Loading
Loading tool
The tool is loaded only when you open it.
All processing for this tool happens in your browser. Your input is not sent to a server.
About this tool
Extract selectable text from one PDF page using a region and explicit column boundaries. Review the inferred cells before exporting. No upload or automatic OCR.
Common uses
- Convert a selectable-text report table into protected CSV.
- Adjust region and column boundaries and inspect possible merged or missing cells.
How to use it
- 1.Choose a text PDF — Load page preview
- 2.Percent coordinates follow the displayed page, from its top left. Boundaries must increase from 0 to 100, separated by commas. Blank rows infer baseline groups; wholly empty rows cannot be inferred. Explicit row boundaries preserve empty rows. Column/row-spanning items go into their center cell and are flagged; text items are not split. Extract table
- 3.I reviewed the region, cells and uncertainties. Export may differ from the original table. Save protected CSV
Limits and notes
- One PDF ≤10 MiB, ≤10 pages; one page per extraction. ≤4000 text items/200000 UTF-16 units, 500 rows/32 columns/16000 cells, CSV ≤2 MiB, JSON ≤8 MiB. Preview ≤1200px per edge/1.44 million pixels. File read ≤30s; each inspect/page/extraction phase ≤45s. Bounded PDF preflight: 20000 objects, decoded content ≤16 MiB, embedded images ≤16 million each/64 million total. These limits do not cap all native decoder memory.
- Layout is heuristic, not a semantic table parser. Only left-to-right horizontal text as displayed is supported; vertical/RTL/skewed text, Type3 fonts, inline images, encryption, forms, actions, annotations and unsupported content/image codecs are rejected. A textless scan cannot be extracted: use OCR separately, then review its recognition. Text order, fragment joining, merged cells and entirely missing rows need manual review. Font fallback may affect the preview. CSV always quotes every field and prefixes formula-triggering values with an apostrophe, changing those exported strings; JSON retains original cells. JSON contains full extracted text and chosen geometry. No persistence, remote assets or uploads; only installed PDF.js font/CMap assets from this site. New file/page/settings/locale, cancel, clear or leaving invalidates results. PDFs generated by the searchable scanned PDF tool use Type3 fonts and are currently unsupported here.
Frequently asked questions
Can this extract a scanned table?
No selectable text was found in this page or region. Textless scans need separate OCR; no table was invented.
Why is an apostrophe added to some CSV cells?
Formula-triggering values are prefixed to reduce spreadsheet formula execution. This changes those exported strings; the JSON report preserves original cell text.