PDF Table to Excel & CSV
Pricing
from $2.00 / 1,000 tables
PDF Table to Excel & CSV
Extract tables from PDF links into rows, one Excel file per PDF and a CSV per table. Tables that run across pages are joined into one. Billed only for tables delivered: $2 per 1,000.
Pricing
from $2.00 / 1,000 tables
Rating
0.0
(0)
Developer
Brandon Ob
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
a day ago
Last modified
Categories
Share
PDF Table to Excel & CSV: extract tables from PDF links
Paste links to PDF files and get every table back as rows you can export, plus one Excel file per PDF and one CSV per table. A table that runs over a page break comes back as one table, not three, and a header repeated on each page is dropped. You are charged per table delivered, never for a PDF that failed or had no tables.
What you get
One dataset item per table, with the rows as objects keyed by the header. Real output from the default run (first 3 pages of a public California layoff-notice report; 3 pages, 1 table, 122 rows):
| Notice Date | Effective | Received | Company | City | No. Of | Layoff/Closure |
|---|---|---|---|---|---|---|
| 06/22/2015 | 03/25/2016 | 07/01/2015 | Maxim Integrated Product | San Jose | 150 | Closure Permanent |
| 06/30/2015 | 08/29/2015 | 07/01/2015 | McGraw-Hill Education | Monterey | 137 | Layoff Unknown at this time |
| 07/17/2015 | 09/18/2015 | 07/21/2015 | Boeing Company | Huntington Beach | 65 | Layoff Unknown at this time |
Each item also has pdfUrl, table, tablesInPdf, pages, columns, rowCount, csvUrl, excelUrl and extractedAt. The excelUrl file has one sheet per table with a frozen header row. The key-value store record SUMMARY lists PDFs that could not be read and PDFs where no tables were found.
How to use
- Add PDF links, one per line. The file must open in a browser without a login.
- Optional: the pages to read (
1-3,7; empty reads every page). - Pick the table detection: lines for tables drawn with ruled borders, text for tables laid out with spaces and no borders.
- Run it, then open the Tables view or download the dataset as CSV, JSON or Excel. For one spreadsheet per PDF, use the
excelUrlcolumn.
Other options: join tables that continue on the next page (on by default), use the first row as the header (on by default), ignore tiny tables with fewer data rows than a minimum, a password for protected PDFs, and a maximum number of tables.
Pricing
Pay per event: $2.00 per 1,000 tables (the dataset-item event) plus a $0.00005 Actor start. A 100-page report with one long table that continues across pages is one table. You set a maximum cost per run and the Actor stops cleanly at it.
Limits (honest)
- Works on PDFs that contain real text. Scanned pages (pictures of paper) have no text to read and come back with no tables; there is no OCR.
- Table detection is a best effort. Ruled tables work best; borderless tables need the text setting and can merge or split columns when the spacing is uneven. Check the first run on your own layout.
- Cells that span several columns or rows are not split out. Text in a merged cell appears once, in the first column it covers.
- Dates that the PDF sets one character at a time (for example
0 3 / 2 5 / 2 0 16) are closed up to03/25/2016. Other cell text is returned as the PDF has it, with extra whitespace collapsed. - Joining across pages needs the same number of columns on both pages and the table to be the first one on the next page.
- Files up to 50 MB. Public links only.
- Check that you are allowed to use the PDFs you submit.
FAQ
My table came back empty or in pieces. Try the other Table detection setting, lower Minimum data rows, or switch off Join tables that continue on the next page to see page by page.
Something broke or you need a feature? Open an issue on the Issues tab. I usually reply within a day.
