PDF to Excel
Operator-list parsing · Border detection · Merged cells · Font styles · Form fields
⧉
Drop your PDF here
or browse file
✓ No upload✓ No watermark✓ No sign-up
Feeding PDFs to an AI support agent?
helpGPT answers your customers' questions from your own docs, with citations and honest refusals.
101 Operator List ParsingEvery path op, colour change (RGB/CMYK/gray) and CTM matrix transform is traced — no rasterization needed.
202 Border-first DetectionH and V line segments are extracted, tolerance-clustered into a grid, and intersected to find cell bounding boxes precisely.
303 Merged Cell DetectionEach expected interior border segment is checked for existence. Missing ones produce XLSX merges entries automatically.
404 Cell Background ColoursFilled rectangles whose bounds match a cell are mapped to Excel cell fills, preserving colour coding from the original PDF.
505 Coordinate FallbackFor borderless tables, items are row-clustered by Y-tolerance then column-histogrammed by X — robust to ragged alignment.
606 Data Type InferenceEvery cell is tested for boolean, date (8 patterns), percentage, currency, number with format, or plain string.
707 Form Field ExtractionAcroForm widgets (Tx, Btn, Ch) are read from page annotations and exported to a dedicated Form Fields sheet with labels.
Move PDF tables into editable spreadsheets
One sheet per page, or combine everything onto a single sheet. Column alignment is inferred from text positions — clean tables convert best.
Safe for tables and statements
Detected tables move sheet by sheet into an editable workbook locally, so invoices, bank tables and research data stay in your browser.
7-STAGE EXTRACTION PIPELINE
- 1Operator List ParsingEvery path op, colour change (RGB/CMYK/gray) and CTM matrix transform is traced — no rasterization needed.
- 2Border-first DetectionH and V line segments are extracted, tolerance-clustered into a grid, and intersected to find cell bounding boxes precisely.
- 3Merged Cell DetectionEach expected interior border segment is checked for existence. Missing ones produce XLSX merges entries automatically.
- 4Cell Background ColoursFilled rectangles whose bounds match a cell are mapped to Excel cell fills, preserving colour coding from the original PDF.
- 5Coordinate FallbackFor borderless tables, items are row-clustered by Y-tolerance then column-histogrammed by X — robust to ragged alignment.
- 6Data Type InferenceEvery cell is tested for boolean, date (8 patterns), percentage, currency, number with format, or plain string.
- 7Form Field ExtractionAcroForm widgets (Tx, Btn, Ch) are read from page annotations and exported to a dedicated Form Fields sheet with labels.
Was this useful?
Frequently asked questions
Does it detect tables automatically?
Column alignment is inferred from text positions — clean tables convert best, messy scans need OCR first.
One sheet per page?
Your choice — each page on its own sheet or all tables combined.