PDF Text Extractor — Text, Pages & Metadata, Scan Detection

Extract the text of PDF files from their URLs: the whole document or page by page, with page count and metadata. Scanned PDFs without a text layer are detected and not charged.

What you get

Field	Meaning
url	the URL you gave
status	ok, no_text_layer, encrypted, not_pdf, too_large, download_failed, parse_error
error	machine-readable detail when status is not ok (e.g. http_404, timeout, password_required, blocked_address), otherwise null
pageCount	pages in the document
pagesExtracted	pages actually extracted (see Max pages per PDF)
charCount	characters of extracted text
needsOcr	true when at least half of the pages have no text layer (typical of scans)
text	the text, pages separated by a blank line — or, with Text page by page on, pages: [{ page, text }]
metadata	title, author, subject, creator, producer, creationDate, modDate (ISO 8601)

{
  "url": "https://example.com/invoice.pdf",
  "status": "ok",
  "error": null,
  "pageCount": 2,
  "pagesExtracted": 2,
  "charCount": 77,
  "needsOcr": false,
  "text": "Invoice 2026-001\nTotal due: 1,234.56 EUR\n\nPage two\nThank you for your business.",
  "metadata": { "title": "Test invoice", "author": "Test author", "subject": null, "creator": null, "producer": "…", "creationDate": "2026-01-15T10:00:00.000Z", "modDate": null }
}

A run summary (counts per status, duplicates, charged events, limits used) is saved in the key-value store as SUMMARY — open it from the Output tab, Run summary.

Very long texts (results over about 8 MB) are stored in the key-value store as a separate record; the result then has text: null and textRecordKey with the record name.

Input

Field	Default
pdfUrls	—	public http(s) links, one per line
outputPerPage	false	page-by-page output
maxPagesPerPdf	500	only the first N pages are extracted and charged (max 2,000)
maxFileSizeMb	50	larger files are not downloaded (max 200)

Pricing

Pay per event: $0.003 per PDF processed successfully, up to 100 pages. Longer documents: one more event for every started block of 100 pages beyond the first 100.

Pages extracted	Events	Price
1–100	1	$0.003
101–200	2	$0.006
201–300	3	$0.009
500	5	$0.015

Not charged: scans without text (no_text_layer), password-protected files, files that are not PDFs, files over the size limit, failed downloads, broken PDFs, duplicate URLs. The Actor respects your spending limit: it stops cleanly before a PDF it could not pay for, and never delivers a result you have not paid for or charges for one you do not receive.

Safety and privacy

• PDFs are parsed with pdf.js with script evaluation disabled (isEvalSupported: false) and a version patched for CVE-2024-4367. Scripts and XFA forms inside the PDF are never run, and no system fonts are used.
• Only http and https. Internal and private network addresses (loopback, private ranges, link-local, cloud metadata endpoints) are refused, also after redirects (max 5).
• Downloads stop as soon as they exceed the size limit; the first bytes must be a PDF header; there is a time limit per file.
• Nothing you submit is written to the logs — no URLs, no text, only counts. Results stay in your own dataset.

Limits, stated plainly

• No OCR in this version: scans are detected (needsOcr, no_text_layer) and not charged, but their text is not recognized.
• Text order follows the PDF's internal structure; complex layouts (columns, tables) may come out in reading order that differs from the visual one.
• Only publicly reachable URLs: files behind a login are not supported.

Licenses

pdf.js (pdfjs-dist) — Apache License 2.0. Apify SDK — Apache License 2.0.