# pdf-reader-mcp

An MCP server that can extract text from PDFs and images. It supports:

- Native text extraction from PDFs (embedded text)
- OCR for scanned PDFs and images using Tesseract

## Requirements

- Node.js >= 18.18
- For best OCR on PDFs, install Poppler (for `pdftoppm`) and Tesseract (CLI):

### macOS

```bash
brew install poppler tesseract
```

### Windows

1. **Tesseract**: Download and install from [GitHub releases](https://github.com/UB-Mannheim/tesseract/wiki) or use chocolatey:
   ```cmd
   choco install tesseract
   ```
2. **Poppler**: Download from [poppler-windows](https://github.com/oschwartz10612/poppler-windows/releases/) and add to PATH, or use chocolatey:
   ```cmd
   choco install poppler
   ```

### Linux (Ubuntu/Debian)

```bash
sudo apt-get install poppler-utils tesseract-ocr
```

**Note**: Poppler is optional; without it we fall back to slower whole-file OCR. Tesseract CLI is preferred for OCR; if it is not installed, the server falls back to `tesseract.js` (which downloads language data on first run).

## Install

```bash
npm install
```

## Build

```bash
npm run build
```

## Run (stdio)

```bash
npm start
```

This server uses stdio transport per MCP. Configure your MCP client with the command `node dist/index.js` in this folder.

Example MCP client configuration (JSON):

```
{
	"mcpServers": {
		"pdf-reader-mcp": {
			"command": "node",
			"args": ["dist/index.js"],
			"env": { }
		}
	}
}
```

## Tool: extract_text_from_pdf

Inputs (one of):

- `path`: Path to a local PDF or image file
- `base64`: Base64-encoded content (optionally with data URL prefix)

Options:

- `ocr` (boolean, default false): Force using OCR. When false, the server tries native extraction then falls back to OCR if needed.

Output: The extracted text as a single text item.