Extract Text from PDF: PDF to Markdown & JSON Converter for AI & RAG

Extract text from PDF, DOCX, PPTX, XLSX, HTML, and images into clean Markdown or structured JSON. Headings, tables, and page structure are preserved. Scanned documents are handled with OCR. Includes a RAG-ready chunked output mode for feeding LLM pipelines and vector stores directly.

Powered by Docling, IBM's open-source document-conversion library (MIT licensed, 51,000+ GitHub stars, 2.4M+ monthly downloads).

Great for: extract text from PDF, PDF text extractor, PDF to Markdown, document parser, extract tables from PDF, RAG/LLM ingestion pipelines, AI knowledge bases, research paper processing, and converting legacy DOCX/PPTX archives to Markdown.

What it does

Give it a list of document URLs, or upload a file directly: it extracts the text from each one and returns clean structured output. No files are stored beyond the run.

Output formats

outputFormat	What you get
markdown	Clean Markdown text
json	Full structured document (headings, tables, layout, bounding boxes)
both (default)	Markdown + JSON
chunks	RAG-ready semantic chunks, each with its heading context: ready to embed and index

Example input

{
  "documents": [
    "https://example.com/report.pdf",
    "https://example.com/slides.pptx"
  ],
  "outputFormat": "both",
  "ocrEnabled": false
}

Output

One row per document:

Field	Description
documentUrl	The document converted
format	Detected format (pdf, docx, pptx, xlsx, html, image)
pageCount	Pages in the document
markdown	Markdown output (if requested)
json	Structured JSON output (if requested)
chunks / chunkCount	RAG chunks (if outputFormat: "chunks")
status	ok or error

OCR

Turn on ocrEnabled for scanned PDFs and images with no embedded text layer. It's slower and billed separately. Leave it off for normal digital PDFs/DOCX/PPTX, which convert faster without it.

Pricing

Billed per page successfully converted (fair for a 2-page doc vs. a 200-page one) plus a per-document OCR surcharge when OCR is used. Failed documents are not charged.

FAQ

How do I extract text from a PDF? Paste the PDF's URL in documents (or upload the file) and run. The markdown field in the output holds the extracted text, ready to use as-is.

How do I convert a PDF to Markdown? Same run: Markdown is the default output format. The markdown field is clean Markdown with headings and tables preserved.

Can it extract tables from a PDF? Yes. Tables are detected and preserved as Markdown tables in the markdown output, and as structured data in the json output.

Does this work for scanned PDFs? Yes. Turn on ocrEnabled. It's slower than normal text extraction, so it's billed at a separate rate; leave it off for regular digital PDFs.

What is the "chunks" output for? It splits the document into semantically coherent pieces (respecting headings and structure) sized for embedding into a vector database: the standard input shape for RAG pipelines.

How is this different from just using Docling myself? Docling requires installing Python + ML dependencies and managing compute. This runs it as a hosted API: no setup, pay only for what you convert.

Please note

Only convert documents you have the right to process. Documents are processed transiently and not retained beyond the run.

Built with Apify Python SDK + Docling.