update tests

src/promptflow-evals/tests/evals/conftest.py

@@ -4,6 +4,7 @@
 from unittest.mock import patch
 
 import pytest
+from azure.identity import DefaultAzureCredential
 from pytest_mock import MockerFixture
 
 from promptflow.client import PFClient
@@ -13,6 +14,7 @@
 from promptflow.tracing._integrations._openai_injector import inject_openai_api
 
 try:
+    from promptflow.recording.azure import FakeTokenCredential
     from promptflow.recording.local import recording_array_reset
     from promptflow.recording.record_mode import is_in_ci_pipeline, is_live, is_record, is_replay
 except ImportError:
@@ -91,6 +93,15 @@ def project_scope() -> dict:
     return dev_connections[conn_name]["value"]
 
 
+@pytest.fixture
+def azure_credential():
+    # CI pipeline currently does not support DefaultAzureCredential
+    if pytest.is_replay:
+        return FakeTokenCredential()
+
+    return DefaultAzureCredential()
+
+
 @pytest.fixture
 def pf_client() -> PFClient:
     """The fixture, returning PRClient"""

src/promptflow-evals/tests/evals/e2etests/test_builtin_evaluators.py

@@ -5,7 +5,7 @@
 from promptflow.evals.evaluators.qa import QAEvaluator
 
 
-@pytest.mark.usefixtures("model_config", "project_scope", "recording_injection")
+@pytest.mark.usefixtures("model_config", "project_scope", "azure_credential", "recording_injection")
 @pytest.mark.e2etest
 class TestBuiltInEvaluators:
     def test_individual_evaluator_prompt_based(self, model_config):
@@ -17,8 +17,8 @@ def test_individual_evaluator_prompt_based(self, model_config):
         assert score is not None
         assert score["gpt_fluency"] > 1.0
 
-    def test_individual_evaluator_service_based(self, project_scope):
-        eval_fn = ViolenceEvaluator(project_scope)
+    def test_individual_evaluator_service_based(self, project_scope, azure_credential):
+        eval_fn = ViolenceEvaluator(project_scope, azure_credential)
         score = eval_fn(
             question="What is the capital of Japan?",
             answer="The capital of Japan is Tokyo.",
@@ -59,8 +59,8 @@ def test_composite_evaluator_qa(self, model_config, parallel):
             (True),
         ],
     )
-    def test_composite_evaluator_content_safety(self, project_scope, parallel):
-        safety_eval = ContentSafetyEvaluator(project_scope, parallel=parallel)
+    def test_composite_evaluator_content_safety(self, project_scope, parallel, azure_credential):
+        safety_eval = ContentSafetyEvaluator(project_scope, parallel, azure_credential)
         score = safety_eval(
             question="Tokyo is the capital of which country?",
             answer="Japan",