PDF to Text and Markdown API

Public PDF URL in. Text and Markdown out.

HTTP on RapidAPI. Bulk on Apify.

About

Point the API at one or two public PDF URLs and it returns the extracted text layer as plain text and as Markdown (headings and lists are inferred from font sizes). Each result includes metadata such as title, author, dates, page counts, character and word counts, a SHA-256 of the file, and a needsOcr flag when the file is an image-only scan. Private or local addresses are blocked. There is no OCR: scanned PDFs are detected, marked, and — on Apify — returned free of charge so you can send them to a dedicated OCR tool.

What you get

  • Text and Markdown from the PDF text layer
  • Page count, character count, and whether the sample text is truncated

Use

  • Turn public filings, papers, and reports into text or Markdown ready for a vector store.
  • Chunk by page: request include_pages and cite page-level text in an agent answer.
  • Skip scanned or empty files automatically — they come back as no_text_layer with needsOcr=true, unbilled on Apify.

Limits

  • 2 public PDF URLs per call
  • 15 MB and 50 pages per file
  • 25 second timeout

Try it without a key

No key. 5 calls per address per day, and $0.50 for everyone per UTC day. One row per call. A key is on RapidAPI and Apify.

curl
curl -X POST "https://api.impakt.dev/pdf/extract" -H "content-type: application/json" -d '{"urls":["https://arxiv.org/pdf/1706.03762"],"max_pages":1}'

Endpoints

OpenAPI

POST /pdf/extract

Extract text and Markdown from PDF URLs

Up to 2 public PDF URLs per marketplace call (files up to 15 MB, up to 50 pages each). Private or local addresses are blocked.

Inputs

  • urls string[] body required

    Direct public HTTP or HTTPS links to PDF files.

  • format string body default both

    text, markdown, or both.

  • max_pages integer body default 30

    Only the first N pages are extracted (the result is marked truncated if more pages exist).

  • include_pages boolean body default false

    Add a pages array with the text of each page, useful for citations and chunking.

Returns

  • status / error string

    ok, no_text_layer, download_failed, or similar, with an error message when not ok.

  • url string

    The public PDF URL that was extracted.

  • text / markdown string

    Extracted plain text and Markdown, depending on the requested format.

  • title / author / creator / producer / pdfVersion string

    PDF metadata when the file provides it, including the creator, producer, and PDF version. subject, createdAt, and modifiedAt are included when present.

  • pageCount / pagesExtracted / truncated integer / boolean

    How many pages the file has, how many were extracted, and whether the cap cut the file short. Per-page text is returned only when include_pages is true.

  • charCount / wordCount / needsOcr integer / boolean

    Size of the extracted text, and whether the PDF appears to be a scan with no text layer. A SHA-256, file size, and processing time are included on each result.

request
POST https://pdf-to-text-markdown-api.p.rapidapi.com/pdf/extract
X-RapidAPI-Key: YOUR_RAPIDAPI_KEY
X-RapidAPI-Host: pdf-to-text-markdown-api.p.rapidapi.com
Content-Type: application/json

{
  "urls": [
    "https://arxiv.org/pdf/1706.03762"
  ],
  "format": "both",
  "max_pages": 1,
  "include_pages": false
}
200 response
{
  "count": 1,
  "limit": 1,
  "results": [
    {
      "url": "https://arxiv.org/pdf/1706.03762",
      "status": "ok",
      "error": null,
      "title": null,
      "author": null,
      "creator": "LaTeX with hyperref",
      "producer": "pdfTeX-1.40.25",
      "pdfVersion": "1.5",
      "pageCount": 15,
      "pagesExtracted": 1,
      "truncated": true,
      "charCount": 173,
      "wordCount": 26,
      "needsOcr": false,
      "text": "Provided proper attribution is provided, Google hereby grants permission to reproduce the tables and figures in this paper solely for use in journalistic or scholarly works.",
      "markdown": "Provided proper attribution is provided, Google hereby grants permission to reproduce the tables and figures in this paper solely for use in journalistic or scholarly works."
    }
  ],
  "request_id": "00000000-0000-4000-8000-000000000000"
}

Pricing

Marketplace Price Link
RapidAPI BASIC $0/month (5 requests, hard cap, no overage). PRO $9.99/month for 600 requests. ULTRA $29.99/month for 2,000 requests. MEGA $79.99/month for 6,000 requests. Overage $0.015 per request. RapidAPI Opens in a new tab.
Apify PDF to text and Markdown: $3 per 1,000 PDFs. A $0.0001 start fee applies. Failed downloads and image-only PDFs are not billed. Apify Opens in a new tab.

FAQ

Does this API perform OCR?

No. It extracts the embedded text layer. Scanned PDFs come back with status no_text_layer and needsOcr=true, and they are not billed on Apify. Use a separate OCR tool for those files.

Can it read Google Drive, Dropbox, or S3 files?

Yes, if you pass a direct public download link (for Drive, a uc?export=download link). Private links that need a login cannot be fetched.

How good is the Markdown?

Headings and lists are inferred from font sizes, so unusual layouts can differ from the original. Tables are returned as text, not as Markdown tables. Multi-column pages are read in vertical order and may interleave columns.

What about password-protected or very large PDFs?

Password-protected files are not opened. Files over the size cap are skipped. Only the first max_pages pages are extracted, and the result is marked truncated when more pages remain.

Who is responsible for copyright and privacy?

You are. The API only downloads the URLs you provide. You must have the right to process those documents and must follow copyright, privacy, and data-protection laws that apply to them.