v0.4.5

fixed reproducing images issue:

• images in CMYK color-space
• transparent images

v0.4.4

detect vector graphic and convert to bitmap

v0.4.3

support paragraph alignment

v0.4.2

• correct block bbox due to unnamed font in PDF
• set borders style for each cell

v0.4.1

• align stream table borders
• consider explicit border in stream table
• consider page rotation
• parsing pages with multi-processing

v0.4.0

new feature: nested table (support 1 layer only)

v0.3.5

enhancement: text format parsing

• consider text shading instead if out of default highlight colors
• font name localization: Chinese
• do not adjust spacing between Chinese and Latin/number

v0.3.4

fixed issue: operator name conflicts with text when extracting path

v0.3.3

• extract annotations over multi-lines precisely
• fix docx.image.exceptions.UnrecognizedImageError
• set font size precisely by setting character scaling
• CS transformation between PDF and PyMuPDF
• consider cell shading for implicit table
• fixed issue on empty page, text style

v0.3.2

parse and recreate text blocks in vertical direction