fix: record gen_ai.request.stop_sequences

packages/opentelemetry-instrumentation-anthropic/opentelemetry/instrumentation/anthropic/span_utils.py

@@ -32,6 +32,15 @@ def _map_finish_reason(anthropic_reason):
     return _FINISH_REASON_MAP.get(anthropic_reason, anthropic_reason)
 
 
+def _normalize_stop_sequences(stop_sequences):
+    """Normalize stop_sequences to a list of strings, or None if nothing usable was passed."""
+    if isinstance(stop_sequences, str):
+        return [stop_sequences]
+    if isinstance(stop_sequences, (list, tuple)):
+        return [s for s in stop_sequences if isinstance(s, str)] or None
+    return None
+
+
 async def _process_image_item(item, trace_id, span_id, message_index, content_index):
     source = item.get("source", {})
     media_type = source.get("media_type", "image/unknown")
@@ -138,6 +147,11 @@ async def aset_input_attributes(span, kwargs):
     set_span_attribute(
         span, GenAIAttributes.GEN_AI_REQUEST_PRESENCE_PENALTY, kwargs.get("presence_penalty")
     )
+    set_span_attribute(
+        span,
+        GenAIAttributes.GEN_AI_REQUEST_STOP_SEQUENCES,
+        _normalize_stop_sequences(kwargs.get("stop_sequences")),
+    )
     set_span_attribute(span, SpanAttributes.GEN_AI_IS_STREAMING, kwargs.get("stream"))
 
     if should_send_prompts():

packages/opentelemetry-instrumentation-anthropic/tests/test_semconv_span_attrs.py

@@ -101,6 +101,75 @@ def test_input_messages_multi_turn():
     assert messages[2] == {"role": "user", "parts": [{"type": "text", "content": "How are you?"}]}
 
 
+def test_request_stop_sequences_attribute():
+    """gen_ai.request.stop_sequences must be recorded as a list of strings."""
+    span = make_span()
+    kwargs = {
+        "model": "claude-3-opus-20240229",
+        "messages": [{"role": "user", "content": "Hello"}],
+        "max_tokens": 1024,
+        "stop_sequences": ["END", "###"],
+    }
+    asyncio.run(aset_input_attributes(span, kwargs))
+
+    assert list(span.attributes[GenAIAttributes.GEN_AI_REQUEST_STOP_SEQUENCES]) == ["END", "###"]
+
+
+def test_request_stop_sequences_bare_string_is_one_item():
+    """A bare-string stop_sequences must be recorded as a single item, not split into characters."""
+    span = make_span()
+    kwargs = {
+        "model": "claude-3-opus-20240229",
+        "messages": [{"role": "user", "content": "Hello"}],
+        "max_tokens": 1024,
+        "stop_sequences": "END",
+    }
+    asyncio.run(aset_input_attributes(span, kwargs))
+
+    assert list(span.attributes[GenAIAttributes.GEN_AI_REQUEST_STOP_SEQUENCES]) == ["END"]
+
+
+def test_request_stop_sequences_tuple_filters_non_strings():
+    """Only str items from a tuple stop_sequences must be recorded."""
+    span = make_span()
+    kwargs = {
+        "model": "claude-3-opus-20240229",
+        "messages": [{"role": "user", "content": "Hello"}],
+        "max_tokens": 1024,
+        "stop_sequences": ("END", 123, None, "###"),
+    }
+    asyncio.run(aset_input_attributes(span, kwargs))
+
+    assert list(span.attributes[GenAIAttributes.GEN_AI_REQUEST_STOP_SEQUENCES]) == ["END", "###"]
+
+
+def test_request_stop_sequences_not_set_when_absent():
+    """gen_ai.request.stop_sequences must not be set when stop_sequences is not passed."""
+    span = make_span()
+    kwargs = {
+        "model": "claude-3-opus-20240229",
+        "messages": [{"role": "user", "content": "Hello"}],
+        "max_tokens": 1024,
+    }
+    asyncio.run(aset_input_attributes(span, kwargs))
+
+    assert GenAIAttributes.GEN_AI_REQUEST_STOP_SEQUENCES not in span.attributes
+
+
+def test_request_stop_sequences_not_set_when_no_string_items():
+    """gen_ai.request.stop_sequences must not be set when stop_sequences has no str items."""
+    span = make_span()
+    kwargs = {
+        "model": "claude-3-opus-20240229",
+        "messages": [{"role": "user", "content": "Hello"}],
+        "max_tokens": 1024,
+        "stop_sequences": [1, None],
+    }
+    asyncio.run(aset_input_attributes(span, kwargs))
+
+    assert GenAIAttributes.GEN_AI_REQUEST_STOP_SEQUENCES not in span.attributes
+
+
 def test_system_instructions_attribute():
     """gen_ai.system_instructions must be a JSON array of typed parts, NOT a plain string.