feat: add Bedrock InvokeModelWithResponseStream instrumentation

plugins/node/opentelemetry-instrumentation-aws-sdk/src/services/bedrock-runtime.ts

@@ -101,7 +101,9 @@
       case 'ConverseStream':
         return this.requestPreSpanHookConverse(request, config, diag, true);
       case 'InvokeModel':
-        return this.requestPreSpanHookInvokeModel(request, config, diag);
+        return this.requestPreSpanHookInvokeModel(request, config, diag, false);
+      case 'InvokeModelWithResponseStream':
+        return this.requestPreSpanHookInvokeModel(request, config, diag, true);
     }
 
     return {
@@ -157,7 +159,8 @@
   private requestPreSpanHookInvokeModel(
     request: NormalizedRequest,
     config: AwsSdkInstrumentationConfig,
-    diag: DiagLogger
+    diag: DiagLogger,
+    isStream: boolean
   ): RequestMetadata {
     let spanName: string | undefined;
     const spanAttributes: Attributes = {
@@ -312,6 +315,7 @@
     return {
       spanName,
       isIncoming: false,
+      isStream,
       spanAttributes,
     };
   }
@@ -346,6 +350,13 @@
         );
       case 'InvokeModel':
         return this.responseHookInvokeModel(response, span, tracer, config);
+      case 'InvokeModelWithResponseStream':
+        return this.responseHookInvokeModelWithResponseStream(
+          response,
+          span,
+          tracer,
+          config
+        );
     }
   }
 
@@ -579,4 +590,134 @@
       }
     }
   }
+
+  private async responseHookInvokeModelWithResponseStream(
+    response: NormalizedResponse,
+    span: Span,
+    tracer: Tracer,
+    config: AwsSdkInstrumentationConfig
+  ): Promise<any> {
+    const stream = response.data?.body;
+    const modelId = response.request.commandInput?.modelId;
+    if (!stream || !span.isRecording()) return;
+
+    const wrappedStream = instrumentAsyncIterable(
+      stream,
+      async (chunk: { chunk?: { bytes?: Uint8Array } }) => {
+        const parsedChunk = parseChunk(chunk?.chunk?.bytes);
+
+        if (!parsedChunk) return;
+
+        if (modelId.includes('amazon.titan')) {
+          recordTitanAttributes(parsedChunk);
+        } else if (modelId.includes('anthropic.claude')) {
+          recordClaudeAttributes(parsedChunk);
+        } else if (modelId.includes('amazon.nova')) {
+          recordNovaAttributes(parsedChunk);
+        }
+      }
+    );
+    // Replace the original response body with our instrumented stream.
+    // - Defers span.end() until the entire stream is consumed
+    // This ensures downstream consumers still receive the full stream correctly,
+    // while OpenTelemetry can record span attributes from streamed data.
+    response.data.body = (async function* () {
+      try {
+        for await (const item of wrappedStream) {
+          yield item;
+        }
+      } finally {
+        span.end();
+      }
+    })();
+    return response.data;
+
+    // Tap into the stream at the chunk level without modifying the chunk itself.
+    function instrumentAsyncIterable<T>(
+      stream: AsyncIterable<T>,
+      onChunk: (chunk: T) => void
+    ): AsyncIterable<T> {
+      return {
+        [Symbol.asyncIterator]: async function* () {
+          for await (const chunk of stream) {
+            onChunk(chunk);
+            yield chunk;
+          }
+        },
+      };
+    }
+
+    function parseChunk(bytes?: Uint8Array): any {
+      if (!bytes || !(bytes instanceof Uint8Array)) return null;
+      try {
+        const str = Buffer.from(bytes).toString('utf-8');
+        return JSON.parse(str);
+      } catch (err) {
+        console.warn('Failed to parse streamed chunk', err);
+        return null;
+      }
+    }
+
+    function recordNovaAttributes(parsedChunk: any) {
+      if (parsedChunk.metadata?.usage !== undefined) {
+        if (parsedChunk.metadata?.usage.inputTokens !== undefined) {
+          span.setAttribute(
+            ATTR_GEN_AI_USAGE_INPUT_TOKENS,
+            parsedChunk.metadata.usage.inputTokens
+          );
+        }
+        if (parsedChunk.metadata?.usage.outputTokens !== undefined) {
+          span.setAttribute(
+            ATTR_GEN_AI_USAGE_OUTPUT_TOKENS,
+            parsedChunk.metadata.usage.outputTokens
+          );
+        }
+      }
+      if (parsedChunk.messageStop?.stopReason !== undefined) {
+        span.setAttribute(ATTR_GEN_AI_RESPONSE_FINISH_REASONS, [
+          parsedChunk.messageStop.stopReason,
+        ]);
+      }
+    }
+
+    function recordClaudeAttributes(parsedChunk: any) {
+      if (parsedChunk.message?.usage?.input_tokens !== undefined) {
+        span.setAttribute(
+          ATTR_GEN_AI_USAGE_INPUT_TOKENS,
+          parsedChunk.message.usage.input_tokens
+        );
+      }
+      if (parsedChunk.message?.usage?.output_tokens !== undefined) {
+        span.setAttribute(
+          ATTR_GEN_AI_USAGE_OUTPUT_TOKENS,
+          parsedChunk.message.usage.output_tokens
+        );
+      }
+      if (parsedChunk.delta?.stop_reason !== undefined) {
+        span.setAttribute(ATTR_GEN_AI_RESPONSE_FINISH_REASONS, [
+          parsedChunk.delta.stop_reason,
+        ]);
+      }
+    }
+
+    function recordTitanAttributes(parsedChunk: any) {
+      if (parsedChunk.inputTextTokenCount !== undefined) {
+        span.setAttribute(
+          ATTR_GEN_AI_USAGE_INPUT_TOKENS,
+          parsedChunk.inputTextTokenCount
+        );
+      }
+      if (parsedChunk.totalOutputTextTokenCount !== undefined) {
+        span.setAttribute(
+          ATTR_GEN_AI_USAGE_OUTPUT_TOKENS,
+          parsedChunk.totalOutputTextTokenCount
+        );
+      }
+      if (parsedChunk.completionReason !== undefined) {
+        span.setAttribute(ATTR_GEN_AI_RESPONSE_FINISH_REASONS, [
+          parsedChunk.completionReason,
+        ]);
+      }
+    }
+  }
 }

plugins/node/opentelemetry-instrumentation-aws-sdk/test/bedrock-runtime.test.ts

@@ -36,6 +36,7 @@ import {
   ConverseStreamCommand,
   ConversationRole,
   InvokeModelCommand,
+  InvokeModelWithResponseStreamCommand,
 } from '@aws-sdk/client-bedrock-runtime';
 import { AwsCredentialIdentity } from '@aws-sdk/types';
 import * as path from 'path';
@@ -79,6 +80,7 @@ const sanitizeRecordings = (scopes: Definition[]) => {
 describe('Bedrock', () => {
   nockBack.fixtures = path.join(__dirname, 'mock-responses');
   let credentials: AwsCredentialIdentity | undefined;
+
   if (nockBack.currentMode === 'dryrun') {
     credentials = {
       accessKeyId: 'testing',
@@ -642,4 +644,184 @@ describe('Bedrock', () => {
       });
     });
   });
+
+  describe('InvokeModelWithStreams', () => {
+    it('adds amazon titan model attributes to span', async () => {
+      const modelId = 'amazon.titan-text-lite-v1';
+      const prompt = '\n\nHuman: Hello, How are you today? \n\nAssistant:';
+
+      const body = {
+        inputText: prompt,
+        textGenerationConfig: {
+          maxTokenCount: 10,
+          temperature: 0.8,
+          topP: 1,
+          stopSequences: ['|'],
+        },
+      };
+      const command = new InvokeModelWithResponseStreamCommand({
+        modelId,
+        body: JSON.stringify(body),
+        contentType: 'application/json',
+        accept: 'application/json',
+      });
+
+      const response = await client.send(command);
+      console.log('response', response);
+
+      let collectedText = '';
+      if (!response.body) return;
+      for await (const chunk of response.body) {
+        if (chunk?.chunk?.bytes instanceof Uint8Array) {
+          const parsed = JSON.parse(decodeChunk(chunk));
+          collectedText += parsed.outputText;
+        }
+      }
+      expect(collectedText).toBe(" Hello there! I'm doing well. Thank you");
+
+      const invokeModelSpans: ReadableSpan[] =
+        getInvokeModelWithResponseStreamSpans();
+
+      expect(invokeModelSpans.length).toBe(1);
+      expect(invokeModelSpans[0].attributes).toMatchObject({
+        [ATTR_GEN_AI_SYSTEM]: GEN_AI_SYSTEM_VALUE_AWS_BEDROCK,
+        [ATTR_GEN_AI_REQUEST_MODEL]: modelId,
+        [ATTR_GEN_AI_REQUEST_MAX_TOKENS]: 10,
+        [ATTR_GEN_AI_REQUEST_TEMPERATURE]: 0.8,
+        [ATTR_GEN_AI_REQUEST_TOP_P]: 1,
+        [ATTR_GEN_AI_REQUEST_STOP_SEQUENCES]: ['|'],
+        [ATTR_GEN_AI_USAGE_INPUT_TOKENS]: 13,
+        [ATTR_GEN_AI_USAGE_OUTPUT_TOKENS]: 10,
+        [ATTR_GEN_AI_RESPONSE_FINISH_REASONS]: ['LENGTH'],
+      });
+    });
+    it('adds claude model attributes to span', async () => {
+      const modelId = 'anthropic.claude-3-5-sonnet-20240620-v1:0';
+      const prompt = '\n\nHuman: Hello, How are you today? \n\nAssistant:';
+
+      const body = {
+        anthropic_version: 'bedrock-2023-05-31',
+        max_tokens: 12,
+        top_k: 250,
+        stop_sequences: ['|'],
+        temperature: 0.8,
+        top_p: 1,
+        messages: [
+          {
+            role: 'user',
+            content: [
+              {
+                type: 'text',
+                text: prompt,
+              },
+            ],
+          },
+        ],
+      };
+
+      const command = new InvokeModelWithResponseStreamCommand({
+        modelId,
+        body: JSON.stringify(body),
+        contentType: 'application/json',
+        accept: 'application/json',
+      });
+
+      const response = await client.send(command);
+
+      let collectedText = '';
+      if (!response.body) return;
+      for await (const chunk of response.body) {
+        if (chunk?.chunk?.bytes instanceof Uint8Array) {
+          const parsed = JSON.parse(decodeChunk(chunk));
+          if (
+            parsed.type === 'content_block_delta' &&
+            parsed.delta?.type === 'text_delta'
+          ) {
+            collectedText += parsed.delta.text;
+          }
+        }
+      }
+
+      expect(collectedText).toBe(
+        "Hello! I'm doing well, thank you for asking."
+      );
+
+      const invokeModelSpans: ReadableSpan[] =
+        getInvokeModelWithResponseStreamSpans();
+
+      expect(invokeModelSpans.length).toBe(1);
+      expect(invokeModelSpans[0].attributes).toMatchObject({
+        [ATTR_GEN_AI_SYSTEM]: GEN_AI_SYSTEM_VALUE_AWS_BEDROCK,
+        [ATTR_GEN_AI_REQUEST_MODEL]: modelId,
+        [ATTR_GEN_AI_REQUEST_MAX_TOKENS]: 12,
+        [ATTR_GEN_AI_REQUEST_TEMPERATURE]: 0.8,
+        [ATTR_GEN_AI_REQUEST_TOP_P]: 1,
+        [ATTR_GEN_AI_REQUEST_STOP_SEQUENCES]: ['|'],
+        [ATTR_GEN_AI_USAGE_INPUT_TOKENS]: 22,
+        [ATTR_GEN_AI_USAGE_OUTPUT_TOKENS]: 1,
+        [ATTR_GEN_AI_RESPONSE_FINISH_REASONS]: ['max_tokens'],
+      });
+    });
+
+    it('adds amazon nova model attributes to span', async () => {
+      const modelId = 'amazon.nova-pro-v1:0';
+      const prompt = 'Say this is a test';
+      const nativeRequest: any = {
+        messages: [{ role: 'user', content: [{ text: prompt }] }],
+        inferenceConfig: {
+          max_new_tokens: 10,
+          temperature: 0.8,
+          top_p: 1,
+          stopSequences: ['|'],
+        },
+      };
+      const command = new InvokeModelWithResponseStreamCommand({
+        modelId,
+        body: JSON.stringify(nativeRequest),
+      });
+
+      const response = await client.send(command);
+
+      let collectedText = '';
+      if (!response.body) return;
+      for await (const chunk of response.body) {
+        if (chunk?.chunk?.bytes instanceof Uint8Array) {
+          const parsed = JSON.parse(decodeChunk(chunk));
+          if (parsed.contentBlockDelta?.delta) {
+            collectedText += parsed.contentBlockDelta?.delta.text;
+          }
+        }
+      }
+
+      expect(collectedText).toBe(
+        "Certainly! If you're indicating that this interaction"
+      );
+
+      const invokeModelSpans: ReadableSpan[] =
+        getInvokeModelWithResponseStreamSpans();
+
+      expect(invokeModelSpans.length).toBe(1);
+      expect(invokeModelSpans[0].attributes).toMatchObject({
+        [ATTR_GEN_AI_SYSTEM]: GEN_AI_SYSTEM_VALUE_AWS_BEDROCK,
+        [ATTR_GEN_AI_REQUEST_MODEL]: modelId,
+        [ATTR_GEN_AI_REQUEST_MAX_TOKENS]: 10,
+        [ATTR_GEN_AI_REQUEST_TEMPERATURE]: 0.8,
+        [ATTR_GEN_AI_REQUEST_TOP_P]: 1,
+        [ATTR_GEN_AI_REQUEST_STOP_SEQUENCES]: ['|'],
+        [ATTR_GEN_AI_USAGE_INPUT_TOKENS]: 5,
+        [ATTR_GEN_AI_USAGE_OUTPUT_TOKENS]: 10,
+        [ATTR_GEN_AI_RESPONSE_FINISH_REASONS]: ['max_tokens'],
+      });
+    });
+  });
+
+  function getInvokeModelWithResponseStreamSpans(): ReadableSpan[] {
+    return getTestSpans().filter((s: ReadableSpan) => {
+      return s.name === 'BedrockRuntime.InvokeModelWithResponseStream';
+    });
+  }
+
+  function decodeChunk(chunk: any) {
+    return Buffer.from(chunk.chunk.bytes).toString('utf-8');
+  }
 });