Session.request() got an unexpected keyword argument 'output_content_format'

[bb-at-ss]

• Package Name: azure-ai-documentintelligence python sdk
• Package Version: 1.0.0
• Operating System: Windows 11
• Python Version: 3.12

Describe the bug
I've been using Azure Doc Intel in production for a couple of months and now we wanted to add some features, ideally extracting text directly into markdown now. However, when trying to read in the pdf as bytes and use then use the Azure Doc Intel client to read the pdf, extract the text, and return it as formatted markdown we get the following error: Session.request() got an unexpected keyword argument 'output_content_format .

I've searched the code base high-and-low, tried the enum, tried several variations of how to input the pdf file bytes, etc., but the error persists. Commenting it out and running it on our test set of pdfs that have previously been through the pipeline and it works as expected.

To Reproduce

client = DocumentAnalysisClient(
            endpoint=os.environ["DOCUMENTINTELLIGENCE_ENDPOINT"], 
            credential=AzureKeyCredential(os.environ["DOCUMENTINTELLIGENCE_API_KEY"])
        )
with open(pdf_path, "rb") as f:
      file_bytes = f.read()

poller = client.begin_analyze_document(
                    document=AnalyzeDocumentRequest(bytes_source=file_bytes),
                    model_id="prebuilt-layout",
                    output_content_format='markdown'
          )


**Expected behavior**
Extract the text from the pdf in a markdown format. 

[pvaneck]

Hey, @bb-at-ss. Can you confirm that you are actually using version 1.0.0 of azure-ai-documentintelligence? I see in your code snippet, that you are using DocumentAnalysisClient, which indicates that this is likely azure-ai-formrecognizer (which azure-ai-doumentintelligence supersedes). I think only azure-ai-documentintelligences supports the output_content_format keyword argument.

For azure-ai-documentintelligence, check out the migration guide and a sample here.

[github-actions]

Hi @bb-at-ss. Thank you for opening this issue and giving us the opportunity to assist. To help our team better understand your issue and the details of your scenario please provide a response to the question asked above or the information requested above. This will help us more accurately address your issue.

[bb-at-ss]

Yes. 100% I'm using 1.0.0 because I made sure to check and update the module before submitting the issue as part of basic trouble shooting. Here again from VS code this morning.

pip show azure-ai-documentintelligence           

azure-ai-documentintelligence           1.0.0

# more modules:

azure-ai-formrecognizer                 3.3.3
azure-ai-inference                      1.0.0b7
azure-common                            1.1.28
azure-core                                  1.32.0
azure-cosmos                              4.9.0
azure-identity                              1.19.0

I tried both strings and the Enum there. Many examples have either, but I continue to encounter this bug. Right now I'm working around it no problem, but I am here to answer more questions if that helps.

[kristapratico]

@bb-at-ss can you share a full example that reproduces this (including imports)? To echo what @pvaneck mentioned, the azure-ai-documentintelligence library has no DocumentAnalysisClient (which is shown in your example). This should be DocumentIntelligenceClient.

[github-actions]

Hi @bb-at-ss. Thank you for opening this issue and giving us the opportunity to assist. To help our team better understand your issue and the details of your scenario please provide a response to the question asked above or the information requested above. This will help us more accurately address your issue.

[bb-at-ss]

This is probably on my end then. I will have to look into this, but this can probably be closed.

[github-actions]

Hi @bb-at-ss. Thank you for opening this issue and giving us the opportunity to assist. We believe that this has been addressed. If you feel that further discussion is needed, please add a comment with the text "/unresolve" to remove the "issue-addressed" label and continue the conversation.

[kristapratico]

@bb-at-ss let us know if you still run into errors, happy to help.