openlit 1.33.11__py3-none-any.whl → 1.33.12__py3-none-any.whl

openlit/instrumentation/anthropic/utils.py

@@ -0,0 +1,251 @@
+"""
+Anthropic OpenTelemetry instrumentation utility functions
+"""
+import time
+
+from opentelemetry.sdk.resources import SERVICE_NAME, TELEMETRY_SDK_NAME, DEPLOYMENT_ENVIRONMENT
+from opentelemetry.trace import Status, StatusCode
+
+from openlit.__helpers import (
+    calculate_ttft,
+    response_as_dict,
+    calculate_tbt,
+    extract_and_format_input,
+    get_chat_model_cost,
+    create_metrics_attributes,
+    otel_event,
+    concatenate_all_contents
+)
+from openlit.semcov import SemanticConvetion
+
+def process_chunk(self, chunk):
+    """
+    Process a chunk of response data and update state.
+    """
+
+    end_time = time.time()
+    # Record the timestamp for the current chunk
+    self._timestamps.append(end_time)
+
+    if len(self._timestamps) == 1:
+        # Calculate time to first chunk
+        self._ttft = calculate_ttft(self._timestamps, self._start_time)
+
+    chunked = response_as_dict(chunk)
+
+    # Collect message IDs and input token from events
+    if chunked.get('type') == 'message_start':
+        self._response_id = chunked.get('message').get('id')
+        self._input_tokens = chunked.get('message').get('usage').get('input_tokens')
+        self._response_model = chunked.get('message').get('model')
+        self._response_role = chunked.get('message').get('role')
+
+    # Collect message IDs and aggregated response from events
+    if chunked.get('type') == 'content_block_delta':
+        if chunked.get('delta').get('text'):
+            self._llmresponse += chunked.get('delta').get('text')
+        elif chunked.get('delta').get('partial_json'):
+            self._tool_arguments += chunked.get('delta').get('partial_json')
+
+    if chunked.get('type') == 'content_block_start':
+        if chunked.get('content_block').get('id'):
+            self._tool_id = chunked.get('content_block').get('id')
+        if chunked.get('content_block').get('name'):
+            self._tool_name = chunked.get('content_block').get('name')
+
+    # Collect output tokens and stop reason from events
+    if chunked.get('type') == 'message_delta':
+        self._output_tokens = chunked.get('usage').get('output_tokens')
+        self._finish_reason = chunked.get('delta').get('stop_reason')
+
+def common_chat_logic(scope, pricing_info, environment, application_name, metrics,
+                        event_provider, capture_message_content, disable_metrics, version, is_stream):
+    """
+    Process chat request and generate Telemetry
+    """
+
+    scope._end_time = time.time()
+    if len(scope._timestamps) > 1:
+        scope._tbt = calculate_tbt(scope._timestamps)
+
+    formatted_messages = extract_and_format_input(scope._kwargs.get('messages', ''))
+    request_model = scope._kwargs.get('model', 'claude-3-opus-20240229')
+
+    cost = get_chat_model_cost(request_model, pricing_info, scope._input_tokens, scope._output_tokens)
+
+    # Set Span attributes (OTel Semconv)
+    scope._span.set_attribute(TELEMETRY_SDK_NAME, 'openlit')
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_OPERATION, SemanticConvetion.GEN_AI_OPERATION_TYPE_CHAT)
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_SYSTEM, SemanticConvetion.GEN_AI_SYSTEM_ANTHROPIC)
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_MODEL, request_model)
+    scope._span.set_attribute(SemanticConvetion.SERVER_PORT, scope._server_port)
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_MAX_TOKENS, scope._kwargs.get('max_tokens', -1))
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_STOP_SEQUENCES, scope._kwargs.get('stop_sequences', []))
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_TEMPERATURE, scope._kwargs.get('temperature', 1.0))
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_TOP_K, scope._kwargs.get('top_k', 1.0))
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_TOP_P, scope._kwargs.get('top_p', 1.0))
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_RESPONSE_FINISH_REASON, [scope._finish_reason])
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_RESPONSE_ID, scope._response_id)
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_RESPONSE_MODEL, scope._response_model)
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_USAGE_INPUT_TOKENS, scope._input_tokens)
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_USAGE_OUTPUT_TOKENS, scope._output_tokens)
+    scope._span.set_attribute(SemanticConvetion.SERVER_ADDRESS, scope._server_address)
+
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_OUTPUT_TYPE,
+                              'text' if isinstance(scope._llmresponse, str) else 'json')
+
+    scope._span.set_attribute(DEPLOYMENT_ENVIRONMENT, environment)
+    scope._span.set_attribute(SERVICE_NAME, application_name)
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_REQUEST_IS_STREAM, is_stream)
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_CLIENT_TOKEN_USAGE, scope._input_tokens + scope._output_tokens)
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_USAGE_COST, cost)
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_SERVER_TBT, scope._tbt)
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_SERVER_TTFT, scope._ttft)
+    scope._span.set_attribute(SemanticConvetion.GEN_AI_SDK_VERSION, version)
+
+    # To be removed one the change to log events (from span events) is complete
+    prompt = concatenate_all_contents(formatted_messages)
+    if capture_message_content:
+        scope._span.add_event(
+            name=SemanticConvetion.GEN_AI_CONTENT_PROMPT_EVENT,
+            attributes={
+                SemanticConvetion.GEN_AI_CONTENT_PROMPT: prompt,
+            },
+        )
+        scope._span.add_event(
+            name=SemanticConvetion.GEN_AI_CONTENT_COMPLETION_EVENT,
+            attributes={
+                SemanticConvetion.GEN_AI_CONTENT_COMPLETION: scope._llmresponse,
+            },
+        )
+
+    choice_event_body = {
+        'finish_reason': scope._finish_reason,
+        'index': 0,
+        'message': {
+            **({'content': scope._llmresponse} if capture_message_content else {}),
+            'role': scope._response_role
+        }
+    }
+
+    if scope._tool_calls:
+        choice_event_body['message'].update({
+            'tool_calls': {
+                'function': {
+                    'name': scope._tool_calls.get('name', ''),
+                    'arguments': scope._tool_calls.get('input', '')
+                },
+                'id': scope._tool_calls.get('id', ''),
+                'type': 'function'
+            }
+        })
+
+    # Emit events
+    for role in ['user', 'system', 'assistant', 'tool']:
+        if formatted_messages.get(role, {}).get('content', ''):
+            event = otel_event(
+                name=getattr(SemanticConvetion, f'GEN_AI_{role.upper()}_MESSAGE'),
+                attributes={
+                    SemanticConvetion.GEN_AI_SYSTEM: SemanticConvetion.GEN_AI_SYSTEM_ANTHROPIC
+                },
+                body = {
+                    # pylint: disable=line-too-long
+                    **({'content': formatted_messages.get(role, {}).get('content', '')} if capture_message_content else {}),
+                    'role': formatted_messages.get(role, {}).get('role', []),
+                    **({
+                        'tool_calls': {
+                            'function': {
+                                # pylint: disable=line-too-long
+                                'name': (scope._tool_calls[0].get('function', {}).get('name', '') if scope._tool_calls else ''),
+                                'arguments': (scope._tool_calls[0].get('function', {}).get('arguments', '') if scope._tool_calls else '')
+                            },
+                            'id': (scope._tool_calls[0].get('id', '') if scope._tool_calls else ''),
+                            'type': 'function'
+                        }
+                    } if role == 'assistant' else {}),
+                    **({
+                        'id': (scope._tool_calls[0].get('id', '') if scope._tool_calls else '')
+                    } if role == 'tool' else {})
+                }
+            )
+            event_provider.emit(event)
+
+    choice_event = otel_event(
+        name=SemanticConvetion.GEN_AI_CHOICE,
+        attributes={
+            SemanticConvetion.GEN_AI_SYSTEM: SemanticConvetion.GEN_AI_SYSTEM_ANTHROPIC
+        },
+        body=choice_event_body
+    )
+    event_provider.emit(choice_event)
+
+    scope._span.set_status(Status(StatusCode.OK))
+
+    if not disable_metrics:
+        metrics_attributes = create_metrics_attributes(
+            service_name=application_name,
+            deployment_environment=environment,
+            operation=SemanticConvetion.GEN_AI_OPERATION_TYPE_CHAT,
+            system=SemanticConvetion.GEN_AI_SYSTEM_ANTHROPIC,
+            request_model=request_model,
+            server_address=scope._server_address,
+            server_port=scope._server_port,
+            response_model=scope._response_model,
+        )
+
+        metrics['genai_client_usage_tokens'].record(scope._input_tokens + scope._output_tokens, metrics_attributes)
+        metrics['genai_client_operation_duration'].record(scope._end_time - scope._start_time, metrics_attributes)
+        metrics['genai_server_tbt'].record(scope._tbt, metrics_attributes)
+        metrics['genai_server_ttft'].record(scope._ttft, metrics_attributes)
+        metrics['genai_requests'].add(1, metrics_attributes)
+        metrics['genai_completion_tokens'].add(scope._output_tokens, metrics_attributes)
+        metrics['genai_prompt_tokens'].add(scope._input_tokens, metrics_attributes)
+        metrics['genai_cost'].record(cost, metrics_attributes)
+
+def process_streaming_chat_response(self, pricing_info, environment, application_name, metrics,
+                                    event_provider, capture_message_content=False, disable_metrics=False, version=''):
+    """
+    Process chat request and generate Telemetry
+    """
+    if self._tool_id != '':
+        self._tool_calls = {
+            'id': self._tool_id,
+            'name': self._tool_name,
+            'input': self._tool_arguments
+        }
+
+    common_chat_logic(self, pricing_info, environment, application_name, metrics,
+                        event_provider, capture_message_content, disable_metrics, version, is_stream=True)
+
+def process_chat_response(response, request_model, pricing_info, server_port, server_address,
+                          environment, application_name, metrics, event_provider, start_time,
+                          span, capture_message_content=False, disable_metrics=False, version='1.0.0', **kwargs):
+    """
+    Process chat request and generate Telemetry
+    """
+
+    self = type('GenericScope', (), {})()
+    response_dict = response_as_dict(response)
+
+    # pylint: disable = no-member
+    self._start_time = start_time
+    self._end_time = time.time()
+    self._span = span
+    self._llmresponse = response_dict.get('content', {})[0].get('text', '')
+    self._response_role = response_dict.get('message', {}).get('role', 'assistant')
+    self._input_tokens = response_dict.get('usage').get('input_tokens')
+    self._output_tokens = response_dict.get('usage').get('output_tokens')
+    self._response_model = response_dict.get('model', '')
+    self._finish_reason = response_dict.get('stop_reason', '')
+    self._response_id = response_dict.get('id', '')
+    self._timestamps = []
+    self._ttft, self._tbt = self._end_time - self._start_time, 0
+    self._server_address, self._server_port = server_address, server_port
+    self._kwargs = kwargs
+    #pylint: disable=line-too-long
+    self._tool_calls = (lambda c: c[1] if len(c) > 1 and c[1].get('type') == 'tool_use' else None)(response_dict.get('content', []))
+
+    common_chat_logic(self, pricing_info, environment, application_name, metrics,
+                        event_provider, capture_message_content, disable_metrics, version, is_stream=False)
+
+    return response

openlit/instrumentation/ollama/utils.py

@@ -129,7 +129,6 @@ def common_chat_logic(scope, pricing_info, environment, application_name, metric
     }
 
     if scope._tool_calls:
-        print(scope._tool_calls)
         function_call = scope._tool_calls[0]
         choice_event_body["message"].update({
             "tool_calls": {

{openlit-1.33.11.dist-info → openlit-1.33.12.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: openlit
-Version: 1.33.11
+Version: 1.33.12
 Summary: OpenTelemetry-native Auto instrumentation library for monitoring LLM Applications and GPUs, facilitating the integration of observability into your GenAI-driven projects
 License: Apache-2.0
 Keywords: OpenTelemetry,otel,otlp,llm,tracing,openai,anthropic,claude,cohere,llm monitoring,observability,monitoring,gpt,Generative AI,chatGPT,gpu

{openlit-1.33.11.dist-info → openlit-1.33.12.dist-info}/RECORD

@@ -1,4 +1,4 @@
-openlit/__helpers.py,sha256=mu3guF9gH4GEkXhKj3ufhqKo-oitkENY-gXGTqFpdEQ,12281
+openlit/__helpers.py,sha256=TBFkzFFAOWLy2YjMOU--VMtQGb9kspSiJRwk4Twts3g,10675
 openlit/__init__.py,sha256=87uE_wi6YF2FKkmVy0VLmRKeNMuJ9e6XrycIbad9T6A,23755
 openlit/evals/__init__.py,sha256=nJe99nuLo1b5rf7pt9U9BCdSDedzbVi2Fj96cgl7msM,380
 openlit/evals/all.py,sha256=oWrue3PotE-rB5WePG3MRYSA-ro6WivkclSHjYlAqGs,7154
@@ -12,15 +12,16 @@ openlit/guard/prompt_injection.py,sha256=3e4DKxB7QDzM-xPCpwEuureiH_2s_OTJ9BSckkn
 openlit/guard/restrict_topic.py,sha256=KTuWa7XeMsV4oXxOrD1CYZV0wXWxTfA0H3p_6q_IOsk,6444
 openlit/guard/sensitive_topic.py,sha256=RgVw_laFERv0nNdzBsAd2_3yLomMOK-gVq-P7oj1bTk,5552
 openlit/guard/utils.py,sha256=x0-_hAtNa_ogYR2GfnwiBF1rlqaXtaJ-rJeGguTDe-Q,7663
-openlit/instrumentation/ag2/__init__.py,sha256=3CTrFvM4PyD8fdg456CNKHz4uMYUf2PVsp_L-3ovSHs,1839
-openlit/instrumentation/ag2/ag2.py,sha256=o6wuvcUbboK2__RRpfmAqJ6w-1mjW5YHiaTI8xV5S_o,6908
-openlit/instrumentation/ai21/__init__.py,sha256=uEAd9wWjEc9aR3IRAGciJqLD50eqvpnL6aTcZqQQIWg,2727
-openlit/instrumentation/ai21/ai21.py,sha256=VmChUo-5V6qn2h4cxLPGPfEZhuYazcUj7dIfHWzapFI,6869
-openlit/instrumentation/ai21/async_ai21.py,sha256=G7AesXLU4rB1t-KJsiZ65aCF71c4LRpAhday2tvF9dk,6973
-openlit/instrumentation/ai21/utils.py,sha256=75Yy1Bi0qi1EqRsvxbiEo0qgoWr2w_UAVkPPCWlDVV0,19566
-openlit/instrumentation/anthropic/__init__.py,sha256=d0NVazGdZRQ-cI0SRErP5OAoe1mUY9q0xhD9WqEUPFQ,1927
-openlit/instrumentation/anthropic/anthropic.py,sha256=bpxHFyfCa50GgNUIWkghwcQ3vK7pEUhPv1NFPhvoaEI,22372
-openlit/instrumentation/anthropic/async_anthropic.py,sha256=6EzI3SM7dqEfCD3vZgyOGHcMOyeBrEoE7VwOI5O-A8s,22458
+openlit/instrumentation/ag2/__init__.py,sha256=KgyLJBmwAxRWu7Z0S8FDDK4TZ13EFoAAIalvG5Oq4wc,1839
+openlit/instrumentation/ag2/ag2.py,sha256=TFqYLyHxGjXMR4s8xoBdysTjZaQYZd8tbf_srw5RYUw,6908
+openlit/instrumentation/ai21/__init__.py,sha256=QXMByKCUhFITUIwUR01m0Fjpr20txV_GWcRJ66dTu_Q,2703
+openlit/instrumentation/ai21/ai21.py,sha256=j18n7X1gL15aZlYvqd64r_6hjhiPg7KHyQwQwzyYutg,6811
+openlit/instrumentation/ai21/async_ai21.py,sha256=Qn4Z0M2TZRN4fkgcy3D4DwIUAqYj_1HKztapVbgf3kw,6915
+openlit/instrumentation/ai21/utils.py,sha256=nLN4VuczcP6x0mq7tAgBkVxjt1ZY8hFMZ1KFUdayQx0,19705
+openlit/instrumentation/anthropic/__init__.py,sha256=QEsiwdxcQDzzlVYR4_x7KTdf0-UJDJt8FjwNQMspnxM,1929
+openlit/instrumentation/anthropic/anthropic.py,sha256=mZbNepBzOtQBCGJCDWaVTpXukii6jjt_fgSOY1RfMoU,5147
+openlit/instrumentation/anthropic/async_anthropic.py,sha256=DBHFdGKBhnI21Tuh_430kc-Zjaq54yUvfqmCmb1EQqg,5233
+openlit/instrumentation/anthropic/utils.py,sha256=nfIOV7eTtAReNN5QN0oG6NaqftulGjbgMrALOaDHQ8A,11897
 openlit/instrumentation/assemblyai/__init__.py,sha256=AS6tEzcyEG7RP6bNDW_Kf4_rN-u-O1BNjJ3khX3AEUs,1565
 openlit/instrumentation/assemblyai/assemblyai.py,sha256=18AQ7wrigCZd9RXwNZ36mn9fc3M3p2WkAHlT_sD5M3c,6292
 openlit/instrumentation/astra/__init__.py,sha256=mEHT_p4q3bl9IiXAFqutGEvw37k6CCuWbmlI9Gx3xKY,8314
@@ -89,7 +90,7 @@ openlit/instrumentation/multion/multion.py,sha256=SrO25cv7dob1H4r00MsAa13Bj-WNsE
 openlit/instrumentation/ollama/__init__.py,sha256=JjxSqEegmRoRqIVz7ZAq9dLyXPZ2DqV2wGmgXCENNpw,3004
 openlit/instrumentation/ollama/async_ollama.py,sha256=LhDQPy3wLyNO9JWksUEeCx-DK9oIV3K98Cgwnp4RfKg,6538
 openlit/instrumentation/ollama/ollama.py,sha256=wVyaX0quoiiCj1J3tyTiQx5Du5CmaWmt9e_lpCr7s6A,6434
-openlit/instrumentation/ollama/utils.py,sha256=4QrOtBHHxfDY0iZbymk149-fGwv8rvmqGQOJD73t5LY,14582
+openlit/instrumentation/ollama/utils.py,sha256=zXsWNqfnZLssrcb-GNbWeZeqTKVzQb1bes8vzgl-gbQ,14549
 openlit/instrumentation/openai/__init__.py,sha256=dfgMBHd2wAT24uckVBBqTy7pzN34ESzeymKzkUy6t58,4893
 openlit/instrumentation/openai/async_openai.py,sha256=8-I6SMT1bFxNGLf2GW1Yfpny-lCvIcR5coKvytKL_yE,50635
 openlit/instrumentation/openai/openai.py,sha256=t1-9KY1CCqeRno2_gXd3PxFkWm1ySwFweno4st2T3bE,50469
@@ -119,7 +120,7 @@ openlit/otel/events.py,sha256=VrMjTpvnLtYRBHCiFwJojTQqqNpRCxoD4yJYeQrtPsk,3560
 openlit/otel/metrics.py,sha256=URL7gzQbnxaNQJSX7oHRa15v6xi1GFmANn-5uFNL-aY,6378
 openlit/otel/tracing.py,sha256=tjV2bEbEDPUB1Z46gE-UsJsb04sRdFrfbhIDkxViZc0,3103
 openlit/semcov/__init__.py,sha256=kUd-ZSmXkXBo-osVve4ce_XEgr0fgEN7nXxoNm7kfEQ,12798
-openlit-1.33.11.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-openlit-1.33.11.dist-info/METADATA,sha256=61sZfrJudmaPtlHZeZF0a5uCx0P41aBenq-2-Rqmlak,23512
-openlit-1.33.11.dist-info/WHEEL,sha256=XbeZDeTWKc1w7CSIyre5aMDU_-PohRwTQceYnisIYYY,88
-openlit-1.33.11.dist-info/RECORD,,
+openlit-1.33.12.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+openlit-1.33.12.dist-info/METADATA,sha256=t2L1uO5RcNpbAuY5qKha7WQWoB6TCT6L0sc0jG1K5cY,23512
+openlit-1.33.12.dist-info/WHEEL,sha256=XbeZDeTWKc1w7CSIyre5aMDU_-PohRwTQceYnisIYYY,88
+openlit-1.33.12.dist-info/RECORD,,