PDF to Text & Markdown Extractor with OCR (Tables, Bulk)

Give it PDF URLs (one, thousands, or a dataset/CSV column) and get back clean text, LLM-ready Markdown, tables, metadata and the text of every page. Scanned PDFs are OCR'd automatically with Tesseract running inside the Actor: no API key, no cloud OCR service, your documents are not sent anywhere else.

What does PDF to Text Extractor do?

PDF to Text Extractor converts PDF URLs in bulk into plain text, LLM-ready Markdown, tables, metadata and per-page content. Scanned pages are OCR'd automatically with Tesseract inside the Actor (no API key, no cloud OCR), which makes it useful for RAG pipelines and document data extraction.

How to use PDF to Text Extractor

1. Click Try for free on this page (a free Apify account is enough to start).
2. Fill in the input (see the examples below) or keep the prefilled example.
3. Click Start and wait for the run to finish.
4. Download the results from the Output tab as JSON, CSV, Excel or HTML, or read them with the Apify API.
5. Optional: create a Schedule to run it daily or weekly, and connect a webhook or an integration (Make, Zapier, n8n, Slack).

Who uses it

• AI / RAG teams: turn reports, manuals and papers into Markdown chunks with page markers for citations.
• Legal, procurement and compliance: pull the text of contracts, tenders and filings to search or classify them.
• Data teams: extract tables from statistical reports and price lists.
• Automation builders: chain it after a scraper that finds PDF links (dataset input with urlField).

Features

• Text per page from the PDF text layer (fast, exact), joined into one text field.
• Markdown for LLMs: headings detected by font size, paragraphs re-flowed, bullet lists, tables as Markdown tables, <!-- page N --> markers.
• Tables (ruled tables) as rows of cells.
• OCR: auto OCRs only pages without a text layer (scans, photos), always OCRs every page, never turns it off. 12 languages (English, Spanish, Portuguese, French, German, Italian, Dutch, Polish, Russian, Turkish, Chinese simplified, Japanese).
• Metadata: title, author, subject, keywords, creator, producer, creation and modification dates.
• Page ranges (1-5, 8, 10-), max pages per PDF, password-protected PDFs (your password), duplicates skipped.
• Respects robots.txt on every redirect hop. Retries on timeouts and 429/5xx.
• Very large documents: the full text and Markdown go to downloadable files when they do not fit in one dataset item.

Input example

{
  "urls": ["https://www.irs.gov/pub/irs-pdf/fw9.pdf", "https://example.com/scanned-invoice.pdf"],
  "ocrMode": "auto",
  "ocrLanguages": ["eng", "spa"],
  "markdown": true,
  "extractTables": true,
  "pageRange": ""
}

Output

{
  "status": "ok",
  "url": "https://www.irs.gov/pub/irs-pdf/fw9.pdf",
  "fileSizeBytes": 140815,
  "pageCount": 6,
  "pagesExtracted": 6,
  "truncated": false,
  "ocrPages": 0,
  "scanned": false,
  "metaTitle": "Form W-9 (Rev. March 2024)",
  "metaAuthor": "SE:W:CAR:MP",
  "metaCreatedAt": "2024-03-04T10:19:41",
  "wordCount": 6210,
  "charCount": 37687,
  "text": "Form W-9 (Rev. March 2024) Request for Taxpayer Identification Number...",
  "markdown": "<!-- page 1 -->\n# W-9 Request for Taxpayer\n...",
  "tables": [{ "page": 1, "rows": [["Social security number", null, "..."]] }],
  "pages": [{ "page": 1, "text": "...", "chars": 6512, "ocr": false, "tables": 2 }],
  "billedPdfEvents": 1
}

Failed items (not a PDF, HTTP errors, password needed, blocked by robots.txt, no text at all) come with status failed or no_text and an error, and are not charged.

Pricing

Pay per event:

• USD 0.003 per PDF (pdf). Each started block of 50 pages counts as one PDF (a 120-page report = 3 events = USD 0.009).
• USD 0.008 per OCR page (ocr-page), only for pages that really were OCR'd.
• Start: USD 0.00005.

A typical 10-page digital PDF costs USD 0.003; a 10-page scan costs USD 0.083. Failed and empty PDFs are free, and the Actor never processes pages your max charge per run cannot pay for.

Notes and limits

• Only public URLs (or PDFs your password opens). No logins.
• OCR quality depends on the scan; 300 DPI is the default. Handwriting is not supported.
• Table detection works best on tables with ruling lines; tables drawn only with spacing come out as text.
• Headings are guessed from font size; forms and multi-column layouts may need a light clean-up.
• OCR speed is about 2-6 s per page at the default 2 GB memory (more memory = more CPU = faster).

Why use PDF to Text Extractor

• OCR included, no extra account: scanned pages are read with Tesseract inside the Actor. No OpenAI, Google or AWS key, and your documents are not sent to a third-party OCR service.
• You only pay OCR where it is needed: in auto mode only pages without a text layer are OCR'd; digital pages are read directly and cost nothing extra.
• LLM-ready Markdown: headings, lists, tables and <!-- page N --> markers, ready for RAG chunking with page citations.
• Bulk input: a list of URLs, a dataset from another Actor (for example a scraper that collected PDF links) or a CSV / Google Sheets column.
• Failed files are free: non-PDFs, HTTP errors, password-protected files without the password and empty PDFs are listed but never charged.

Cost examples

Prices at the current pay-per-event rates (see Pricing). You can set a maximum charge per run in the run options.

Job	Approximate cost
100 digital PDFs of up to 50 pages	about $0.30
1,000 digital PDFs of up to 50 pages	about $3.00
100 scanned pages (OCR)	about $0.80 + $0.003 per PDF

Use it from your code, an integration or an AI agent

Python (pip install apify-client):

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("latamdatatools/pdf-text-ocr-extractor").call(run_input={
  "urls": [
    "https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf",
    "https://www.irs.gov/pub/irs-pdf/fw9.pdf"
  ]
})
for item in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(item)

JavaScript / Node.js (npm install apify-client):

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: '<YOUR_APIFY_TOKEN>' });
const run = await client.actor('latamdatatools/pdf-text-ocr-extractor').call({
  "urls": [
    "https://www.w3.org/WAI/ER/tests/xhtml/testfiles/resources/pdf/dummy.pdf",
    "https://www.irs.gov/pub/irs-pdf/fw9.pdf"
  ]
});
const { items } = await client.dataset(run.defaultDatasetId).listItems();
console.log(items);

No code:

• Schedules: run it daily or weekly from Apify Console (Schedules), with the same input saved as a Task.
• Integrations: send results to Google Sheets, Slack, email, Make, Zapier, n8n or Airbyte from the Integrations tab, or call your own webhook when a run finishes.
• AI agents: add it as a tool in Claude, Cursor, VS Code and other MCP clients through the Apify MCP server (https://mcp.apify.com), so the agent can run PDF to Text Extractor and read the results.

FAQ

Can it read PDFs I have on my computer? Upload them to any public link (a storage bucket, an Apify key-value store or a file host) and paste the URLs. Private, logged-in links are not supported.

Which OCR languages can I combine? Any of the 12 bundled languages, for example ["eng", "spa"] for bilingual documents.

Can I call it from my own code or from an AI agent? Yes. Start runs and fetch results with the Apify API or the JavaScript / Python clients, or use it as a tool from Claude, Cursor and other MCP clients through the Apify MCP server (mcp.apify.com).

Can I run it automatically? Yes. Create a Schedule in Apify Console (for example every morning) and add a webhook or an integration to get the results where you need them.

Which export formats are available? JSON, JSONL, CSV, Excel (XLSX), XML, RSS and HTML table, from the Output tab or the API.

What does it cost? See the Pricing section above. You only pay for the events listed there.

Changelog

• 0.1 (2026-10): first version.