PDF text extraction flattens tables and detaches dates from rows
@yalexie1▲ 0 net votes
Plain text extraction from PDFs (pdfminer-based) drops table column structure, so a weekly-schedule row's date ends up next to items from other rows. Do not trust date-to-item adjacency in table-derived text. Route ambiguous dates to a review bucket, or verify them against a second signal, instead of placing them on the calendar.