Add function to extract Github display name automatically

packages/base/src/helpers/__tests__/ci.test.ts

@@ -2,11 +2,17 @@ import fs from 'fs'
 
 import upath from 'upath'
 
-import {getCIEnv, getCIMetadata, getCISpanTags, isInteractive} from '../ci'
+import {
+  getCIEnv,
+  getCIMetadata,
+  getCISpanTags,
+  getGithubJobDisplayNameFromLogs,
+  githubWellKnownDiagnosticDirs,
+  isInteractive,
+} from '../ci'
 import {SpanTags} from '../interfaces'
 import {
   CI_ENV_VARS,
-  CI_JOB_ID,
   CI_NODE_LABELS,
   CI_NODE_NAME,
   GIT_HEAD_SHA,
@@ -427,6 +433,170 @@ describe('isInteractive', () => {
   })
 })
 
+describe('getGithubJobDisplayNameFromLogs', () => {
+  const mockedFs = fs as jest.Mocked<typeof fs>
+
+  afterEach(() => {
+    jest.resetAllMocks()
+  })
+
+  const getNotFoundFsError = (): Error => {
+    const error = new Error('not found')
+    Object.assign(error, {code: 'ENOENT'})
+
+    return error
+  }
+
+  const mockLogFileDirent = (logFileName: string): fs.Dirent => {
+    return {
+      name: logFileName,
+      isFile: () => true,
+      isDirectory: () => false,
+      isBlockDevice: () => false,
+      isCharacterDevice: () => false,
+      isSymbolicLink: () => false,
+      isFIFO: () => false,
+      isSocket: () => false,
+      parentPath: '',
+      path: '',
+    }
+  }
+
+  const sampleLogContent = (jobDisplayName: string): string => `
+    [2025-09-15 10:14:00Z INFO Worker] Waiting to receive the job message from the channel.
+    [2025-09-15 10:14:00Z INFO ProcessChannel] Receiving message of length 22985, with hash 'abcdef'
+    [2025-09-15 10:14:00Z INFO Worker] Message received.
+    [2025-09-15 10:14:00Z INFO Worker] Job message:
+    {
+      "jobId": "95a4619c-e316-542f-8a21-74cd5a8ac9ca",
+      "jobDisplayName": "${jobDisplayName}",
+      "jobName": "__default"
+    }`
+
+  const sampleLogFileName = 'Worker_20251014-083000.log'
+  const sampleJobDisplayName = 'build-and-test'
+
+  const mockReaddirSync = (targetDir: fs.PathLike, logFileName: string) => {
+    fs.readdirSync = jest.fn().mockImplementation((pathToRead: fs.PathLike): fs.Dirent[] => {
+      if (pathToRead === targetDir) {
+        return [mockLogFileDirent(logFileName)]
+      }
+      throw getNotFoundFsError()
+    })
+  }
+
+  test('should find and return the job display name (SaaS)', () => {
+    const targetDir = githubWellKnownDiagnosticDirs[0] // SaaS directory
+    const logContent = sampleLogContent(sampleJobDisplayName)
+
+    mockReaddirSync(targetDir, sampleLogFileName)
+    jest.spyOn(fs, 'readFileSync').mockReturnValue(logContent)
+
+    const result = getGithubJobDisplayNameFromLogs()
+
+    expect(result).toBe(sampleJobDisplayName)
+    expect(mockedFs.readdirSync).toHaveBeenCalledWith(targetDir, {withFileTypes: true})
+    expect(mockedFs.readFileSync).toHaveBeenCalledWith(`${targetDir}/${sampleLogFileName}`, 'utf-8')
+  })
+
+  test('should find and return the job display name (self-hosted)', () => {
+    const targetDir = githubWellKnownDiagnosticDirs[1] // self-hosted directory
+    const logContent = sampleLogContent(sampleJobDisplayName)
+
+    mockReaddirSync(targetDir, sampleLogFileName)
+    jest.spyOn(fs, 'readFileSync').mockReturnValue(logContent)
+
+    const result = getGithubJobDisplayNameFromLogs()
+
+    expect(result).toBe(sampleJobDisplayName)
+    expect(mockedFs.readdirSync).toHaveBeenCalledWith(targetDir, {withFileTypes: true})
+    expect(mockedFs.readFileSync).toHaveBeenCalledWith(`${targetDir}/${sampleLogFileName}`, 'utf-8')
+  })
+
+  test('should check multiple log files until the display name is found', () => {
+    const logContent1 = 'no job display name here'
+    const logContent2 = 'nor here'
+    const logContent3 = sampleLogContent('my job name')
+
+    fs.readdirSync = jest.fn().mockImplementation((pathToRead: fs.PathLike): fs.Dirent[] => {
+      return [mockLogFileDirent('Worker_1.log'), mockLogFileDirent('Worker_2.log'), mockLogFileDirent('Worker_3.log')]
+    })
+    jest.spyOn(fs, 'readFileSync').mockReturnValue(logContent1)
+    jest.spyOn(fs, 'readFileSync').mockReturnValue(logContent2)
+    jest.spyOn(fs, 'readFileSync').mockReturnValue(logContent3)
+
+    const result = getGithubJobDisplayNameFromLogs()
+
+    expect(result).toBe('my job name')
+  })
+
+  test('should throw an error if no diagnostic log directories are found', () => {
+    fs.readdirSync = jest.fn().mockImplementation((pathToRead: fs.PathLike): fs.Dirent[] => {
+      throw getNotFoundFsError()
+    })
+
+    expect(() => {
+      getGithubJobDisplayNameFromLogs()
+    }).toThrow('could not find Github diagnostic log files')
+  })
+
+  test('should throw an error if no worker log files are found in any directory', () => {
+    fs.readdirSync = jest.fn().mockImplementation((pathToRead: fs.PathLike): fs.Dirent[] => {
+      return [mockLogFileDirent('random_file_1'), mockLogFileDirent('random_file_2')]
+    })
+
+    expect(() => {
+      getGithubJobDisplayNameFromLogs()
+    }).toThrow('could not find Github diagnostic log files')
+  })
+
+  test('should throw an error if log files are found but none contain the display name', () => {
+    const targetDir = githubWellKnownDiagnosticDirs[0]
+    const logContent = 'This log does not have the job display name.'
+
+    mockReaddirSync(targetDir, sampleLogFileName)
+    jest.spyOn(fs, 'readFileSync').mockReturnValue(logContent)
+
+    expect(() => {
+      getGithubJobDisplayNameFromLogs()
+    }).toThrow('could not find jobDisplayName attribute in Github diagnostic logs')
+  })
+
+  test('should throw an error if reading a directory throws an unexpected error', () => {
+    const accessDeniedError = new Error('access denied')
+    Object.assign(accessDeniedError, {code: 'EACCES'})
+
+    fs.readdirSync = jest.fn().mockImplementation((pathToRead: fs.PathLike): fs.Dirent[] => {
+      throw accessDeniedError
+    })
+
+    expect(() => {
+      getGithubJobDisplayNameFromLogs()
+    }).toThrow(`error reading Github diagnostic log files: access denied`)
+  })
+
+  test('should re-throw for unexpected errors', () => {
+    const err = Error('some error')
+
+    fs.readdirSync = jest.fn().mockImplementation((pathToRead: fs.PathLike): fs.Dirent[] => {
+      throw err
+    })
+
+    expect(() => {
+      getGithubJobDisplayNameFromLogs()
+    }).toThrow(new Error('error reading Github diagnostic log files: some error'))
+
+    const stringErr = 'hello error'
+    fs.readdirSync = jest.fn().mockImplementation((pathToRead: fs.PathLike): fs.Dirent[] => {
+      // eslint-disable-next-line @typescript-eslint/only-throw-error
+      throw stringErr
+    })
+    expect(() => {
+      getGithubJobDisplayNameFromLogs()
+    }).toThrow(stringErr)
+  })
+})
+
 const getTags = (): SpanTags => {
   return {
     ...getCISpanTags(),

packages/base/src/helpers/ci.ts

@@ -1,3 +1,7 @@
+import fs from 'fs'
+
+import upath from 'upath'
+
 import {Metadata, SpanTag, SpanTags} from './interfaces'
 import {
   CI_ENV_VARS,
@@ -56,13 +60,18 @@ export const CI_ENGINES = {
   TEAMCITY: 'teamcity',
 }
 
-export const PROVIDER_TO_DISPLAY_NAME = {
-  github: 'GitHub Actions',
-}
+export const envDDGithubJobName = 'DD_GITHUB_JOB_NAME'
 
 // DD_GITHUB_JOB_NAME is an override that is required for adding custom tags and metrics
 // to GHA jobs if the 'name' property is used. It's ok for it to be missing in case the name property is not used.
-const envAllowedToBeMissing = ['DD_GITHUB_JOB_NAME']
+const envAllowedToBeMissing = [envDDGithubJobName]
+
+export const githubWellKnownDiagnosticDirs = [
+  '/home/runner/actions-runner/cached/_diag', // for SaaS
+  '/home/runner/actions-runner/_diag', // for self-hosted
+]
+
+const githubJobDisplayNameRegex = /"jobDisplayName":\s*"([^"]+)"/
 
 // Receives a string with the form 'John Doe <[email protected]>'
 // and returns { name: 'John Doe', email: '[email protected]' }
@@ -901,7 +910,7 @@ export const getCIEnv = (): {ciEnv: Record<string, string>; provider: string} =>
         'GITHUB_RUN_ID',
         'GITHUB_RUN_ATTEMPT',
         'GITHUB_JOB',
-        'DD_GITHUB_JOB_NAME',
+        envDDGithubJobName,
       ]),
       provider: 'github',
     }
@@ -1020,3 +1029,61 @@ const filterEnv = (values: string[]): Record<string, string> => {
 export const isInteractive = ({stream = process.stdout}: {stream?: NodeJS.WriteStream} = {}) => {
   return Boolean(!('CI' in process.env) && process.env.TERM !== 'dumb' && stream && stream.isTTY)
 }
+
+export const shouldGetGithubJobDisplayName = (): boolean => {
+  return getCIProvider() !== CI_ENGINES.GITHUB && process.env.DD_GITHUB_JOB_NAME !== ''
+}
+
+/**
+ * Extracts the job display name from the GitHub Actions diagnostic log files.
+ *
+ * @returns The job display name, or an empty string if not found.
+ */
+export const getGithubJobDisplayNameFromLogs = (): string => {
+  let foundDiagDir = ''
+  let workerLogFiles: string[] = []
+
+  // 1. Iterate through well known directories to check for worker logs
+  for (const currentDir of githubWellKnownDiagnosticDirs) {
+    try {
+      const files = fs.readdirSync(currentDir, {withFileTypes: true})
+      const potentialLogs = files
+        .filter((file) => file.isFile() && file.name.startsWith('Worker_') && file.name.endsWith('.log'))
+        .map((file) => file.name)
+
+      if (potentialLogs.length > 0) {
+        foundDiagDir = currentDir
+        workerLogFiles = potentialLogs
+        break
+      }
+    } catch (error) {
+      // If the directory does not exist, just try the next one
+      if (error instanceof Error && 'code' in error && error.code === 'ENOENT') {
+        continue
+      }
+      if (error instanceof Error) {
+        throw new Error(`error reading Github diagnostic log files: ${error.message}`)
+      } else {
+        throw error
+      }
+    }
+  }
+  if (workerLogFiles.length === 0 || foundDiagDir === '') {
+    throw new Error('could not find Github diagnostic log files')
+  }
+
+  // 2. Get the job display name via regex
+  for (const logFile of workerLogFiles) {
+    const filePath = upath.join(foundDiagDir, logFile)
+    const content = fs.readFileSync(filePath, 'utf-8')
+
+    const match = content.match(githubJobDisplayNameRegex)
+
+    if (match && match[1]) {
+      // match[1] is the captured group
+      return match[1]
+    }
+  }
+
+  throw Error('could not find jobDisplayName attribute in Github diagnostic logs')
+}

packages/datadog-ci/src/commands/measure/measure.ts

@@ -2,7 +2,12 @@ import type {AxiosError} from 'axios'
 
 import {BaseCommand} from '@datadog/datadog-ci-base'
 import {FIPS_ENV_VAR, FIPS_IGNORE_ERROR_ENV_VAR} from '@datadog/datadog-ci-base/constants'
-import {getCIEnv} from '@datadog/datadog-ci-base/helpers/ci'
+import {
+  envDDGithubJobName,
+  getCIEnv,
+  getGithubJobDisplayNameFromLogs,
+  shouldGetGithubJobDisplayName,
+} from '@datadog/datadog-ci-base/helpers/ci'
 import {toBoolean} from '@datadog/datadog-ci-base/helpers/env'
 import {enableFips} from '@datadog/datadog-ci-base/helpers/fips'
 import {retryRequest} from '@datadog/datadog-ci-base/helpers/retry'
@@ -97,9 +102,25 @@ export class MeasureCommand extends BaseCommand {
       return 1
     }
 
+    let githubDisplayName = ''
+    try {
+      if (shouldGetGithubJobDisplayName()) {
+        this.context.stdout.write('Inferring github job name\n')
+        githubDisplayName = getGithubJobDisplayNameFromLogs()
+      }
+    } catch (error) {
+      this.context.stdout.write(
+        `could not infer job display name, defaulting to env variables ${error instanceof Error ? error.message : ''}\n`
+      )
+    }
+
     try {
       const {provider, ciEnv} = getCIEnv()
 
+      if (githubDisplayName !== '' && !Object.values(ciEnv).includes(envDDGithubJobName)) {
+        ciEnv[envDDGithubJobName] = githubDisplayName
+      }
+
       const exitStatus = await this.sendMeasures(ciEnv, this.level === 'pipeline' ? 0 : 1, provider, measures)
       if (exitStatus !== 0 && this.noFail) {
         this.context.stderr.write(