Document Parser

OpenClaw Skill - High-precision document parsing

Extract structured data from PDF, images, and Word documents.

Features

• ✅ Multi-format Support: PDF, Images (JPG/PNG), Word documents
• ✅ Layout Analysis: Automatically detect and structure document elements
• ✅ Table Recognition: Extract tables with HTML and Markdown outputs
• ✅ OCR Support: Recognize text in scanned documents and images
• ✅ Seal Detection: Detect stamps and seals in documents
• ✅ TOC Extraction: Extract table of contents from documents
• ✅ Cross-page Merge: Automatically merge content across pages

Quick Start

Installation

# Install via ClawHub
openclaw skills install document-parser

# Or manual installation (local development)
cd E:\skills\document-parser
pip install -r requirements.txt

Configuration

Option 1: Environment Variables (Recommended)

# Windows PowerShell
$env:DOCUMENT_PARSER_API_KEY="your_api_key"
setx DOCUMENT_PARSER_API_KEY "your_api_key"

# Optional: Custom API endpoint
$env:DOCUMENT_PARSER_BASE_URL="http://your-server:8088/taidp/v1/idp/general_parse"

Option 2: Configuration File

cd E:\skills\document-parser
copy config.example.json config.json
# Edit config.json with your API Key

Usage

Parse a Document

# Basic parsing
document-parser parse "C:\docs\report.pdf"

# Enable layout analysis and table recognition
document-parser parse "C:\docs\report.pdf" --layout --table

# Specify output format
document-parser parse "C:\docs\scan.jpg" --output markdown

# Specify page range
document-parser parse "C:\docs\book.pdf" --pages 1-5,10-15

Query Task Status

document-parser status <task_id>

Parameters

Parameter	Type	Required	Description
file	string	Yes	PDF/Image/Word file path
--layout	flag	No	Enable layout analysis
--table	flag	No	Enable table recognition
--seal	flag	No	Enable seal recognition
--output	string	No	Output format: json/markdown/both
--pages	string	No	Page range, e.g., "1-5,8,10-12"

Output Format

Returns structured JSON containing:

{
  "code": 10000,
  "message": "Success",
  "request_id": "xxx-xxx-xxx",
  "elapsed_time": 8.5,
  "pages": [
    {
      "page_no": 1,
      "height": 1654,
      "width": 1169,
      "elements": [...]
    }
  ],
  "markdown": "Parsed Markdown text"
}

Supported Document Elements

Type	Description
DocumentTitle	Document title
LevelTitle	Section heading
Paragraph	Text paragraph
Table	Table
Image	Image
PageHeader	Header
PageFooter	Footer
Seal	Seal/Stamp
Formula	Mathematical formula
TableOfContents	Table of contents

Error Codes

Code	Message	Description
10000	Success	Parsing successful
10001	Missing parameter	Missing required parameter
10002	Invalid parameter	Invalid parameter value
10003	Invalid file	Unsupported file format
10004	Failed to recognize	Recognition failed
10005	Internal error	Internal server error

Examples

Parse PDF and extract tables

document-parser parse "C:\docs\financial_report.pdf" --table --output markdown

Parse scanned document with OCR

document-parser parse "C:\docs\scanned_contract.jpg" --layout

Parse Word document

document-parser parse "C:\docs\manual.docx" --output json

Dependencies

• Python 3.8+
• requests>=2.28.0
• python-docx>=0.8.11
• Pillow>=9.0.0

License

MIT License

Support

File an issue or contact the author for help.