Extract tables from a PDF to CSV
Open a PDF, see every table it found, and download the one you want as CSV or JSON. Your file never leaves the browser.
1 Open a PDFdrag it in, or click to choose
2 Pick the tableevery table found is previewed free
The tables found in your PDF will appear here, one per page, with a preview of the cells before you export anything.
3 Export itCSV or JSON, once you have a table
How Gridlift finds tables in a PDF
Open a PDF
Open a PDF by dragging it onto the drop area or choosing a file with the picker. The document is read and parsed directly in this page, so it stays in your browser.
Review detected tables
Review the detected tables listed for each page, then select one to preview its rows, columns, and extracted values before deciding how to export the result.
Download your data
Every detected table can be previewed for free. The free preview shows the first ten data rows of the selected table. Seeing every row and downloading CSV or JSON is available through a one-time unlock.
How detection works
Lattice detection looks for ruled lines in the PDF and uses those lines to reconstruct table grids. Stream detection instead examines whitespace column gaps between text, which helps recover tables that have no visible ruling.
When detection mode is auto, the page runs lattice detection first. It falls back to stream detection only when lattice found nothing on that page, keeping ruled tables and whitespace-aligned tables on their appropriate paths.
Headers and merged cells
Header detection considers at most three header rows. When headers are nested, their parts are joined with the separator " > ", producing readable column paths that retain the hierarchy represented by the source table.
If column names repeat, each later duplicate receives a numbered suffix — "(2)", then "(3)", and so on — so exported fields remain distinguishable. Merged cells can either repeat their value across covered positions or be left empty, depending on the selected merge policy.
Export options
Every detected table can be previewed for free, with the free preview showing the first ten data rows of the selected table. The one-time unlock reveals every row and produces CSV or JSON downloads.
The unlock-produced CSV supports comma, semicolon, and tab delimiters, uses CRLF line endings, and can optionally include a UTF-8 BOM. The unlock-produced JSON can contain either the selected table or all tables detected in the document.
- Is my PDF uploaded anywhere?
- The PDF is read and processed in your browser, and the file itself never leaves the browser. If you choose to pay, a Stripe-hosted checkout page opens in a new browser tab; Stripe is a third party you visit to complete payment.
- Which PDFs work best?
- Text-based PDFs work best because their characters are available as a readable text layer. Scanned or image-only pages have no text layer, and this app does not provide OCR.
- What if no tables are found?
- If no tables are found, check whether the table uses ruled lines. Tables without ruled lines fall to the whitespace-based detector, and very tight column gaps can defeat it.
- How are multi-page tables handled?
- Each page is detected separately, so page boundaries remain clear. When exporting all tables to JSON, every detected table keeps an addressable identity for reliable downstream use.
- Is Gridlift free?
- Opening a PDF, detecting every table on every page, and previewing the first ten rows of any table are free, with no page or table limit. Seeing every row and downloading CSV or JSON is available through a one-time unlock.