Extract tables from PDF files and export as CSV. Table detection based on text positioning — 100% local, no upload.
Table detection uses text positioning heuristics. Scanned PDFs (images) require OCR and may not produce accurate results. For best results, use PDFs with selectable text.
A PDF table is not a table — it is a scatter of text items, each carrying a string and an(x, y) position from its text matrix. To recover a spreadsheet, this tool clusters items by Y into rows, sorts each row by X into columns, merges cells that are nearly touching, and emits the result as CSV rows and columns. The heuristics handle clean, text-based tables well; scanned pages have no text items to cluster, so they require OCR first.
Rows are recovered from shared Y baselines, columns from left-to-right X order, and cell boundaries from gaps between items. The CSV keeps numeric cells unquoted so Excel imports them as real numbers.
A one-page invoice from a web shop has a 3-column line-item table plus a totals block.
Extract tables from PDF files and export as CSV. Table detection based on text positioning. 100% local — no upload, no signup, free.
The PDF to Excel Converter finds tables inside PDF documents and exports them as CSV files you can open in Excel, Google Sheets, or Numbers. Built on Mozilla's pdf.js renderer, it walks every page of your document, reads the position of each text run, and reconstructs rows and columns from those coordinates. Detected tables appear in a scrollable preview with sticky headers, labeled by page number and row and column counts, and a progress bar tracks multi-page files in real time. The export quotes every cell and escapes embedded quotes, so commas inside values survive intact. Files stay on your machine; the parsing worker is loaded locally and nothing is transmitted.
Accountants and analysts pull figures from bank statements, invoices, and annual reports into spreadsheets without retyping them. Researchers extract results tables from published papers for meta-analysis. Procurement and operations staff convert supplier price lists and inventory reports that arrive as PDF attachments. Students grab data tables from textbooks and government publications for coursework. Consultants working with confidential client documents benefit most from the architecture: because parsing runs in the browser through pdf.js, a sensitive financial PDF never travels to a conversion server that might store it. The trade-off to know upfront is that the detector reads text coordinates, so it shines with digital, text-based PDFs and struggles with scans, which is covered below.
(1) Select a PDF; the page accepts files up to the size limit displayed on the upload panel and rejects anything that is not a PDF. (2) Click Extract Tables and pdf.js parses each page's text content with coordinates. The detector groups text items into a row when their vertical positions sit within 3 pixels of each other, sorts each row left to right by horizontal position, and splits cells wherever neighboring items sit more than 20 pixels apart. Rows with fewer than two columns are discarded as prose. (3) Review the per-page tables, then click Download CSV; each section is prefixed with a '# Page N' comment and cells are quote-escaped for clean spreadsheet import.
No, and knowing why saves you time. A scanned PDF is a stack of images; it contains no text objects for pdf.js to read, so the coordinate-based detector finds nothing and the tool reports that no tables were detected. The same applies to PDFs where the table was drawn as vector lines with text placed loosely around it. The fix is to run the scan through an OCR tool first, which converts the image into selectable text while preserving approximate positions, then bring that output here. For native digital PDFs, exports from Word, Excel, LaTeX, or reporting tools, extraction is reliable, including multi-row financial tables, since row grouping tolerates three pixels of vertical jitter between text runs.
It analyzes text positioning on each PDF page to identify tabular data. Cells are detected by their x/y coordinates and grouped into rows and columns based on alignment. The extracted table is exported as CSV, which opens directly in Excel, Google Sheets, or Numbers. Accuracy depends on how well-structured the original PDF table is.
No. This tool extracts text from digitally created PDFs (where text is encoded as characters). Scanned PDFs (images of pages) require OCR first. For scanned tables, use the AI Data Extractor tool which includes OCR. Both tools run 100% in-browser—no file upload to cloud servers.
This tool is also known by these tasks — each link opens the same tool with a focused guide:
What do you call a crab that plays baseball?
No paywalls, no signups, no data sold. Built by a solo developer who believes useful tools should be accessible to everyone.
☕Support me on Ko-fi— keep tools free100% of proceeds go towards hosting & building more free tools.