Add httpx_client parameter in LangfuseConnector

[alex-stoica]

The current environment variables LANGFUSE_SECRET_KEY / LANGFUSE_API_KEY, LANGFUSE_PUBLIC_KEY, and LANGFUSE_HOST may not be sufficient for tracing in Langfuse.

For example, the following code:

tracer = Langfuse()
tracer.create_prompt(
    name="greeting-prompt",
    prompt="..."
)

might return an error:

...
ConnectError: [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed

However, this works correctly if the env variables are set correctly:

Langfuse(
    httpx_client=httpx.Client(verify=os.getenv('LANGFUSE_PATH_TO_CERTIFICATE'))
)

To address this, please update the wrapper in haystack_integrations.components.connectors.langfuse.LangfuseConnector to accept httpx_client as a parameter.
The code fix would only affect the lines

• haystack-core-integrations/integrations/langfuse/src/haystack_integrations/components/connectors/langfuse/langfuse_connector.py

  Line 108 in 61ac2f4

  self.tracer = LangfuseTracer(tracer=Langfuse(), name=name, public=public)

  
  self.tracer = LangfuseTracer(tracer=Langfuse(), name=name, public=public)
  -> tracer=Langfuse(httpx_client=httpx_client or httpx.Client(verify=os.getenv('LANGFUSE_PATH_TO_CERTIFICATE')))
• https://github.com/deepset-ai/haystack-core-integrations/blob/61ac2f457c53697e8f906d9f44c39995a008f140/integrations/langfuse/src/haystack_integrations/components/connectors/langfuse/langfuse_connector.py#L108` -> def __init__(self, name: str, public: bool = False, httpx_client: Optional[httpx.Client] = None)

Thanks!

[alex-stoica]

Tried locally to overwrite the LangfuseConnector in this way:

class LangfuseConnector:
    def __init__(self, name: str, public: bool = False, httpx_client: Optional[httpx.Client] = None):
        self.name = name
        self.tracer = LangfuseTracer(
            tracer=Langfuse(httpx_client=httpx_client or httpx.Client(verify=os.getenv('LANGFUSE_PATH_TO_CERTIFICATE')))                     
        )
        tracing.enable_tracing(self.tracer)

    @component.output_types(name=str, trace_url=str)
    def run(self, invocation_context: Optional[Dict[str, Any]] = None):
        logger.debug(
            "Langfuse tracer invoked with the following context: '{invocation_context}'",
            invocation_context=invocation_context,
        )
        return {"name": self.name, "trace_url": self.tracer.get_trace_url()}

and it solved the certification related issue that is usually showed as Unexpected error occurred. Please check your request and contact support: https://langfuse.com/support.