PDF to Markdown & Text with OCR

Give it PDF links. Get clean Markdown and plain text, ready for RAG, LLMs, search or spreadsheets. Scanned pages are read with OCR, including Chinese (Traditional and Simplified), Japanese and Korean.

中文、日本語、한국어

• 繁體中文 OCR： 給 PDF 網址，轉成 Markdown 和純文字，掃描檔用 OCR 讀取，支援繁體中文、简體中文、日文、韓文與英文。每頁 US$0.002，OCR 頁 US$0.006。
• 简体中文 PDF： 提供 PDF 链接，转换为 Markdown 和纯文本，扫描件使用 OCR 识别。每页 US$0.002，OCR 页 US$0.006。
• 日本語 PDF： PDF の URL から Markdown とテキストに変換します。スキャン PDF は OCR で読み取ります。1 ページ US$0.002、OCR ページ US$0.006。
• 한국어 PDF: PDF URL을 Markdown과 텍스트로 변환합니다. 스캔 PDF는 OCR로 읽습니다. 페이지당 US$0.002, OCR 페이지 US$0.006.

What it does

• Keeps headings, paragraphs and ruled tables in Markdown.
• Uses OCR only on pages with no text layer, so normal pages stay fast and cheap. You can force OCR on or off.
• Reads English, Traditional Chinese, Simplified Chinese, Japanese and Korean.
• Tells you which files failed and why. One bad file does not stop the others.

Use cases

• RAG and LLM pipelines: turn reports, manuals and contracts into Markdown before chunking and embedding.
• Scanned documents in Asian languages: read Traditional Chinese, Simplified Chinese, Japanese or Korean scans that normal PDF parsers return empty.
• Search indexes: get one row per page, so a hit can point to the page it came from.
• Data entry: pull ruled tables out of PDF reports without copying by hand.

How to use it

1. Add public PDF links to PDF URLs.
2. Pick OCR languages if your scans are not in English.
3. Run it. You get one row per PDF, or one row per page if you turn on One row per page.

Pricing

• US$0.002 per page with a text layer.
• US$0.006 per page read with OCR.
• A file that cannot be converted costs nothing.
• Each run also has a US$0.00005 start fee.

Output

Field	Meaning
url	The PDF link
totalPages, pagesConverted, ocrPages	Page counts
markdown	Markdown version
text	Plain text version
error	Why the file failed, otherwise empty

With One row per page, each row also has page and ocr. A very large PDF is split into one row per page automatically, because a dataset row has a size limit.

Limits

• Links must be public. Password-protected PDFs are reported as failed.
• 100 MB per file; 500 pages per file by default.
• Layout is detected from the page. Two-column papers can have lines mixed across columns, and tables without ruled lines come out as text.
• Images and charts are not described. OCR accuracy depends on scan quality.
• OCR output is plain text with no headings.

Run it from code

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")
run = client.actor("inkbridge/pdf-to-markdown-text-with-ocr").call(run_input={
    "pdfUrls": [{"url": "https://example.com/report.pdf"}],
    "onePerPage": True,
})
for row in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(row["url"], row.get("page"), row.get("markdown", "")[:200])

FAQ

Will a normal PDF be charged the OCR price? Not in the default mode. OCR runs only on pages with no text layer, and only those pages cost US$0.006.

Can I cap the cost? Yes. Limit the pages per PDF in the input, or set a maximum charge for the run in the run options.

What if one file fails? The others still run. The failed file gets a row with the reason in error, and it costs nothing.

Does it work with private links? No. Links must be reachable without a login.

Data

Files are processed only to produce your output, which stays in your own Apify storage. Convert only documents you have the right to use.

Questions or a file that fails? Open an issue on this Actor's Issues tab.