Module document_processor 

Document Processing for Skills using Vision Models

Implements OpenAI-style document processing by converting PDFs, DOCX, and spreadsheets to rendered images for vision model analysis. This preserves layout, formatting, and visual information that would be lost in text extraction.

Supported Formats

• PDF: Multi-page documents converted to page-by-page PNGs
• DOCX/DOC: Word documents rendered per-page
• Spreadsheets: Excel/CSV files rendered as visual tables
• Images: Direct vision model processing

Architecture

Document → Renderer → PNG Images → Vision Model → Structured Data

Inspired by OpenAI’s implementation in ChatGPT’s Code Interpreter.

Structs

DocumentMetadata

Document metadata

DocumentProcessor

Main document processor

DocumentProcessorConfig

Document processing configuration

ImageDimensions

Image dimensions

PageImage

Single page image data

ProcessedDocument

Processed document result

Enums

DocumentType

Document type classification