PDF Text Extractor: PDF to Markdown, Text, Pages & Metadata

Turn any PDF URL into clean, structured text in seconds, with no API key and no setup. Give it direct PDF links, or a web page that links to PDFs, and get:

• Full text as Markdown (headings rebuilt from font sizes, wrapped lines joined into paragraphs, hyphenated line breaks fixed, bullet lists) or plain text with the original line breaks, or both
• One row per document, per page, or per chunk: chunk mode splits the text into RAG-ready pieces (size and overlap you choose) with page numbers, ready for a vector database, embeddings or an LLM
• PDF metadata: title, author, subject, keywords, creator app, producer, creation and modification dates, PDF version, language, page count, file size, SHA-256, fillable form and tagged-PDF flags
• Document outline (table of contents / bookmarks) and every hyperlink inside the PDF
• Scanned-page detection: pages without a text layer are listed (noTextLayerPages) and likelyScanned flags image-only documents, so you know which files need OCR
• PDF link discovery: put in a web page (an annual-report page, a government forms page, a documentation index) and every linked PDF is extracted
• Page ranges (1-5, 8, 10-), max pages, password-protected PDFs, max file size, optional Apify Proxy

Built on Mozilla's pdf.js, the PDF engine inside Firefox, so it reads real-world PDFs from Word, LaTeX, InDesign, Google Docs, scanners and government form tools.

Pricing: $1 per 1,000 PDFs

	This Actor	Typical PDF text extractors on Apify
Per PDF	$0.001 ($1 per 1,000)	$0.0035 to $0.02 per PDF
Per page	$0.00002 ($0.02 per 1,000 pages)	up to $0.001 per page
Actor start	$0.0005 per GB of memory (512 MB default)	$0.005 per run
Failed URLs, web pages scanned for links	free	often charged

A 10-page report costs $0.0012. A 150-page manual costs $0.004. Markdown, plain text, chunks, metadata, outline and links are all included in the same price. Set Max PDFs in total to cap a run; your Apify maximum cost per run is always respected.

Use cases

• AI, RAG and LLM pipelines: feed PDFs to ChatGPT, Claude, LangChain, LlamaIndex or a vector database as clean Markdown chunks with page numbers for citations.
• Document search and archiving: index reports, manuals, papers and policies; keep the SHA-256 to detect changed files.
• Financial and ESG research: pull the text of annual reports, filings and investor presentations straight from investor-relations pages.
• Legal, tender and government documents: extract forms, notices, regulations and tender documents in bulk from public portals.
• Academic research: get the text, outline and references links of arXiv and journal PDFs.
• Monitoring: schedule a daily run on a page that publishes PDFs and get the new documents' text by webhook, Slack, Google Sheets, Make or Zapier.

Input examples

Two PDFs as Markdown, one row each:

{ "urls": ["https://www.irs.gov/pub/irs-pdf/fw9.pdf", "https://arxiv.org/pdf/1706.03762"] }

Every PDF linked on a web page:

{ "urls": ["https://www.irs.gov/forms-pubs/about-form-1040"], "maxPdfsPerPage": 25 }

RAG chunks of about 1,500 characters with 200 characters of overlap:

{ "urls": ["https://arxiv.org/pdf/1706.03762"], "outputMode": "chunks", "chunkSize": 1500, "chunkOverlap": 200 }

First 3 pages only, one row per page, Markdown and plain text:

{ "urls": ["https://www.irs.gov/pub/irs-pdf/p17.pdf"], "outputMode": "pages", "pages": "1-3", "format": "both" }

Output

Document row (text shortened):

{
  "url": "https://www.irs.gov/pub/irs-pdf/fw9.pdf", "status": "ok", "fileName": "fw9.pdf", "fileSizeBytes": 140815,
  "sha256": "2d420cbb4123dcf1fb82595b2359cfbb5d81f00b9df9d359fcc7af361d093f53",
  "numPages": 6, "pagesExtracted": 6, "title": "Form W-9 (Rev. March 2024)",
  "subject": "Request for Taxpayer Identification Number and Certification", "author": "SE:W:CAR:MP",
  "creator": "Designer 6.5", "producer": "Designer 6.5", "createdAt": "2024-03-06T13:18:13.000Z", "modifiedAt": "2024-03-06T13:18:13.000Z",
  "pdfVersion": "1.7", "language": "en-US", "hasForm": true, "isTagged": false,
  "wordCount": 6282, "charCount": 37924, "noTextLayerPages": [], "likelyScanned": false,
  "outline": null, "links": [],
  "markdown": "# Form W-9\n\n(Rev. March 2024)\n\n## Request for Taxpayer Identification Number and Certification\n\n...",
  "extractedAt": "2026-11-03T05:19:49.029Z"
}

Page rows add page, wordCount, noTextLayer and that page's links. Chunk rows have chunkIndex, chunkCount, pageStart, pageEnd, text and charCount.

URLs that fail (404, not a PDF, too large, wrong password) get a free row with status: "failed" and an error, are listed in the FAILURES record of the key-value store, and the rest of the run continues. SUMMARY holds the run totals. Very large documents (over about 4 million characters) are saved to the key-value store and the row links them as markdownUrl / textUrl, because dataset rows are limited to 9 MB.

The dataset has three views: PDFs (overview table), Text and Metadata.

How it works and limits

• Text comes from the PDF's own text layer. Scanned PDFs (images of text) have no text layer: those pages are listed in noTextLayerPages and are not OCRed. Most PDFs made by software (reports, forms, papers, invoices, manuals) have a text layer.
• Markdown structure is rebuilt from layout (font sizes, line spacing, bullets). Tables come out as text lines, and multi-column layouts follow the PDF's reading order.
• Downloads use a clear user agent, three PDFs at a time, and retries with back-off. Only use it on documents you are allowed to access.
• Default memory is 512 MB, enough for PDFs up to about 100 MB. Raise memory for very large files.

FAQ

Do I need an API key or account anywhere? No. Only public URLs are fetched.

How fast is it? Small PDFs take well under a second each; a 140-page publication takes a few seconds at 512 MB.

Can I extract password-protected PDFs? Yes, if you know the password: set PDF password.

Can I use it from AI agents? Yes, through the Apify API, the Apify MCP server, or the Python and JavaScript clients:

from apify_client import ApifyClient
client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("best_apis/pdf-text-extractor").call(run_input={"urls": ["https://arxiv.org/pdf/1706.03762"], "outputMode": "chunks"})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item["pageStart"], item["text"][:80])

curl -X POST "https://api.apify.com/v2/acts/best_apis~pdf-text-extractor/run-sync-get-dataset-items?token=<YOUR_APIFY_TOKEN>" \
  -H "Content-Type: application/json" -d '{"urls":["https://www.irs.gov/pub/irs-pdf/fw9.pdf"]}'

Is my data stored? Results stay in your own Apify storage. The Actor keeps nothing.

Related Actors

• Sitemap URL Extractor: find every PDF URL a website lists in its sitemaps.
• Broken Link Checker: find dead links, including links to PDFs.
• SEC EDGAR Scraper: company filings and financial data.
• Federal Register API: US rules and notices, with their PDF links.