PDF Text Extractor & Parser — OCR for Scans, Changed Pages
Pricing
from $2.40 / 1,000 pdfs
PDF Text Extractor & Parser — OCR for Scans, Changed Pages
Paste PDF links and get the text of every page, plus title, dates and page count. Scanned pages are read with OCR (19 languages). Run it again and get only the PDFs and pages whose text changed. Good for price lists, policies, menus and reports. $3 per 1,000 PDFs.
Pricing
from $2.40 / 1,000 pdfs
Rating
0.0
(0)
Developer
Northbell
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
5 days ago
Last modified
Categories
Features
Share
PDF Text Extractor & Parser — Text, OCR for Scans & Only Changed Pages
Paste PDF links and get the text of every page as clean JSON, with the title, subject, creation and modification dates, page count and file size. Scanned pages are read with OCR (19 languages, including Japanese, Chinese, Korean and Arabic). Turn on One row per page for long documents or RAG chunks. Schedule it with Only new and changed PDFs and you get back only the PDFs whose text changed since the last run — with the numbers of the pages that changed — so you can watch price lists, policies, terms, menus, tenders and reports without reading them again.
Try it now. The input is already filled in. Press Start: the IRS Form W-9 PDF — 6 pages, 6,285 words of text in about 7 seconds, for $0.013 (measured on 28 Oct 2026). The row:
title | pageCount | words | hasText | ocrPages |
|---|---|---|---|---|
| Form W-9 (Rev. March 2024) | 6 | 6,285 | true | [] |
The row also has the full text, author, subject, dates and a SHA-256 fingerprint. Run it again and status says whether the PDF changed, and changedPages which pages.
What you get
| Row | Fields |
|---|---|
pdf | url, title, author, subject, keywords, creator, producer, createdAt, modifiedAt, pdfVersion, pageCount, pagesRead, fileBytes, sha256, status (new / changed / unchanged), changedPages, removedPages, hasText, textChars, words, ocrPages, ocrConfidence, text, note |
page | url, page, pageCount, text, chars, words, ocr, changed (with One row per page) |
error | url, message — free |
author is the PDF's own Author field, as the file states it.
Use it for
- Watching documents that change quietly: supplier price lists, terms of service, privacy policies, school menus, council minutes, tender documents: schedule it daily and get a row only when the text changes, with the changed pages
- AI and RAG: page-level text ready for chunking and embedding, refreshed only where it changed
- Document inventories: title, dates and page counts for a list of PDFs
- Search and analysis: full text of reports and filings for keyword search or LLM summaries
Input
{ "urls": ["https://www.irs.gov/pub/irs-pdf/fw9.pdf"], "splitPages": true, "onlyChanged": true }
Measured on 23 October 2026: the 6-page IRS Form W-9 (6,285 words) in 1.4 seconds.
How it reads
- Each site's robots.txt is read first (the rules for all robots and those addressed to Claude crawlers); links it does not allow are skipped with a free error row. Links to private networks are not read.
- The file is downloaded over plain HTTP (no browser, no login) and read with PDF.js. Scanned pages are read with OCR (Tesseract, inside the run — the file is not sent to another service): a page without a text layer has its picture read in the language(s) you pick under OCR languages.
ocrPageslists those pages andocrConfidencetheir average confidence; OCR text can contain mistakes. Turn off Read scanned pages (OCR) to skip them. - Only new and changed PDFs compares each page's text with this Actor's last run on the same link. A PDF whose file is byte-for-byte the same is not even re-read.
Pricing
$3 per 1,000 PDFs ($0.003 each, page rows included), plus $3 per 1,000 pages read with OCR ($0.003 per scanned page where OCR found text) and $0.01 per run. Error rows, PDFs with no text at all and unchanged PDFs (with Only new and changed PDFs) are free. The $0.01 run-start fee is the one charge every run pays, even one that returns nothing.
For AI agents
{"urls": ["https://example.com/report.pdf"], "splitPages": true} returns one pdf row with the details and one page row per page with its text. Rows have type: pdf, page or error.