Document Intelligence Loader not functional

Question

Document Intelligence Loader not functional

Billy Zhou 30

I encountered an issue with the AzureAIDocumentIntelligenceLoader API when converting a PDF to Markdown format. The API marked all PDF headers (both primary and secondary) with a single "#" symbol, while I expected primary headers to have one "#" and secondary headers to have two "##".

Could you give me some pointers on using this API? below are some details:

Python code to call AzureAIDocumentIntelligenceLoader API : User's image

Screen shot of source pdf: User's image

Actual API Results (primary header start with one "#", but sub header also start with one "#") User's image

Expected Results(primary header start with one "#", sub headers start with two "#" and so on): User's image

Ramr-msft 17,836 Reputation points

2024-01-12T16:40:56.1566667+00:00

Thanks for the question, The API does not have built-in functionality to differentiate between primary and secondary headers in the way you’re expecting. One possible workaround could be to post-process the API’s output. You could write a script that traverses the Markdown document and adjusts the header levels based on your specific rules. For example, you could use a Python to parse the Markdown into a syntax tree, adjust the header levels, and then serialize it back into Markdown.

Your answer

Ramr-msft 17,836 Reputation points

2024-01-12T16:40:56.1566667+00:00

Thanks for the question, The API does not have built-in functionality to differentiate between primary and secondary headers in the way you’re expecting. One possible workaround could be to post-process the API’s output. You could write a script that traverses the Markdown document and adjusts the header levels based on your specific rules. For example, you could use a Python to parse the Markdown into a syntax tree, adjust the header levels, and then serialize it back into Markdown.

Share via

Document Intelligence Loader not functional

Your answer