$ pdf --tables --csv
Text and filesExtract PDF Tables
Turn tables from PDFs into CSV or XLSX by character position — with a preview of the result. All in the browser, no document upload.
- no upload
- CSV and XLSX
- free
When to convert a PDF to a spreadsheet
A table from an electronic invoice you need in Excel. A bank statement PDF with rows you want in a tracking spreadsheet. Product specifications with a dimensions table that has to land in a catalogue. In all those cases, copying each cell by hand is the long way — the table extractor saves the work.
How it works (and the heuristic limit)
pdfjs exposes each character of the PDF with its coordinates (x, y) on the page. The tool groups characters into rows (similar Y) and into columns (significant gaps between X positions). The result is a two-dimensional grid that becomes the table. On well-structured documents — invoices, statements, receipts — it works well.
For scanned PDFs (image without embedded text), there are no character positions to work with. In that case the path is OCR: the Image OCR tool extracts text from an image converted first via PDF to JPG. It is not the same as table extraction, but it is the best alternative for scans.
Why "no upload" changes everything
Services like iLovePDF "PDF to Excel" and Adobe online upload the PDF to the cloud and return the XLSX. Here every step of parsing happens in your browser, in memory. The resulting spreadsheet is generated locally. Important for statements with account numbers and balances, invoices with a customer CNPJ, specifications with prices and margins. None of that leaves your machine.
## faq
Frequently asked questions
Is the PDF uploaded to a server?
No. The PDF is read in browser memory, the position of each character is analysed and the table is rebuilt locally. No upload, no queue, no cloud copy — useful for invoices, statements and receipts with sensitive data.
When does it work well?
On PDFs with well-formed tables that were generated digitally (not scanned): electronic invoices, bank statements in PDF, utility bills, product specifications. The detection uses the geographic position of every character on the page, grouping by coordinates — it works when position is meaningful.
When does it NOT work?
Scanned PDFs (no embedded text) have nothing to detect — use our OCR first. Tables with merged cells, column breaks across multiple pages and complex layouts (matrices with rows of different heights) produce imprecise results; manual review is worth it.
CSV or XLSX?
CSV is the choice when you will import into Google Sheets, Numbers, LibreOffice Calc or a database. XLSX is better when you will send it to someone who opens it directly in Microsoft Excel. Both preserve the column and row count.
Does it work with password-protected PDFs?
No. We detect and warn, but we do not try to decrypt — it would be dishonest. If you have the password, open the PDF in your reader, save an unlocked copy and bring it here.
## other tools