agenta 0.27.7a2__tar.gz → 0.27.8a2__tar.gz

{agenta-0.27.7a2 → agenta-0.27.8a2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: agenta
-Version: 0.27.7a2
+Version: 0.27.8a2
 Summary: The SDK for agenta is an open-source LLMOps platform.
 Home-page: https://agenta.ai
 Keywords: LLMOps,LLM,evaluation,prompt engineering

{agenta-0.27.7a2 → agenta-0.27.8a2}/agenta/client/backend/evaluations/client.py

@@ -26,7 +26,6 @@ class EvaluationsClient:
     def fetch_evaluation_ids(
         self,
         *,
-        app_id: str,
         resource_type: str,
         resource_ids: typing.Optional[typing.Union[str, typing.Sequence[str]]] = None,
         request_options: typing.Optional[RequestOptions] = None,
@@ -35,7 +34,6 @@ class EvaluationsClient:
         Fetches evaluation ids for a given resource type and id.
 
         Arguments:
-        app_id (str): The ID of the app for which to fetch evaluations.
         resource_type (str): The type of resource for which to fetch evaluations.
         resource_ids List[ObjectId]: The IDs of resource for which to fetch evaluations.
 
@@ -47,8 +45,6 @@ class EvaluationsClient:
 
         Parameters
         ----------
-        app_id : str
-
         resource_type : str
 
         resource_ids : typing.Optional[typing.Union[str, typing.Sequence[str]]]
@@ -70,7 +66,6 @@ class EvaluationsClient:
             base_url="https://yourhost.com/path/to/api",
         )
         client.evaluations.fetch_evaluation_ids(
-            app_id="app_id",
             resource_type="resource_type",
         )
         """
@@ -78,7 +73,6 @@ class EvaluationsClient:
             "evaluations/by_resource",
             method="GET",
             params={
-                "app_id": app_id,
                 "resource_type": resource_type,
                 "resource_ids": resource_ids,
             },
@@ -714,7 +708,6 @@ class AsyncEvaluationsClient:
     async def fetch_evaluation_ids(
         self,
         *,
-        app_id: str,
         resource_type: str,
         resource_ids: typing.Optional[typing.Union[str, typing.Sequence[str]]] = None,
         request_options: typing.Optional[RequestOptions] = None,
@@ -723,7 +716,6 @@ class AsyncEvaluationsClient:
         Fetches evaluation ids for a given resource type and id.
 
         Arguments:
-        app_id (str): The ID of the app for which to fetch evaluations.
         resource_type (str): The type of resource for which to fetch evaluations.
         resource_ids List[ObjectId]: The IDs of resource for which to fetch evaluations.
 
@@ -735,7 +727,6 @@ class AsyncEvaluationsClient:
 
         Parameters
         ----------
-        app_id : str
 
         resource_type : str
 
@@ -763,7 +754,6 @@ class AsyncEvaluationsClient:
 
         async def main() -> None:
             await client.evaluations.fetch_evaluation_ids(
-                app_id="app_id",
                 resource_type="resource_type",
             )
 
@@ -774,7 +764,6 @@ class AsyncEvaluationsClient:
             "evaluations/by_resource",
             method="GET",
             params={
-                "app_id": app_id,
                 "resource_type": resource_type,
                 "resource_ids": resource_ids,
             },

{agenta-0.27.7a2 → agenta-0.27.8a2}/agenta/client/backend/testsets/client.py

@@ -28,7 +28,6 @@ class TestsetsClient:
         file: core.File,
         upload_type: typing.Optional[str] = OMIT,
         testset_name: typing.Optional[str] = OMIT,
-        app_id: typing.Optional[str] = OMIT,
         request_options: typing.Optional[RequestOptions] = None,
     ) -> TestSetSimpleResponse:
         """
@@ -51,8 +50,6 @@ class TestsetsClient:
 
         testset_name : typing.Optional[str]
 
-        app_id : typing.Optional[str]
-
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration.
 
@@ -77,7 +74,6 @@ class TestsetsClient:
             data={
                 "upload_type": upload_type,
                 "testset_name": testset_name,
-                "app_id": app_id,
             },
             files={
                 "file": file,
@@ -173,18 +169,16 @@ class TestsetsClient:
 
     def create_testset(
         self,
-        app_id: str,
         *,
         name: str,
         csvdata: typing.Sequence[typing.Dict[str, typing.Optional[typing.Any]]],
         request_options: typing.Optional[RequestOptions] = None,
     ) -> TestSetSimpleResponse:
         """
-        Create a testset with given name and app_name, save the testset to MongoDB.
+        Create a testset with given name, save the testset to MongoDB.
 
         Args:
         name (str): name of the test set.
-        app_name (str): name of the application.
         testset (Dict[str, str]): test set data.
 
         Returns:
@@ -192,8 +186,6 @@ class TestsetsClient:
 
         Parameters
         ----------
-        app_id : str
-
         name : str
 
         csvdata : typing.Sequence[typing.Dict[str, typing.Optional[typing.Any]]]
@@ -215,13 +207,12 @@ class TestsetsClient:
             base_url="https://yourhost.com/path/to/api",
         )
         client.testsets.create_testset(
-            app_id="app_id",
             name="name",
             csvdata=[{"key": "value"}],
         )
         """
         _response = self._client_wrapper.httpx_client.request(
-            f"testsets/{jsonable_encoder(app_id)}",
+            "testsets",
             method="POST",
             json={
                 "name": name,
@@ -405,7 +396,7 @@ class TestsetsClient:
         raise ApiError(status_code=_response.status_code, body=_response_json)
 
     def get_testsets(
-        self, *, app_id: str, request_options: typing.Optional[RequestOptions] = None
+        self, *, request_options: typing.Optional[RequestOptions] = None
     ) -> typing.List[TestSetOutputResponse]:
         """
         Get all testsets.
@@ -420,8 +411,6 @@ class TestsetsClient:
 
         Parameters
         ----------
-        app_id : str
-
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration.
 
@@ -438,16 +427,11 @@ class TestsetsClient:
             api_key="YOUR_API_KEY",
             base_url="https://yourhost.com/path/to/api",
         )
-        client.testsets.get_testsets(
-            app_id="app_id",
-        )
+        client.testsets.get_testsets()
         """
         _response = self._client_wrapper.httpx_client.request(
             "testsets",
             method="GET",
-            params={
-                "app_id": app_id,
-            },
             request_options=request_options,
         )
         try:
@@ -557,7 +541,6 @@ class AsyncTestsetsClient:
         file: core.File,
         upload_type: typing.Optional[str] = OMIT,
         testset_name: typing.Optional[str] = OMIT,
-        app_id: typing.Optional[str] = OMIT,
         request_options: typing.Optional[RequestOptions] = None,
     ) -> TestSetSimpleResponse:
         """
@@ -580,8 +563,6 @@ class AsyncTestsetsClient:
 
         testset_name : typing.Optional[str]
 
-        app_id : typing.Optional[str]
-
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration.
 
@@ -614,7 +595,6 @@ class AsyncTestsetsClient:
             data={
                 "upload_type": upload_type,
                 "testset_name": testset_name,
-                "app_id": app_id,
             },
             files={
                 "file": file,
@@ -718,18 +698,16 @@ class AsyncTestsetsClient:
 
     async def create_testset(
         self,
-        app_id: str,
         *,
         name: str,
         csvdata: typing.Sequence[typing.Dict[str, typing.Optional[typing.Any]]],
         request_options: typing.Optional[RequestOptions] = None,
     ) -> TestSetSimpleResponse:
         """
-        Create a testset with given name and app_name, save the testset to MongoDB.
+        Create a testset with given name, save the testset to MongoDB.
 
         Args:
         name (str): name of the test set.
-        app_name (str): name of the application.
         testset (Dict[str, str]): test set data.
 
         Returns:
@@ -737,8 +715,6 @@ class AsyncTestsetsClient:
 
         Parameters
         ----------
-        app_id : str
-
         name : str
 
         csvdata : typing.Sequence[typing.Dict[str, typing.Optional[typing.Any]]]
@@ -765,7 +741,6 @@ class AsyncTestsetsClient:
 
         async def main() -> None:
             await client.testsets.create_testset(
-                app_id="app_id",
                 name="name",
                 csvdata=[{"key": "value"}],
             )
@@ -774,7 +749,7 @@ class AsyncTestsetsClient:
         asyncio.run(main())
         """
         _response = await self._client_wrapper.httpx_client.request(
-            f"testsets/{jsonable_encoder(app_id)}",
+            "testsets",
             method="POST",
             json={
                 "name": name,
@@ -974,7 +949,7 @@ class AsyncTestsetsClient:
         raise ApiError(status_code=_response.status_code, body=_response_json)
 
     async def get_testsets(
-        self, *, app_id: str, request_options: typing.Optional[RequestOptions] = None
+        self, *, request_options: typing.Optional[RequestOptions] = None
     ) -> typing.List[TestSetOutputResponse]:
         """
         Get all testsets.
@@ -989,8 +964,6 @@ class AsyncTestsetsClient:
 
         Parameters
         ----------
-        app_id : str
-
         request_options : typing.Optional[RequestOptions]
             Request-specific configuration.
 
@@ -1012,9 +985,7 @@ class AsyncTestsetsClient:
 
 
         async def main() -> None:
-            await client.testsets.get_testsets(
-                app_id="app_id",
-            )
+            await client.testsets.get_testsets()
 
 
         asyncio.run(main())
@@ -1022,9 +993,6 @@ class AsyncTestsetsClient:
         _response = await self._client_wrapper.httpx_client.request(
             "testsets",
             method="GET",
-            params={
-                "app_id": app_id,
-            },
             request_options=request_options,
         )
         try:

{agenta-0.27.7a2 → agenta-0.27.8a2}/agenta/sdk/agenta_init.py

@@ -1,4 +1,3 @@
-import logging
 import toml
 from os import getenv
 from typing import Optional, Callable, Any
@@ -11,10 +10,6 @@ from agenta.sdk.tracing import Tracing
 from agenta.client.exceptions import APIRequestError
 
 
-logger = logging.getLogger(__name__)
-logger.setLevel(logging.DEBUG)
-
-
 class AgentaSingleton:
     """Singleton class to save all the "global variables" for the sdk."""
 
@@ -164,7 +159,7 @@ class Config:
         try:
             self.push(config_name="default", overwrite=overwrite, **kwargs)
         except Exception as ex:
-            logger.warning(
+            log.warning(
                 "Unable to push the default configuration to the server. %s", str(ex)
             )
 
@@ -185,7 +180,7 @@ class Config:
                 overwrite=overwrite,
             )
         except Exception as ex:
-            logger.warning(
+            log.warning(
                 "Failed to push the configuration to the server with error: %s", ex
             )
 
@@ -212,14 +207,14 @@ class Config:
                         config_name=config_name,
                     )
             except Exception as ex:
-                logger.warning(
+                log.warning(
                     "Failed to pull the configuration from the server with error: %s",
                     str(ex),
                 )
         try:
             self.set(**{"current_version": config.current_version, **config.parameters})
         except Exception as ex:
-            logger.warning("Failed to set the configuration with error: %s", str(ex))
+            log.warning("Failed to set the configuration with error: %s", str(ex))
 
     def all(self):
         """Returns all the parameters for the app variant"""

{agenta-0.27.7a2 → agenta-0.27.8a2}/agenta/sdk/decorators/routing.py

@@ -143,9 +143,7 @@ class entrypoint:
                 _MIDDLEWARES = False
 
         except:  # pylint: disable=bare-except
-            log.error("------------------------------------")
-            log.error("Agenta SDK - failed to secure route: %s", route_path)
-            log.error("------------------------------------")
+            log.warning("Agenta SDK - failed to secure route: %s", route_path)
         ### --- Update Middleware --- #
 
         DEFAULT_PATH = "generate"
@@ -357,9 +355,7 @@ class entrypoint:
         *args,
         **func_params,
     ):
-        log.info("---------------------------")
-        log.info(f"Agenta SDK - running route: {repr(self.route_path or '/')}")
-        log.info("---------------------------")
+        log.info("Agenta SDK - handling route: %s", repr(self.route_path or "/"))
 
         tracing_context.set(routing_context.get())
 
@@ -377,7 +373,7 @@ class entrypoint:
 
     async def handle_success(self, result: Any, inline_trace: bool):
         data = None
-        trace = dict()
+        trace = None
 
         with suppress():
             data = self.patch_result(result)
@@ -392,21 +388,17 @@ class entrypoint:
         return BaseResponse(data=data, tree=trace)
 
     def handle_failure(self, error: Exception):
-        log.error("--------------------------------------------------")
-        log.error("Agenta SDK - handling application exception below:")
-        log.error("--------------------------------------------------")
-        log.error(format_exc().strip("\n"))
-        log.error("--------------------------------------------------")
+        log.warning("--------------------------------------------------")
+        log.warning("Agenta SDK - handling application exception below:")
+        log.warning("--------------------------------------------------")
+        log.warning(format_exc().strip("\n"))
+        log.warning("--------------------------------------------------")
 
         status_code = error.status_code if hasattr(error, "status_code") else 500
         message = str(error)
         stacktrace = format_exception(error, value=error, tb=error.__traceback__)  # type: ignore
         detail = {"message": message, "stacktrace": stacktrace}
 
-        log.error(f"----------------------------------")
-        log.error(f"Agenta SDK - exiting with failure: {status_code}")
-        log.error(f"----------------------------------")
-
         raise HTTPException(status_code=status_code, detail=detail)
 
     def patch_result(self, result: Any):
@@ -683,10 +675,7 @@ class entrypoint:
 
         loop = get_event_loop()
 
-        with routing_context_manager(
-            config=args_config_params,
-            environment="terminal",
-        ):
+        with routing_context_manager(config=args_config_params):
             result = loop.run_until_complete(
                 self.execute_function(
                     func,
@@ -695,30 +684,23 @@ class entrypoint:
                 )
             )
 
-        SHOW_DETAILS = True
-        SHOW_DATA = False
-        SHOW_TRACE = False
-
         if result.trace:
             log.info("\n========= Result =========\n")
 
             log.info(f"trace_id: {result.trace['trace_id']}")
-            if SHOW_DETAILS:
-                log.info(f"latency:  {result.trace.get('latency')}")
-                log.info(f"cost:     {result.trace.get('cost')}")
-                log.info(f"usage:   {list(result.trace.get('usage', {}).values())}")
-
-            if SHOW_DATA:
-                log.info(" ")
-                log.info(f"data:")
-                log.info(dumps(result.data, indent=2))
-
-            if SHOW_TRACE:
-                log.info(" ")
-                log.info(f"trace:")
-                log.info(f"----------------")
-                log.info(dumps(result.trace.get("spans", []), indent=2))
-                log.info(f"----------------")
+            log.info(f"latency:  {result.trace.get('latency')}")
+            log.info(f"cost:     {result.trace.get('cost')}")
+            log.info(f"usage:   {list(result.trace.get('usage', {}).values())}")
+
+            log.info(" ")
+            log.info("data:")
+            log.info(dumps(result.data, indent=2))
+
+            log.info(" ")
+            log.info("trace:")
+            log.info("----------------")
+            log.info(dumps(result.trace.get("spans", []), indent=2))
+            log.info("----------------")
 
             log.info("\n==========================\n")
 

{agenta-0.27.7a2 → agenta-0.27.8a2}/agenta/sdk/litellm/litellm.py

@@ -60,7 +60,11 @@ def litellm_handler():
             )
 
             if not self.span:
-                log.error("LiteLLM callback error: span not found.")
+                log.warning("Agenta SDK - litellm tracing failed")
+                return
+
+            if not self.span.is_recording():
+                log.error("Agenta SDK - litellm span not recording.")
                 return
 
             self.span.set_attributes(
@@ -86,43 +90,11 @@ def litellm_handler():
             end_time,
         ):
             if not self.span:
-                log.error("LiteLLM callback error: span not found.")
+                log.warning("Agenta SDK - litellm tracing failed")
                 return
 
-            try:
-                result = []
-                for choice in response_obj.choices:
-                    message = choice.message.__dict__
-                    result.append(message)
-
-                outputs = {"completion": result}
-                self.span.set_attributes(
-                    attributes={"outputs": outputs},
-                    namespace="data",
-                )
-
-            except Exception as e:
-                pass
-
-            self.span.set_attributes(
-                attributes={"total": kwargs.get("response_cost")},
-                namespace="metrics.unit.costs",
-            )
-
-            self.span.set_attributes(
-                attributes=(
-                    {
-                        "prompt": response_obj.usage.prompt_tokens,
-                        "completion": response_obj.usage.completion_tokens,
-                        "total": response_obj.usage.total_tokens,
-                    }
-                ),
-                namespace="metrics.unit.tokens",
-            )
-
-            self.span.set_status(status="OK")
-
-            self.span.end()
+            if not self.span.is_recording():
+                return
 
         def log_success_event(
             self,
@@ -131,8 +103,14 @@ def litellm_handler():
             start_time,
             end_time,
         ):
+            if kwargs.get("stream"):
+                return
+
             if not self.span:
-                log.error("LiteLLM callback error: span not found.")
+                log.warning("Agenta SDK - litellm tracing failed")
+                return
+
+            if not self.span.is_recording():
                 return
 
             try:
@@ -178,7 +156,10 @@ def litellm_handler():
             end_time,
         ):
             if not self.span:
-                log.error("LiteLLM callback error: span not found.")
+                log.warning("Agenta SDK - litellm tracing failed")
+                return
+
+            if not self.span.is_recording():
                 return
 
             self.span.record_exception(kwargs["exception"])
@@ -195,43 +176,11 @@ def litellm_handler():
             end_time,
         ):
             if not self.span:
-                log.error("LiteLLM callback error: span not found.")
+                log.warning("Agenta SDK - litellm tracing failed")
                 return
 
-            try:
-                result = []
-                for choice in response_obj.choices:
-                    message = choice.message.__dict__
-                    result.append(message)
-
-                outputs = {"completion": result}
-                self.span.set_attributes(
-                    attributes={"outputs": outputs},
-                    namespace="data",
-                )
-
-            except Exception as e:
-                pass
-
-            self.span.set_attributes(
-                attributes={"total": kwargs.get("response_cost")},
-                namespace="metrics.unit.costs",
-            )
-
-            self.span.set_attributes(
-                attributes=(
-                    {
-                        "prompt": response_obj.usage.prompt_tokens,
-                        "completion": response_obj.usage.completion_tokens,
-                        "total": response_obj.usage.total_tokens,
-                    }
-                ),
-                namespace="metrics.unit.tokens",
-            )
-
-            self.span.set_status(status="OK")
-
-            self.span.end()
+            if not self.span.is_recording():
+                return
 
         async def async_log_success_event(
             self,
@@ -241,7 +190,10 @@ def litellm_handler():
             end_time,
         ):
             if not self.span:
-                log.error("LiteLLM callback error: span not found.")
+                log.warning("Agenta SDK - litellm tracing failed")
+                return
+
+            if not self.span.is_recording():
                 return
 
             try:
@@ -287,7 +239,10 @@ def litellm_handler():
             end_time,
         ):
             if not self.span:
-                log.error("LiteLLM callback error: span not found.")
+                log.warning("Agenta SDK - litellm tracing failed")
+                return
+
+            if not self.span.is_recording():
                 return
 
             self.span.record_exception(kwargs["exception"])

{agenta-0.27.7a2 → agenta-0.27.8a2}/agenta/sdk/middleware/auth.py

@@ -133,10 +133,10 @@ class AuthorizationMiddleware(BaseHTTPMiddleware):
             return await call_next(request)
 
         except:  # pylint: disable=bare-except
-            log.error("------------------------------------------------------")
-            log.error("Agenta SDK - handling auth middleware exception below:")
-            log.error("------------------------------------------------------")
-            log.error(format_exc().strip("\n"))
-            log.error("------------------------------------------------------")
+            log.warning("------------------------------------------------------")
+            log.warning("Agenta SDK - handling auth middleware exception below:")
+            log.warning("------------------------------------------------------")
+            log.warning(format_exc().strip("\n"))
+            log.warning("------------------------------------------------------")
 
             return Deny()

{agenta-0.27.7a2 → agenta-0.27.8a2}/agenta/sdk/tracing/context.py

@@ -14,11 +14,11 @@ def tracing_context_manager():
     token = tracing_context.set(_tracing_context)
     try:
         yield
-    except Exception as e:
-        log.error("----------------------------------------------")
-        log.error("Agenta SDK - handling tracing exception below:")
-        log.error("----------------------------------------------")
-        log.error(format_exc().strip("\n"))
-        log.error("----------------------------------------------")
+    except:  # pylint: disable=bare-except
+        log.warning("----------------------------------------------")
+        log.warning("Agenta SDK - handling tracing exception below:")
+        log.warning("----------------------------------------------")
+        log.warning(format_exc().strip("\n"))
+        log.warning("----------------------------------------------")
     finally:
         tracing_context.reset(token)

{agenta-0.27.7a2 → agenta-0.27.8a2}/agenta/sdk/tracing/inline.py

@@ -1118,13 +1118,15 @@ def calculate_costs(span_idx: Dict[str, SpanDTO]):
             and span.meta
             and span.metrics
         ):
+            model = span.meta.get("response.model")
+            prompt_tokens = span.metrics.get("unit.tokens.prompt", 0.0)
+            completion_tokens = span.metrics.get("unit.tokens.completion", 0.0)
+
             try:
                 costs = cost_calculator.cost_per_token(
-                    model=span.meta.get("response.model"),
-                    prompt_tokens=span.metrics.get("unit.tokens.prompt", 0.0),
-                    completion_tokens=span.metrics.get("unit.tokens.completion", 0.0),
-                    call_type=span.node.type.name.lower(),
-                    response_time_ms=span.time.span // 1_000,
+                    model=model,
+                    prompt_tokens=prompt_tokens,
+                    completion_tokens=completion_tokens,
                 )
 
                 if not costs:
@@ -1137,5 +1139,8 @@ def calculate_costs(span_idx: Dict[str, SpanDTO]):
                 span.metrics["unit.costs.completion"] = completion_cost
                 span.metrics["unit.costs.total"] = total_cost
 
-            except:
-                pass
+            except:  # pylint: disable=bare-except
+                print("Failed to calculate costs:")
+                print(
+                    f"model={model}, prompt_tokens={prompt_tokens}, completion_tokens={completion_tokens}"
+                )