Extract PDF Text
Share tool
Only this public address is included. Your files, inputs and results stay out of the link.
Copy PDF text into readable paragraphs. Inspect the source, correct reading order and review line-break cleanup before copying or exporting TXT and Markdown.
Preparing your tool…
A visual example
Read one column before the next
Example: a two-column page is extracted row by row instead of column by column.
- Extracted order
The river
The mountain
reaches the lake.
has a snowy peak.
Interleaved extraction
The right-column heading appears before the left column has finished.
- Corrected order
The river reaches the lake.
The mountain has a snowy peak.
Reviewed reading order
After correcting the order, each thought stays together.
Check the source layout first, then review how line breaks and paragraphs are joined.
Illustrative example. Column and paragraph suggestions are heuristic. This tool uses selectable PDF text, not OCR; tables, code and rotated text need particular care.
How to use it
- Choose or drop a PDF, or load the synthetic study notes.
- Suggest reading order and check the source preview. Correct individual line positions and mark headings, lists or protected code/table lines.
- Choose paragraph joining and optional repeated-margin removal. Confirm each hard-hyphen break separately; real compound words remain unchanged by default.
- Create the cleaned preview, compare the original and change log, and undo up to five recent edits if needed.
- Copy the reviewed text or prepare TXT/Markdown. Generated bytes are reopened as UTF-8 and compared before download.
The three-page synthetic PDF contains German columns with Datenver- / arbeitung, a real E-Mail-Konto compound, English paragraphs, lists, code and repeating margins. Confirm only the Datenver- break; code and real hyphens must remain.
Details & limits
Selectable text only; no OCR. Up to 25 MiB, 100 pages, 50,000 fragments and 2 million text characters. A 60-second worker limit applies to each operation. Complex compression and unusually large expanded contents may be rejected. Text export is limited to 16 MiB. Columns, paragraphs and margin repetition are heuristic; tables, code and rotated text need review. Embedded images: unfiltered/Flate pixels or checked JPEG, up to 8,192 px per side. Inline images, JPEG 2000, JBIG2, CCITT and unusual JPEG variants are unsupported.
Will it correctly reconstruct every column, table or scanned page?
No. Reading order and cleanup are inspectable suggestions. Reorder lines and protect complex content manually. Pages without extractable text are listed separately; scans and outlined letters are not claimed as successful extraction.