dd-trace 5.124.0 → 5.126.0

package/packages/dd-trace/src/llmobs/index.js

@@ -2,7 +2,6 @@
 
 const { channel } = require('dc-polyfill')
 
-const { readDatadogTags, writeDatadogTags } = require('../carrier')
 const { registerTelemetryFlusher } = require('../flush')
 const log = require('../log')
 const { createServerlessDeliveryTracker } = require('../serverless')
@@ -25,7 +24,7 @@ const {
   PROPAGATED_TRACE_ID_KEY,
 } = require('./constants/tags')
 const { storage } = require('./storage')
-const { agentNameWireSafe, appendOptionalPropagatedTag, resolveAgentAttribution, stripTagsetEntry } = require('./util')
+const { agentNameWireSafe, resolveAgentAttribution } = require('./util')
 const telemetry = require('./telemetry')
 const LLMObsSpanProcessor = require('./span_processor')
 const LLMObsEvalMetricsWriter = require('./writers/evaluations')
@@ -64,6 +63,13 @@ let unregisterTelemetryFlusher
 /** @type {import('../config/config-base')} */
 let globalTracerConfig
 
+/**
+ * @typedef {object} TraceTagInjection
+ * @property {import('../opentracing/span_context')} spanContext
+ * @property {Array<string | undefined>} [traceTagReplacements]
+ * @property {number} [optionalTraceTagCount]
+ */
+
 /**
  * @param {@type import('../config/config-base')} config
  */
@@ -163,7 +169,8 @@ function retireWriters (retiredSpanWriter, retiredEvalWriter) {
 
 // since LLMObs traces can extend between services and be the same trace,
 // we need to propagate the parent id, mlApp, session id, and sampling rate/decision.
-function handleLLMObsInjection ({ carrier }) {
+/** @param {TraceTagInjection} injection */
+function handleLLMObsInjection (injection) {
   // Respect the standard propagator's gate: when trace tag propagation is
   // disabled, don't write `x-datadog-tags` for LLMObs either.
   if (globalTracerConfig.DD_TRACE_X_DATADOG_TAGS_MAX_LENGTH === 0) return
@@ -201,37 +208,30 @@ function handleLLMObsInjection ({ carrier }) {
     mlObsSpanTags, parent
   )
 
-  // `_injectTags` only writes `x-datadog-tags` when the trace has `_dd.p.*`
-  // tags, so it may be undefined here — coalesce before appending.
-  const existing = readDatadogTags(carrier)
-  let tags = existing || ''
-  if (parentId) tags += `${tags ? ',' : ''}${PROPAGATED_PARENT_ID_KEY}=${parentId}`
-  if (mlApp) tags += `${tags ? ',' : ''}${PROPAGATED_ML_APP_KEY}=${mlApp}`
-  if (sessionId) tags += `${tags ? ',' : ''}${PROPAGATED_SESSION_ID_KEY}=${sessionId}`
-  if (sampleRate != null) tags += `${tags ? ',' : ''}${PROPAGATED_SAMPLE_RATE_KEY}=${sampleRate}`
-  if (samplingDecision != null) tags += `${tags ? ',' : ''}${PROPAGATED_SAMPLING_DECISION_KEY}=${samplingDecision}`
-  if (propagatedTraceId != null) tags += `${tags ? ',' : ''}${PROPAGATED_TRACE_ID_KEY}=${propagatedTraceId}`
-  // When a local agent attribution is resolved, strip any stale upstream pagent entries that
-  // `_injectTags` may have already written into the carrier (it propagates all `_dd.p.*` from
-  // `_trace.tags`). This ensures the downstream sees a consistent id-only or id+name pair
-  // rather than a mix from different hops. The id is always digit-safe; an unsafe name is wiped.
-  if (parentAgentSpanId) {
-    tags = stripTagsetEntry(tags, PROPAGATED_PARENT_AGENT_ID_KEY)
-    tags = stripTagsetEntry(tags, PROPAGATED_PARENT_AGENT_NAME_KEY)
+  const traceTagReplacements = []
+  if (parentId) traceTagReplacements.push(PROPAGATED_PARENT_ID_KEY, parentId)
+  if (mlApp) traceTagReplacements.push(PROPAGATED_ML_APP_KEY, mlApp)
+  if (sessionId) traceTagReplacements.push(PROPAGATED_SESSION_ID_KEY, sessionId)
+  if (sampleRate != null) traceTagReplacements.push(PROPAGATED_SAMPLE_RATE_KEY, sampleRate.toString())
+  if (samplingDecision != null) {
+    traceTagReplacements.push(PROPAGATED_SAMPLING_DECISION_KEY, samplingDecision.toString())
   }
-  const maxLength = globalTracerConfig.DD_TRACE_X_DATADOG_TAGS_MAX_LENGTH
-  const tagsWithId = appendOptionalPropagatedTag(
-    tags, PROPAGATED_PARENT_AGENT_ID_KEY, parentAgentSpanId, null, maxLength
-  )
-  // Only append the name when the id fit: a name without an id is unresolvable by the backend.
-  if (tagsWithId === tags) {
-    tags = tagsWithId
-  } else {
-    tags = appendOptionalPropagatedTag(
-      tagsWithId, PROPAGATED_PARENT_AGENT_NAME_KEY, parentAgentName, agentNameWireSafe, maxLength
-    )
+  if (propagatedTraceId != null) traceTagReplacements.push(PROPAGATED_TRACE_ID_KEY, propagatedTraceId)
+
+  let optionalTraceTagCount = 0
+  if (parentAgentSpanId) {
+    traceTagReplacements.push(PROPAGATED_PARENT_AGENT_ID_KEY, parentAgentSpanId)
+    optionalTraceTagCount++
+    if (parentAgentName && agentNameWireSafe(parentAgentName)) {
+      traceTagReplacements.push(PROPAGATED_PARENT_AGENT_NAME_KEY, parentAgentName)
+    } else {
+      traceTagReplacements.push(PROPAGATED_PARENT_AGENT_NAME_KEY, undefined)
+    }
+    optionalTraceTagCount++
   }
-  if (tags !== existing) writeDatadogTags(carrier, tags)
+
+  injection.traceTagReplacements = traceTagReplacements
+  injection.optionalTraceTagCount = optionalTraceTagCount
 }
 
 /**

package/packages/dd-trace/src/llmobs/plugins/ai/ddTelemetry.js

@@ -4,7 +4,6 @@ const { channel } = require('dc-polyfill')
 const BaseLLMObsPlugin = require('../base')
 const { getModelProvider } = require('../../../../../datadog-plugin-ai/src/utils')
 
-const toolCreationCh = channel('tracing:orchestrion:ai:tool:start')
 const setAttributesCh = channel('dd-trace:vercel-ai:span:setAttributes')
 
 const { MODEL_NAME, MODEL_PROVIDER, NAME } = require('../../constants/tags')
@@ -21,16 +20,10 @@ const {
   getTelemetryMetadata,
 } = require('./util')
 
-/**
- * @typedef {Record<string, unknown> & { description?: string, id?: string }} AvailableToolArgs
- */
-
 /**
  * @typedef {string | number | boolean | null | undefined | string[] | number[] | boolean[]} TagValue
  * @typedef {Record<string, TagValue>} SpanTags
  *
- * @typedef {{ name?: string, description?: string }} ToolForModel
- *
  * @typedef {{ type: 'text' | 'reasoning' | 'redacted-reasoning', text?: string, data?: string }} TextPart
  * @typedef {{ type: 'tool-call', toolName: string, toolCallId: string, args?: unknown, input?: unknown }} ToolCallPart
  * @typedef {(
@@ -76,58 +69,14 @@ class DdTelemetryPlugin extends BaseLLMObsPlugin {
   static integration = 'ai'
   static prefix = 'tracing:dd-trace:vercel-ai'
 
-  /**
-   * The available tools within the runtime scope of this integration.
-   * This essentially acts as a global registry for all tools made through the Vercel AI SDK.
-   * @type {Set<AvailableToolArgs>}
-   */
-  #availableTools
-
-  /**
-   * A mapping of tool call IDs to tool names.
-   * This is used to map the tool call ID to the tool name for the output message.
-   * @type {Record<string, string>}
-   */
-  #toolCallIdsToName
-
   constructor (...args) {
     super(...args)
 
-    this.#toolCallIdsToName = {}
-    this.#availableTools = new Set()
-    toolCreationCh.subscribe(ctx => {
-      const toolArgs = ctx.arguments
-      const tool = toolArgs[0] ?? {}
-      this.#availableTools.add(tool)
-    })
-
     setAttributesCh.subscribe(({ ctx, attributes }) => {
       Object.assign(ctx.attributes, attributes)
     })
   }
 
-  /**
-   * Does a best-effort attempt to find the right tool name for the given tool description.
-   * This is because the Vercel AI SDK does not tag tools by name properly, but
-   * rather by the index they were passed in. Tool names appear nowhere in the span tags.
-   *
-   * We use the tool description as the next best identifier for a tool.
-   *
-   * @param {string} toolName
-   * @param {string | undefined} toolDescription
-   * @returns {string | undefined}
-   */
-  findToolName (toolName, toolDescription) {
-    if (Number.isNaN(Number.parseInt(toolName, 10))) return toolName
-
-    for (const availableTool of this.#availableTools) {
-      const description = availableTool.description
-      if (description === toolDescription && availableTool.id) {
-        return availableTool.id
-      }
-    }
-  }
-
   /**
    * @override
    */
@@ -232,7 +181,13 @@ class DdTelemetryPlugin extends BaseLLMObsPlugin {
     const lastUserPrompt =
       promptInfo.prompt ??
       promptInfo.messages.reverse().find(message => message.role === 'user')?.content
-    const prompt = Array.isArray(lastUserPrompt) ? lastUserPrompt.map(part => part.text ?? '').join('') : lastUserPrompt
+    let prompt = lastUserPrompt
+    if (Array.isArray(lastUserPrompt)) {
+      prompt = ''
+      for (const part of lastUserPrompt) {
+        if (typeof part.text === 'string') prompt += part.text
+      }
+    }
 
     const output = tags['ai.response.object']
 
@@ -248,7 +203,13 @@ class DdTelemetryPlugin extends BaseLLMObsPlugin {
     const lastUserPrompt =
       promptInfo.prompt ??
       promptInfo.messages.reverse().find(message => message.role === 'user')?.content
-    const prompt = Array.isArray(lastUserPrompt) ? lastUserPrompt.map(part => part.text ?? '').join('') : lastUserPrompt
+    let prompt = lastUserPrompt
+    if (Array.isArray(lastUserPrompt)) {
+      prompt = ''
+      for (const part of lastUserPrompt) {
+        if (typeof part.text === 'string') prompt += part.text
+      }
+    }
 
     const output = tags['ai.response.text']
 
@@ -263,16 +224,14 @@ class DdTelemetryPlugin extends BaseLLMObsPlugin {
    * @param {SpanTags} tags
    */
   setLLMOperationTags (span, tags) {
-    const toolsForModel = tags['ai.prompt.tools']?.map(getJsonStringValue)
-
     const inputMessages = getJsonStringValue(tags['ai.prompt.messages'], [])
     const parsedInputMessages = []
     for (const message of inputMessages) {
-      const formattedMessages = this.formatMessage(message, toolsForModel)
+      const formattedMessages = this.formatMessage(message)
       parsedInputMessages.push(...formattedMessages)
     }
 
-    const outputMessage = this.formatOutputMessage(tags, toolsForModel)
+    const outputMessage = this.formatOutputMessage(tags)
 
     this._tagger.tagLLMIO(span, parsedInputMessages, outputMessage)
 
@@ -284,8 +243,8 @@ class DdTelemetryPlugin extends BaseLLMObsPlugin {
   }
 
   setToolTags (span, tags) {
-    const toolCallId = tags['ai.toolCall.id']
-    const name = getToolNameFromTags(tags) ?? this.#toolCallIdsToName[toolCallId]
+    const name = getToolNameFromTags(tags)
+
     if (name) this._tagger._setTag(span, NAME, name)
 
     const input = tags['ai.toolCall.args']
@@ -294,7 +253,7 @@ class DdTelemetryPlugin extends BaseLLMObsPlugin {
     this._tagger.tagTextIO(span, input, output)
   }
 
-  formatOutputMessage (tags, toolsForModel) {
+  formatOutputMessage (tags) {
     const outputMessageText = tags['ai.response.text'] ?? tags['ai.response.object']
     const outputMessageToolCalls = getJsonStringValue(tags['ai.response.toolCalls'], [])
 
@@ -302,13 +261,9 @@ class DdTelemetryPlugin extends BaseLLMObsPlugin {
     for (const toolCall of outputMessageToolCalls) {
       const toolArgs = toolCall.args ?? toolCall.input
       const toolCallArgs = typeof toolArgs === 'string' ? getJsonStringValue(toolArgs, {}) : toolArgs
-      const toolDescription = toolsForModel?.find(tool => toolCall.toolName === tool.name)?.description
-      const name = this.findToolName(toolCall.toolName, toolDescription)
-      this.#toolCallIdsToName[toolCall.toolCallId] = name
-
       formattedToolCalls.push({
         arguments: toolCallArgs,
-        name,
+        name: toolCall.toolName,
         toolId: toolCall.toolCallId,
         type: toolCall.toolCallType ?? 'function',
       })
@@ -328,11 +283,10 @@ class DdTelemetryPlugin extends BaseLLMObsPlugin {
    * need to split into multiple messages.
    *
    * @param {AiSdkMessage} message
-   * @param {ToolForModel[] | null | undefined} toolsForModel
    * @returns {Array<{role: string, content: string, toolId?: string,
    *   toolCalls?: Array<{arguments: string, name: string, toolId: string, type: string}>}>}
    */
-  formatMessage (message, toolsForModel) {
+  formatMessage (message) {
     const { role, content } = message
 
     if (role === 'system') {
@@ -357,12 +311,9 @@ class DdTelemetryPlugin extends BaseLLMObsPlugin {
         if (['text', 'reasoning', 'redacted-reasoning'].includes(type)) {
           finalContent += part.text ?? part.data
         } else if (type === 'tool-call') {
-          const toolDescription = toolsForModel?.find(tool => part.toolName === tool.name)?.description
-          const name = this.findToolName(part.toolName, toolDescription)
-
           toolCalls.push({
             arguments: part.args ?? part.input,
-            name,
+            name: part.toolName,
             toolId: part.toolCallId,
             type: 'function',
           })

package/packages/dd-trace/src/llmobs/plugins/ai/vercelTelemetry.js

@@ -38,11 +38,15 @@ function formatLanguageModelInputMessages (instructions, messages) {
   const inputMessages = []
 
   if (instructions) {
-    const systemPrompt = typeof instructions === 'string'
-      ? instructions
-      : Array.isArray(instructions)
-        ? instructions.map(instruction => instruction.content).join('')
-        : instructions.content
+    let systemPrompt = instructions
+    if (typeof instructions !== 'string') {
+      if (Array.isArray(instructions)) {
+        systemPrompt = ''
+        for (const instruction of instructions) systemPrompt += instruction.content
+      } else {
+        systemPrompt = instructions.content
+      }
+    }
 
     inputMessages.push({ role: 'system', content: systemPrompt })
   }
@@ -53,13 +57,13 @@ function formatLanguageModelInputMessages (instructions, messages) {
     if (role === 'system') {
       inputMessages.push({ role, content })
     } else if (role === 'user') {
-      const userMessageContent =
-      typeof content === 'string'
-        ? content
-        : content
-          .filter(part => part.type === 'text')
-          .map(part => part.text)
-          .join('')
+      let userMessageContent = content
+      if (typeof content !== 'string') {
+        userMessageContent = ''
+        for (const part of content) {
+          if (part.type === 'text') userMessageContent += part.text
+        }
+      }
 
       inputMessages.push({ role, content: userMessageContent })
     } else if (role === 'assistant') {
@@ -292,7 +296,13 @@ class VercelAiTelemetryPlugin extends BaseLLMObsPlugin {
     const { event, result } = ctx
 
     const lastUserPrompt = event.messages.findLast(message => message.role === 'user')?.content
-    const input = Array.isArray(lastUserPrompt) ? lastUserPrompt.map(part => part.text ?? '').join('') : lastUserPrompt
+    let input = lastUserPrompt
+    if (Array.isArray(lastUserPrompt)) {
+      input = ''
+      for (const part of lastUserPrompt) {
+        if (part.type === 'text') input += part.text
+      }
+    }
 
     const output =
       ctx.isStream

package/packages/dd-trace/src/llmobs/plugins/langchain/index.js

@@ -46,6 +46,13 @@ class BaseLangChainLLMObsPlugin extends LLMObsPlugin {
     }
   }
 
+  error (ctx) {
+    // NodeInterrupt or GraphInterrupt "errors" for control flow, do not mark as real errors
+    if (ctx.error?.is_bubble_up) return
+
+    super.error(ctx)
+  }
+
   getLLMObsSpanRegisterOptions (ctx) {
     const span = ctx.currentStore?.span
     const spanContext = span?.context()

package/packages/dd-trace/src/llmobs/util.js

@@ -337,37 +337,6 @@ function resolveAgentAttribution (tags, span) {
   return { name: tags[PARENT_AGENT_NAME], spanId: tags[PARENT_AGENT_SPAN_ID] }
 }
 
-/**
- * Removes all `key=value` entries for the given key from a comma-separated tagset string.
- *
- * @param {string} tags - Existing tagset string (may be empty).
- * @param {string} key
- * @returns {string}
- */
-function stripTagsetEntry (tags, key) {
-  if (!tags.includes(key)) return tags
-  return tags.split(',').filter(entry => !entry.startsWith(`${key}=`)).join(',')
-}
-
-/**
- * Appends `key=value` to the tagset string with a comma separator, but only when `value` is
- * truthy, passes the optional `safeguard` predicate, and fits within `maxTagSetLength`. Returns
- * the original `tags` unchanged when the value is absent, unsafe, or would overflow the budget.
- *
- * @param {string} tags - Existing tagset string (may be empty).
- * @param {string} key
- * @param {string | undefined} value
- * @param {((v: string) => boolean) | null} [safeguard]
- * @param {number} [maxTagSetLength]
- * @returns {string}
- */
-function appendOptionalPropagatedTag (tags, key, value, safeguard, maxTagSetLength) {
-  if (!value || (safeguard && !safeguard(value))) return tags
-  const entry = `${tags ? ',' : ''}${key}=${value}`
-  if (maxTagSetLength != null && tags.length + entry.length > maxTagSetLength) return tags
-  return `${tags}${entry}`
-}
-
 function spanHasError (span) {
   const spanContext = span.context()
   return !!(spanContext.getTag('error') || spanContext.getTag('error.type'))
@@ -518,7 +487,6 @@ function formatAudioPart (data, mimeType) {
 
 module.exports = {
   agentNameWireSafe,
-  appendOptionalPropagatedTag,
   audioMimeTypeFromFormat,
   encodeUnicode,
   findGenAIAncestorSpanId,
@@ -527,7 +495,6 @@ module.exports = {
   normalizeLlmObsTraceId,
   formatAudioPart,
   resolveAgentAttribution,
-  stripTagsetEntry,
   validateCostTags,
   validateKind,
   getFunctionArguments,

package/packages/dd-trace/src/llmobs/writers/base.js

@@ -17,13 +17,20 @@ const {
 } = require('../constants/writers')
 const { parseResponseAndLog } = require('./util')
 
+const MAX_BUFFER_EVENTS = 1000
+
 class LLMObsBuffer {
-  constructor ({ events, size, routing = {}, isDefault = false, limit = 1000 }) {
+  /**
+   * @param {{
+   *   events: Record<string, unknown>[],
+   *   size: number,
+   *   routing?: { apiKey?: string, site?: string }
+   * }} options
+   */
+  constructor ({ events, size, routing = {} }) {
     this.events = events
     this.size = size
     this.routing = routing
-    this.isDefault = isDefault
-    this.limit = limit
   }
 
   clear () {
@@ -46,7 +53,7 @@ class BaseLLMObsWriter {
     this._eventType = eventType
 
     /** @type {LLMObsBuffer} */
-    this._buffer = new LLMObsBuffer({ events: [], size: 0, isDefault: true })
+    this._buffer = new LLMObsBuffer({ events: [], size: 0 })
 
     /** @type {import('../../config/config-base')} */
     this._config = config
@@ -102,8 +109,8 @@ class BaseLLMObsWriter {
   append (event, routing, byteLength) {
     const buffer = this._getBuffer(routing)
 
-    if (buffer.events.length >= buffer.limit) {
-      logger.warn(`${this.constructor.name} event buffer full (limit is ${buffer.limit}), dropping event`)
+    if (buffer.events.length >= MAX_BUFFER_EVENTS) {
+      logger.warn(`${this.constructor.name} event buffer full (limit is ${MAX_BUFFER_EVENTS}), dropping event`)
       telemetry.recordDroppedPayload(1, this._eventType, 'buffer_full')
       return false
     }
@@ -158,30 +165,48 @@ class BaseLLMObsWriter {
     }
 
     for (const [apiKey, buffer] of this.#multiTenantBuffers) {
-      if (buffer.events.length === 0) continue
-      const events = buffer.events
-      buffer.clear()
+      if (buffer.events.length === 0) {
+        this.#multiTenantBuffers.delete(apiKey)
+        continue
+      }
       const site = buffer.routing.site || this._config.site
-      const options = {
-        headers: {
-          'Content-Type': 'application/json',
-          'DD-API-KEY': apiKey,
-        },
-        method: 'POST',
-        timeout: this._timeout,
-        url: new URL(format({
-          protocol: 'https:',
-          hostname: `${this._intake}.${site}`,
-        })),
-        path: this._baseEndpoint,
+      const maskedApiKey = `****${apiKey.slice(-4)}`
+      let options
+      let url
+      try {
+        options = {
+          headers: {
+            'Content-Type': 'application/json',
+            'DD-API-KEY': apiKey,
+          },
+          method: 'POST',
+          timeout: this._timeout,
+          url: new URL(format({
+            protocol: 'https:',
+            hostname: `${this._intake}.${site}`,
+          })),
+          path: this._baseEndpoint,
+        }
+        url = this.#buildUrl(options.url.href, options.path)
+      } catch (error) {
+        buffer.clear()
+        this.#multiTenantBuffers.delete(apiKey)
+        logger.error(
+          'Failed to route LLMObs %s events for API key %s: %s',
+          this._eventType,
+          maskedApiKey,
+          error.message
+        )
+        continue
       }
-      const url = this.#buildUrl(options.url.href, options.path)
-      const maskedApiKey = apiKey ? `****${apiKey.slice(-4)}` : ''
+
+      const events = buffer.events
+      buffer.clear()
+      this.#multiTenantBuffers.delete(apiKey)
       log.debug('Encoding and flushing multi-tenant buffer for %s', maskedApiKey)
       requests.push({ events, options, url })
     }
 
-    this.#cleanupEmptyBuffers()
     return requests
   }
 
@@ -194,14 +219,6 @@ class BaseLLMObsWriter {
     })
   }
 
-  #cleanupEmptyBuffers () {
-    for (const [key, buffer] of this.#multiTenantBuffers) {
-      if (buffer.events.length === 0) {
-        this.#multiTenantBuffers.delete(key)
-      }
-    }
-  }
-
   makePayload (events) {}
 
   /**

package/packages/dd-trace/src/log/index.js

@@ -29,7 +29,13 @@ const log = {
       const stack = logRecord.stack.split('\n')
       const fn = stack[1].replace(/^\s+at ([^\s]+) .+/, '$1')
       const options = { depth: 2, breakLength: Infinity, compact: true, maxArrayLength: Infinity }
-      const params = args.map(a => inspect(a, options)).join(', ')
+      let params = ''
+      let isFirstParameter = true
+      for (const argument of args) {
+        if (!isFirstParameter) params += ', '
+        params += inspect(argument, options)
+        isFirstParameter = false
+      }
 
       stack[0] = `Trace: ${fn}(${params})`