Version 0.2.0 has been released

[JorjMcKie]

This version offers a significantly improved way for extracting data from document by using package PyMuPDF-Layout.

After installation of that package and including the import import pymupdf.layout in your script before importing pymupdf4llm, the layout feature will automatically be used.

Please consider trying it out. Among the many improvements are:

• Significantly improved layout recognition, including things like page headers and footers, list items, titles and section headers.
• List item hierarchies are now supported.
• Greatly improved table recognition while maintaining an unchanged API.
• Support new output formats: there now are the following alternatives:
  • .to_markdown() - as before, however with some new parameters that allow page header and footer suppression. Other parameters are now obsolete or not yet adapted to the layout feature.
  • .to_text() - output of plain text. Table content its shown using package tabulate.
  • .to_json() - output the document's metadata and the selected pages in JSON format.
• Before processing a page, we check whether it requires OCR and in that case automatically invoke Tesseract.