Python: Add OpenTelemetry to Python SK (#6914)

### Motivation and Context
We want observability into usage of SK

### Description
Add OpenTelemetry to Python SK

### Contribution Checklist
- [ ] The code builds clean without any errors or warnings
- [ ] The PR follows the [SK Contribution
Guidelines](https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md)
and the [pre-submission formatting
script](https://github.com/microsoft/semantic-kernel/blob/main/CONTRIBUTING.md#development-scripts)
raises no violations

---------

Co-authored-by: Tao Chen <[email protected]>

python/pyproject.toml

@@ -26,6 +26,11 @@ openai = ">=1.0"
 
 # openapi and swagger
 openapi_core = ">=0.18,<0.20"
+
+# OpenTelemetry
+opentelemetry-api = "^1.24.0"
+opentelemetry-sdk = "^1.24.0"
+
 prance = "^23.6.21.0"
 
 # templating

python/semantic_kernel/connectors/ai/azure_ai_inference/services/azure_ai_inference_chat_completion.py

@@ -7,7 +7,7 @@
 from functools import reduce
 from typing import TYPE_CHECKING, Any
 
-from semantic_kernel.connectors.telemetry import SEMANTIC_KERNEL_USER_AGENT
+from semantic_kernel.utils.telemetry.user_agent import SEMANTIC_KERNEL_USER_AGENT
 
 if sys.version_info >= (3, 12):
     from typing import override  # pragma: no cover

python/semantic_kernel/connectors/ai/azure_ai_inference/services/azure_ai_inference_text_embedding.py

@@ -20,9 +20,9 @@
 from semantic_kernel.connectors.ai.azure_ai_inference.azure_ai_inference_settings import AzureAIInferenceSettings
 from semantic_kernel.connectors.ai.azure_ai_inference.services.azure_ai_inference_base import AzureAIInferenceBase
 from semantic_kernel.connectors.ai.embeddings.embedding_generator_base import EmbeddingGeneratorBase
-from semantic_kernel.connectors.telemetry import SEMANTIC_KERNEL_USER_AGENT
 from semantic_kernel.exceptions.service_exceptions import ServiceInitializationError
 from semantic_kernel.utils.experimental_decorator import experimental_class
+from semantic_kernel.utils.telemetry.user_agent import SEMANTIC_KERNEL_USER_AGENT
 
 if TYPE_CHECKING:
     from semantic_kernel.connectors.ai.prompt_execution_settings import PromptExecutionSettings

python/semantic_kernel/connectors/ai/chat_completion_client_base.py

@@ -50,7 +50,7 @@ async def get_chat_message_content(
         Returns:
             A string representing the response from the LLM.
         """
-        results = await self.get_chat_message_contents(chat_history, settings, **kwargs)
+        results = await self.get_chat_message_contents(chat_history=chat_history, settings=settings, **kwargs)
         if results:
             return results[0]
         # this should not happen, should error out before returning an empty list

python/semantic_kernel/connectors/ai/open_ai/services/azure_config_base.py

@@ -9,10 +9,10 @@
 
 from semantic_kernel.connectors.ai.open_ai.const import DEFAULT_AZURE_API_VERSION
 from semantic_kernel.connectors.ai.open_ai.services.open_ai_handler import OpenAIHandler, OpenAIModelTypes
-from semantic_kernel.connectors.telemetry import APP_INFO, prepend_semantic_kernel_to_user_agent
 from semantic_kernel.const import USER_AGENT
 from semantic_kernel.exceptions import ServiceInitializationError
 from semantic_kernel.kernel_pydantic import HttpsUrl
+from semantic_kernel.utils.telemetry.user_agent import APP_INFO, prepend_semantic_kernel_to_user_agent
 
 logger: logging.Logger = logging.getLogger(__name__)
 

python/semantic_kernel/connectors/ai/open_ai/services/open_ai_chat_completion_base.py

@@ -5,7 +5,7 @@
 import sys
 from collections.abc import AsyncGenerator
 from functools import reduce
-from typing import TYPE_CHECKING, Any
+from typing import TYPE_CHECKING, Any, ClassVar
 
 if sys.version_info >= (3, 12):
     from typing import override  # pragma: no cover
@@ -38,6 +38,7 @@
 from semantic_kernel.filters.auto_function_invocation.auto_function_invocation_context import (
     AutoFunctionInvocationContext,
 )
+from semantic_kernel.utils.telemetry.decorators import trace_chat_completion
 
 if TYPE_CHECKING:
     from semantic_kernel.connectors.ai.prompt_execution_settings import PromptExecutionSettings
@@ -56,6 +57,8 @@ class InvokeTermination(Exception):
 class OpenAIChatCompletionBase(OpenAIHandler, ChatCompletionClientBase):
     """OpenAI Chat completion class."""
 
+    MODEL_PROVIDER_NAME: ClassVar[str] = "openai"
+
     # region Overriding base class methods
     # most of the methods are overridden from the ChatCompletionClientBase class, otherwise it is mentioned
 
@@ -64,6 +67,7 @@ def get_prompt_execution_settings_class(self) -> type["PromptExecutionSettings"]
         return OpenAIChatPromptExecutionSettings
 
     @override
+    @trace_chat_completion(MODEL_PROVIDER_NAME)
     async def get_chat_message_contents(
         self,
         chat_history: ChatHistory,

python/semantic_kernel/connectors/ai/open_ai/services/open_ai_config_base.py

@@ -9,9 +9,9 @@
 
 from semantic_kernel.connectors.ai.open_ai.services.open_ai_handler import OpenAIHandler
 from semantic_kernel.connectors.ai.open_ai.services.open_ai_model_types import OpenAIModelTypes
-from semantic_kernel.connectors.telemetry import APP_INFO, prepend_semantic_kernel_to_user_agent
 from semantic_kernel.const import USER_AGENT
 from semantic_kernel.exceptions import ServiceInitializationError
+from semantic_kernel.utils.telemetry.user_agent import APP_INFO, prepend_semantic_kernel_to_user_agent
 
 logger: logging.Logger = logging.getLogger(__name__)
 

python/semantic_kernel/connectors/ai/open_ai/services/open_ai_text_completion_base.py

@@ -3,7 +3,7 @@
 import logging
 import sys
 from collections.abc import AsyncGenerator
-from typing import TYPE_CHECKING, Any
+from typing import TYPE_CHECKING, Any, ClassVar
 
 if sys.version_info >= (3, 12):
     from typing import override  # pragma: no cover
@@ -26,6 +26,7 @@
 from semantic_kernel.connectors.ai.text_completion_client_base import TextCompletionClientBase
 from semantic_kernel.contents.streaming_text_content import StreamingTextContent
 from semantic_kernel.contents.text_content import TextContent
+from semantic_kernel.utils.telemetry.decorators import trace_text_completion
 
 if TYPE_CHECKING:
     from semantic_kernel.connectors.ai.prompt_execution_settings import PromptExecutionSettings
@@ -34,11 +35,14 @@
 
 
 class OpenAITextCompletionBase(OpenAIHandler, TextCompletionClientBase):
+    MODEL_PROVIDER_NAME: ClassVar[str] = "openai"
+
     @override
     def get_prompt_execution_settings_class(self) -> type["PromptExecutionSettings"]:
         return OpenAITextPromptExecutionSettings
 
     @override
+    @trace_text_completion(MODEL_PROVIDER_NAME)
     async def get_text_contents(
         self,
         prompt: str,

python/semantic_kernel/connectors/ai/text_completion_client_base.py

@@ -40,7 +40,7 @@ async def get_text_content(self, prompt: str, settings: "PromptExecutionSettings
         Returns:
             TextContent: A string or list of strings representing the response(s) from the LLM.
         """
-        result = await self.get_text_contents(prompt, settings)
+        result = await self.get_text_contents(prompt=prompt, settings=settings)
         if result:
             return result[0]
         # this should not happen, should error out before returning an empty list

python/semantic_kernel/connectors/memory/astradb/astra_client.py

@@ -5,9 +5,9 @@
 import aiohttp
 
 from semantic_kernel.connectors.memory.astradb.utils import AsyncSession
-from semantic_kernel.connectors.telemetry import APP_INFO
 from semantic_kernel.exceptions import ServiceResponseException
 from semantic_kernel.utils.experimental_decorator import experimental_class
+from semantic_kernel.utils.telemetry.user_agent import APP_INFO
 
 ASTRA_CALLER_IDENTITY: str
 SEMANTIC_KERNEL_VERSION = APP_INFO.get("Semantic-Kernel-Version")

python/semantic_kernel/connectors/openapi_plugin/openapi_function_execution_parameters.py

@@ -28,7 +28,7 @@ class OpenAPIFunctionExecutionParameters(KernelBaseModel):
 
     def model_post_init(self, __context: Any) -> None:
         """Post initialization method for the model."""
-        from semantic_kernel.connectors.telemetry import HTTP_USER_AGENT
+        from semantic_kernel.utils.telemetry.user_agent import HTTP_USER_AGENT
 
         if self.server_url_override:
             parsed_url = urlparse(self.server_url_override)

python/semantic_kernel/connectors/openapi_plugin/openapi_runner.py

@@ -17,10 +17,10 @@
 )
 from semantic_kernel.connectors.openapi_plugin.models.rest_api_operation_payload import RestApiOperationPayload
 from semantic_kernel.connectors.openapi_plugin.models.rest_api_operation_run_options import RestApiOperationRunOptions
-from semantic_kernel.connectors.telemetry import APP_INFO, prepend_semantic_kernel_to_user_agent
 from semantic_kernel.exceptions.function_exceptions import FunctionExecutionException
 from semantic_kernel.functions.kernel_arguments import KernelArguments
 from semantic_kernel.utils.experimental_decorator import experimental_class
+from semantic_kernel.utils.telemetry.user_agent import APP_INFO, prepend_semantic_kernel_to_user_agent
 
 logger: logging.Logger = logging.getLogger(__name__)
 

python/semantic_kernel/connectors/utils/document_loader.py

@@ -6,8 +6,8 @@
 
 from httpx import AsyncClient, HTTPStatusError, RequestError
 
-from semantic_kernel.connectors.telemetry import HTTP_USER_AGENT
 from semantic_kernel.exceptions import ServiceInvalidRequestError
+from semantic_kernel.utils.telemetry.user_agent import HTTP_USER_AGENT
 
 logger: logging.Logger = logging.getLogger(__name__)
 

python/semantic_kernel/core_plugins/sessions_python_tool/sessions_python_plugin.py

@@ -10,7 +10,6 @@
 from httpx import AsyncClient, HTTPStatusError
 from pydantic import ValidationError
 
-from semantic_kernel.connectors.telemetry import HTTP_USER_AGENT, version_info
 from semantic_kernel.const import USER_AGENT
 from semantic_kernel.core_plugins.sessions_python_tool.sessions_python_settings import (
     ACASessionsSettings,
@@ -20,6 +19,7 @@
 from semantic_kernel.exceptions.function_exceptions import FunctionExecutionException, FunctionInitializationError
 from semantic_kernel.functions.kernel_function_decorator import kernel_function
 from semantic_kernel.kernel_pydantic import HttpsUrl, KernelBaseModel
+from semantic_kernel.utils.telemetry.user_agent import HTTP_USER_AGENT, version_info
 
 logger = logging.getLogger(__name__)
 

python/semantic_kernel/functions/kernel_function_from_prompt.py

@@ -187,7 +187,8 @@ async def _invoke_internal(self, context: FunctionInvocationContext) -> None:
         if isinstance(prompt_render_result.ai_service, TextCompletionClientBase):
             try:
                 texts = await prompt_render_result.ai_service.get_text_contents(
-                    unescape(prompt_render_result.rendered_prompt), prompt_render_result.execution_settings
+                    prompt=unescape(prompt_render_result.rendered_prompt),
+                    settings=prompt_render_result.execution_settings,
                 )
             except Exception as exc:
                 raise FunctionExecutionException(f"Error occurred while invoking function {self.name}: {exc}") from exc

python/semantic_kernel/utils/telemetry/const.py

@@ -0,0 +1,28 @@
+# Copyright (c) Microsoft. All rights reserved.
+#
+# Constants for tracing activities with semantic conventions.
+
+# Activity tags
+SYSTEM = "gen_ai.system"
+OPERATION = "gen_ai.operation.name"
+CHAT_COMPLETION_OPERATION = "chat.completions"
+TEXT_COMPLETION_OPERATION = "text.completions"
+MODEL = "gen_ai.request.model"
+MAX_TOKENS = "gen_ai.request.max_tokens"  # nosec
+TEMPERATURE = "gen_ai.request.temperature"
+TOP_P = "gen_ai.request.top_p"
+RESPONSE_ID = "gen_ai.response.id"
+FINISH_REASON = "gen_ai.response.finish_reason"
+PROMPT_TOKENS = "gen_ai.response.prompt_tokens"  # nosec
+COMPLETION_TOKENS = "gen_ai.response.completion_tokens"  # nosec
+ADDRESS = "server.address"
+PORT = "server.port"
+ERROR_TYPE = "error.type"
+
+# Activity events
+PROMPT_EVENT = "gen_ai.content.prompt"
+COMPLETION_EVENT = "gen_ai.content.completion"
+
+# Activity event attributes
+PROMPT_EVENT_PROMPT = "gen_ai.prompt"
+COMPLETION_EVENT_COMPLETION = "gen_ai.completion"