A PDF table extractor that lets you check the result
The PDF Table Extractor finds the tables inside a PDF and turns them into rows and columns you can edit, then export to Excel, CSV, TSV, JSON or Markdown. It is built for the documents people actually struggle with: bank and card statements, price lists, invoices with line items, timetables, lab reports and the tables at the back of annual reports.
Copying a table straight out of a PDF viewer usually gives you one long line of words, because a PDF does not store “cells” at all — only pieces of text placed at exact positions on the page. This tool reads those positions, groups the text into rows, works out where the columns are, and shows the result in a grid so you can merge a column that was split, split one that was merged, fix a cell or delete a stray row before you export.
Everything runs in your browser with Mozilla’s open-source PDF engine. Your statement or report is never uploaded, which matters when the numbers are private.
How to extract a table from a PDF
- Drop your PDF onto the upload area, or click it to choose a file. Every page is scanned for tables straight away.
- Optional: type page numbers such as
3-5in Pages, or change Find tables using if the result is not right. - Pick a table from the Table list. Click a cell to select its row and column, then use Merge column with next, Split column, Delete column or Delete row. Type in any cell to correct it, and press Undo if you change your mind.
- Tick First row is the header if the detection guessed wrong — it controls JSON keys and the Markdown header.
- Choose whether to export the selected table or all tables, then press Excel (.xlsx), CSV, TSV, JSON or Markdown. Copy for spreadsheet puts the table on your clipboard ready to paste into Excel or Google Sheets.
PDF Table Extractor features
Ruled and borderless tables
Grid lines drawn in the PDF are read from its drawing instructions and used as exact cell borders. Tables without lines are found from text that lines up in columns across several rows.
Wrapped rows joined
A description that wraps onto a second line, with no date or amount of its own, is joined back into the row above — the classic bank-statement problem.
Editable grid with undo
Merge or split columns, edit cells and delete rows, with up to 40 steps of undo. Splitting uses the real gap between the words on the page when there is one.
Numbers understood
Western (1,234.56) and Indian (1,23,456.00) digit grouping, bracket negatives such as (2,500.50), minus signs and currency symbols are all recognised.
Real Excel numbers
In the .xlsx file numeric cells are stored as numbers with a matching format, so totals and formulas work immediately. All tables can go into one workbook, one sheet each.
Private by design
The PDF is processed on your device; the spreadsheet library is only downloaded when you ask for an .xlsx file.
How table detection works
First, every piece of text is placed on a line by its baseline. Lines are then cut into cells wherever there is a gap wider than about four-fifths of the text height, or wherever a vertical ruling line crosses between two words. A run of neighbouring lines that each have two or more cells becomes a candidate table.
Columns are the x-ranges covered by cells in the fullest rows of the block. Every word is then assigned to the column it overlaps most, which is why a right-aligned amount and a left-aligned heading still end up in the same column. Rows whose words fall neatly into those columns are added even when their own gaps are narrow. When the PDF contains drawn grid lines, they define the rows and columns directly instead.
Which detection setting to use
| Setting | Best for | What it does |
|---|---|---|
| Ruling lines + text alignment | Most documents | Uses drawn grids where they exist and column alignment everywhere else |
| Ruling lines only | Forms and reports full of text | Ignores borderless layouts, so paragraphs and address blocks are never mistaken for tables |
| Whole page as one table | Statements and lists the detector misses | Treats every line on the page as a row and lines the columns up across the whole page |
Tips for clean tables
- Two columns came out as one? Select a cell in that column and press Split column. If the words sit close together in the PDF, the split happens at the first space instead — check the result.
- A column was cut in half? Select its left part and press Merge column with next.
- Scanned PDF? A scan contains pictures, not text, so there is nothing to detect. Turn the pages into images with PDF to JPG and read them with Image to Text first.
- Need plain CSV with more options? PDF to CSV offers delimiter, quoting, BOM and one combined file with page and table columns.
- Leading zeros (PIN codes, account numbers) and numbers longer than 15 digits are kept as text so Excel cannot damage them.
PDF Table Extractor FAQ
How do I extract a table from a PDF to Excel?
Upload the PDF, check the table in the grid, then press “Excel (.xlsx)”. Choose “All tables” under Export to get every table in one workbook, one sheet per table. Amounts become real numbers you can add up.
Is this PDF table extractor free and private?
Yes. There is no sign-up or watermark, and the PDF never leaves your device — detection, editing and export all happen in your browser.
Why did a table not get detected?
The page may be a scanned image, the “table” may be a picture, or the text may be rotated. For borderless tables with very tight columns, try “Whole page as one table”, then delete the rows you do not need.
Can it handle tables that continue over several pages?
Each page’s part of the table is found separately and listed as its own table. Export “All tables” to CSV or Excel, or use PDF to CSV to combine them into one file with a page column.
Does it keep merged cells?
A spanning cell’s text is placed in the first column it overlaps; the other columns in that row stay empty. Formatting such as colours, bold and cell borders is not exported.
Which number formats does it recognise?
Grouped numbers like 1,234.56 and Indian lakh grouping like 1,23,456.00, negatives written as (2,500.50) or −2,500.50, European 1.234,56, and values with ₹, $, £ or € signs.