Documentation request around Azure endpoint difference and how best to work around it

[ekaj2]

Description

If you just pass your endpoint to the Azure ChatCompletionsClient, it will not work properly with Mirascope (or likely any OpenAI-compatible client). Requests will return something like:

File ".../site-packages/mirascope/core/base/_create.py", line 232, in inner
    response = create(stream=False, **call_kwargs)
               ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File ".../site-packages/mirascope/core/base/_utils/_get_create_fn_or_async_create_fn.py", line 135, in create_or_stream
    return cast(_NonStreamedResponse, sync_func(**kwargs))
                                      ^^^^^^^^^^^^^^^^^^^
  File ".../site-packages/azure/ai/inference/_patch.py", line 656, in complete
    map_error(status_code=response.status_code, response=response, error_map=error_map)
  File ".../site-packages/azure/core/exceptions.py", line 163, in map_error
    raise error
azure.core.exceptions.ResourceNotFoundError: (404) Resource not found
Code: 404
Message: Resource not found

Discussed on Slack an easy way to increase DX with an example in the docs or a note about it in the Azure custom client section:

from azure.ai.inference import ChatCompletionsClient
from azure.core.credentials import AzureKeyCredential
from mirascope.core import azure


@azure.call(
    "gpt-4o-mini",
    client=ChatCompletionsClient(
        endpoint="https://my-endpoint.openai.azure.com/openai/deployments/gpt-4o-mini/",
        credential=AzureKeyCredential("...")
    )
)
def hello():
    return "Hello, how are you?"

print(hello())

Ref: https://mirascope-community.slack.com/archives/C072S9P1R38/p1736630556580189