fix(llmobs): safely output format bedrock cohere rerank spans [backport 3.17]

ddtrace/llmobs/_integrations/bedrock.py

@@ -371,13 +371,14 @@ def _extract_output_message(response) -> List[Message]:
         """Extract output messages from the stored response.
         Anthropic allows for chat messages, which requires some special casing.
         """
-        if isinstance(response["text"], str):
-            return [Message(content=response["text"])]
-        if isinstance(response["text"], list):
-            if isinstance(response["text"][0], str):
-                return [Message(content=str(content)) for content in response["text"]]
-            if isinstance(response["text"][0], dict):
-                return [Message(content=response["text"][0].get("text", ""))]
+        resp_text = response.get("text", "")
+        if isinstance(resp_text, str):
+            return [Message(content=resp_text)]
+        if resp_text and isinstance(resp_text, list):
+            if isinstance(resp_text[0], str):
+                return [Message(content=str(content)) for content in resp_text]
+            if isinstance(resp_text[0], dict):
+                return [Message(content=resp_text[0].get("text", ""))]
         return []
 
     def _get_base_url(self, **kwargs: Dict[str, Any]) -> Optional[str]:

releasenotes/notes/fix-bedrock-llmobs-cohere-a1bcb48e9b252d77.yaml

@@ -0,0 +1,4 @@
+---
+fixes:
+  - |
+    LLM Observability: Resolves an issue in the bedrock integration where invoking cohere rerank models would result in missing spans due to output formatting index errors.

tests/contrib/botocore/bedrock_cassettes/cohere_rerank_invoke.yaml

@@ -0,0 +1,52 @@
+interactions:
+- request:
+    body: '{"query": "What is the capital of the United States?", "documents": ["Carson
+      City is the capital city of the American state of Nevada.", "The Commonwealth
+      of the Northern Mariana Islands is a group of islands in the Pacific Ocean.
+      Its capital is Saipan.", "Washington, D.C. (also known as simply Washington
+      or D.C., and officially as the District of Columbia) is the capital of the United
+      States. It is a federal district.", "Capitalization or capitalisation in English
+      grammar is the use of a capital letter at the start of a word. English usage
+      varies from capitalization in other languages.", "Capital punishment has existed
+      in the United States since beforethe United States was a country. As of 2017,
+      capital punishment is legal in 30 of the 50 states."], "api_version": 2, "top_n":
+      3}'
+    headers:
+      Content-Length:
+      - '790'
+      User-Agent:
+      - !!binary |
+        Qm90bzMvMS4zNC40OSBtZC9Cb3RvY29yZSMxLjM0LjQ5IHVhLzIuMCBvcy9tYWNvcyMyNC42LjAg
+        bWQvYXJjaCNhcm02NCBsYW5nL3B5dGhvbiMzLjExLjEzIG1kL3B5aW1wbCNDUHl0aG9uIGNmZy9y
+        ZXRyeS1tb2RlI2xlZ2FjeSBCb3RvY29yZS8xLjM0LjQ5
+      X-Amz-Date:
+      - !!binary |
+        MjAyNTEwMzFUMjEwODQ2Wg==
+      amz-sdk-invocation-id:
+      - !!binary |
+        NDNlZWJhN2EtNGY1Yy00ZDI1LWFmNzUtYjY0NDNjNmUzYzM0
+      amz-sdk-request:
+      - !!binary |
+        YXR0ZW1wdD0x
+    method: POST
+    uri: https://bedrock-runtime.us-east-1.amazonaws.com/model/cohere.rerank-v3-5%3A0/invoke
+  response:
+    body:
+      string: '{"results":[{"index":2,"relevance_score":0.8742601},{"index":0,"relevance_score":0.1728413},{"index":4,"relevance_score":0.10793502}]}'
+    headers:
+      Connection:
+      - keep-alive
+      Content-Length:
+      - '134'
+      Content-Type:
+      - application/json
+      Date:
+      - Fri, 31 Oct 2025 21:08:47 GMT
+      X-Amzn-Bedrock-Invocation-Latency:
+      - '109'
+      x-amzn-RequestId:
+      - de4d85e7-fe35-4809-bda9-7675d0eb091f
+    status:
+      code: 200
+      message: OK
+version: 1

tests/contrib/botocore/test_bedrock_llmobs.py

@@ -26,10 +26,15 @@
 )
 class TestLLMObsBedrock:
     @staticmethod
-    def expected_llmobs_span_event(span, n_output, message=False, metadata=None, token_metrics=None):
+    def expected_llmobs_span_event(
+        span, n_output, input_message=False, output_message=False, metadata=None, token_metrics=None
+    ):
         expected_input = [{"content": mock.ANY}]
-        if message:
+        if input_message:
             expected_input = [{"content": mock.ANY, "role": "user"}]
+        expected_output = []
+        if output_message:
+            expected_output = [{"content": mock.ANY} for _ in range(n_output)]
 
         # Use empty dicts as defaults for _expected_llmobs_llm_span_event to avoid None issues
         expected_parameters = metadata if metadata is not None else {}
@@ -40,7 +45,7 @@ def expected_llmobs_span_event(span, n_output, message=False, metadata=None, tok
             model_name=span.get_tag("bedrock.request.model"),
             model_provider=span.get_tag("bedrock.request.model_provider"),
             input_messages=expected_input,
-            output_messages=[{"content": mock.ANY} for _ in range(n_output)],
+            output_messages=expected_output,
             metadata=expected_parameters,
             token_metrics=expected_token_metrics,
             tags={"service": "aws.bedrock-runtime", "ml_app": "<ml-app-name>"},
@@ -86,7 +91,7 @@ def _test_llmobs_invoke(cls, provider, bedrock_client, mock_tracer, llmobs_event
 
         assert len(llmobs_events) == 1
         assert llmobs_events[0] == cls.expected_llmobs_span_event(
-            span, n_output, message="message" in provider, metadata=expected_metadata
+            span, n_output, input_message="message" in provider, output_message=True, metadata=expected_metadata
         )
         LLMObs.disable()
 
@@ -121,7 +126,7 @@ def _test_llmobs_invoke_stream(
 
         assert len(llmobs_events) == 1
         assert llmobs_events[0] == cls.expected_llmobs_span_event(
-            span, n_output, message="message" in provider, metadata=expected_metadata
+            span, n_output, input_message="message" in provider, output_message=True, metadata=expected_metadata
         )
 
     def test_llmobs_ai21_invoke(self, ddtrace_global_config, bedrock_client, mock_tracer, llmobs_events):
@@ -156,6 +161,24 @@ def test_llmobs_cohere_multi_output_invoke(self, ddtrace_global_config, bedrock_
     def test_llmobs_meta_invoke(self, ddtrace_global_config, bedrock_client, mock_tracer, llmobs_events):
         self._test_llmobs_invoke("meta", bedrock_client, mock_tracer, llmobs_events)
 
+    def test_llmobs_cohere_rerank_invoke(self, ddtrace_global_config, bedrock_client, mock_tracer, llmobs_events):
+        cassette_name = "cohere_rerank_invoke.yaml"
+        model = "cohere.rerank-v3-5:0"
+        prompt_data = "What is the capital of the United States?"
+        documents = [
+            "Carson City is the capital city of the American state of Nevada.",
+            "The Commonwealth of the Northern Mariana Islands's capital is Saipan.",
+        ]
+        body = json.dumps({"query": prompt_data, "documents": documents, "api_version": 2, "top_n": 3})
+        with get_request_vcr().use_cassette(cassette_name):
+            response = bedrock_client.invoke_model(body=body, modelId=model)
+            json.loads(response.get("body").read())
+        span = mock_tracer.pop_traces()[0][0]
+
+        assert len(llmobs_events) == 1
+        assert llmobs_events[0] == self.expected_llmobs_span_event(span, 1)
+        LLMObs.disable()
+
     def test_llmobs_amazon_invoke_stream(self, ddtrace_global_config, bedrock_client, mock_tracer, llmobs_events):
         self._test_llmobs_invoke_stream("amazon", bedrock_client, mock_tracer, llmobs_events)