Text extraction

​

Overview

​

Description

• Read and extract text from PDF documents.
• Capture and process the textual content of a webpage.
• Identify and read text from images (OCR-based).
• Convert speech-to-text from audio files.

​

Input Parameters

• extractionType (string, required) Defines the type of extraction to perform. Supported values:
  • "pdf-extraction" – Extracts text from PDF documents.
  • "web-extraction" – Extracts text and metadata from a webpage.
  • "image-extraction" – Extracts text from images using OCR (Optical Character Recognition).
  • "audio-extraction" – Converts spoken content from audio files into text.
• webUrl (string, required for web-extraction) The URL of the web page to extract content from. Example:
  Copy

  https://example.com/article
  

• attachments (string, required for pdf/image/audio extraction) Comma-separated list of file IDs or variable references to uploaded files. Each file represents one source of extraction. Example:
  Copy

  file1.pdf,file2.pdf
  

  or
  Copy

  {{nodeId.output.file1}},{{nodeId.output.file2}}
  

{{nodeId.input.<key>}}

​

Output Parameters

• processingCount Number of files, pages, or segments processed during extraction.
• processingTime Total time taken for extraction, returned in ISO timestamp format.
• processingId A unique identifier assigned to this specific extraction request.
• content The main text extracted from the provided source. For audio extraction, this will contain the full transcription.
• markdown A markdown-formatted version of the extracted text for easy readability and formatting.
• linksOnPage[] A list of links found on the page (only applicable for web-extraction).
• metadata.title The page or document title.
• metadata.keywords Extracted keywords or tags found in metadata.
• metadata.description Short description or summary found in the document or webpage metadata.
• metadata.robots Robot meta instructions (e.g., index,follow).
• metadata.ogTitle Open Graph title of the page.
• metadata.ogDescription Open Graph description of the page.
• metadata.ogImage URL of the Open Graph image associated with the webpage.
• metadata.ogSiteName The name of the website where content was extracted.
• metadata.screenshot Screenshot image reference captured during extraction (for web-extraction).
• speaker_labels[].speaker For audio extractions — identifies the speaker label (e.g., Speaker 1, Speaker 2).
• speaker_labels[].text Text spoken by each identified speaker segment.
• timestamps[].startTime Timestamp indicating when a specific word or phrase began in the audio file.
• timestamps[].endTime Timestamp indicating when a specific word or phrase ended.
• timestamps[].word Word or phrase detected during speech-to-text transcription.
• timestamps[].duration (string) Duration of the spoken segment.

{{nodeId.output.content}}                 → Extracted text
{{nodeId.output.metadata.title}}          → Page or document title
{{nodeId.output.linksOnPage[0]}}          → First link on the page
{{nodeId.output.speaker_labels[0].text}}  → First speaker’s text segment

​

Output Type

"text"

​

Example Usage

​

Example 1: Extract Text from a PDF

{
  "extractionType": "pdf-extraction",
  "attachments": "file123.pdf"
}

{
  "content": "This is the text extracted from the PDF document.",
  "processingId": "proc_67892",
  "processingTime": "2025-10-27T12:05:43Z",
  "processingCount": 1
}

​

Example 2: Extract Text from Image

{
  "extractionType": "image-extraction",
  "attachments": "image123.png"
}

{
  "content": "Here is the article content extracted from the page...",
  "processingTime": "2025-10-27T12:06:11Z",
  "processingId": "mistral-ocr-1761738927697",
  "processingCount": 41
}

​

Example 3: Extract Text from Audio

{
  "extractionType": "audio-extraction",
  "attachments": "nature_audio.mp3"
}

{
  "content": "The evening sky was painted with shades of orange and violet. As the city slowly settled into its nightly rhythm. Streetlights flickered to life, casting warm pools of light on the sidewalks where people hurried home, their faces half hidden behind scarves and tired smiles. Somewhere, a musician played a soft tune on a saxophone, its notes drifting between buildings like fragments of a forgotten dream. In the distance, the sound of traffic blended with laughter spilling out of a nearby cafe. Inside, a, uh, writer sat alone with a cup of coffee gone cold, his laptop open but untouched. He watched the rain begin to fall against the window, tracing patterns that matched the rhythm of his thoughts. Every drop felt like a reminder that even in a crowded city, moments of stillness could be found in the smallest details. A quiet song, a stranger's laughter, or the glow of street lights reflecting on wet pavement. And in that quiet, he finally began to type M. Not to impress anyone or chase perfection, but simply to capture the fleeting beauty of the world as he saw it.",
  "processingTime": "2025-10-27T12:05:43Z",
  "processingId": "dc9509af-3783-48bd-8819-85126a8e66a8",
  "processingCount": 188
}

​

How to Use in a No-Code Workflow

1. Add the Text Extraction Node Drag and drop the node into your workflow.
2. Select Extraction Type Choose the appropriate extraction type:
   • "pdf-extraction" for PDF files
   • "web-extraction" for web pages
   • "image-extraction" for scanned images or screenshots
   • "audio-extraction" for speech-to-text conversion
3. Provide Input
   • For PDFs, images, or audio: add file references under attachments.
   • For webpages: enter the target URL in webUrl.
4. Run the Node The node will extract text, metadata, or transcription automatically.
5. Access Results Use variable syntax to retrieve outputs such as:
   Copy

   {{textExtraction.output.content}}
   {{textExtraction.output.metadata.title}}
   

6. Connect to Next Nodes You can pass the extracted text to other nodes like Text Generation, AI Analysis, or Summarization for further automation.

​

Best Practices

• Ensure the correct extraction type is selected before execution.
• When extracting from multiple files, separate file IDs with commas.
• For web extractions, make sure the URL is publicly accessible.
• Use OCR-friendly images (clear text visibility) for best results.
• For large audio files, split them into smaller segments to improve performance.
• Always review extracted content for accuracy before using it in production workflows.

​

Example Workflow Integration

• Step 1: The File Upload Node provides the PDF file reference.
• Step 2: The Text Extraction Node extracts all text from the PDF.
• Step 3: The Text Generation Node summarizes the extracted content into key points.
• Step 4: The Email Node sends the summary to a user.

{{fileUpload.output.fileId}}        →  {{textExtraction.input.attachments}}
{{textExtraction.output.content}}   →  {{textGeneration.input.prompt}}

​

Common Errors

• “Missing extractionType” Cause: No extraction type was provided. Solution: Always specify one of "pdf-extraction", "web-extraction", "image-extraction", or "audio-extraction".
• “Missing webUrl” Cause: The webUrl field is required for web-extraction. Solution: Add a valid, publicly accessible URL.
• “Missing attachments” Cause: No file references provided for file-based extraction. Solution: Add valid file IDs or variable references in the attachments field.
• “Unsupported file format” Cause: The file type is not supported for extraction. Solution: Use PDF, image, or audio files in standard formats.
• “Empty output” Cause: The AI was unable to extract content from the provided source. Solution: Recheck file quality, accessibility, or try a different extraction type.