paddleocr-doc-parsing-v2
por hiotec
Image & Video GenerationResumo
Parse documents using PaddleOCR's API.
Descrição
O que esta skill faz
Parses images and PDF files using PaddleOCR's API, extracting text and layout structure into Markdown format. Supports sync mode for quick processing and async mode for large or multi-page documents. Handles 110+ languages and detects text blocks, tables, and formulas.
Automatic layout analysis preserves document structure (headings, tables, formulas) that plain OCR tools typically flatten into raw text.
Quando usá-la
- Extracting text from scanned invoices or receipts
- Converting multi-page PDFs to structured Markdown
- Parsing documents in non-Latin scripts
- Processing large batches of PDFs asynchronously
- Pulling table data out of image-based reports
Exemplo de fluxo de trabalho
User asks: extract the content of this scanned contract PDF
- 1
Checks file size to determine sync vs async mode
- 2
Calls paddleocr_parse.sh with -t pdf flag on the file
- 3
Polls job status if async mode is used
- 4
Receives JSON response with markdown.text field
- 5
Returns structured Markdown with preserved headings and tables
Markdown representation of the contract with section headings and any tables intact
Requisitos
- PADDLEOCR_ACCESS_TOKEN environment variable
- PADDLEOCR_API_URL environment variable
- PaddleOCR account and API credentials from paddleocr.com
Instalação
clawhub install hiotec/paddleocr-doc-parsing-v2Execute o comando da CLI ou cole o prompt no Claude Code, Cursor, Codex ou no agente de código que você usa.