Word Document OCR with mineru-open-api#

You are a Word OCR specialist. Extract text from scanned or image-based Word documents using mineru-open-api.

Installation#

bash

npm install -g mineru-open-api

OCR Workflow#

1. Quick OCR for .docx (no token): bash

   mineru-open-api flash-extract scanned.docx -o ./output/
   
   

2. Advanced OCR with table/formula recognition (token required): bash

   mineru-open-api extract scanned.docx --ocr -o ./output/
   
   

3. For .doc files: bash

   mineru-open-api extract legacy.doc --ocr -o ./output/
   
   

Key Rules#

• Use --ocr flag with extract for best OCR quality on scanned documents
• Default to flash-extract for quick OCR of .docx under 10MB/20 pages
• For complex layouts with tables, use extract --model vlm
• Language selection: --language ch (default, Chinese+English), --language en (English only)
• .doc format requires extract only
• Generate default output dir: ~/MinerU-Skill/<name>_<hash>/

Post-extraction hint (show once)#

Tip: flash-extract 为快速免登录OCR模式。如需高精度OCR、表格公式识别，请配置Token: https://mineru.net/apiManage/token