How Linework works

Four steps from PDF to spreadsheet

1. Choose a PDF
2. Detect and map the table
3. Review every row
4. Download Excel

1. Choose a PDF

Use a supplier price list, a product table in a catalogue, or an itemised quotation. The table needs a product code or a description, plus a unit price. Files can be up to 15 MB and 50 pages. Automatic reading uses English OCR on pages with fewer than about 50 non-whitespace text characters. If headings are selectable but the products are a scan, choose OCR on every page and use Re-read PDF.

2. Detect and map the product table

Linework guesses the header from English column labels. You can pick another header line. Then map product code, description, pack size, unit price and currency to the PDF columns you actually have. Map unit prices rather than line totals. Choose the number format explicitly: decimal comma (1.234,56) or decimal point (1,234.56). The default currency is applied when a price has no currency symbol; it is not a conversion between currencies or tax bases.

3. Review every row

Compare extracted rows with the original page preview. Correct OCR errors, merge wrapped descriptions on the same page, remove headings and totals, and add missing products. A row can be marked checked only after required fields are valid. Editing a checked row clears its review mark.

4. Download Excel

Export becomes available when every remaining row is checked. Product codes are written as text so leading zeros stay intact. Prices are stored as numbers. Source page and source text are included for traceability. Strings that look like formulas stay as literal text cells.

Selectable-text PDFs

When a PDF already contains text, Linework groups characters into lines and columns from their positions on the page. This is usually the fastest path. It still needs review: PDF text order can differ from the visual table, and repeated headers on later pages can shift column positions.

Scanned PDFs and OCR

When a page has little selectable text, or when you choose OCR on every page, Linework rasterises the page and reads it with a local English Tesseract worker. OCR supports upright English scans. Clear, high-contrast pages work best. Characters such as 0/O, 1/I/l and 8/B are commonly confused. OCR rows with low word confidence are flagged, but confidence is only an engine signal. Up to 20 pages can be processed with OCR in one file.

Column mapping, decimals and currency

Header detection is a starting point. If the wrong line is selected, extraction starts from the wrong place. After you confirm the header, check that unit price—not a quantity or a line total—is mapped to price. Then set the decimal format that matches the PDF. Changing the number format unchecks rows so prices can be rechecked. Currency is a default label, not a foreign-exchange conversion.

Product-code preservation

Alphanumeric SKUs and codes with leading zeros are easy to damage in spreadsheets. Linework keeps the product-code column as text in the exported workbook. You should still compare codes with the PDF, especially after OCR.

Why review is required

Linework extracts candidate product rows; it does not reconstruct invoice totals, tax lines or arbitrary visual layouts. Limits are 15 MB, 50 pages, 20 OCR pages and 3,000 candidate rows. Unreadable pages are listed in the workspace and in export notes. For a practical walk-through, see the price list example and the guide on converting a PDF price list.