OCR PDF Extractor: PDF to Text, Image to Text & PDF to Markdown (Hebrew, Arabic + 11 languages)

Turn PDFs and images into clean Markdown or plain text, with tables. Digital PDF pages are read straight from the text layer (fast, exact, cheap). Only scanned pages go through OCR (Tesseract 5). It handles right-to-left Hebrew and Arabic properly, including PDFs whose Hebrew text layer is stored backwards.

What you can use it for

• RAG and LLM pipelines: turn PDFs into clean Markdown with tables, ready for chunking and embeddings.
• Document processing: extract text from scanned contracts, invoices and letters (image to text OCR) into JSON.
• Hebrew and Arabic documents: get readable right-to-left text, including PDFs whose Hebrew text layer is stored backwards.
• Data entry automation: pull tables out of digital PDF reports as rows you can load into a spreadsheet.
• Archives: batch-convert hundreds of PDFs or TIFF scans into searchable text.

How to use it

1. Add PDF or image URLs under File URLs (or upload files to an Apify key-value store and paste the record URLs).
2. Pick the OCR languages, e.g. eng + heb.
3. Keep mode on auto: the text layer is used where it exists and OCR only where it doesn't.
4. Click Start and download Markdown, text or JSON.

Why this one

• Hybrid extraction: text layer first, OCR fallback page by page. You pay the OCR price only for scanned pages.
• RTL done right: Hebrew and Arabic come out in readable (logical) order, and reversed "visual order" Hebrew text layers are detected and fixed.
• Tables: tables in digital PDFs are returned as rows and as Markdown tables, ready for LLM/RAG pipelines.
• 13 OCR languages: English, Hebrew, Arabic, Russian, German, French, Spanish, Italian, Portuguese, Dutch, Polish, Turkish, Ukrainian. Combine them, e.g. ["heb","eng"].
• Formats: PDF, PNG, JPEG, TIFF (multi-page), GIF, BMP, WebP. Up to 50 MB per file.

Input

{
  "fileUrls": [{"url": "https://example.com/contract.pdf"}],
  "languages": ["eng", "heb"], "mode": "auto", "outputFormat": "markdown", "extractTables": true
}

• mode: auto (recommended), ocr (force OCR on every page) or text (never OCR).
• maxPagesPerFile (default 200), ocrDpi (default 300), includePageDetails (default true).
• Have local files? Upload them to an Apify key-value store and pass the record URLs.

Output (one item per file)

{
  "url": "https://example.com/contract.pdf", "pageCount": 12, "ocrPages": 3, "textLayerPages": 9, "tableCount": 2,
  "content": "<!-- page 1 (text) -->\n\nQuarterly Report 2026 ...\n\n| Region | Q2 | Q3 |\n|---|---|---|\n| North | 120 | 150 |",
  "pages": [{"page": 1, "method": "text", "text": "...", "tables": [[["Region","Q2","Q3"],["North","120","150"]]], "ocrConfidence": null}]
}

Good to know

• Processing happens inside the run. Files are not kept beyond the run's default storage retention.
• OCR accuracy depends on scan quality. ocrConfidence (0–100) per page helps you spot pages to re-check.
• Handwriting recognition is not supported.

Pricing

Pay per event, no monthly rent:

• $0.001 per text-layer page (digital PDFs)
• $0.006 per OCR page (scans and images)
• $0.002 per run start

Example	Cost
A 10-page digital PDF	$0.012
A 10-page scanned PDF	$0.062
1,000 digital pages	$1.00
1,000 scanned pages	$6.00

Failed downloads and unsupported files are not charged.

You can set a maximum cost per run in the run options; the Actor stops cleanly when it is reached. New to Apify? The free plan includes $5 of platform credit every month, so you can try this Actor at no cost.

API, integrations and AI agents (MCP)

Run it from your own code with the Apify API and get the results in one call:

curl -X POST "https://api.apify.com/v2/acts/manageable_gully~pdf-ocr-extractor/run-sync-get-dataset-items?token=<YOUR_APIFY_TOKEN>" \
  -H "Content-Type: application/json" -d @input.json

from apify_client import ApifyClient
run = ApifyClient("<YOUR_APIFY_TOKEN>").actor("manageable_gully/pdf-ocr-extractor").call(run_input={...})

• No-code: use the Apify app in n8n, Make or Zapier ("Run Actor", then "Get dataset items"), or send results to Google Sheets, Slack or a webhook.
• AI agents (MCP): add https://mcp.apify.com?tools=manageable_gully/pdf-ocr-extractor to Claude, Cursor, VS Code or any MCP client and the agent can call this Actor as a tool.
• Schedules: run it hourly, daily or weekly from Apify Console.

FAQ

Does it read handwriting? No. Tesseract OCR handles printed text. Handwriting is not supported.

Can I convert a PDF to Markdown for ChatGPT or Claude? Yes. Set outputFormat to markdown. Headings, paragraphs and tables are preserved, and each page is marked so you can cite page numbers.

What image formats are supported? PNG, JPEG, TIFF (including multi-page), GIF, BMP and WebP, plus PDF. Up to 50 MB per file.

Can AI agents use it? Yes. It is available as a tool through the Apify MCP server (https://mcp.apify.com?tools=manageable_gully/pdf-ocr-extractor) and the Apify API, and it returns structured JSON that agents can read directly.

More tools from WebIntel Tools

• Web Page to Markdown: clean Markdown and RAG chunks from any URL, for LLMs
• Website Screenshot & URL to PDF: full-page screenshots and PDFs of URLs or your own HTML
• RSS Feed Reader & Monitor: read and monitor RSS, Atom and JSON feeds, only new items

See all WebIntel Tools Actors.

Made by WebIntel Tools. This Actor was built and is maintained with AI assistance. Report issues on the Issues tab; they are monitored daily.