PATH` | `./docs` | Directory containing docs to index |
| `API_DOCS_MODEL_NAME` | `all-MiniLM-L6-v2` | Embedding model to use |
| `API_DOCS_EMBEDDING_DEVICE` | `cpu` | Device for embeddings (cpu/cuda/auto) |
| `API_DOCS_CHROMA_PERSIST_DIR` | `.api-docs/chroma` | ChromaDB storage directory |
| `API_DOCS_DEFAULT_LIMIT` | `10` | Default result limit |
| `API_DOCS_VERBOSE` | `false` | Enable verbose logging |

### YAML Config

You can also use a `config.yaml` file:

```yaml
index_path: ./docs
model_name: all-MiniLM-L6-v2
embedding_device: cpu
chroma_persist_dir: .api-docs/chroma
default_limit: 10
verbose: false
```

## Commands

### index

Index documentation files for search.

```bash
api-docs index <path> [OPTIONS]

Options:
  --type TEXT     Type of docs: openapi, readme, code, or all (default: all)
  --recursive     Recursively search directories
  --batch-size    Documents per batch (default: 32)
  --help          Show this message and exit
```

### search

Search indexed documentation.

```bash
api-docs search <query> [OPTIONS]

Options:
  --limit INTEGER    Maximum results (default: 10)
  --type TEXT        Filter by source type
  --json             Output as JSON
  --help             Show this message and exit
```

### list

List indexed documents.

```bash
api-docs list [OPTIONS]

Options:
  --type TEXT    Filter by source type
  --json         Output as JSON
  --help         Show this message and exit
```

### config

Manage configuration.

```bash
api-docs config [OPTIONS]

Options:
  --show    Show current configuration
  --reset   Reset to defaults
  --help    Show this message and exit
```

### interactive

Enter interactive search mode with history and navigation.

```bash
api-docs interactive
```

## Supported Formats

### OpenAPI/Swagger

Parses OpenAPI 3.0+ and Swagger 2.0 specifications in YAML or JSON format. Extracts:
- Endpoint paths and methods
- Operation IDs and summaries
- Parameters and request bodies
- Response schemas
- Tags and descriptions

### README Files

Parses Markdown files including:
- Section headers
- Paragraph text
- Code blocks
- Lists and tables

### Code Comments

Supports Python, JavaScript, and TypeScript files:
- Google-style docstrings
- JSDoc comments
- Function and class documentation
- Inline comments (configurable)

## Interactive Mode

Interactive mode provides:
- Real-time search as you type
- Arrow key navigation through history
- Result pagination
- Syntax highlighting
- Source attribution

Keyboard shortcuts:
- `↑` / `↓`: Navigate history
- `Enter`: Submit search
- `Tab`: Complete suggestion
- `Ctrl+C`: Exit
- `F1`: Toggle help

## Architecture

```
local-api-docs-search/
├── src/
│   ├── main.py           # CLI entry point
│   ├── cli/
│   │   ├── commands.py   # CLI command definitions
│   │   └── interactive.py # Interactive mode
│   ├── indexer/
│   │   ├── base.py       # Base indexer interface
│   │   ├── openapi.py    # OpenAPI spec indexer
│   │   ├── readme.py     # README file indexer
│   │   └── code.py       # Code comment indexer
│   ├── search/
│   │   ├── embeddings.py # Embedding model management
│   │   ├── vectorstore.py # ChromaDB operations
│   │   └── searcher.py   # Search logic
│   ├── models/
│   │   └── document.py   # Document models
│   └── utils/
│       ├── config.py     # Configuration management
│       └── formatters.py # Output formatting
├── tests/
├── pyproject.toml
└── README.md
```

## Development

```bash
# Run tests
pytest tests/ -v --cov=src

# Run linting
black src/ tests/
ruff check src/ tests/

# Type checking
mypy src/
```

## Error Handling

| Error | Solution |
|-------|----------|
| No embedding model found | Model downloads automatically on first use. Use `--force-download` to re-download |
| Invalid OpenAPI spec | Validation errors are logged with line numbers; valid parts are still indexed |
| ChromaDB collection not found | Collection is created automatically; prompt to index if empty |
| No documents indexed | Run `api-docs index` first with appropriate path |
| MemoryError with large docs | Use batch processing; consider quantized embeddings |

## Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Run tests and linting
5. Submit a pull request

## License

MIT License - see LICENSE file for details.