Extract Tables
Move tables from documents into structured files. Accepts PDF, JPG or PNG up to 20 MB. Produces CSV or JSON. Processing runs on demand with Cloudflare Workers AI without FileClear AI file storage.
Input: PDF, JPG or PNG up to 20 MB ยท Output: CSV or JSON
Quality boundary
Detected tables include page context and structured cells. Merged cells, borderless layouts, and photographed tables require review.
How this file is processed
Cloudflare processes the request on demand; FileClear AI does not save the uploaded bytes or generated result in R2. Review the result before consequential use.
Read the image-to-CSV table extraction tutorial
Useful next actions
Move PDF tables into JSON or CSV without losing context.
Table extraction is reliable only when rows, columns, headers, units, and page references stay connected. A spreadsheet that looks tidy can still contain a shifted value or an incorrect total.
Identify how the table is stored
Selectable text can often be parsed directly, while scanned or photographed tables require OCR first. Borderless tables, merged cells, repeated headers, multi-page rows, and footnotes need additional review. Choose CSV for portable rows and columns, or JSON when nested table and source metadata matter.
Validate before analysis
Compare the first, middle, and final rows with the source page. Confirm headers, units, decimal separators, negative values, blank cells, and totals. Keep page references in the export and treat formulas beginning with spreadsheet control characters carefully before opening data from an untrusted document.
Quick quality check
- Confirm headers and units
- Review merged cells
- Compare totals
- Keep source references
- Choose CSV or JSON intentionally