Add:Experimental tone library for scanning metadata

2024-12-20 19:06:06 +01:00 · 2022-09-11 15:35:06 -05:00 · 2022-09-11 15:35:06 -05:00 · c16e6d19ae
commit c16e6d19ae
parent 0e98620939
12 changed files with 235 additions and 277 deletions
--- a/3
+++ b/3
@ -6,7 +6,9 @@ RUN npm ci && npm cache clean --force
 RUN npm run generate

 ### STAGE 1: Build server ###
+FROM sandreas/tone:v0.0.9 AS tone
 FROM node:16-alpine
+
 ENV NODE_ENV=production
 RUN apk update && \
    apk add --no-cache --update \
@ -14,6 +16,7 @@ RUN apk update && \
    tzdata \
    ffmpeg

+COPY --from=tone /usr/local/bin/tone /usr/local/bin/
 COPY --from=build /client/dist /client/dist
 COPY index.js package* /
 COPY server server
--- a/client/pages/config/index.vue
+++ b/client/pages/config/index.vue
@ -172,17 +172,15 @@
          </div>

          <div class="flex items-center py-2">
-            <div class="flex items-center">
-              <ui-toggle-switch v-model="showExperimentalFeatures" />
-              <ui-tooltip :text="tooltips.experimentalFeatures">
-                <p class="pl-4">
-                  Experimental Features
-                  <a href="https://github.com/advplyr/audiobookshelf/discussions/75" target="_blank">
-                    <span class="material-icons icon-text text-sm">info_outlined</span>
-                  </a>
-                </p>
-              </ui-tooltip>
-            </div>
+            <ui-toggle-switch v-model="showExperimentalFeatures" />
+            <ui-tooltip :text="tooltips.experimentalFeatures">
+              <p class="pl-4">
+                Experimental Features
+                <a href="https://github.com/advplyr/audiobookshelf/discussions/75" target="_blank">
+                  <span class="material-icons icon-text text-sm">info_outlined</span>
+                </a>
+              </p>
+            </ui-tooltip>
          </div>

          <div class="flex items-center py-2">
@ -195,15 +193,15 @@
            </ui-tooltip>
          </div>

-          <!-- <div class="flex items-center py-2">
-            <ui-toggle-switch v-model="newServerSettings.scannerUseSingleThreadedProber" :disabled="updatingServerSettings" @input="(val) => updateSettingsKey('scannerUseSingleThreadedProber', val)" />
-            <ui-tooltip :text="tooltips.scannerUseSingleThreadedProber">
+          <div class="flex items-center py-2">
+            <ui-toggle-switch v-model="newServerSettings.scannerUseTone" :disabled="updatingServerSettings" @input="(val) => updateSettingsKey('scannerUseTone', val)" />
+            <ui-tooltip text="Tone library for metadata">
              <p class="pl-4">
-                Scanner use old single threaded audio prober
+                Use Tone library for metadata
                <span class="material-icons icon-text text-sm">info_outlined</span>
              </p>
            </ui-tooltip>
-          </div> -->
+          </div>
        </div>
      </div>
    </div>
--- a/docker-compose.yml
+++ b/docker-compose.yml
@ -3,11 +3,11 @@ version: "3.7"

 services:
  audiobookshelf:
-    image: ghcr.io/advplyr/audiobookshelf
+    image: audiobookshelf-test
    ports:
      - 13378:80
    volumes:
-      - ./audiobooks:/audiobooks
-      - ./metadata:/metadata
-      - ./config:/config
+      - ./media/audiobooks:/audiobooks
+      - ./test/metadata:/metadata
+      - ./test/config:/config
    restart: unless-stopped
--- a/package-lock.json
+++ b/package-lock.json
@ -13,6 +13,7 @@
        "express": "^4.17.1",
        "graceful-fs": "^4.2.10",
        "htmlparser2": "^8.0.1",
+        "node-tone": "^1.0.1",
        "socket.io": "^4.4.1",
        "xml2js": "^0.4.23"
      },
@ -594,6 +595,11 @@
        "node": ">= 0.6"
      }
    },
+    "node_modules/node-tone": {
+      "version": "1.0.1",
+      "resolved": "https://registry.npmjs.org/node-tone/-/node-tone-1.0.1.tgz",
+      "integrity": "sha512-wi7L0taDZMN6tM5l85TDKHsYzdhqJTtPNgvgpk2zHeZzPt6ZIUZ9vBLTJRRDpm0xzCvbsvFHjAaudeQjLHTE4w=="
+    },
    "node_modules/object-assign": {
      "version": "4.1.1",
      "resolved": "https://registry.npmjs.org/object-assign/-/object-assign-4.1.1.tgz",
@ -1360,6 +1366,11 @@
      "resolved": "https://registry.npmjs.org/negotiator/-/negotiator-0.6.3.tgz",
      "integrity": "sha512-+EUsqGPLsM+j/zdChZjsnX51g4XrHFOIXwfnCVPGlQk/k5giakcKsuxCObBRu6DSm9opw/O6slWbJdghQM4bBg=="
    },
+    "node-tone": {
+      "version": "1.0.1",
+      "resolved": "https://registry.npmjs.org/node-tone/-/node-tone-1.0.1.tgz",
+      "integrity": "sha512-wi7L0taDZMN6tM5l85TDKHsYzdhqJTtPNgvgpk2zHeZzPt6ZIUZ9vBLTJRRDpm0xzCvbsvFHjAaudeQjLHTE4w=="
+    },
    "object-assign": {
      "version": "4.1.1",
      "resolved": "https://registry.npmjs.org/object-assign/-/object-assign-4.1.1.tgz",
@ -1602,4 +1613,4 @@
      "integrity": "sha512-fDlsI/kFEx7gLvbecc0/ohLG50fugQp8ryHzMTuW9vSa1GJ0XYWKnhsUx7oie3G98+r56aTQIUB4kht42R3JvA=="
    }
  }
-}
+}
--- a/package.json
+++ b/package.json
@ -34,7 +34,8 @@
    "express": "^4.17.1",
    "graceful-fs": "^4.2.10",
    "htmlparser2": "^8.0.1",
+    "node-tone": "^1.0.1",
    "socket.io": "^4.4.1",
    "xml2js": "^0.4.23"
  }
-}
+}
--- a/server/objects/metadata/AudioMetaTags.js
+++ b/server/objects/metadata/AudioMetaTags.js
@ -87,6 +87,10 @@ class AudioMetaTags {
    this.tagOverdriveMediaMarker = payload.file_tag_overdrive_media_marker || null
  }

+  setDataFromTone(tags) {
+    // TODO: Implement
+  }
+
  updateData(payload) {
    const dataMap = {
      tagAlbum: payload.file_tag_album || null,
--- a/server/objects/settings/ServerSettings.js
+++ b/server/objects/settings/ServerSettings.js
@ -17,7 +17,8 @@ class ServerSettings {
    this.scannerDisableWatcher = false
    this.scannerPreferOverdriveMediaMarker = false
    this.scannerUseSingleThreadedProber = true
-    this.scannerMaxThreads = 0 // 0 = defaults to CPUs * 2
+    this.scannerMaxThreads = 0 // Currently not being used
+    this.scannerUseTone = false

    // Metadata - choose to store inside users library item folder
    this.storeCoverWithItem = false
@ -82,6 +83,7 @@ class ServerSettings {
      this.scannerUseSingleThreadedProber = true
    }
    this.scannerMaxThreads = isNullOrNaN(settings.scannerMaxThreads) ? 0 : Number(settings.scannerMaxThreads)
+    this.scannerUseTone = !!settings.scannerUseTone

    this.storeCoverWithItem = !!settings.storeCoverWithItem
    this.storeMetadataWithItem = !!settings.storeMetadataWithItem
@ -139,6 +141,7 @@ class ServerSettings {
      scannerPreferOverdriveMediaMarker: this.scannerPreferOverdriveMediaMarker,
      scannerUseSingleThreadedProber: this.scannerUseSingleThreadedProber,
      scannerMaxThreads: this.scannerMaxThreads,
+      scannerUseTone: this.scannerUseTone,
      storeCoverWithItem: this.storeCoverWithItem,
      storeMetadataWithItem: this.storeMetadataWithItem,
      rateLimitLoginRequests: this.rateLimitLoginRequests,
--- a/server/scanner/MediaFileScanner.js
+++ b/server/scanner/MediaFileScanner.js
@ -3,9 +3,8 @@ const Path = require('path')
 const AudioFile = require('../objects/files/AudioFile')
 const VideoFile = require('../objects/files/VideoFile')

-const MediaProbePool = require('./MediaProbePool')
-
 const prober = require('../utils/prober')
+const toneProber = require('../utils/toneProber')
 const Logger = require('../Logger')
 const { LogLevel } = require('../utils/constants')

@ -59,7 +58,15 @@ class MediaFileScanner {

  async scan(mediaType, libraryFile, mediaMetadataFromScan, verbose = false) {
    var probeStart = Date.now()
-    var probeData = await prober.probe(libraryFile.metadata.path, verbose)
+
+    var probeData = null
+    if (global.ServerSettings.scannerUseTone) {
+      Logger.debug(`[MediaFileScanner] using tone to probe audio file "${libraryFile.metadata.path}"`)
+      probeData = await toneProber.probe(libraryFile.metadata.path, true)
+    } else {
+      probeData = await prober.probe(libraryFile.metadata.path, verbose)
+    }
+
    if (probeData.error) {
      Logger.error(`[MediaFileScanner] ${probeData.error} : "${libraryFile.metadata.path}"`)
      return null
@ -105,35 +112,18 @@ class MediaFileScanner {
  async executeMediaFileScans(libraryItem, mediaLibraryFiles, scanData) {
    const mediaType = libraryItem.mediaType

-    if (!global.ServerSettings.scannerUseSingleThreadedProber) { // New multi-threaded scanner
-      var scanStart = Date.now()
-      const probeResults = await new Promise((resolve) => {
-        // const probePool = new MediaProbePool(mediaType, mediaLibraryFiles, scanData, global.ServerSettings.scannerMaxThreads)
-        const itemBatch = MediaProbePool.initBatch(libraryItem, mediaLibraryFiles, scanData)
-        itemBatch.on('done', resolve)
-        MediaProbePool.runBatch(itemBatch)
-      })
-
-      return {
-        audioFiles: probeResults.audioFiles || [],
-        videoFiles: probeResults.videoFiles || [],
-        elapsed: Date.now() - scanStart,
-        averageScanDuration: probeResults.averageTimePerMb
-      }
-    } else { // Old single threaded scanner
-      var scanStart = Date.now()
-      var mediaMetadataFromScan = scanData.media.metadata || null
-      var proms = []
-      for (let i = 0; i < mediaLibraryFiles.length; i++) {
-        proms.push(this.scan(mediaType, mediaLibraryFiles[i], mediaMetadataFromScan))
-      }
-      var results = await Promise.all(proms).then((scanResults) => scanResults.filter(sr => sr))
-      return {
-        audioFiles: results.filter(r => r.audioFile).map(r => r.audioFile),
-        videoFiles: results.filter(r => r.videoFile).map(r => r.videoFile),
-        elapsed: Date.now() - scanStart,
-        averageScanDuration: this.getAverageScanDurationMs(results)
-      }
+    var scanStart = Date.now()
+    var mediaMetadataFromScan = scanData.media.metadata || null
+    var proms = []
+    for (let i = 0; i < mediaLibraryFiles.length; i++) {
+      proms.push(this.scan(mediaType, mediaLibraryFiles[i], mediaMetadataFromScan))
+    }
+    var results = await Promise.all(proms).then((scanResults) => scanResults.filter(sr => sr))
+    return {
+      audioFiles: results.filter(r => r.audioFile).map(r => r.audioFile),
+      videoFiles: results.filter(r => r.videoFile).map(r => r.videoFile),
+      elapsed: Date.now() - scanStart,
+      averageScanDuration: this.getAverageScanDurationMs(results)
    }
  }

--- a/server/scanner/MediaProbeData.js
+++ b/server/scanner/MediaProbeData.js
@ -66,15 +66,20 @@ class MediaProbeData {
    this.sampleRate = audioStream.sample_rate
    this.chapters = data.chapters || []

-    var metatags = {}
-    for (const key in data) {
-      if (data[key] && key.startsWith('file_tag')) {
-        metatags[key] = data[key]
+    if (data.tags) { // New for tone library data (toneProber.js)
+      this.audioFileMetadata = new AudioFileMetadata()
+      this.audioFileMetadata.setDataFromTone(data.tags)
+    } else { // Data from ffprobe (prober.js)
+      var metatags = {}
+      for (const key in data) {
+        if (data[key] && key.startsWith('file_tag')) {
+          metatags[key] = data[key]
+        }
      }
-    }

-    this.audioFileMetadata = new AudioFileMetadata()
-    this.audioFileMetadata.setData(metatags)
+      this.audioFileMetadata = new AudioFileMetadata()
+      this.audioFileMetadata.setData(metatags)
+    }

    // Track ID3 tag might be "3/10" or just "3"
    if (this.audioFileMetadata.tagTrack) {
--- a/server/scanner/MediaProbePool.js
+++ b/server/scanner/MediaProbePool.js
@ -1,209 +0,0 @@
-const os = require('os')
-const Path = require('path')
-const { EventEmitter } = require('events')
-const { Worker } = require("worker_threads")
-const Logger = require('../Logger')
-const AudioFile = require('../objects/files/AudioFile')
-const VideoFile = require('../objects/files/VideoFile')
-const MediaProbeData = require('./MediaProbeData')
-
-class LibraryItemBatch extends EventEmitter {
-  constructor(libraryItem, libraryFiles, scanData) {
-    super()
-
-    this.id = libraryItem.id
-    this.mediaType = libraryItem.mediaType
-    this.mediaMetadataFromScan = scanData.media.metadata || null
-    this.libraryFilesToScan = libraryFiles
-
-    // Results
-    this.totalElapsed = 0
-    this.totalProbed = 0
-    this.audioFiles = []
-    this.videoFiles = []
-  }
-
-  done() {
-    this.emit('done', {
-      videoFiles: this.videoFiles,
-      audioFiles: this.audioFiles,
-      averageTimePerMb: Math.round(this.totalElapsed / this.totalProbed)
-    })
-  }
-}
-
-class MediaProbePool {
-  constructor() {
-    this.MaxThreads = 0
-    this.probeWorkerScript = null
-
-    this.itemBatchMap = {}
-
-    this.probesRunning = []
-    this.probeQueue = []
-  }
-
-  tick() {
-    if (this.probesRunning.length < this.MaxThreads) {
-      if (this.probeQueue.length > 0) {
-        const pw = this.probeQueue.shift()
-        // console.log('Unqueued probe - Remaining is', this.probeQueue.length, 'Currently running is', this.probesRunning.length)
-        this.startTask(pw)
-      } else if (!this.probesRunning.length) {
-        // console.log('No more probes to run')
-      }
-    }
-  }
-
-  async startTask(task) {
-    this.probesRunning.push(task)
-
-    const itemBatch = this.itemBatchMap[task.batchId]
-
-    await task.start().then((taskResult) => {
-      itemBatch.libraryFilesToScan = itemBatch.libraryFilesToScan.filter(lf => lf.ino !== taskResult.libraryFile.ino)
-
-      var fileSizeMb = taskResult.libraryFile.metadata.size / (1024 * 1024)
-      var elapsedPerMb = Math.round(taskResult.elapsed / fileSizeMb)
-
-      const probeData = new MediaProbeData(taskResult.data)
-
-      if (itemBatch.mediaType === 'video') {
-        if (!probeData.videoStream) {
-          Logger.error('[MediaProbePool] Invalid video file no video stream')
-        } else {
-          itemBatch.totalElapsed += elapsedPerMb
-          itemBatch.totalProbed++
-
-          var videoFile = new VideoFile()
-          videoFile.setDataFromProbe(libraryFile, probeData)
-          itemBatch.videoFiles.push(videoFile)
-        }
-      } else {
-        if (!probeData.audioStream) {
-          Logger.error('[MediaProbePool] Invalid audio file no audio stream')
-        } else {
-          itemBatch.totalElapsed += elapsedPerMb
-          itemBatch.totalProbed++
-
-          var audioFile = new AudioFile()
-          audioFile.trackNumFromMeta = probeData.trackNumber
-          audioFile.discNumFromMeta = probeData.discNumber
-          if (itemBatch.mediaType === 'book') {
-            const { trackNumber, discNumber } = this.getTrackAndDiscNumberFromFilename(itemBatch.mediaMetadataFromScan, taskResult.libraryFile)
-            audioFile.trackNumFromFilename = trackNumber
-            audioFile.discNumFromFilename = discNumber
-          }
-          audioFile.setDataFromProbe(taskResult.libraryFile, probeData)
-
-          itemBatch.audioFiles.push(audioFile)
-        }
-      }
-
-      this.probesRunning = this.probesRunning.filter(tq => tq.mediaPath !== task.mediaPath)
-      this.tick()
-    }).catch((error) => {
-      itemBatch.libraryFilesToScan = itemBatch.libraryFilesToScan.filter(lf => lf.ino !== taskResult.libraryFile.ino)
-
-      Logger.error('[MediaProbePool] Task failed', error)
-      this.probesRunning = this.probesRunning.filter(tq => tq.mediaPath !== task.mediaPath)
-      this.tick()
-    })
-
-    if (!itemBatch.libraryFilesToScan.length) {
-      itemBatch.done()
-      delete this.itemBatchMap[itemBatch.id]
-    }
-  }
-
-  buildTask(libraryFile, batchId) {
-    return {
-      batchId,
-      mediaPath: libraryFile.metadata.path,
-      start: () => {
-        return new Promise((resolve, reject) => {
-          const startTime = Date.now()
-
-          const worker = new Worker(this.probeWorkerScript)
-          worker.on("message", ({ data }) => {
-            if (data.error) {
-              reject(data.error)
-            } else {
-              resolve({
-                data,
-                elapsed: Date.now() - startTime,
-                libraryFile
-              })
-            }
-          })
-          worker.postMessage({
-            mediaPath: libraryFile.metadata.path
-          })
-        })
-      }
-    }
-  }
-
-  initBatch(libraryItem, libraryFiles, scanData) {
-    this.MaxThreads = global.ServerSettings.scannerMaxThreads || (os.cpus().length * 2)
-    this.probeWorkerScript = Path.join(global.appRoot, 'server/utils/probeWorker.js')
-
-    Logger.debug(`[MediaProbePool] Run item batch ${libraryItem.id} with`, libraryFiles.length, 'files and max concurrent of', this.MaxThreads)
-
-    const itemBatch = new LibraryItemBatch(libraryItem, libraryFiles, scanData)
-    this.itemBatchMap[itemBatch.id] = itemBatch
-
-    return itemBatch
-  }
-
-  runBatch(itemBatch) {
-    for (const libraryFile of itemBatch.libraryFilesToScan) {
-      const probeTask = this.buildTask(libraryFile, itemBatch.id)
-
-      if (this.probesRunning.length < this.MaxThreads) {
-        this.startTask(probeTask)
-      } else {
-        this.probeQueue.push(probeTask)
-      }
-    }
-  }
-
-  getTrackAndDiscNumberFromFilename(mediaMetadataFromScan, audioLibraryFile) {
-    const { title, author, series, publishedYear } = mediaMetadataFromScan
-    const { filename, path } = audioLibraryFile.metadata
-    var partbasename = Path.basename(filename, Path.extname(filename))
-
-    // Remove title, author, series, and publishedYear from filename if there
-    if (title) partbasename = partbasename.replace(title, '')
-    if (author) partbasename = partbasename.replace(author, '')
-    if (series) partbasename = partbasename.replace(series, '')
-    if (publishedYear) partbasename = partbasename.replace(publishedYear)
-
-    // Look for disc number
-    var discNumber = null
-    var discMatch = partbasename.match(/\b(disc|cd) ?(\d\d?)\b/i)
-    if (discMatch && discMatch.length > 2 && discMatch[2]) {
-      if (!isNaN(discMatch[2])) {
-        discNumber = Number(discMatch[2])
-      }
-
-      // Remove disc number from filename
-      partbasename = partbasename.replace(/\b(disc|cd) ?(\d\d?)\b/i, '')
-    }
-
-    // Look for disc number in folder path e.g. /Book Title/CD01/audiofile.mp3
-    var pathdir = Path.dirname(path).split('/').pop()
-    if (pathdir && /^cd\d{1,3}$/i.test(pathdir)) {
-      var discFromFolder = Number(pathdir.replace(/cd/i, ''))
-      if (!isNaN(discFromFolder) && discFromFolder !== null) discNumber = discFromFolder
-    }
-
-    var numbersinpath = partbasename.match(/\d{1,4}/g)
-    var trackNumber = numbersinpath && numbersinpath.length ? parseInt(numbersinpath[0]) : null
-    return {
-      trackNumber,
-      discNumber
-    }
-  }
-}
-module.exports = new MediaProbePool()
--- a/server/utils/prober.js
+++ b/server/utils/prober.js
@ -200,12 +200,6 @@ function parseTags(format, verbose) {
    }
  }

-  // var keysToLookOutFor = ['file_tag_genre1', 'file_tag_genre2', 'file_tag_series', 'file_tag_seriespart', 'file_tag_movement', 'file_tag_movementname', 'file_tag_wwwaudiofile', 'file_tag_contentgroup', 'file_tag_releasetime', 'file_tag_isbn']
-  // keysToLookOutFor.forEach((key) => {
-  //   if (tags[key]) {
-  //     Logger.debug(`Notable! ${key} => ${tags[key]}`)
-  //   }
-  // })
  return tags
 }

--- a/server/utils/toneProber.js
+++ b/server/utils/toneProber.js
@ -0,0 +1,158 @@
+const tone = require('node-tone')
+const MediaProbeData = require('../scanner/MediaProbeData')
+const Logger = require('../Logger')
+
+/*
+Sample dump from tone
+{
+  "audio": {
+    "bitrate": 17,
+    "format": "MPEG-4 Part 14",
+    "formatShort": "MPEG-4",
+    "sampleRate": 44100.0,
+    "duration": 209284.0,
+    "channels": {
+      "count": 2,
+      "description": "Stereo (2/0.0)"
+    },
+    "frames": {
+      "offset": 42168,
+      "length": 446932
+    "metaFormat": [
+      "mp4"
+    ]
+  },
+  "meta": {
+    "album": "node-tone",
+    "albumArtist": "advplyr",
+    "artist": "advplyr",
+    "composer": "Composer 5",
+    "comment": "testing out tone metadata",
+    "encodingTool": "audiobookshelf",
+    "genre": "abs",
+    "itunesCompilation": "no",
+    "itunesMediaType": "audiobook",
+    "itunesPlayGap": "noGap",
+    "narrator": "Narrator 5",
+    "recordingDate": "2022-09-10T00:00:00",
+    "title": "Test 5",
+    "trackNumber": 5,
+    "chapters": [
+      {
+        "start": 0,
+        "length": 500,
+        "title": "chapter 1"
+      },
+      {
+        "start": 500,
+        "length": 500,
+        "title": "chapter 2"
+      },
+      {
+        "start": 1000,
+        "length": 208284,
+        "title": "chapter 3"
+      }
+    ],
+    "embeddedPictures": [
+      {
+        "code": 14,
+        "mimetype": "image/png",
+        "data": "..."
+    },
+    "additionalFields": {
+      "test": "Test 5"
+    }
+  },
+  "file": {
+    "size": 530793,
+    "created": "2022-09-10T13:32:51.1942586-05:00",
+    "modified": "2022-09-10T14:09:19.366071-05:00",
+    "accessed": "2022-09-11T13:00:56.5097533-05:00",
+    "path": "C:\\Users\\Coop\\Documents\\NodeProjects\\node-tone\\samples",
+    "name": "m4b.m4b"
+  }
+
+*/
+
+function bitrateKilobitToBit(bitrate) {
+  if (isNaN(bitrate) || !bitrate) return 0
+  return Number(bitrate) * 1000
+}
+
+function msToSeconds(ms) {
+  if (isNaN(ms) || !ms) return 0
+  return Number(ms) / 1000
+}
+
+function parseProbeDump(dumpPayload) {
+  const audioMetadata = dumpPayload.audio
+  const audioChannels = audioMetadata.channels || {}
+  const audio_stream = {
+    bit_rate: bitrateKilobitToBit(audioMetadata.bitrate), // tone uses Kbps but ffprobe uses bps so convert to bits
+    codec: null,
+    time_base: null,
+    language: null,
+    channel_layout: audioChannels.description || null,
+    channels: audioChannels.count || null,
+    sample_rate: audioMetadata.sampleRate || null
+  }
+
+  let chapterIndex = 0
+  const chapters = (dumpPayload.meta.chapters || []).map(chap => {
+    return {
+      id: chapterIndex++,
+      start: msToSeconds(chap.start),
+      end: msToSeconds(chap.start + chap.length),
+      title: chap.title || ''
+    }
+  })
+
+  var video_stream = null
+  if (dumpPayload.meta.embeddedPictures && dumpPayload.meta.embeddedPictures.length) {
+    const mimetype = dumpPayload.meta.embeddedPictures[0].mimetype
+    video_stream = {
+      codec: mimetype === 'image/png' ? 'png' : 'jpeg'
+    }
+  }
+
+  const tags = { ...dumpPayload.meta }
+  delete tags.chapters
+  delete tags.embeddedPictures
+
+  const fileMetadata = dumpPayload.file
+  var sizeBytes = !isNaN(fileMetadata.size) ? Number(fileMetadata.size) : null
+  var sizeMb = sizeBytes !== null ? Number((sizeBytes / (1024 * 1024)).toFixed(2)) : null
+  return {
+    format: audioMetadata.format || 'Unknown',
+    duration: msToSeconds(audioMetadata.duration),
+    size: sizeBytes,
+    sizeMb,
+    bit_rate: audio_stream.bit_rate,
+    audio_stream,
+    video_stream,
+    chapters,
+    tags
+  }
+}
+
+module.exports.probe = (filepath, verbose = false) => {
+  if (process.env.TONE_PATH) {
+    ffprobe.TONE_PATH = process.env.TONE_PATH
+  }
+
+  return tone.dump(filepath).then((dumpPayload) => {
+    if (verbose) {
+      Logger.debug(`[toneProber] dump for file "${filepath}"`, dumpPayload)
+    }
+    const rawProbeData = parseProbeDump(dumpPayload)
+    const probeData = new MediaProbeData()
+    probeData.setData(rawProbeData)
+    return probeData
+  }).catch((error) => {
+    Logger.error(`[toneProber] Failed to probe file at path "${filepath}"`, error)
+    return {
+      error
+    }
+  })
+}