📄 ParseFlow

Universal document parsing library for PDF, Word, and Excel files

ParseFlow is a comprehensive document parsing solution that supports PDF, Word (docx), and Excel (xlsx/xls) files. It provides both a standalone library and an MCP (Model Context Protocol) server for AI assistants.

中文文档 | Examples | GitHub

---

✨ Features

📄 PDF Support

• ✅ Text extraction with multiple strategies (raw, formatted, clean)
• ✅ Page-specific and range-based extraction
• ✅ Metadata retrieval (title, author, dates, page count)
• ✅ Full-text search with context
• ✅ Image extraction (placeholder)
• ✅ Table of contents (TOC) extraction (placeholder)

📝 Word (docx) Support

• ✅ Text extraction
• ✅ HTML conversion
• ✅ Metadata retrieval
• ✅ Text search with context

📊 Excel (xlsx/xls) Support

• ✅ Multi-sheet data extraction
• ✅ Multiple output formats (JSON, CSV, Text)
• ✅ Sheet-specific extraction
• ✅ Cell-based search
• ✅ Range extraction
• ✅ Workbook metadata

🤖 MCP Server

• ✅ 9 tools for AI assistants (5 PDF + 2 Word + 2 Excel)
• ✅ Works with Claude Desktop and other MCP clients
• ✅ Path security with allowlist support

---

📦 Installation

Core Library

npm install parseflow-core

MCP Server (Global)

npm install -g parseflow-mcp-server

Or use with npx:

npx parseflow-mcp-server

---

🚀 Quick Start

PDF Parsing

import { PDFParser } from 'parseflow-core';

const parser = new PDFParser();

// Extract all text
const text = await parser.extractText('document.pdf');

// Extract specific page
const page5 = await parser.extractPage('document.pdf', 5);

// Search
const results = await parser.search('document.pdf', 'keyword');

// Get metadata
const metadata = await parser.getMetadata('document.pdf');

Word Parsing

import { WordParser } from 'parseflow-core';

const parser = new WordParser();

// Extract text
const result = await parser.extractText('report.docx');
console.log(result.text);

// Convert to HTML
const html = await parser.extractHTML('report.docx');

// Search
const matches = await parser.searchText('report.docx', 'budget');

Excel Parsing

import { ExcelParser } from 'parseflow-core';

const parser = new ExcelParser();

// Extract all sheets (JSON format)
const data = await parser.extractData('spreadsheet.xlsx');

// Extract specific sheet
const sales = await parser.extractData('data.xlsx', {
  sheetName: 'Q4 Sales',
  format: 'json'
});

// Search in cells
const results = await parser.searchText('data.xlsx', 'revenue');

---

🛠️ MCP Server Usage

Configuration for Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "parseflow": {
      "command": "npx",
      "args": ["-y", "parseflow-mcp-server"],
      "env": {
        "PARSEFLOW_ALLOWED_PATHS": "C:\\Documents;D:\\Projects"
      }
    }
  }
}

Available Tools

PDF Tools

• extract_text - Extract text from PDF files
• search_pdf - Search for keywords in PDF
• get_metadata - Get PDF metadata
• extract_images - Extract images from PDF
• get_toc - Get table of contents

Word Tools

• extract_word - Extract text/HTML from Word documents
• search_word - Search in Word documents

Excel Tools

• extract_excel - Extract data from Excel spreadsheets
• search_excel - Search in Excel cells

Example Usage in Claude

"请读取 report.docx 文件的内容"
→ Uses extract_word tool

"在 sales.xlsx 中查找 '产品A'"
→ Uses search_excel tool

"提取 document.pdf 的元数据"
→ Uses get_metadata tool

---

📚 Documentation

• Office Examples - Word and Excel usage examples
• Release Guide - How to publish new versions
• Contributing - Contribution guidelines
• Security Policy - Security vulnerability reporting
• Code of Conduct - Community guidelines

---

🏗️ Project Structure

ParseFlow/
├── packages/
│   ├── pdf-parser-core/      # Core library (parseflow-core)
│   │   ├── src/
│   │   │   ├── parser.ts     # PDF parser
│   │   │   ├── WordParser.ts # Word parser
│   │   │   └── ExcelParser.ts # Excel parser
│   │   └── package.json
│   └── mcp-server/           # MCP server (parseflow-mcp-server)
│       ├── src/
│       │   ├── index.ts      # Server entry
│       │   └── tools/        # MCP tools
│       └── package.json
├── docs/                     # Documentation
├── examples/                 # Usage examples
├── tests/                    # Test files
└── scripts/                  # Build scripts

---

🧪 Testing

# Run all tests
pnpm test

# Test coverage
pnpm test:coverage

# Run specific test
pnpm test parser.test.ts

Test Files

• Word测试文件.docx - Word test document
• Excel测试文件.xlsx - Excel test workbook (3 sheets)
• PDF测试文档.pdf - PDF test document

---

🔧 Development

# Install dependencies
pnpm install

# Build all packages
pnpm build

# Watch mode
pnpm dev

# Lint
pnpm lint

# Type check
pnpm type-check

---

📈 Roadmap

v1.1.0 (Current)

• ✅ Word (docx) support
• ✅ Excel (xlsx/xls) support
• ✅ 9 MCP tools

v1.2.0 (Planned)

• Encrypted PDF support
• OCR text recognition
• PowerPoint (pptx) support
• Batch processing optimization

v2.0.0 (Future)

• Plugin system
• More document formats (CSV, TXT, RTF)
• Advanced table extraction
• Document conversion

---

🤝 Contributing

We welcome contributions! Please see CONTRIBUTING.md for details.

Ways to Contribute

• 🐛 Report bugs
• 💡 Suggest features
• 📝 Improve documentation
• 🔧 Submit pull requests

---

📦 Packages

Package	Version	Description
parseflow-core	1.0.1	Core parsing library
parseflow-mcp-server	1.0.2	MCP server for AI

---

🔗 Links

• npm Core: https://www.npmjs.com/package/parseflow-core
• npm MCP: https://www.npmjs.com/package/parseflow-mcp-server
• GitHub: https://github.com/Libres-coder/ParseFlow
• Issues: https://github.com/Libres-coder/ParseFlow/issues
• MCP Registry: https://registry.modelcontextprotocol.io/

---

📄 License

MIT License - see LICENSE file for details.

---

🙏 Acknowledgments

• pdf-parse - PDF parsing
• pdf-lib - PDF manipulation
• mammoth - Word document parsing
• xlsx - Excel spreadsheet parsing
• MCP SDK - Model Context Protocol

---

📊 Stats

• Test Coverage: 83%+
• Supported Formats: 3 (PDF, Word, Excel)
• MCP Tools: 9
• Dependencies: Minimal and well-maintained

---

💬 Community

• Issues: GitHub Issues
• Discussions: GitHub Discussions

---

Made with ❤️ by Libres-coder

Status: 🎉 Production Ready (v1.1.0)