PDF to Excel

PDFs store text as a stream of positioned characters — each glyph carries its x/y coordinates on the page, but the file has no notion of "table cells". This tool reconstructs the table in three passes:

1. Read the page text. pdfjs-dist returns every text item with its position and width on the page.
2. Group items into rows. Items whose y-coordinates are within a few points of each other are treated as the same row, then sorted top-to-bottom.
3. Split each row into cells. Within a row we scan left-to-right; if there's a horizontal gap wider than ~8 PDF points we start a new cell, otherwise we concatenate the text.

The result is one .xlsx workbook with either a single sheet (stacked) or one sheet per page (separate), your choice.

Works well: bank statements, invoices, tax forms, exam mark sheets, simple data tables, anything originally exported from a spreadsheet. The rows are crisp, column gaps are obvious, and there's a clear text layer.

Tricky: articles laid out in multiple columns (e.g. magazines), tables with merged cells or rotated text, scanned pages without an OCR layer, and forms where fields aren't aligned to a grid. For these cases, expect to do some clean-up in Excel after import — or pre-process the PDF through /tools/pdf-ocr to add a text layer.

Two modes are offered for the workbook layout:

• All pages → one sheet: the rows from every selected page are concatenated into one worksheet, with an empty separator row between pages. Use this when a single table runs across multiple pages.
• One sheet per page: each selected page becomes its own worksheet named "Page 1", "Page 2" and so on. Use this when each page has an independent table.

The page range picker lets you skip cover pages, indices and other non-table content — try ranges like 3-15 or 2-4,7,9.

Both the PDF parsing (pdfjs-dist) and the .xlsx assembly (JSZip + a hand-written Open XML SpreadsheetML emitter) run locally in your browser. There's no upload step, no temporary server-side copy, and no analytics that ship document contents. Verify in DevTools → Network: the page should make zero outbound requests for your file during extraction.

Use this tool when you need a fast, one-off PDF task and want the document to stay in your browser. Desktop editors or command-line tools are better for heavily encrypted files, regulated review workflows, or very large batch jobs that need repeatable automation.

• Password-protected, digitally signed, or archival PDFs may need a specialist workflow before editing.
• Large scans can use a lot of memory, especially on phones or older laptops.
• Check the downloaded file before replacing the original, because compression, OCR, or conversion can change visual details.