Skip to main content

Document Text Extractor

The Document Text Extractor node extracts embedded text from PDFs and performs local OCR on scanned PDF pages and image files. It uses pdf-parse for PDF text extraction and page rendering, and Tesseract.js for OCR.

Quick Start​

  1. Provide a PDF or image in an n8n binary field, normally data.
  2. Add the Document Text Extractor node.
  3. Select Automatic for mixed digital and scanned PDFs.
  4. Read the result from documentText (or the configured destination field).

Images always use OCR. In Automatic mode, each PDF page with fewer than the configured minimum non-whitespace characters is rendered and OCRed; pages with usable embedded text are not OCRed.

Supported Inputs​

  • PDF, including password-protected PDFs
  • Image MIME types supported by Tesseract.js, including common PNG, JPEG, BMP, and WebP inputs
  • Maximum input size: 25 MB
  • Maximum configurable PDF page count: 100

Numeric limits and mode/layout selections are validated at runtime, before binary retrieval. Imported workflows and expression results must satisfy the same limits as the editor.

OCR language data may be downloaded by Tesseract.js when a language is first used, so workers require outbound access unless language data is already cached.

Output​

{
  "documentText": {
    "text": "Combined document text",
    "method": "mixed",
    "pages": [
      {
        "pageNumber": 1,
        "text": "Embedded text",
        "method": "pdfText",
        "confidence": null
      },
      {
        "pageNumber": 2,
        "text": "OCR text",
        "method": "ocr",
        "confidence": 93
      }
    ],
    "pageCount": 2,
    "processedPageCount": 2,
    "truncated": false,
    "textTruncated": false,
    "sourceFileName": "document.pdf",
    "sourceMimeType": "application/pdf"
  }
}

Existing input JSON fields are preserved outside the configurable destination field. The source binary is omitted by default and can be retained with Keep Input Binary.

More Information​