Fix column mapping conversion for ProxyClient with target functions

[Copilot]

When using _use_pf_client=True with a target function, user-provided column mappings referencing target outputs (e.g., ${target.response}) fail because the conversion chain stops at ${run.outputs.} instead of completing to ${data.__outputs.} format required by ProxyClient.

Changes

• Column mapping conversion: Added conversion step in _preprocess_data() to transform ${run.outputs.*} → ${data.__outputs.*} when using ProxyClient with target functions
• Test coverage: Added test_column_mapping_with_target_and_pf_client to validate the fix

Technical Details

The conversion chain now properly handles all three stages:

1. User input: ${target.response}
2. _process_column_mappings(): ${run.outputs.response}
3. New: ProxyClient conversion: ${data.__outputs.response}

Without step 3, ProxyClient cannot map columns from the temporary JSONL file containing target outputs.

Example

evaluate(
    data="data.csv",
    target=lambda query: {"response": f"Answer: {query}"},
    evaluators={"coherence": CoherenceEvaluator(model_config=config)},
    evaluator_config={
        "default": {
            "column_mapping": {
                "query": "${data.query}",
                "response": "${target.response}",  # This now works
            }
        }
    },
    _use_pf_client=True,
)

Warning

Firewall rules blocked me from connecting to one or more addresses (expand for details)

I tried to connect to the following addresses, but was blocked by firewall rules:

• pypi.org
  • Triggering command: /home/REDACTED/work/azure-sdk-for-python/azure-sdk-for-python/.venv/bin/python /home/REDACTED/work/azure-sdk-for-python/azure-sdk-for-python/.venv/bin/python /home/REDACTED/work/azure-sdk-for-python/azure-sdk-for-python/.venv/lib/python3.9/site-packages/pip/__pip-REDACTED__.py install --ignore-installed --no-user --prefix /tmp/pip-build-env-9p571cym/overlay --no-warn-script-location --no-binary :none: --only-binary :none: -i REDACTED -- setuptools>=40.8.0 (dns block)
• scanning-api.github.com
  • Triggering command: /home/REDACTED/work/_temp/ghcca-node/node/bin/node /home/REDACTED/work/_temp/ghcca-node/node/bin/node --enable-source-maps /home/REDACTED/work/_temp/copilot-developer-action-main/dist/index.js (dns block)

If you need me to access, download, or install something from one of these locations, you can either:

• Configure Actions setup steps to set up my environment, which run before the firewall is enabled
• Add the appropriate URLs or hosts to the custom allowlist in this repository's Copilot coding agent settings (admins only)

Original prompt

---

This section details on the original issue you should resolve

<issue_title>azure-ai-evaluation: column mapping error when set _use_pf_client to true</issue_title>
<issue_description>- Package Name: azure-ai-evaluation

• Package Version: 1.10.0
• Operating System: Windows 11
• Python Version: 3.12

Describe the bug
Error when set _use_pf_client to true when invoking evaluate function.

To Reproduce
data.csv:

query
hello

main.py

from azure.ai.evaluation import (
    evaluate,
    CoherenceEvaluator,
    AzureOpenAIModelConfiguration,
)
from langchain_openai import AzureChatOpenAI
from azure.identity import AzureCliCredential, get_bearer_token_provider
from typing import TypedDict

endpoint = "xxx"


class TargetFunctionOutput(TypedDict):
    response: str


def eval_target(query: str) -> TargetFunctionOutput:
    llm = AzureChatOpenAI(
        azure_deployment="gpt-4.1-mini",
        model_name="gpt-4.1-mini",
        openai_api_version="2024-10-21",
        temperature=1,
        max_retries=3,
        max_tokens=None,
        timeout=None,
        azure_endpoint=endpoint,
        openai_api_type="azure",
        azure_ad_token_provider=get_bearer_token_provider(
            AzureCliCredential(), "https://cognitiveservices.azure.com/.default"
        ),
    )
    return TargetFunctionOutput(response=llm.invoke(query).content)


if __name__ == "__main__":
    evaluate(
        data="data.csv",
        target=eval_target,
        evaluators={
            "coherence": CoherenceEvaluator(
                model_config=AzureOpenAIModelConfiguration(
                    azure_deployment="gpt-4.1-mini",
                    azure_endpoint=endpoint,
                    api_version="2024-10-21",
                )
            )
        },
        evaluator_config={
            "default": {
                "column_mapping": {
                    "query": "${data.query}",
                    "response": "${target.response}",
                },
            }
        },
        _use_pf_client=True,
    )

</issue_description>

Comments on the Issue (you are @copilot in this section)

@w-javed @Wixee Can you provide the actual error you are seeing. Thanks! @w-javed Thanks for the info. We'll investigate this and get back to you asap.

• Fixes azure-ai-evaluation: column mapping error when set _use_pf_client to true #42752

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.