Update tests to be compatible with new OpenAI, MistralAI and MCP versions #2094
Conversation
PR Change SummaryUpdated tests to ensure compatibility with the latest versions of OpenAI, MistralAI, and MCP.
Modified Files
How can I customize these reviews?Check out the Hyperlint AI Reviewer docs for more information on how to customize the review. If you just want to ignore it on this PR, you can add the Note specifically for link checks, we only check the first 30 links in a file and we cache the results for several hours (for instance, if you just added a page, you might experience this). Our recommendation is to add |
There was a problem hiding this comment.
@medaminezghal Thanks Mohamed! The only thing I'm not sure about is how we should handle ResourceLink.
pydantic_ai_slim/pydantic_ai/mcp.py
Outdated
| @@ -239,6 +239,13 @@ def _map_tool_result_part( | |||
| ) | |||
| else: | |||
| assert_never(resource) | |||
| elif isinstance(part, mcp_types.ResourceLink): | |||
| return { | |||
There was a problem hiding this comment.
Is the expectation that this resource link is sent straight to the model as JSON, or are we supposed to handle it somehow and pass the actual resource?
If it's supposed to be JSON, we can use part.model_dump() here.
There was a problem hiding this comment.
@DouweM I'm not sure either. So I have open an question in Github about the new ResourceLink and that's what I get.
There was a problem hiding this comment.
@medaminezghal I've asked a followup question: modelcontextprotocol/modelcontextprotocol#872 (reply in thread)
There was a problem hiding this comment.
@DouweM Since there is no support to resource subscribing, so it's better to make it return the content of the resource?
There was a problem hiding this comment.
@medaminezghal Yeah that's what I'm thinking. I was gonna wait for confirmation but I expect that's the way we'll go, so if you feel like adding it please do!
There was a problem hiding this comment.
elif isinstance(part, mcp_types.ResourceLink):
filepath = part.uri[7:]
with open(filepath, "rb") as f:
file_data = f.read()
return messages.BinaryContent(data=base64.b64encode(file_data), media_type=part.mimeType)
What about this solution?
There was a problem hiding this comment.
@medaminezghal That'll only work if the resource is on the local filesystem. I think we'll need to use the resources/read endpoint: https://modelcontextprotocol.io/specification/2025-06-18/server/resources#reading-resources. In the Python SDK, that's the await session.read_resource call from this example: https://github.com/modelcontextprotocol/python-sdk#writing-mcp-clients.
There was a problem hiding this comment.
@DouweM What about this solution?
Do you have any suggestion?
|
@medaminezghal Can you please have a look at the failing coverage check? |
I've updated known model list with new OpenAI version, fixed MistralAI test issue in new version, added new ResourceLink MCP type and fixed tests for new MCP version.