magic-pdf

1.3.12 verified Sat May 09 auth: no python

A practical tool for converting PDF to Markdown, part of the MinerU project by OpenDataLab. Current version is 1.3.12, requires Python >=3.10, <3.14. The package is actively maintained with frequent releases.

pip install magic-pdf

Common errors

error ModuleNotFoundError: No module named 'magic_pdf' ↓

cause Package not installed or imported with wrong name (hyphen instead of underscore).

fix

Run 'pip install magic-pdf' and use 'import magic_pdf'.

error AttributeError: module 'magic_pdf' has no attribute 'parse' ↓

cause Using deprecated function name 'parse' after v1.3.0.

fix

Use 'magic_pdf.parse_pdf()' instead.

error KeyError: 'markdown' ↓

cause Accessing 'markdown' key on result from older version (<1.3.0).

fix

Check version: if <1.3.0 use result['text'], else use result['markdown'].

Warnings

gotcha The package name on PyPI is 'magic-pdf', but the import uses underscore: 'magic_pdf'. ↓

fix Use 'pip install magic-pdf' and 'import magic_pdf'.

breaking Version 1.3.0 changed the output structure: the Markdown content is now under 'markdown' key instead of 'text'. ↓

fix Access result['markdown'] for v1.3.0+, or result['text'] for older versions.

deprecated The function 'magic_pdf.parse' was deprecated in v1.3.0 in favor of 'magic_pdf.parse_pdf'. ↓

fix Use 'parse_pdf' instead of 'parse'.

Imports

magic_pdf
wrong
```
import magic_pdf as pdf
```
correct
```
import magic_pdf
```
Common mistake: importing 'magic-pdf' with hyphen (invalid Python syntax). Use underscore.

parse_pdf

wrong

from magic_pdf.pipeline import parse_pdf

correct

from magic_pdf import parse_pdf

parse_pdf is directly under magic_pdf, not a submodule.

Quickstart

Convert a PDF to Markdown and print the result.

import magic_pdf
result = magic_pdf.parse_pdf('sample.pdf', output_dir='./output')
print(result['markdown'])