OmniDocHow to Extract Tables from a PDF File
Learn how to extract tables from a PDF file to Excel, Word or PowerPoint, including scanned PDFs, with checks that catch misaligned rows and numbers.
A table in a PDF looks like data, but the file stores it as text and lines placed at fixed positions on the page. When you try to pull that table out, the rows and columns often come apart, numbers lose their formatting, or the header ends up in the middle of the data. This guide is for anyone who needs the numbers or text in a table from a report, contract, paper or invoice and wants them in a usable format.
The short answer is to find out whether the PDF is digital or scanned, export the table pages to the format you need, and then check the result against the original. Digital PDFs usually export cleanly. Scanned PDFs need text recognition first. The table below shows which path fits your situation.
Extract tables from a PDF in the format you need, not just the one you happen to have on hand. The steps below work for any target format, and they start from the table itself.
How do I extract tables from a PDF? The short answer
Pick the route based on whether the PDF contains real text or only an image of the page, and on where you plan to use the table.
| Your situation | What to do | Why |
|---|---|---|
| Digital PDF, and you need numbers to calculate with | Export the table pages to Excel from the PDF editor | Rows and columns map to spreadsheet cells, so formulas work |
| Digital PDF, and you need to read or edit the text | Export to Word | The table stays a table you can change |
| Scanned PDF, of any target format | Run OCR first, then export | Export works from recognized text, not a picture of the page |
| Table is in a language you don't read | Translate the PDF with tables kept, then extract | Tables stay in place, and the bilingual option lets you compare |
| Table spans many pages | Split out only the table pages, then export once | Fewer broken headers and split rows to repair |
Before you start
Gather a few things before you export. Each one prevents a common problem later.
- The original PDF, unchanged. Keep a copy of the source file. You will compare the extracted table against it, and that is the only reliable way to catch a shifted number.
- The page numbers that hold the table. Note them now. Table data often starts on one page and continues on the next, and knowing the range helps you split the file cleanly.
- Your target format. Excel suits numbers you will calculate with. Word suits tables you will read, edit or place in a document. PowerPoint suits a table you will show on a slide. Choosing before you export saves a second conversion.
- Whether the PDF is digital or scanned. Try to select a line of text on the table with your cursor. If the words highlight, the text is real and a direct export should work. If nothing highlights, the page is probably an image, so you need OCR first. A scanned page can also have a text layer added by a scanner, so the highlight test is the quickest way to know.
- A rough idea of the table's layout. Note whether it has merged cells, multi-line headers or footnotes. These features are a common reason extracted columns shift, so you will know what to check afterward.
Extracting a table from a PDF, step by step
These steps use the PDF editor in the OmniDoc desktop client, which you can download at download.omnidoc.top. The general method also applies if you use another tool, since the logic is the same.
- Confirm the PDF type. Select a line of text in the table. If it highlights, the PDF is digital. If it doesn't, treat it as scanned and continue to the OCR step.
- Run OCR on a scanned PDF. Use the OCR function in the PDF editor and let it recognize the text on the table pages. Afterward, compare a few cells with the original, because recognition can misread characters such as 1 and l or 0 and O.
- Split out the table pages. Use the page operations to extract only the pages with the table. A smaller file means fewer stray headers, page numbers and body paragraphs in your export.
- Export to your target format. Export the file to Excel, Word or PowerPoint, depending on the choice you made earlier. Export each table range once if you can, rather than in pieces.
- Open the output and look at the structure. Check whether each row sits in one row and each column in one column. Note any place where a value has moved into the wrong cell.
- Clean up the layout. In a spreadsheet, such as UniCell in the desktop client, split or merge cells to match the original, move stray values back into place and restore the header row. Fix only what differs from the source.
- Translate, if you need another language. If the table must be in a different language, upload the PDF to omnidoc.top, choose translation-only or bilingual output, and download the result. Tables, formulas and pictures are kept in the output, so you can extract the table from the translated file using the same steps.
- Save a final copy. Save the cleaned file under a new name and keep the original PDF untouched, so you can always compare against the source.
Check the result
Extracted tables fail in predictable ways. Use this table to find the problem quickly.
| Check | How to see it | If it's wrong |
|---|---|---|
| Row count | Count the data rows in the original and in the output | Look for a row that was split in two or merged with the next one, and rejoin it |
| Column alignment | Compare the first and last columns against the source | Check for merged cells in the original and recreate them in the output |
| Numbers | Compare totals and a sample of values with the original | Fix decimal separators, negative signs and percentages, which often arrive as text |
| Header row | Confirm the column names match the source, including any header that repeats across pages | Move the repeated header into a single row and delete the duplicates |
| Footnotes and units | Look for notes below the table and unit labels such as "thousands" | Add the note or unit back to the output so the numbers keep their meaning |
| Characters | Search for symbols and accented letters | Re-run OCR on that page, or correct the characters by hand |
Machine translation and OCR can both make mistakes, so check any table that carries important numbers, legal terms or medical values against the original page before you rely on it.
❓ FAQ
1. Can I get a table out of a PDF without using Excel?
Yes. You can export the table to Word or PowerPoint, which keeps it as a table in a document or on a slide. If you only need the text, you can also copy the table from a digital PDF and paste it into any spreadsheet or text editor, then fix the layout there.
2. Why do my columns shift after I extract a table?
A common cause is a merged cell or a header that spans several columns in the original. The export may read the text correctly but place it in the wrong column. Compare the output with the source, rebuild the merged cells, and check that each value sits in the same column as its header.
3. Will the table layout stay the same after translation?
OmniDoc translation keeps tables, formulas and pictures in place, and bilingual output lets you check each translated cell beside the original. Even so, check the numbers and any terms that must stay exact, because translation can change how text fits in a cell.
4. What should I do if copied text from a PDF looks like random characters?
This usually means the PDF uses a font that does not map its characters to standard text. Try running OCR on the page so the text is recognized from the image. Then compare the output with the original, because recognition is not perfect and some characters may still need manual correction.
📝 Summary
Start by checking whether your PDF is digital or scanned, since that decides whether you can export the table directly or need OCR first. Choose Excel, Word or PowerPoint based on how you plan to use the data, and split out only the table pages before exporting. Finish by checking rows, columns, numbers and headers against the original, because that comparison catches most extraction errors.
