v5.27.0 proposal

.github/workflows/appsec.yml

@@ -205,11 +205,12 @@ jobs:
 
   next:
     strategy:
+      fail-fast: false
       matrix:
         version:
           - 18
           - latest
-        range: ['9.5.0', '11.1.4', '13.2.0', '14.2.6']
+        range: ['9.5.0', '11.1.4', '13.2.0', '>=14.0.0 <=14.2.6', '>=14.2.7 <15']
     runs-on: ubuntu-latest
     env:
       PLUGINS: next

.github/workflows/plugins.yml

@@ -561,6 +561,25 @@ jobs:
       - uses: actions/checkout@v4
       - uses: ./.github/actions/plugins/test-and-upstream
 
+  langchain:
+    runs-on: ubuntu-latest
+    env:
+      PLUGINS: langchain
+    steps:
+      - uses: actions/checkout@v4
+      - uses: ./.github/actions/testagent/start
+      - uses: ./.github/actions/node/setup
+      - uses: ./.github/actions/install
+      - uses: ./.github/actions/node/18 # langchain doesn't support Node 16
+      - run: yarn test:plugins:ci
+        shell: bash
+      - uses: ./.github/actions/node/latest
+      - run: yarn test:plugins:ci
+        shell: bash
+      - uses: codecov/codecov-action@v3
+      - if: always()
+        uses: ./.github/actions/testagent/logs
+
   limitd-client:
     runs-on: ubuntu-latest
     services:
@@ -734,11 +753,12 @@ jobs:
   # TODO: fix performance issues and test more Node versions
   next:
     strategy:
+      fail-fast: false
       matrix:
         version:
           - 18
           - latest
-        range: ['9.5.0', '11.1.4', '13.2.0', '14.2.6']
+        range: ['9.5.0', '11.1.4', '13.2.0', '>=14.0.0 <=14.2.6', '>=14.2.7 <15']
     runs-on: ubuntu-latest
     env:
       PLUGINS: next

.gitlab/requirements_block.json

@@ -6,6 +6,9 @@
   {"name": "unsupported 2.x.x glibc x64","filepath": "/some/path", "args": [], "envars": [], "host": {"os": "linux", "arch": "arm64", "libc": "glibc:2.16.9"}},
   {"name": "unsupported 2.x.x glibc x86","filepath": "/some/path", "args": [], "envars": [], "host": {"os": "linux", "arch": "x86", "libc": "glibc:2.17"}},
   {"name": "npm","filepath": "/pathto/node", "args": ["/pathto/node", "/pathto/npm-cli.js"], "envars": [], "host": {"os": "linux", "arch": "x64", "libc": "glibc:2.40"}},
+  {"name": "npm-symlink","filepath": "/pathto/node", "args": ["/pathto/node", "/pathto/npm"], "envars": [], "host": {"os": "linux", "arch": "x64", "libc": "glibc:2.40"}},
   {"name": "yarn","filepath": "/pathto/node", "args": ["/pathto/node", "/pathto/yarn.js"], "envars": [], "host": {"os": "linux", "arch": "x64", "libc": "glibc:2.40"}},
-  {"name": "pnpm","filepath": "/pathto/node", "args": ["/pathto/node", "/pathto/pnpm.cjs"], "envars": [], "host": {"os": "linux", "arch": "x64", "libc": "glibc:2.40"}}
+  {"name": "yarn-symlink","filepath": "/pathto/node", "args": ["/pathto/node", "/pathto/yarn"], "envars": [], "host": {"os": "linux", "arch": "x64", "libc": "glibc:2.40"}},
+  {"name": "pnpm","filepath": "/pathto/node", "args": ["/pathto/node", "/pathto/pnpm.cjs"], "envars": [], "host": {"os": "linux", "arch": "x64", "libc": "glibc:2.40"}},
+  {"name": "pnpm-symlink","filepath": "/pathto/node", "args": ["/pathto/node", "/pathto/pnpm"], "envars": [], "host": {"os": "linux", "arch": "x64", "libc": "glibc:2.40"}}
 ]

docs/test.ts

@@ -342,6 +342,7 @@ tracer.use('kafkajs');
 tracer.use('knex');
 tracer.use('koa');
 tracer.use('koa', httpServerOptions);
+tracer.use('langchain');
 tracer.use('mariadb', { service: () => `my-custom-mariadb` })
 tracer.use('memcached');
 tracer.use('microgateway-core');

index.d.ts

@@ -179,6 +179,7 @@ interface Plugins {
   "kafkajs": tracer.plugins.kafkajs
   "knex": tracer.plugins.knex;
   "koa": tracer.plugins.koa;
+  "langchain": tracer.plugins.langchain;
   "mariadb": tracer.plugins.mariadb;
   "memcached": tracer.plugins.memcached;
   "microgateway-core": tracer.plugins.microgateway_core;
@@ -1592,6 +1593,12 @@ declare namespace tracer {
      */
     interface kafkajs extends Instrumentation {}
 
+    /**
+     * This plugin automatically instruments the
+     * [langchain](https://js.langchain.com/) module
+     */
+    interface langchain extends Instrumentation {}
+
     /**
      * This plugin automatically instruments the
      * [ldapjs](https://github.com/ldapjs/node-ldapjs/) module.

integration-tests/profiler/index.js

@@ -21,4 +21,5 @@ function busyWait (ms) {
   })
 }
 
-setImmediate(async () => busyWait(500))
+const durationMs = Number.parseInt(process.env.TEST_DURATION_MS ?? '500')
+setImmediate(async () => busyWait(durationMs))

integration-tests/profiler/profiler.spec.js

@@ -544,6 +544,71 @@ describe('profiler', () => {
     })
   })
 
+  context('Profiler API telemetry', () => {
+    beforeEach(async () => {
+      agent = await new FakeAgent().start()
+    })
+
+    afterEach(async () => {
+      proc.kill()
+      await agent.stop()
+    })
+
+    it('sends profiler API telemetry', () => {
+      proc = fork(profilerTestFile, {
+        cwd,
+        env: {
+          DD_TRACE_AGENT_PORT: agent.port,
+          DD_PROFILING_ENABLED: 1,
+          DD_PROFILING_UPLOAD_PERIOD: 1,
+          TEST_DURATION_MS: 2500
+        }
+      })
+
+      let requestCount = 0
+      let pointsCount = 0
+
+      const checkMetrics = agent.assertTelemetryReceived(({ _, payload }) => {
+        const pp = payload.payload
+        assert.equal(pp.namespace, 'profilers')
+        const series = pp.series
+        assert.lengthOf(series, 2)
+        assert.equal(series[0].metric, 'profile_api.requests')
+        assert.equal(series[0].type, 'count')
+        // There's a race between metrics and on-shutdown profile, so metric
+        // value will be between 2 and 3
+        requestCount = series[0].points[0][1]
+        assert.isAtLeast(requestCount, 2)
+        assert.isAtMost(requestCount, 3)
+
+        assert.equal(series[1].metric, 'profile_api.responses')
+        assert.equal(series[1].type, 'count')
+        assert.include(series[1].tags, 'status_code:200')
+
+        // Same number of requests and responses
+        assert.equal(series[1].points[0][1], requestCount)
+      }, timeout, 'generate-metrics')
+
+      const checkDistributions = agent.assertTelemetryReceived(({ _, payload }) => {
+        const pp = payload.payload
+        assert.equal(pp.namespace, 'profilers')
+        const series = pp.series
+        assert.lengthOf(series, 2)
+        assert.equal(series[0].metric, 'profile_api.bytes')
+        assert.equal(series[1].metric, 'profile_api.ms')
+
+        // Same number of points
+        pointsCount = series[0].points.length
+        assert.equal(pointsCount, series[1].points.length)
+      }, timeout, 'distributions')
+
+      return Promise.all([checkProfiles(agent, proc, timeout), checkMetrics, checkDistributions]).then(() => {
+        // Same number of requests and points
+        assert.equal(requestCount, pointsCount)
+      })
+    })
+  })
+
   function forkSsi (args, whichEnv) {
     const profilerEnablingEnv = whichEnv ? { DD_PROFILING_ENABLED: 'auto' } : { DD_INJECTION_ENABLED: 'profiler' }
     return fork(ssiTestFile, args, {

package.json

@@ -1,6 +1,6 @@
 {
   "name": "dd-trace",
-  "version": "5.26.0",
+  "version": "5.27.0",
   "description": "Datadog APM tracing client for JavaScript",
   "main": "index.js",
   "typings": "index.d.ts",

packages/datadog-instrumentations/src/helpers/hooks.js

@@ -19,6 +19,8 @@ module.exports = {
   '@jest/test-sequencer': () => require('../jest'),
   '@jest/transform': () => require('../jest'),
   '@koa/router': () => require('../koa'),
+  '@langchain/core': () => require('../langchain'),
+  '@langchain/openai': () => require('../langchain'),
   '@node-redis/client': () => require('../redis'),
   '@opensearch-project/opensearch': () => require('../opensearch'),
   '@opentelemetry/sdk-trace-node': () => require('../otel-sdk-trace'),
@@ -67,6 +69,7 @@ module.exports = {
   koa: () => require('../koa'),
   'koa-router': () => require('../koa'),
   kafkajs: () => require('../kafkajs'),
+  langchain: () => require('../langchain'),
   ldapjs: () => require('../ldapjs'),
   'limitd-client': () => require('../limitd-client'),
   lodash: () => require('../lodash'),

packages/datadog-instrumentations/src/langchain.js

@@ -0,0 +1,77 @@
+'use strict'
+
+const { addHook } = require('./helpers/instrument')
+const shimmer = require('../../datadog-shimmer')
+
+const tracingChannel = require('dc-polyfill').tracingChannel
+
+const invokeTracingChannel = tracingChannel('apm:langchain:invoke')
+
+function wrapLangChainPromise (fn, type, namespace = []) {
+  return function () {
+    if (!invokeTracingChannel.start.hasSubscribers) {
+      return fn.apply(this, arguments)
+    }
+
+    // Runnable interfaces have an `lc_namespace` property
+    const ns = this.lc_namespace || namespace
+    const resource = [...ns, this.constructor.name].join('.')
+
+    const ctx = {
+      args: arguments,
+      instance: this,
+      type,
+      resource
+    }
+
+    return invokeTracingChannel.tracePromise(fn, ctx, this, ...arguments)
+  }
+}
+
+// langchain compiles into ESM and CommonJS, with ESM being the default and landing in the `.js` files
+// however, CommonJS ends up in `cjs` files, and are required under the hood with `.cjs` files
+// we patch each separately and explicitly to match against exports only once, and not rely on file regex matching
+const extensions = ['js', 'cjs']
+
+for (const extension of extensions) {
+  addHook({ name: '@langchain/core', file: `dist/runnables/base.${extension}`, versions: ['>=0.1'] }, exports => {
+    const RunnableSequence = exports.RunnableSequence
+    shimmer.wrap(RunnableSequence.prototype, 'invoke', invoke => wrapLangChainPromise(invoke, 'chain'))
+    shimmer.wrap(RunnableSequence.prototype, 'batch', batch => wrapLangChainPromise(batch, 'chain'))
+    return exports
+  })
+
+  addHook({
+    name: '@langchain/core',
+    file: `dist/language_models/chat_models.${extension}`,
+    versions: ['>=0.1']
+  }, exports => {
+    const BaseChatModel = exports.BaseChatModel
+    shimmer.wrap(
+      BaseChatModel.prototype,
+      'generate',
+      generate => wrapLangChainPromise(generate, 'chat_model')
+    )
+    return exports
+  })
+
+  addHook({ name: '@langchain/core', file: `dist/language_models/llms.${extension}`, versions: ['>=0.1'] }, exports => {
+    const BaseLLM = exports.BaseLLM
+    shimmer.wrap(BaseLLM.prototype, 'generate', generate => wrapLangChainPromise(generate, 'llm'))
+    return exports
+  })
+
+  addHook({ name: '@langchain/openai', file: `dist/embeddings.${extension}`, versions: ['>=0.1'] }, exports => {
+    const OpenAIEmbeddings = exports.OpenAIEmbeddings
+
+    // OpenAI (and Embeddings in general) do not define an lc_namespace
+    const namespace = ['langchain', 'embeddings', 'openai']
+    shimmer.wrap(OpenAIEmbeddings.prototype, 'embedDocuments', embedDocuments =>
+      wrapLangChainPromise(embedDocuments, 'embedding', namespace)
+    )
+    shimmer.wrap(OpenAIEmbeddings.prototype, 'embedQuery', embedQuery =>
+      wrapLangChainPromise(embedQuery, 'embedding', namespace)
+    )
+    return exports
+  })
+}

packages/datadog-instrumentations/src/next.js

@@ -14,8 +14,14 @@ const queryParsedChannel = channel('apm:next:query-parsed')
 const requests = new WeakSet()
 const nodeNextRequestsToNextRequests = new WeakMap()
 
+// Next.js <= 14.2.6
 const MIDDLEWARE_HEADER = 'x-middleware-invoke'
 
+// Next.js >= 14.2.7
+const NEXT_REQUEST_META = Symbol.for('NextInternalRequestMeta')
+const META_IS_MIDDLEWARE = 'middlewareInvoke'
+const encounteredMiddleware = new WeakSet()
+
 function wrapHandleRequest (handleRequest) {
   return function (req, res, pathname, query) {
     return instrument(req, res, () => handleRequest.apply(this, arguments))
@@ -111,6 +117,11 @@ function getPageFromPath (page, dynamicRoutes = []) {
   return getPagePath(page)
 }
 
+function getRequestMeta (req, key) {
+  const meta = req[NEXT_REQUEST_META] || {}
+  return typeof key === 'string' ? meta[key] : meta
+}
+
 function instrument (req, res, error, handler) {
   if (typeof error === 'function') {
     handler = error
@@ -121,8 +132,9 @@ function instrument (req, res, error, handler) {
   res = res.originalResponse || res
 
   // TODO support middleware properly in the future?
-  const isMiddleware = req.headers[MIDDLEWARE_HEADER]
-  if (isMiddleware || requests.has(req)) {
+  const isMiddleware = req.headers[MIDDLEWARE_HEADER] || getRequestMeta(req, META_IS_MIDDLEWARE)
+  if ((isMiddleware && !encounteredMiddleware.has(req)) || requests.has(req)) {
+    encounteredMiddleware.add(req)
     if (error) {
       errorChannel.publish({ error })
     }
@@ -188,7 +200,7 @@ function finish (ctx, result, err) {
 // however, it is not provided as a class function or exported property
 addHook({
   name: 'next',
-  versions: ['>=13.3.0 <14.2.7'],
+  versions: ['>=13.3.0 <15'],
   file: 'dist/server/web/spec-extension/adapters/next-request.js'
 }, NextRequestAdapter => {
   shimmer.wrap(NextRequestAdapter.NextRequestAdapter, 'fromNodeNextRequest', fromNodeNextRequest => {
@@ -203,7 +215,7 @@ addHook({
 
 addHook({
   name: 'next',
-  versions: ['>=11.1 <14.2.7'],
+  versions: ['>=11.1 <15'],
   file: 'dist/server/serve-static.js'
 }, serveStatic => shimmer.wrap(serveStatic, 'serveStatic', wrapServeStatic))
 
@@ -213,7 +225,7 @@ addHook({
   file: 'dist/next-server/server/serve-static.js'
 }, serveStatic => shimmer.wrap(serveStatic, 'serveStatic', wrapServeStatic))
 
-addHook({ name: 'next', versions: ['>=11.1 <14.2.7'], file: 'dist/server/next-server.js' }, nextServer => {
+addHook({ name: 'next', versions: ['>=11.1 <15'], file: 'dist/server/next-server.js' }, nextServer => {
   const Server = nextServer.default
 
   shimmer.wrap(Server.prototype, 'handleRequest', wrapHandleRequest)
@@ -230,7 +242,7 @@ addHook({ name: 'next', versions: ['>=11.1 <14.2.7'], file: 'dist/server/next-se
 })
 
 // `handleApiRequest` changes parameters/implementation at 13.2.0
-addHook({ name: 'next', versions: ['>=13.2 <14.2.7'], file: 'dist/server/next-server.js' }, nextServer => {
+addHook({ name: 'next', versions: ['>=13.2 <15'], file: 'dist/server/next-server.js' }, nextServer => {
   const Server = nextServer.default
   shimmer.wrap(Server.prototype, 'handleApiRequest', wrapHandleApiRequestWithMatch)
   return nextServer
@@ -264,7 +276,7 @@ addHook({
 
 addHook({
   name: 'next',
-  versions: ['>=13 <14.2.7'],
+  versions: ['>=13 <15'],
   file: 'dist/server/web/spec-extension/request.js'
 }, request => {
   const nextUrlDescriptor = Object.getOwnPropertyDescriptor(request.NextRequest.prototype, 'nextUrl')

packages/datadog-plugin-amqplib/src/consumer.js

@@ -26,8 +26,7 @@ class AmqplibConsumerPlugin extends ConsumerPlugin {
         'amqp.consumerTag': fields.consumerTag,
         'amqp.source': fields.source,
         'amqp.destination': fields.destination
-      },
-      extractedLinks: childOf?._links
+      }
     })
 
     if (

packages/datadog-plugin-aws-sdk/src/services/kinesis.js

@@ -42,8 +42,7 @@ class Kinesis extends BaseAwsSdkPlugin {
               {},
               this.requestTags.get(request) || {},
               { 'span.kind': 'server' }
-            ),
-            extractedLinks: responseExtraction.maybeChildOf._links
+            )
           }
           span = plugin.tracer.startSpan('aws.response', options)
           this.enter(span, store)

packages/datadog-plugin-aws-sdk/src/services/sqs.js

@@ -33,8 +33,7 @@ class Sqs extends BaseAwsSdkPlugin {
             {},
             this.requestTags.get(request) || {},
             { 'span.kind': 'server' }
-          ),
-          extractedLinks: contextExtraction.datadogContext._links
+          )
         }
         parsedMessageAttributes = contextExtraction.parsedAttributes
         span = plugin.tracer.startSpan('aws.response', options)

packages/datadog-plugin-google-cloud-pubsub/src/consumer.js

@@ -22,8 +22,7 @@ class GoogleCloudPubsubConsumerPlugin extends ConsumerPlugin {
       },
       metrics: {
         'pubsub.ack': 0
-      },
-      extractedLinks: childOf?._links
+      }
     })
     if (this.config.dsmEnabled && message?.attributes) {
       const payloadSize = getMessageSize(message)