Can't find a way to use `ImageRefMode.EMBEDDED` in `generate_multimodal_pages`

[sunwoongc]

trafficstars

Question

I tried using the generate_multimodal_pages method from the official documentation example and attempted to apply it. I wanted to export content_md with ImageRefMode.EMBEDDED, but this appears to be a legacy version, and I can't find a way to use ImageRefMode.EMBEDDED.

def _process_page():
    page_ix = page_no - 1
    page = doc_result.pages[page_ix]

    page_cells = _process_page_cells(page=page)
    page_segments = _process_page_segments(doc_items=doc_items, page=page)
    content_md = doc.export_to_markdown(main_text_start=start_ix, main_text_stop=end_ix)
    # No page-tagging since we only process one page at a time
    content_dt = doc.export_to_document_tokens(
        main_text_start=start_ix, main_text_stop=end_ix, add_page_index=False
    )

Do you have any suggestions for using this? It seems that the main issue is that the export_to_markdown method in DoclingDocument doesn’t support the parameters main_text_start and main_text_stop.

[dolfim-ibm]

Thanks for the report, we will have a look and first update the examples.

[dolfim-ibm]

I'm closing this issue in favour of a broader story in #835. New examples for multimodal exports and datasets creation are coming.