[Bug] Vision-LLM conversion only processes the first page of scanned PDFs

[dandelie1998]

Description:

I am using markitdown to convert scanned PDF documents into Markdown with options markitdown-ocr.
However, I found that when processing a multi-page scanned PDF, the tool only extracts the content of the first page and ignores the remaining pages.

screenshots

code

result

Environment Information

• OS: Ubuntu 24.04.2 LTS

• python: Python 3.12.13

• markitdown info:

  • Name: markitdown
    Version: 0.1.5
    Summary: Utility tool for converting various files to Markdown
    Home-page:
    Author:
    Author-email: Adam Fourney adamfo@microsoft.com
    License-Expression: MIT
    Location: /root/miniconda3/envs/markitdown/lib/python3.12/site-packages
    Requires: beautifulsoup4, charset-normalizer, defusedxml, magika, markdownify, requests
    Required-by: markitdown-ocr

• llm_client: I use the api from bailian, aliyun

  • And I've tested several models, the results are the same.

• the tested file: 👇
  4-2南实党委〔 2024〕20号-关于印发《南湖实验室采购管理办法实施细则（试行）》的通知.pdf