openlit 1.34.30__py3-none-any.whl → 1.34.31__py3-none-any.whl

openlit/instrumentation/transformers/utils.py

@@ -1,6 +1,7 @@
 """
 HF Transformers OpenTelemetry instrumentation utility functions
 """
+
 import time
 
 from opentelemetry.trace import Status, StatusCode
@@ -13,6 +14,7 @@ from openlit.__helpers import (
 )
 from openlit.semcov import SemanticConvention
 
+
 def format_content(content):
     """
     Format content to a consistent structure.
@@ -21,8 +23,12 @@ def format_content(content):
         return content
     elif isinstance(content, list):
         # Check if its a list of chat messages (like in the test case)
-        if (len(content) > 0 and isinstance(content[0], dict) and
-            "role" in content[0] and "content" in content[0]):
+        if (
+            len(content) > 0
+            and isinstance(content[0], dict)
+            and "role" in content[0]
+            and "content" in content[0]
+        ):
             # Handle chat message format like Groq
             formatted_messages = []
             for message in content:
@@ -31,7 +37,7 @@ def format_content(content):
 
                 if isinstance(msg_content, list):
                     content_str = ", ".join(
-                        f'{item["type"]}: {item["text"] if "text" in item else item.get("image_url", str(item))}'
+                        f"{item['type']}: {item['text'] if 'text' in item else item.get('image_url', str(item))}"
                         if isinstance(item, dict) and "type" in item
                         else str(item)
                         for item in msg_content
@@ -58,9 +64,20 @@ def format_content(content):
     else:
         return str(content)
 
-def common_chat_logic(scope, pricing_info, environment, application_name, metrics,
-    capture_message_content, disable_metrics, version, args, kwargs, is_stream):
 
+def common_chat_logic(
+    scope,
+    pricing_info,
+    environment,
+    application_name,
+    metrics,
+    capture_message_content,
+    disable_metrics,
+    version,
+    args,
+    kwargs,
+    is_stream,
+):
     """
     Process chat request and generate Telemetry
     """
@@ -75,31 +92,60 @@ def common_chat_logic(scope, pricing_info, environment, application_name, metric
     cost = get_chat_model_cost(request_model, pricing_info, input_tokens, output_tokens)
 
     # Common Span Attributes
-    common_span_attributes(scope,
-        SemanticConvention.GEN_AI_OPERATION_TYPE_CHAT, SemanticConvention.GEN_AI_SYSTEM_HUGGING_FACE,
-        scope._server_address, scope._server_port, request_model, request_model,
-        environment, application_name, is_stream, scope._tbt, scope._ttft, version)
+    common_span_attributes(
+        scope,
+        SemanticConvention.GEN_AI_OPERATION_TYPE_CHAT,
+        SemanticConvention.GEN_AI_SYSTEM_HUGGING_FACE,
+        scope._server_address,
+        scope._server_port,
+        request_model,
+        request_model,
+        environment,
+        application_name,
+        is_stream,
+        scope._tbt,
+        scope._ttft,
+        version,
+    )
 
     # Set request parameters from forward_params
     if forward_params.get("temperature") is not None:
-        scope._span.set_attribute(SemanticConvention.GEN_AI_REQUEST_TEMPERATURE, forward_params["temperature"])
+        scope._span.set_attribute(
+            SemanticConvention.GEN_AI_REQUEST_TEMPERATURE, forward_params["temperature"]
+        )
     if forward_params.get("top_k") is not None:
-        scope._span.set_attribute(SemanticConvention.GEN_AI_REQUEST_TOP_K, forward_params["top_k"])
+        scope._span.set_attribute(
+            SemanticConvention.GEN_AI_REQUEST_TOP_K, forward_params["top_k"]
+        )
     if forward_params.get("top_p") is not None:
-        scope._span.set_attribute(SemanticConvention.GEN_AI_REQUEST_TOP_P, forward_params["top_p"])
+        scope._span.set_attribute(
+            SemanticConvention.GEN_AI_REQUEST_TOP_P, forward_params["top_p"]
+        )
     if forward_params.get("max_length") is not None:
-        scope._span.set_attribute(SemanticConvention.GEN_AI_REQUEST_MAX_TOKENS, forward_params["max_length"])
+        scope._span.set_attribute(
+            SemanticConvention.GEN_AI_REQUEST_MAX_TOKENS, forward_params["max_length"]
+        )
 
     # Set token usage and cost attributes
-    scope._span.set_attribute(SemanticConvention.GEN_AI_USAGE_INPUT_TOKENS, input_tokens)
-    scope._span.set_attribute(SemanticConvention.GEN_AI_USAGE_OUTPUT_TOKENS, output_tokens)
-    scope._span.set_attribute(SemanticConvention.GEN_AI_CLIENT_TOKEN_USAGE, input_tokens + output_tokens)
+    scope._span.set_attribute(
+        SemanticConvention.GEN_AI_USAGE_INPUT_TOKENS, input_tokens
+    )
+    scope._span.set_attribute(
+        SemanticConvention.GEN_AI_USAGE_OUTPUT_TOKENS, output_tokens
+    )
+    scope._span.set_attribute(
+        SemanticConvention.GEN_AI_CLIENT_TOKEN_USAGE, input_tokens + output_tokens
+    )
     scope._span.set_attribute(SemanticConvention.GEN_AI_USAGE_COST, cost)
 
     # Span Attributes for Content
     if capture_message_content:
-        scope._span.set_attribute(SemanticConvention.GEN_AI_CONTENT_PROMPT, scope._prompt)
-        scope._span.set_attribute(SemanticConvention.GEN_AI_CONTENT_COMPLETION, scope._completion)
+        scope._span.set_attribute(
+            SemanticConvention.GEN_AI_CONTENT_PROMPT, scope._prompt
+        )
+        scope._span.set_attribute(
+            SemanticConvention.GEN_AI_CONTENT_COMPLETION, scope._completion
+        )
 
         # To be removed once the change to span_attributes (from span events) is complete
         scope._span.add_event(
@@ -119,14 +165,44 @@ def common_chat_logic(scope, pricing_info, environment, application_name, metric
 
     # Record metrics using the standardized helper function
     if not disable_metrics:
-        record_completion_metrics(metrics, SemanticConvention.GEN_AI_OPERATION_TYPE_CHAT,
-            SemanticConvention.GEN_AI_SYSTEM_HUGGING_FACE, scope._server_address, scope._server_port,
-            request_model, request_model, environment, application_name, scope._start_time, scope._end_time,
-            cost, input_tokens, output_tokens, scope._tbt, scope._ttft)
-
-def process_chat_response(instance, response, request_model, pricing_info, server_port, server_address,
-    environment, application_name, metrics, start_time,
-    span, args, kwargs, capture_message_content=False, disable_metrics=False, version="1.0.0"):
+        record_completion_metrics(
+            metrics,
+            SemanticConvention.GEN_AI_OPERATION_TYPE_CHAT,
+            SemanticConvention.GEN_AI_SYSTEM_HUGGING_FACE,
+            scope._server_address,
+            scope._server_port,
+            request_model,
+            request_model,
+            environment,
+            application_name,
+            scope._start_time,
+            scope._end_time,
+            cost,
+            input_tokens,
+            output_tokens,
+            scope._tbt,
+            scope._ttft,
+        )
+
+
+def process_chat_response(
+    instance,
+    response,
+    request_model,
+    pricing_info,
+    server_port,
+    server_address,
+    environment,
+    application_name,
+    metrics,
+    start_time,
+    span,
+    args,
+    kwargs,
+    capture_message_content=False,
+    disable_metrics=False,
+    version="1.0.0",
+):
     """
     Process chat request and generate Telemetry
     """
@@ -146,11 +222,19 @@ def process_chat_response(instance, response, request_model, pricing_info, serve
         scope._prompt = args[0]
     else:
         scope._prompt = (
-            kwargs.get("text_inputs") or
-            (kwargs.get("image") and kwargs.get("question") and
-             ("image: " + kwargs.get("image") + " question:" + kwargs.get("question"))) or
-            kwargs.get("fallback") or
-            ""
+            kwargs.get("text_inputs")
+            or (
+                kwargs.get("image")
+                and kwargs.get("question")
+                and (
+                    "image: "
+                    + kwargs.get("image")
+                    + " question:"
+                    + kwargs.get("question")
+                )
+            )
+            or kwargs.get("fallback")
+            or ""
         )
     scope._prompt = format_content(scope._prompt)
 
@@ -175,13 +259,21 @@ def process_chat_response(instance, response, request_model, pricing_info, serve
             scope._completion = ""
 
     elif task == "automatic-speech-recognition":
-        scope._completion = response.get("text", "") if isinstance(response, dict) else ""
+        scope._completion = (
+            response.get("text", "") if isinstance(response, dict) else ""
+        )
 
     elif task == "image-classification":
-        scope._completion = str(response[0]) if isinstance(response, list) and len(response) > 0 else ""
+        scope._completion = (
+            str(response[0]) if isinstance(response, list) and len(response) > 0 else ""
+        )
 
     elif task == "visual-question-answering":
-        if isinstance(response, list) and len(response) > 0 and isinstance(response[0], dict):
+        if (
+            isinstance(response, list)
+            and len(response) > 0
+            and isinstance(response[0], dict)
+        ):
             scope._completion = response[0].get("answer", "")
         else:
             scope._completion = ""
@@ -193,7 +285,18 @@ def process_chat_response(instance, response, request_model, pricing_info, serve
     scope._tbt = 0
     scope._ttft = scope._end_time - scope._start_time
 
-    common_chat_logic(scope, pricing_info, environment, application_name, metrics,
-                     capture_message_content, disable_metrics, version, args, kwargs, is_stream=False)
+    common_chat_logic(
+        scope,
+        pricing_info,
+        environment,
+        application_name,
+        metrics,
+        capture_message_content,
+        disable_metrics,
+        version,
+        args,
+        kwargs,
+        is_stream=False,
+    )
 
     return response

openlit/instrumentation/vertexai/__init__.py

@@ -10,6 +10,7 @@ from openlit.instrumentation.vertexai.async_vertexai import async_send_message
 
 _instruments = ("google-cloud-aiplatform >= 1.38.1",)
 
+
 class VertexAIInstrumentor(BaseInstrumentor):
     """
     An instrumentor for VertexAI client library.
@@ -32,60 +33,124 @@ class VertexAIInstrumentor(BaseInstrumentor):
         wrap_function_wrapper(
             "vertexai.generative_models",
             "GenerativeModel.generate_content",
-            send_message(version, environment, application_name,
-                tracer, pricing_info, capture_message_content, metrics, disable_metrics),
+            send_message(
+                version,
+                environment,
+                application_name,
+                tracer,
+                pricing_info,
+                capture_message_content,
+                metrics,
+                disable_metrics,
+            ),
         )
 
         wrap_function_wrapper(
             "vertexai.generative_models",
             "ChatSession.send_message",
-            send_message(version, environment, application_name,
-                tracer, pricing_info, capture_message_content, metrics, disable_metrics),
+            send_message(
+                version,
+                environment,
+                application_name,
+                tracer,
+                pricing_info,
+                capture_message_content,
+                metrics,
+                disable_metrics,
+            ),
         )
 
         # sync language models
         wrap_function_wrapper(
             "vertexai.language_models",
             "ChatSession.send_message",
-            send_message(version, environment, application_name,
-                tracer, pricing_info, capture_message_content, metrics, disable_metrics),
+            send_message(
+                version,
+                environment,
+                application_name,
+                tracer,
+                pricing_info,
+                capture_message_content,
+                metrics,
+                disable_metrics,
+            ),
         )
 
         wrap_function_wrapper(
             "vertexai.language_models",
             "ChatSession.send_message_streaming",
-            send_message(version, environment, application_name,
-                tracer, pricing_info, capture_message_content, metrics, disable_metrics),
+            send_message(
+                version,
+                environment,
+                application_name,
+                tracer,
+                pricing_info,
+                capture_message_content,
+                metrics,
+                disable_metrics,
+            ),
         )
 
         # async generative models
         wrap_function_wrapper(
             "vertexai.generative_models",
             "GenerativeModel.generate_content_async",
-            async_send_message(version, environment, application_name,
-                tracer, pricing_info, capture_message_content, metrics, disable_metrics),
+            async_send_message(
+                version,
+                environment,
+                application_name,
+                tracer,
+                pricing_info,
+                capture_message_content,
+                metrics,
+                disable_metrics,
+            ),
         )
 
         wrap_function_wrapper(
             "vertexai.generative_models",
             "ChatSession.send_message_async",
-            async_send_message(version, environment, application_name,
-                tracer, pricing_info, capture_message_content, metrics, disable_metrics),
+            async_send_message(
+                version,
+                environment,
+                application_name,
+                tracer,
+                pricing_info,
+                capture_message_content,
+                metrics,
+                disable_metrics,
+            ),
         )
 
         # async language models
         wrap_function_wrapper(
             "vertexai.language_models",
             "ChatSession.send_message_async",
-            async_send_message(version, environment, application_name,
-                tracer, pricing_info, capture_message_content, metrics, disable_metrics),
+            async_send_message(
+                version,
+                environment,
+                application_name,
+                tracer,
+                pricing_info,
+                capture_message_content,
+                metrics,
+                disable_metrics,
+            ),
         )
 
         wrap_function_wrapper(
             "vertexai.language_models",
             "ChatSession.send_message_streaming_async",
-            async_send_message(version, environment, application_name,
-                tracer, pricing_info, capture_message_content, metrics, disable_metrics),
+            async_send_message(
+                version,
+                environment,
+                application_name,
+                tracer,
+                pricing_info,
+                capture_message_content,
+                metrics,
+                disable_metrics,
+            ),
         )
 
     def _uninstrument(self, **kwargs):

openlit/instrumentation/vertexai/async_vertexai.py

@@ -19,8 +19,17 @@ from openlit.semcov import SemanticConvention
 # Initialize logger for logging potential issues and operations
 logger = logging.getLogger(__name__)
 
-def async_send_message(version, environment, application_name, tracer,
-             pricing_info, capture_message_content, metrics, disable_metrics):
+
+def async_send_message(
+    version,
+    environment,
+    application_name,
+    tracer,
+    pricing_info,
+    capture_message_content,
+    metrics,
+    disable_metrics,
+):
     """
     Generates a telemetry wrapper for VertexAI AsyncMessages calls.
     """
@@ -31,16 +40,16 @@ def async_send_message(version, environment, application_name, tracer,
         """
 
         def __init__(
-                self,
-                wrapped,
-                span,
-                span_name,
-                kwargs,
-                server_address,
-                server_port,
-                request_model,
-                args,
-            ):
+            self,
+            wrapped,
+            span,
+            span_name,
+            kwargs,
+            server_address,
+            server_port,
+            request_model,
+            args,
+        ):
             self.__wrapped__ = wrapped
             self._span = span
             self._span_name = span_name
@@ -88,7 +97,7 @@ def async_send_message(version, environment, application_name, tracer,
                             metrics=metrics,
                             capture_message_content=capture_message_content,
                             disable_metrics=disable_metrics,
-                            version=version
+                            version=version,
                         )
                 except Exception as e:
                     handle_exception(self._span, e)
@@ -108,7 +117,16 @@ def async_send_message(version, environment, application_name, tracer,
             awaited_wrapped = await wrapped(*args, **kwargs)
             span = tracer.start_span(span_name, kind=SpanKind.CLIENT)
 
-            return TracedAsyncStream(awaited_wrapped, span, span_name, kwargs, server_address, server_port, request_model, args)
+            return TracedAsyncStream(
+                awaited_wrapped,
+                span,
+                span_name,
+                kwargs,
+                server_address,
+                server_port,
+                request_model,
+                args,
+            )
 
         else:
             with tracer.start_as_current_span(span_name, kind=SpanKind.CLIENT) as span:
@@ -130,7 +148,7 @@ def async_send_message(version, environment, application_name, tracer,
                         capture_message_content=capture_message_content,
                         disable_metrics=disable_metrics,
                         version=version,
-                        **kwargs
+                        **kwargs,
                     )
 
                 except Exception as e: