Remove embeddings capture via events

Author: drewby

instrumentation-genai/opentelemetry-instrumentation-openai-v2/src/opentelemetry/instrumentation/openai_v2/patch.py

@@ -19,9 +19,6 @@
 from openai import Stream
 
 from opentelemetry._events import Event, EventLogger
-from opentelemetry.semconv._incubating.attributes import (
-    event_attributes as EventAttributes,
-)
 from opentelemetry.semconv._incubating.attributes import (
     gen_ai_attributes as GenAIAttributes,
 )
@@ -438,50 +435,6 @@ def _set_embeddings_response_attributes(
         )
         # Don't set output tokens for embeddings as all tokens are input tokens
 
-    # Emit events for embeddings if content capture is enabled
-    if capture_content:
-        # Emit input event
-        input_event_attributes = {
-            GenAIAttributes.GEN_AI_SYSTEM: GenAIAttributes.GenAiSystemValues.OPENAI.value,
-            EventAttributes.EVENT_NAME: "gen_ai.embeddings.input",
-        }
-        event_logger.emit(
-            Event(
-                name="gen_ai.embeddings.input",
-                attributes=input_event_attributes,
-                body={"content": input_text, "role": "user"},
-            )
-        )
-
-        # Emit output event with embeddings data
-        if getattr(result, "data", None) and len(result.data) > 0:
-            embedding_data = []
-            for item in result.data:
-                if getattr(item, "embedding", None):
-                    embedding_data.append(
-                        {
-                            "index": item.index
-                            if hasattr(item, "index")
-                            else None,
-                            "embedding": item.embedding,
-                            "object": item.object
-                            if hasattr(item, "object")
-                            else "embedding",
-                        }
-                    )
-
-            output_event_attributes = {
-                GenAIAttributes.GEN_AI_SYSTEM: GenAIAttributes.GenAiSystemValues.OPENAI.value,
-                EventAttributes.EVENT_NAME: "gen_ai.embeddings.output",
-            }
-            event_logger.emit(
-                Event(
-                    name="gen_ai.embeddings.output",
-                    attributes=output_event_attributes,
-                    body={"embeddings": embedding_data},
-                )
-            )
-
 
 class ToolCallBuffer:
     def __init__(self, index, tool_call_id, function_name):

instrumentation-genai/opentelemetry-instrumentation-openai-v2/tests/test_async_embeddings.py

@@ -35,71 +35,6 @@
 from opentelemetry.semconv._incubating.metrics import gen_ai_metrics
 
 
-@pytest.mark.asyncio
-@pytest.mark.vcr()
-async def test_async_embeddings_with_content(
-    span_exporter, log_exporter, async_openai_client, instrument_with_content
-):
-    """Test creating embeddings asynchronously with content capture enabled"""
-    model_name = "text-embedding-3-small"
-    input_text = "This is a test for async embeddings"
-
-    response = await async_openai_client.embeddings.create(
-        model=model_name,
-        input=input_text,
-    )
-
-    # Verify spans
-    spans = span_exporter.get_finished_spans()
-    assert len(spans) == 1
-    assert_embedding_attributes(spans[0], model_name, response)
-
-    # Verify logs
-    logs = log_exporter.get_finished_logs()
-    assert (
-        len(logs) == 2
-    )  # Should contain both input and output embeddings events
-
-    # Find input and output events
-    input_event = None
-    output_event = None
-    for log in logs:
-        if (
-            log.log_record.attributes.get(EventAttributes.EVENT_NAME)
-            == "gen_ai.embeddings.input"
-        ):
-            input_event = log
-        elif (
-            log.log_record.attributes.get(EventAttributes.EVENT_NAME)
-            == "gen_ai.embeddings.output"
-        ):
-            output_event = log
-
-    # Verify both events exist
-    assert input_event is not None, "Input embeddings event not found"
-    assert output_event is not None, "Output embeddings event not found"
-
-    # Verify input event content
-    input_content = {"content": input_text, "role": "user"}
-    assert_message_in_logs(
-        input_event, "gen_ai.embeddings.input", input_content, spans[0]
-    )
-
-    # Verify output event content
-    output_content = {
-        "embeddings": [
-            {
-                "index": 0,
-                "embedding": response.data[0].embedding,
-                "object": "embedding",
-            }
-        ]
-    }
-    assert_message_in_logs(
-        output_event, "gen_ai.embeddings.output", output_content, spans[0]
-    )
-
-
 @pytest.mark.asyncio
 @pytest.mark.vcr()
 async def test_async_embeddings_no_content(

instrumentation-genai/opentelemetry-instrumentation-openai-v2/tests/test_embeddings.py

@@ -35,71 +35,6 @@
 from opentelemetry.semconv._incubating.metrics import gen_ai_metrics
 
 
-@pytest.mark.vcr()
-def test_embeddings_with_content(
-    span_exporter, log_exporter, openai_client, instrument_with_content
-):
-    """Test creating embeddings with content capture enabled"""
-    model_name = "text-embedding-3-small"
-    input_text = "This is a test for embeddings"
-
-    response = openai_client.embeddings.create(
-        model=model_name,
-        input=input_text,
-    )
-
-    # Verify spans
-    spans = span_exporter.get_finished_spans()
-    assert len(spans) == 1
-    assert_embedding_attributes(spans[0], model_name, response)
-
-    # Verify logs
-    logs = log_exporter.get_finished_logs()
-    assert (
-        len(logs) == 2
-    )  # Should contain both input and output embeddings events
-
-    # Find input and output events
-    input_event = None
-    output_event = None
-    for log in logs:
-        if (
-            log.log_record.attributes.get(EventAttributes.EVENT_NAME)
-            == "gen_ai.embeddings.input"
-        ):
-            input_event = log
-        elif (
-            log.log_record.attributes.get(EventAttributes.EVENT_NAME)
-            == "gen_ai.embeddings.output"
-        ):
-            output_event = log
-
-    # Verify both events exist
-    assert input_event is not None, "Input embeddings event not found"
-    assert output_event is not None, "Output embeddings event not found"
-
-    # Verify input event content
-    input_content = {"content": input_text, "role": "user"}
-    assert_message_in_logs(
-        input_event, "gen_ai.embeddings.input", input_content, spans[0]
-    )
-
-    # Verify output event content
-    # We don't need to specify exact values since the test provides a helper function to verify
-    output_content = {
-        "embeddings": [
-            {
-                "index": 0,
-                "embedding": response.data[0].embedding,
-                "object": "embedding",
-            }
-        ]
-    }
-    assert_message_in_logs(
-        output_event, "gen_ai.embeddings.output", output_content, spans[0]
-    )
-
-
 @pytest.mark.vcr()
 def test_embeddings_no_content(
     span_exporter, log_exporter, openai_client, instrument_no_content