v0.5.8

• New documentation link: https://pdf2docx.readthedocs.io/
• Fixed GUI relevant issues #211, #222, #224
• Fixed extra spaces issue at the end of Chinese characters lines #205, #239, #240
• Fixed compatible issues relevant with PyMuPDF upgrading #248, #250
• Fixed empty font name issue #256
• Fixed spacing issue caused by shapes partly out of page #256

v0.5.7

• Support in-memory file such as BytesIO when reading PDF or writing docx: #108, #177, #223
• Fixed tabulation issue when multiple tabs exist: #157
• Fixed cannot find builtin font with name 'Arial' issue: #216, #235, #237, #241
• Fixed importing issue caused by python-docx 1.0.0: #233, #234
• Fixed font name encoding issue: #194, #246
• Fixed duplicated columns (xml) issue: #245

v0.5.6

• Workaround for UnicodeDecodeError issue: #144, #155
• Fixed table parsing (small images as dashed borders) issue: #138, #158
• Improved table parsing (allow one-cell-table with shading): #149

v0.5.5

• fixed closePath issue for PyMuPDF 1.20+: #146, #147

v0.5.4

• Ignore hidden text #132
• Fixed property name deprecated by PyMuPDF 1.20+ #139, #141, #143

v0.5.3

• section parsing enhancement
• table parsing enhancement: #105, #107
• vector graphic parsing enhancement: recursive xy cut
• upgrade with PyMuPDF>=1.19.0
• supported Python 3.10
• fixed images parsing issues: #110, #125, #123
• fixed importing GUI issue for non-GUI platform: #118, #121

v0.5.2

• new layout structure: Page -> Section -> Column -> blocks
• extended parsing scope: document -> page -> section
• corrected font name with font-tools
• changed from exact line spacing to relative line spacing
• supported encrypted pdf #86
• added user interface #88

v0.5.1

• support hyperlink #59
• create API documentation
• enhancement: parse page layout recursively

v0.5.0

• extract pdf path with PyMuPDF (>=1.18.0) API
• support floating picture
• enhance paragraph alignment and vertical spacing
• global settings on page parsing
• enhance CLI commands

v0.4.6

• enhance parsing stream table, e.g. consider explicit shading when construct stream table
• enhance parsing lattice table, e.g. consider stream border deep into cell when parsing lattice table, detect semantic type of stroke/fill shapes in advance
• enhance text alignment detection
• fixed issue on vector graphics criterion
• switch to pytest and integrate code coverage result