Refactor event system and improve crew testing

- Extract base CrewEvent class to a new base_events.py module
- Update event imports across multiple event-related files
- Modify CrewTestStartedEvent to use eval_llm instead of openai_model_name
- Add LLM creation validation in crew testing method
- Improve type handling and event consistency

src/crewai/crew.py

@@ -1201,6 +1201,10 @@ class Crew(BaseModel):
     ) -> None:
         """Test and evaluate the Crew with the given inputs for n iterations concurrently using concurrent.futures."""
         try:
+            eval_llm = create_llm(eval_llm)
+            if not eval_llm:
+                raise ValueError("Failed to create LLM instance.")
+
             crewai_event_bus.emit(
                 self,
                 CrewTestStartedEvent(
@@ -1211,7 +1215,7 @@ class Crew(BaseModel):
                 ),
             )
             test_crew = self.copy()
-            evaluator = CrewEvaluator(test_crew, eval_llm)
+            evaluator = CrewEvaluator(test_crew, eval_llm)  # type: ignore[arg-type]
 
             for i in range(1, n_iterations + 1):
                 evaluator.set_iteration(i)

src/crewai/utilities/events/agent_events.py

@@ -4,7 +4,7 @@ from crewai.agents.agent_builder.base_agent import BaseAgent
 from crewai.tools.base_tool import BaseTool
 from crewai.tools.structured_tool import CrewStructuredTool
 
-from .crew_events import CrewEvent
+from .base_events import CrewEvent
 
 if TYPE_CHECKING:
     from crewai.agents.agent_builder.base_agent import BaseAgent

src/crewai/utilities/events/base_events.py

@@ -0,0 +1,10 @@
+from datetime import datetime
+
+from pydantic import BaseModel, Field
+
+
+class CrewEvent(BaseModel):
+    """Base class for all crew events"""
+
+    timestamp: datetime = Field(default_factory=datetime.now)
+    type: str

src/crewai/utilities/events/crew_events.py

@@ -1,14 +1,8 @@
-from datetime import datetime
-from typing import Any, Dict, Optional
+from typing import Any, Dict, Optional, Union
 
-from pydantic import BaseModel, Field
+from pydantic import InstanceOf
 
-
-class CrewEvent(BaseModel):
-    """Base class for all crew events"""
-
-    timestamp: datetime = Field(default_factory=datetime.now)
-    type: str
+from crewai.utilities.events.base_events import CrewEvent
 
 
 class CrewKickoffStartedEvent(CrewEvent):
@@ -67,7 +61,7 @@ class CrewTestStartedEvent(CrewEvent):
 
     crew_name: Optional[str]
     n_iterations: int
-    openai_model_name: Optional[str]
+    eval_llm: Optional[Union[str, Any]]
     inputs: Optional[Dict[str, Any]]
     type: str = "crew_test_started"
 

src/crewai/utilities/events/crewai_event_bus.py

@@ -4,7 +4,7 @@ from typing import Any, Callable, Dict, List, Type, TypeVar, cast
 
 from blinker import Signal
 
-from crewai.utilities.events.crew_events import CrewEvent
+from crewai.utilities.events.base_events import CrewEvent
 from crewai.utilities.events.event_types import EventTypes
 
 EventT = TypeVar("EventT", bound=CrewEvent)

src/crewai/utilities/events/event_listener.py

@@ -85,7 +85,7 @@ class EventListener(BaseEventListener):
                 cloned_crew,
                 event.n_iterations,
                 event.inputs,
-                event.openai_model_name,
+                event.eval_llm,
             )
             self.logger.log(
                 f"🚀 Crew '{event.crew_name}' started test",

src/crewai/utilities/events/flow_events.py

@@ -2,7 +2,7 @@ from typing import Any, Dict, Optional, Union
 
 from pydantic import BaseModel
 
-from .crew_events import CrewEvent
+from .base_events import CrewEvent
 
 
 class FlowEvent(CrewEvent):

src/crewai/utilities/events/tool_usage_events.py

@@ -1,7 +1,7 @@
 from datetime import datetime
 from typing import Any, Callable, Dict
 
-from .crew_events import CrewEvent
+from .base_events import CrewEvent
 
 
 class ToolUsageEvent(CrewEvent):