Precision Document Extraction with mineru-open-api
Full-featured document extraction with table/formula recognition, OCR, multi-format output, batch processing, and web crawling.
Why use extract?
- Table recognition — accurately extracts tables from PDFs and images
- Formula recognition — preserves mathematical formulas as LaTeX
- Multi-format output — Markdown, HTML, LaTeX, DOCX, JSON
- Model selection — choose
vlm for highest accuracy or pipeline for zero-hallucination
- Batch processing — process hundreds of files in one command
- Web crawling — convert web pages to structured Markdown
- All file formats — PDF, images, DOC, DOCX, PPT, PPTX, HTML
- Higher limits — much larger file size and page count than quick mode
- 80+ languages — full language coverage across all script families
Installation
npm install -g mineru-open-api
Or via Go (macOS/Linux):
go install github.com/opendatalab/MinerU-Ecosystem/cli/mineru-open-api@latest
Verify installation
mineru-open-api version
Authentication
Create a token at https://mineru.net/apiManage/token, then configure:
mineru-open-api auth # Interactive token setup
export MINERU_TOKEN="your-token" # Or set via environment variable
Token resolution order: --token flag > MINERU_TOKEN env > ~/.mineru/config.yaml.
Quick start
mineru-open-api extract report.pdf # Markdown to stdout
mineru-open-api extract report.pdf -o ./out/ # Save to directory
mineru-open-api extract report.pdf -f md,html,docx -o ./ # Multi-format
mineru-open-api extract report.pdf --model vlm -o ./out/ # High-accuracy mode
mineru-open-api extract *.pdf -o ./results/ # Batch extract
mineru-open-api crawl https://example.com/article # Web page → Markdown
Supported input formats
| Format | Supported |
|---|
PDF (.pdf) | Yes |
Images (.png, .jpg, .jpeg, .jp2, .webp, .gif, .bmp) | Yes |
Word (.doc, .docx) | Yes |
PowerPoint (.ppt, .pptx) | Yes |
HTML (.html) | Yes |
| URLs (remote files) | Yes |
Commands
extract — Precision extraction
mineru-open-api extract <file-or-url> [...] [flags]
Examples
mineru-open-api extract report.pdf # Markdown to stdout
mineru-open-api extract report.pdf -f html # HTML to stdout
mineru-open-api extract report.pdf -o ./out/ # Save to directory
mineru-open-api extract report.pdf -o ./out/ -f md,docx # Multiple formats
mineru-open-api extract report.pdf -f latex -o ./out/ # LaTeX output
mineru-open-api extract report.pdf --model vlm -o ./out/ # High-accuracy mode
mineru-open-api extract report.pdf --ocr -o ./out/ # OCR for scanned docs
mineru-open-api extract report.pdf --language en -o ./out/ # Specify language
mineru-open-api extract report.pdf --pages "1-10" -o ./out/ # Page range
mineru-open-api extract *.pdf -o ./results/ # Batch extract
mineru-open-api extract --list files.txt -o ./results/ # Batch from file list
mineru-open-api extract https://example.com/doc.pdf # Extract from URL
cat doc.pdf | mineru-open-api extract --stdin -o ./out/ # From stdin
extract flags
| Flag | Short | Default | Description |
|---|
--output | -o | (stdout) | Output path (file or directory) |
--format | -f | md | Output formats: md, json, html, latex, docx (comma-separated) |
--model | | (auto) | Model: vlm, pipeline, html (see below) |
--ocr | | false | Enable OCR for scanned documents |
--formula | | true | Enable/disable formula recognition |
--table | | true | Enable/disable table recognition |
--language | | ch | Document language |
--pages | | (all) | Page range, e.g. 1-10,15 |
--timeout | | 900/1800 | Timeout in seconds (single/batch) |
--list | | | Read input list from file (one path per line) |
--concurrency | | 0 | Batch concurrency (0 = server default) |
Model comparison: vlm vs pipeline
| vlm | pipeline |
|---|
| Parsing accuracy | Higher — better at complex layouts, mixed content | Standard |
| Hallucination risk | May produce hallucinated text in rare cases | No hallucination — biggest advantage |
| Best for | Academic papers, complex tables, intricate layouts | General documents where fidelity matters most |
When the user values accuracy and the document has complex formatting, suggest --model vlm. When the user prioritizes reliability and no-hallucination guarantee, suggest --model pipeline (or omit --model to use auto).
crawl — Web page extraction
Fetch web pages and convert to structured Markdown.
mineru-open-api crawl https://example.com/article # Markdown to stdout
mineru-open-api crawl https://example.com/article -f html # HTML to stdout
mineru-open-api crawl https://example.com/article -o ./out/ # Save to file
mineru-open-api crawl url1 url2 -o ./pages/ # Batch crawl
mineru-open-api crawl --list urls.txt -o ./pages/ # Batch from file list
crawl flags
| Flag | Short | Default | Description |
|---|
--output | -o | (stdout) | Output path |
--format | -f | md | Output formats: md, json, html (comma-separated) |
--timeout | | 900/1800 | Timeout in seconds (single/batch) |
--list | | | Read URL list from file (one per line) |
--stdin-list | | false | Read URL list from stdin |
--concurrency | | 0 | Batch concurrency |
auth — Authentication management
mineru-open-api auth # Interactive token setup
mineru-open-api auth --verify # Verify current token is valid
mineru-open-api auth --show # Show current token source and masked value
Supported --language values
Values are organized by script/language family — each value covers all languages in its group.
Standalone language packs
| Value | Included languages | 说明 |
|---|
ch | Chinese, English, Chinese Traditional | 中英文（默认值） |
ch_server | Chinese, English, Chinese Traditional, Japanese | 繁体、手写体 |
en | English | 纯英文 |
japan | Chinese, English, Chinese Traditional, Japanese | 日文为主 |
korean | Korean, English | 韩文 |
chinese_cht | Chinese, English, Chinese Traditional, Japanese | 繁体中文为主 |
ta | Tamil, English | 泰米尔文 |
te | Telugu, English | 泰卢固文 |
ka | Kannada | 卡纳达文 |
el | Greek, English | 希腊文 |
th | Thai, English | 泰文 |
Language family packs
| Value | Script/Family | Included languages |
|---|
latin | Latin script (拉丁语系) | French, German, Afrikaans, Italian, Spanish, Bosnian, Portuguese, Czech, Welsh, Danish, Estonian, Irish, Croatian, Uzbek, Hungarian, Serbian (Latin), Indonesian, Occitan, Icelandic, Lithuanian, Maori, Malay, Dutch, Norwegian, Polish, Slovak, Slovenian, Albanian, Swedish, Swahili, Tagalog, Turkish, Latin, Azerbaijani, Kurdish, Latvian, Maltese, Pali, Romanian, Vietnamese, Finnish, Basque, Galician, Luxembourgish, Romansh, Catalan, Quechua |
arabic | Arabic script (阿拉伯语系) | Arabic, Persian, Uyghur, Urdu, Pashto, Kurdish, Sindhi, Balochi, English |
cyrillic | Cyrillic script (西里尔语系) | Russian, Belarusian, Ukrainian, Serbian (Cyrillic), Bulgarian, Mongolian, Abkhazian, Adyghe, Kabardian, Avar, Dargin, Ingush, Chechen, Lak, Lezgin, Tabasaran, Kazakh, Kyrgyz, Tajik, Macedonian, Tatar, Chuvash, Bashkir, Malian, Moldovan, Udmurt, Komi, Ossetian, Buryat, Kalmyk, Tuvan, Sakha, Karakalpak, English |
east_slavic | East Slavic (东斯拉夫语系) | Russian, Belarusian, Ukrainian, English |
devanagari | Devanagari script (天城文语系) | Hindi, Marathi, Nepali, Bihari, Maithili, Angika, Bhojpuri, Magahi, Santali, Newari, Konkani, Sanskrit, Haryanvi, English |
Global flags
| Flag | Short | Description |
|---|
--token | | API token (overrides env and config) |
--base-url | | API base URL (for private deployments) |
--verbose | -v | Verbose mode, print HTTP details |
Output behavior
- No
-o flag: result goes to stdout; status/progress messages go to stderr
- With
-o flag: result saved to file/directory; progress messages on stderr
- Batch mode (
extract/crawl): requires -o to specify output directory
- Binary formats (
docx): cannot output to stdout, must use -o
- Markdown output includes extracted images saved alongside the
.md file
Agent guidelines
When using this skill on behalf of the user:
- Quote file paths that contain spaces or special characters with double quotes. Example:
mineru-open-api extract "report 01.pdf".
- Don't run commands blindly on errors — explain the exit code and troubleshooting steps.
- Installation questions ("mineru 怎么安装") should be answered with the install instructions above.
- For stdout mode (no
-o), only one text format can be output at a time. If the user wants multiple formats, suggest adding -o.
- If the user hasn't authenticated yet, guide them to create a token at https://mineru.net/apiManage/token and run
mineru-open-api auth.
Default output directory
When the user does NOT specify -o, generate a default output directory:
~/MinerU-Skill/<name>_<hash>/
<name>: derived from the source, then sanitized (replace spaces and shell-unsafe characters with _, collapse consecutive _).
- For URLs: last path segment (e.g.
https://arxiv.org/pdf/2509.22186 → 2509.22186)
- For local files: filename without extension (e.g.
report.pdf → report)
<hash>: first 6 characters of MD5 hash of the full original source.
echo -n "source" | md5sum | cut -c1-6 # Linux
echo -n "source" | md5 | cut -c1-6 # macOS
When the user specifies -o: use the user's path as-is.
Skill upgrade = CLI upgrade
When the user asks to upgrade this skill, re-install the CLI first:
npm install -g mineru-open-api@latest
Exit codes
| Code | Meaning | Recovery |
|---|
| 0 | Success | — |
| 1 | General API or unknown error | Check network; retry; use --verbose |
| 2 | Invalid parameters / usage error | Check command syntax and flag values |
| 3 | Authentication error | Create or refresh token at https://mineru.net/apiManage/token, then run mineru-open-api auth |
| 4 | File too large or page limit exceeded | Split the file or use --pages |
| 5 | Extraction failed | Document may be corrupted; try a different --model |
| 6 | Timeout | Increase with --timeout; large files may need 1600+ seconds |
Troubleshooting
- "no API token found": Run
mineru-open-api auth or set MINERU_TOKEN env variable. Create token at https://mineru.net/apiManage/token.
- Timeout on large files: Increase with
--timeout 1600
- Batch fails partially: Check stderr for per-file status; succeeded files are still saved
- Binary format to stdout: Use
-o flag; docx cannot stream to stdout
- Private deployment: Use
--base-url https://your-server.com/api
- Extraction quality is poor: Try
--model vlm for complex layouts, or --ocr for scanned documents
- Tables not extracted correctly: Try
--model vlm for better table recognition
Reporting Issues
