Selective PDF Page Conversion in Docling Based on Textual Content Matching

[Omar-M-Abdelbary]

I'm currently working on a workflow where I only need to convert specific pages from a PDF document using Docling, based on whether they contain certain predefined headers.
To achieve this, I need a way to access the textual content of each page individually as a variable before conversion, so I can check if any of my target keywords are present. If a page contains one of the desired headers, it should be processed and converted; otherwise, it should be skipped.
I've explored DocumentConverter.convert() and noticed there's a page_range parameter that can limit the conversion to specific pages. However, my challenge is determining which pages to include in advance — based on the actual text content of each page — before performing the final conversion.

Could you please guide me on the recommended way to:

1. Extract or access the raw or structured text content from individual pages before full conversion.
   

2. Perform keyword matching on those pages.
   

3. Pass only the relevant page numbers to convert() based on that filtering.