How to use the debug parameter in the search method of the SearchClient class in the module azure.search.documents.aio._search_client_async.py

[yk7180]

• Package Name: azure-search-documents
• Package Version: 11.6.08b
• Operating System: ubuntu
• Python Version: 3.12

Question

In the official documentation [[1]] regarding the debug parameter for performing index searches using the search method of the SearchClient class in the azure.search.documents.aio._search_client_async.py module of the azure-search-documents library (version 11.6.0b8), it is stated as follows:

Enables a debugging tool that can be used to further explore your Semantic search results. Known values are: "disabled", "speller", "semantic", and "all".

How can I obtain debug information from this debugging tool as an object?

What I Tried

I inferred that setting the debug parameter to a value other than "disabled" would result in the debug information being included in the return value of the search method.

Therefore, I set debug="all" as shown in the [code] and executed the search method, then checked the keys of the returned object of the type AsyncSearchItemPaged. The keys contained in the object are as follows ([output]), and there was no change before and after adding the debug parameter. The key related to the debug information (ex: @search.documentDebugInfo) was not present.

[code]

async def azure_ai_search(search_text: str, top: int, openai_client:OpenAiClient, ai_search_client: AiSearchClient) -> list[Document]:

    vector_query = VectorizedQuery(vector=await generate_embeddings(search_text,openai_client), k_nearest_neighbors=top, fields=ai_search_client.vector_field_name,weight=0.33)

    results = await ai_search_client.search_client.search(
        search_text=search_text,
        query_type=QueryType.SEMANTIC,
        semantic_configuration_name=ai_search_client.semantic_configuration_name if ai_search_client.semantic_configuration_name is not None else "mySemanticSearchConfig",
        vector_queries=[vector_query],
        top=top,
        debug="all"
    )
    keys=[result.keys() async for result in results]
    print(keys[0])

[output]

dict_keys(['url', 'tag', 'content_vector', 'content', 'id', 'metadata', 'document_internal_index', 'last_update', 'source', '@search.score', '@search.reranker_score', '@search.highlights', '@search.captions'])

On the other hand, when checking the JSON data of the API response after executing this, I confirmed the presence of debug information such as "@search.documentDebugInfo".

{"@search.answers":[],"value":[{"@search.score":0.022166667506098747,"@search.rerankerScore":2.606142282485962,"@search.documentDebugInfo":{"semantic":null,"vectors":{"subscores":{"documentBoost":1.0,"text":{"searchScore":8.36231517791748},"vectors":[{"content_vector":{"searchScore":0.6478207111358643,"vectorSimilarity":0.45636302944585094}}]}}},

Why is the debug information that exists in the API response not available as the return value of the search method?

[[1]]: https://learn.microsoft.com/en-us/python/api/azure-search-documents/azure.search.documents.searchclient?view=azure-python-preview

[github-actions]

Thank you for your feedback. Tagging and routing to the team member best able to assist.