msdoc2docx 0.7.0

A specification-driven, pure-Python Word binary .doc to .docx converter

doc2docx

doc2docx is a specification-driven, pure-Python converter for Microsoft Word 97–2003 binary .doc files. It does not invoke Microsoft Word, LibreOffice, COM, Java, or an external conversion executable.

The current M0–M5b implementation supports unencrypted CFB/OLE Word documents, the 0Table/1Table selection mechanism, CLX piece tables, compressed and UTF-16LE text pieces, and the main document story. It reads CHPX/PAPX FKPs and preserves a useful first set of direct character and paragraph properties. It also reads the SttbfFfn font table, emits DOCX font definitions, converts paragraph/character styles with basedOn inheritance, resolves style-relative toggles, and applies simple and complex piece-level PRMs in specification order. It reconstructs tables from DOC cell/row markers, including nested tables, and preserves their column grid, preferred cell widths, basic merges, alignment, row sizing, Word 97 borders and shading, and default or range-specific cell margins. It also follows main-story PlcfSed records through Sed.fcSepx to each Sepx.grpprl, reconstructing single or multiple sections. Section-break types, paper size, portrait/landscape orientation, page margins, header/footer distance, and gutter width are emitted as WordprocessingML section properties. It parses PlcfHdd, removes each story's guard paragraph, preserves the six default/even/first header and footer positions per section, and retains empty-story inheritance. Header/footer parts, relationships, first-page rules, and the DOP facing-pages setting are written as native DOCX package parts.

Usage

python -m pip install msdoc2docx
doc2docx input.doc
doc2docx input.doc -o output.docx --report report.json
doc2docx inspect input.doc --json

To install from a local source checkout instead, run python -m pip install ..

Python API:

from doc2docx import convert

result = convert("input.doc", "output.docx")
print(result.report.to_dict())

M5b currently preserves bold, italic, strike, double strike, capitalization, hidden text, underline, text color, highlight, font size, vertical alignment, paragraph justification, indents, spacing, line spacing, and keep/page-break flags, font names, paragraph/character style references, style inheritance, basic section/page layout, and ordinary paragraph/table content in headers and footers. Conditional table styles, newer color-based table shading, cell spacing, numbering styles, multi-column layout, page-number settings, header/footer textboxes and shapes, other secondary stories, images, live fields, embedded objects, and encrypted documents are intentionally deferred to later iterations. Unsupported or lossy content is reported rather than silently treated as fully converted.

Tests

The test suite uses only the Python standard library:

PYTHONPATH=src python -m unittest discover -v

It includes constructed CFB version 3/version 4 files, regular and mini streams, both Table stream variants, mixed compressed/UTF-16 text pieces, UTF-16 surrogate pair coordinates, CHPX/PAPX and piece-level PRM formatting, font/style table parsing, nested table reconstruction, grids, cell margins and shading, PlcfSed and Sepx page-layout parsing, multi-section break placement, PlcfHdd guard and story mapping, header/footer OPC relationships, malformed input checks, CLI coverage, and end-to-end DOCX package validation.