pdflite

pdf_md converts every page with pdf-oxide and fixes four things conversion gets wrong:

• a word broken across lines by a hyphen is rejoined
• a two-column table reflowed into one stream is swapped for the plain-text layer, when that strands fewer values
• a scanned page with no text layer goes through OCR
• reader chrome from a print-to-PDF is dropped

from pdflite import pdf_md

md = pdf_md('pdfs/attention_is_all_you_need.pdf', out_path='out'); md[:100] # out_path stores images
# pages = pdf_md('statement.pdf', out_path='out', pages=True)  # one entry per page

Dictionary used where Stream expected, treating as empty stream
Dictionary used where Stream expected, treating as empty stream
Dictionary used where Stream expected, treating as empty stream
Dictionary used where Stream expected, treating as empty stream

'Provided proper attribution is provided, Google hereby grants permission to reproduce the tables and'

Each fix is exported on its own: clean_md, fix_layout, orphan_vals, scrambled_layout, needs_ocr, ocr_parse, oxide_parse.

clean_md('a docu-\nment')      # a hyphen split across lines

'a document'

orphan_vals('Opening balance\nFees\n1,240.00\n18.50')   # values cut loose from their labels

2

Install

pip install pdflite

OCR needs liteparse, imported when it runs. No extras: an optional package is imported where it is used and raises saying what to install.