feat(core): Support truncation for LangChain integration request messages

dev-packages/node-integration-tests/suites/tracing/langchain/scenario-message-truncation.mjs

@@ -0,0 +1,72 @@
+import { ChatAnthropic } from '@langchain/anthropic';
+import * as Sentry from '@sentry/node';
+import express from 'express';
+
+function startMockAnthropicServer() {
+  const app = express();
+  app.use(express.json());
+
+  app.post('/v1/messages', (req, res) => {
+    const model = req.body.model;
+
+    res.json({
+      id: 'msg_truncation_test',
+      type: 'message',
+      role: 'assistant',
+      content: [
+        {
+          type: 'text',
+          text: 'Response to truncated messages',
+        },
+      ],
+      model: model,
+      stop_reason: 'end_turn',
+      stop_sequence: null,
+      usage: {
+        input_tokens: 10,
+        output_tokens: 15,
+      },
+    });
+  });
+
+  return new Promise(resolve => {
+    const server = app.listen(0, () => {
+      resolve(server);
+    });
+  });
+}
+
+async function run() {
+  const server = await startMockAnthropicServer();
+  const baseUrl = `http://localhost:${server.address().port}`;
+
+  await Sentry.startSpan({ op: 'function', name: 'main' }, async () => {
+    const model = new ChatAnthropic({
+      model: 'claude-3-5-sonnet-20241022',
+      apiKey: 'mock-api-key',
+      clientOptions: {
+        baseURL: baseUrl,
+      },
+    });
+
+    const largeContent1 = 'A'.repeat(15000); // ~15KB
+    const largeContent2 = 'B'.repeat(15000); // ~15KB
+    const largeContent3 = 'C'.repeat(25000); // ~25KB (will be truncated)
+
+    // Create one very large string that gets truncated to only include Cs
+    await model.invoke(largeContent3 + largeContent2);
+
+    // Create an array of messages that gets truncated to only include the last message (result should again contain only Cs)
+    await model.invoke([
+      { role: 'system', content: largeContent1 },
+      { role: 'user', content: largeContent2 },
+      { role: 'user', content: largeContent3 },
+    ]);
+  });
+
+  await Sentry.flush(2000);
+
+  server.close();
+}
+
+run();

dev-packages/node-integration-tests/suites/tracing/langchain/test.ts

@@ -194,4 +194,55 @@ describe('LangChain integration', () => {
       await createRunner().ignore('event').expect({ transaction: EXPECTED_TRANSACTION_TOOL_CALLS }).start().completed();
     });
   });
+
+  const EXPECTED_TRANSACTION_MESSAGE_TRUNCATION = {
+    transaction: 'main',
+    spans: expect.arrayContaining([
+      expect.objectContaining({
+        data: expect.objectContaining({
+          'gen_ai.operation.name': 'chat',
+          'sentry.op': 'gen_ai.chat',
+          'sentry.origin': 'auto.ai.langchain',
+          'gen_ai.system': 'anthropic',
+          'gen_ai.request.model': 'claude-3-5-sonnet-20241022',
+          // Messages should be present and should include truncated string input (contains only Cs)
+          'gen_ai.request.messages': expect.stringMatching(/^\[\{"role":"user","content":"C+"\}\]$/),
+        }),
+        description: 'chat claude-3-5-sonnet-20241022',
+        op: 'gen_ai.chat',
+        origin: 'auto.ai.langchain',
+        status: 'ok',
+      }),
+      expect.objectContaining({
+        data: expect.objectContaining({
+          'gen_ai.operation.name': 'chat',
+          'sentry.op': 'gen_ai.chat',
+          'sentry.origin': 'auto.ai.langchain',
+          'gen_ai.system': 'anthropic',
+          'gen_ai.request.model': 'claude-3-5-sonnet-20241022',
+          // Messages should be present (truncation happened) and should be a JSON array of a single index (contains only Cs)
+          'gen_ai.request.messages': expect.stringMatching(/^\[\{"role":"user","content":"C+"\}\]$/),
+        }),
+        description: 'chat claude-3-5-sonnet-20241022',
+        op: 'gen_ai.chat',
+        origin: 'auto.ai.langchain',
+        status: 'ok',
+      }),
+    ]),
+  };
+
+  createEsmAndCjsTests(
+    __dirname,
+    'scenario-message-truncation.mjs',
+    'instrument-with-pii.mjs',
+    (createRunner, test) => {
+      test('truncates messages when they exceed byte limit', async () => {
+        await createRunner()
+          .ignore('event')
+          .expect({ transaction: EXPECTED_TRANSACTION_MESSAGE_TRUNCATION })
+          .start()
+          .completed();
+      });
+    },
+  );
 });

packages/core/src/utils/langchain/utils.ts

@@ -23,6 +23,7 @@ import {
   GEN_AI_USAGE_OUTPUT_TOKENS_ATTRIBUTE,
   GEN_AI_USAGE_TOTAL_TOKENS_ATTRIBUTE,
 } from '../ai/gen-ai-attributes';
+import { truncateGenAiMessages } from '../ai/messageTruncation';
 import { LANGCHAIN_ORIGIN, ROLE_MAP } from './constants';
 import type { LangChainLLMResult, LangChainMessage, LangChainSerialized } from './types';
 
@@ -281,7 +282,8 @@ export function extractChatModelRequestAttributes(
 
   if (recordInputs && Array.isArray(langChainMessages) && langChainMessages.length > 0) {
     const normalized = normalizeLangChainMessages(langChainMessages.flat());
-    setIfDefined(attrs, GEN_AI_REQUEST_MESSAGES_ATTRIBUTE, asString(normalized));
+    const truncated = truncateGenAiMessages(normalized);
+    setIfDefined(attrs, GEN_AI_REQUEST_MESSAGES_ATTRIBUTE, asString(truncated));
   }
 
   return attrs;