/pdf2docx

Open source Python library for converting PDF to DOCX.

Primary LanguagePythonGNU General Public License v3.0GPL-3.0

English | 中文

pdf2docx

python-version codecov pypi-version license pypi-downloads

  • Extract data from PDF with PyMuPDF, e.g. text, images and drawings
  • Parse layout with rule, e.g. sections, paragraphs, images and tables
  • Generate docx with python-docx

Features

  • Parse and re-create page layout

    • page margin
    • section and column (1 or 2 columns only)
    • page header and footer [TODO]
  • Parse and re-create paragraph

    • OCR text [TODO]
    • text in horizontal/vertical direction: from left to right, from bottom to top
    • font style, e.g. font name, size, weight, italic and color
    • text format, e.g. highlight, underline, strike-through
    • list style [TODO]
    • external hyper link
    • paragraph horizontal alignment (left/right/center/justify) and vertical spacing
  • Parse and re-create image

    • in-line image
    • image in Gray/RGB/CMYK mode
    • transparent image
    • floating image, i.e. picture behind text
  • Parse and re-create table

    • border style, e.g. width, color
    • shading style, i.e. background color
    • merged cells
    • vertical direction cell
    • table with partly hidden borders
    • nested tables
  • Parsing pages with multi-processing

It can also be used as a tool to extract table contents since both table content and format/style is parsed.

Limitations

  • Text-based PDF file
  • Left to right language
  • Normal reading direction, no word transformation / rotation
  • Rule-based method can't 100% convert the PDF layout

Documentation

Sample

sample_compare.png