PDF Document Loader

​

Node Details

• Name: pdfFile
• Type: Document
• Category: Document Loaders
• Version: 1.0

​

Parameters

1. Pdf File (required)

   • Type: file
   • File Type: .pdf
   • Description: The PDF file(s) to be processed.

2. Text Splitter (optional)

   • Type: TextSplitter
   • Description: A text splitter to be applied to the extracted text.

3. Usage (required)

   • Type: options
   • Options:
     • One document per page (perPage)
     • One document per file (perFile)
   • Default: perPage
   • Description: Determines how the PDF content is split into documents.

4. Use Legacy Build (optional)

   • Type: boolean
   • Description: Whether to use the legacy build of the PDF parsing library.

5. Additional Metadata (optional)

   • Type: json
   • Description: Additional metadata to be added to the extracted documents.

6. Omit Metadata Keys (optional)

   • Type: string
   • Description: Comma-separated list of metadata keys to omit from the final documents. Use * to omit all default metadata keys.

​

Input

The node accepts PDF files either as base64-encoded strings or as file references from storage.

​

Output

The node outputs an array of IDocument objects, each representing a page or a whole PDF file (depending on the Usage setting). Each document contains:

• Page content as text
• Metadata (which can be customized or filtered based on the input parameters)

​

Functionality

1. Loads PDF file(s) from the provided input (either base64-encoded or from file storage).
2. Processes each PDF file:
   • If Usage is set to ‘perFile’, it creates one document per PDF file.
   • If Usage is set to ‘perPage’, it creates one document per page of each PDF file.
3. Applies text splitting if a TextSplitter is provided.
4. Adds or modifies metadata based on the Additional Metadata input.
5. Filters metadata based on the Omit Metadata Keys input.

​

Use Cases

• Extracting text content from PDF files for further processing or analysis.
• Preparing PDF content for ingestion into vector databases or search engines.
• Splitting large PDF documents into smaller, more manageable chunks.
• Customizing metadata for PDF-sourced documents in a document processing pipeline.

​

Notes

• The node uses the PDFLoader from the langchain library for PDF parsing.
• It supports both modern and legacy builds of the PDF parsing library, which can be useful for compatibility with different types of PDF files.
• The node is flexible in handling both single and multiple PDF files in one operation.