Sample content for theme review. Figures are from the linked document. Remove with npm run unseed.
Lines
470
Parsed
383
Volume tiers
17
Discount off list
10%
Evidence
list price

The document

Extraction

Text extracted with pypdfium2, 1 regex per sheet layout, rows to CSV. 383 of 470 lines parsed. The parser splits the remaining 87 multi-line product names wrongly. They are held until it is fixed and a person has checked them. Parsed rows carry confidence A: the sheet states unit, term, list price and contract price.