Merge branch 'main' into raflFaisal/generaliseContainerEnvForAgent

Author: raflFaisal

pyproject.toml

@@ -157,7 +157,10 @@ asyncio_default_fixture_loop_scope = "function"
 asyncio_mode = "auto"
 
 [tool.mypy]
+python_version = "3.9"
 exclude = "tests/"
 plugins = ["pydantic.mypy"]
-strict = true
+# Start with non-strict mode, and swtich to strict mode later.
+# strict = true
 disable_error_code = ["import-not-found", "import-untyped", "unused-ignore"]
+follow_imports = "skip"

src/google/adk/cli/browser/index.html

@@ -12,6 +12,8 @@
 # See the License for the specific language governing permissions and
 # limitations under the License.
 
+from __future__ import annotations
+
 import importlib.util
 import json
 import logging
@@ -22,99 +24,20 @@
 from typing import Optional
 import uuid
 
-from pydantic import BaseModel
-from pydantic import Field
-
 from ..agents import Agent
 from ..artifacts.base_artifact_service import BaseArtifactService
 from ..evaluation.eval_case import EvalCase
-from ..evaluation.eval_case import Invocation
+from ..evaluation.eval_metrics import EvalMetric
+from ..evaluation.eval_metrics import EvalMetricResult
+from ..evaluation.eval_metrics import EvalMetricResultPerInvocation
+from ..evaluation.eval_result import EvalCaseResult
 from ..evaluation.evaluator import EvalStatus
 from ..evaluation.evaluator import Evaluator
 from ..sessions.base_session_service import BaseSessionService
-from ..sessions.session import Session
-from .utils import common
 
 logger = logging.getLogger("google_adk." + __name__)
 
 
-class EvalMetric(common.BaseModel):
-  """A metric used to evaluate a particular aspect of an eval case."""
-
-  metric_name: str
-  """The name of the metric."""
-
-  threshold: float
-  """A threshold value. Each metric decides how to interpret this threshold."""
-
-
-class EvalMetricResult(EvalMetric):
-  """The actual computed score/value of a particular EvalMetric."""
-
-  score: Optional[float] = None
-  eval_status: EvalStatus
-
-
-class EvalMetricResultPerInvocation(common.BaseModel):
-  """Eval metric results per invocation."""
-
-  actual_invocation: Invocation
-  """The actual invocation, usually obtained by inferencing the agent."""
-
-  expected_invocation: Invocation
-  """The expected invocation, usually the reference or golden invocation."""
-
-  eval_metric_results: list[EvalMetricResult] = []
-  """Eval resutls for each applicable metric."""
-
-
-class EvalCaseResult(common.BaseModel):
-  """Case-level evaluation results."""
-
-  eval_set_file: str = Field(
-      deprecated=True,
-      description="This field is deprecated, use eval_set_id instead.",
-  )
-  eval_set_id: str = ""
-  """The eval set id."""
-
-  eval_id: str = ""
-  """The eval case id."""
-
-  final_eval_status: EvalStatus
-  """Final eval status for this eval case."""
-
-  eval_metric_results: list[tuple[EvalMetric, EvalMetricResult]] = Field(
-      deprecated=True,
-      description=(
-          "This field is deprecated, use overall_eval_metric_results instead."
-      ),
-  )
-
-  overall_eval_metric_results: list[EvalMetricResult]
-  """Overall result for each metric for the entire eval case."""
-
-  eval_metric_result_per_invocation: list[EvalMetricResultPerInvocation]
-  """Result for each metric on a per invocation basis."""
-
-  session_id: str
-  """Session id of the session generated as result of inferencing/scraping stage of the eval."""
-
-  session_details: Optional[Session] = None
-  """Session generated as result of inferencing/scraping stage of the eval."""
-
-  user_id: Optional[str] = None
-  """User id used during inferencing/scraping stage of the eval."""
-
-
-class EvalSetResult(common.BaseModel):
-  eval_set_result_id: str
-  eval_set_result_name: str
-  eval_set_id: str
-  eval_case_results: list[EvalCaseResult] = Field(default_factory=list)
-  creation_timestamp: float = 0.0
-
-
 MISSING_EVAL_DEPENDENCIES_MESSAGE = (
     "Eval module is not installed, please install via `pip install"
     " google-adk[eval]`."
@@ -228,8 +151,6 @@ async def run_evals(
   """
   try:
     from ..evaluation.agent_evaluator import EvaluationGenerator
-    from ..evaluation.response_evaluator import ResponseEvaluator
-    from ..evaluation.trajectory_evaluator import TrajectoryEvaluator
   except ModuleNotFoundError as e:
     raise ModuleNotFoundError(MISSING_EVAL_DEPENDENCIES_MESSAGE) from e
 

src/google/adk/cli/browser/main-G4TORPTW.js renamed to src/google/adk/cli/browser/main-GR6432WH.js

@@ -19,8 +19,6 @@
 import logging
 import os
 import tempfile
-from typing import AsyncGenerator
-from typing import Coroutine
 from typing import Optional
 from typing import Tuple
 
@@ -465,7 +463,7 @@ async def _collect_eval_results() -> list[EvalCaseResult]:
 
   # Write eval set results.
   local_eval_set_results_manager = LocalEvalSetResultsManager(
-      agent_dir=os.path.dirname(agent_module_file_path)
+      agents_dir=os.path.dirname(agent_module_file_path)
   )
   eval_set_id_to_eval_results = collections.defaultdict(list)
   for eval_case_result in eval_results:
@@ -610,7 +608,7 @@ async def _lifespan(app: FastAPI):
     )
 
   app = get_fast_api_app(
-      agent_dir=agents_dir,
+      agents_dir=agents_dir,
       session_db_url=session_db_url,
       allow_origins=allow_origins,
       web=True,
@@ -711,7 +709,7 @@ def cli_api_server(
 
   config = uvicorn.Config(
       get_fast_api_app(
-          agent_dir=agents_dir,
+          agents_dir=agents_dir,
           session_db_url=session_db_url,
           allow_origins=allow_origins,
           web=False,