0.12.0-16 misc fixes

backend/src/utils/encodePlan.js

@@ -3,6 +3,16 @@ const { splitArgs } = require('./commandLine');
 
 const DEFAULT_AUDIO_COPY_MASK = ['aac', 'ac3', 'eac3', 'truehd', 'dts', 'dtshd', 'mp3', 'flac'];
 const DEFAULT_AUDIO_FALLBACK = 'av_aac';
+const SUBTITLE_CONFIDENCE_SCORES = Object.freeze({
+  low: 1,
+  medium: 2,
+  high: 3
+});
+const FORCED_SUBTITLE_EVENT_RATIO_THRESHOLD = 0.35;
+const FORCED_SUBTITLE_SIZE_RATIO_THRESHOLD = 0.35;
+const FORCED_SUBTITLE_MAX_EVENT_COUNT = 220;
+const FORCED_SUBTITLE_MIN_EVENT_GAP = 12;
+const FORCED_SUBTITLE_MIN_SIZE_GAP_BYTES = 64 * 1024;
 const ISO2_TO_3_LANGUAGE = {
   de: 'deu',
   en: 'eng',
@@ -271,7 +281,7 @@ function parseMediaInfoFile(mediaInfoJson, fileInfo, index) {
       channels: item?.Channels || item?.Channel_s_ || null
     }));
 
-  const subtitleTracks = tracks
+  const subtitleTracksRaw = tracks
     .filter((item) => {
       const type = String(item?.['@type'] || '').toLowerCase();
       return type === 'text' || type === 'subtitle';
@@ -282,8 +292,14 @@ function parseMediaInfoFile(mediaInfoJson, fileInfo, index) {
       language: normalizeLanguage(item?.Language || item?.Language_String3 || item?.Language_String || 'und'),
       languageLabel: item?.Language_String3 || item?.Language || item?.Language_String || 'und',
       title: item?.Title || null,
-      format: item?.Format || null
+      format: item?.Format || null,
+      defaultFlag: parseBooleanFlag(item?.Default ?? item?.Default_String ?? item?.IsDefault ?? item?.isDefault),
+      forcedFlag: parseBooleanFlagNullable(item?.Forced ?? item?.Forced_String ?? item?.IsForced ?? item?.isForced),
+      sdhFlag: parseSubtitleSdhFlag(item),
+      eventCount: parseSubtitleEventCount(item),
+      streamSizeBytes: parseSubtitleStreamSizeBytes(item)
     }));
+  const subtitleTracks = annotateSubtitleTracks(subtitleTracksRaw);
 
   const videoTracks = tracks
     .filter((item) => String(item?.['@type'] || '').toLowerCase() === 'video')
@@ -350,6 +366,467 @@ function parseTrackIdList(raw) {
     .filter((item) => Number.isFinite(item));
 }
 
+function parseBooleanFlag(value) {
+  return parseBooleanFlagNullable(value) === true;
+}
+
+function parseBooleanFlagNullable(value) {
+  if (typeof value === 'boolean') {
+    return value;
+  }
+  if (typeof value === 'number') {
+    return value === 1;
+  }
+  const raw = String(value || '').trim().toLowerCase();
+  if (!raw) {
+    return null;
+  }
+  if (raw === 'yes' || raw === 'true' || raw === '1') {
+    return true;
+  }
+  if (raw === 'no' || raw === 'false' || raw === '0') {
+    return false;
+  }
+  return null;
+}
+
+function parseSubtitleForcedFlag(track) {
+  if (!track || typeof track !== 'object') {
+    return null;
+  }
+  const candidates = [
+    track?.forcedFlag,
+    track?.forced,
+    track?.Forced,
+    track?.isForced,
+    track?.IsForced,
+    track?.forced_only,
+    track?.forcedOnly,
+    track?.Attributes?.Forced
+  ];
+  for (const candidate of candidates) {
+    const parsed = parseBooleanFlagNullable(candidate);
+    if (parsed === true || parsed === false) {
+      return parsed;
+    }
+  }
+  return null;
+}
+
+function parseSubtitleSdhFlag(track) {
+  if (!track || typeof track !== 'object') {
+    return null;
+  }
+  const candidates = [
+    track?.sdhFlag,
+    track?.hearingImpaired,
+    track?.HearingImpaired,
+    track?.isHearingImpaired,
+    track?.IsHearingImpaired,
+    track?.Hearing_Impaired,
+    track?.closedCaptions,
+    track?.ClosedCaptions,
+    track?.Attributes?.HearingImpaired,
+    track?.Attributes?.ClosedCaptions
+  ];
+  for (const candidate of candidates) {
+    const parsed = parseBooleanFlagNullable(candidate);
+    if (parsed === true || parsed === false) {
+      return parsed;
+    }
+  }
+
+  const serviceKind = String(track?.serviceKind || track?.ServiceKind || '').trim().toLowerCase();
+  if (!serviceKind) {
+    return null;
+  }
+  if (serviceKind.includes('sdh') || serviceKind.includes('cc') || serviceKind.includes('hearing')) {
+    return true;
+  }
+  return null;
+}
+
+function parseSubtitleEventCount(track) {
+  const candidates = [
+    track?.CountOfEvents,
+    track?.countOfEvents,
+    track?.EventCount,
+    track?.eventCount,
+    track?.ElementCount,
+    track?.elementCount
+  ];
+  for (const candidate of candidates) {
+    const numeric = Number(candidate);
+    if (Number.isFinite(numeric) && numeric >= 0) {
+      return Math.trunc(numeric);
+    }
+  }
+  return null;
+}
+
+function parseSubtitleStreamSizeBytes(track) {
+  const numericCandidates = [
+    track?.StreamSize,
+    track?.streamSize,
+    track?.StreamSize_Original,
+    track?.streamSizeOriginal,
+    track?.Bytes,
+    track?.bytes
+  ];
+  for (const candidate of numericCandidates) {
+    const numeric = Number(candidate);
+    if (Number.isFinite(numeric) && numeric > 0) {
+      return Math.trunc(numeric);
+    }
+  }
+
+  const textCandidates = [
+    track?.StreamSize_String,
+    track?.streamSizeString,
+    track?.StreamSize_Original_String,
+    track?.streamSizeOriginalString,
+    track?.Size_String,
+    track?.sizeString
+  ];
+  for (const candidate of textCandidates) {
+    const text = String(candidate || '').trim();
+    if (!text) {
+      continue;
+    }
+    const match = text.match(/([0-9]+(?:[.,][0-9]+)?)\s*([kmgt]?b)/i);
+    if (!match) {
+      continue;
+    }
+    const value = Number(String(match[1]).replace(',', '.'));
+    if (!Number.isFinite(value) || value <= 0) {
+      continue;
+    }
+    const unit = String(match[2] || 'b').toLowerCase();
+    const factorByUnit = {
+      b: 1,
+      kb: 1024,
+      mb: 1024 ** 2,
+      gb: 1024 ** 3,
+      tb: 1024 ** 4
+    };
+    const factor = factorByUnit[unit] || 1;
+    return Math.max(0, Math.trunc(value * factor));
+  }
+  return null;
+}
+
+function normalizeSubtitleConfidence(raw) {
+  const value = String(raw || '').trim().toLowerCase();
+  if (value === 'high' || value === 'medium' || value === 'low') {
+    return value;
+  }
+  return 'low';
+}
+
+function subtitleConfidenceScore(raw) {
+  return SUBTITLE_CONFIDENCE_SCORES[normalizeSubtitleConfidence(raw)] || 0;
+}
+
+function collectSubtitleText(track) {
+  return [
+    track?.title,
+    track?.description,
+    track?.name,
+    track?.format,
+    track?.label
+  ]
+    .map((value) => String(value || '').trim().toLowerCase())
+    .filter(Boolean)
+    .join(' ');
+}
+
+function isLikelyBitmapSubtitleFormat(track) {
+  const text = [
+    track?.format,
+    track?.codec,
+    track?.codecName,
+    track?.title,
+    track?.description,
+    track?.name,
+    track?.label
+  ]
+    .map((value) => String(value || '').trim().toLowerCase())
+    .filter(Boolean)
+    .join(' ');
+  if (!text) {
+    return false;
+  }
+  return (
+    /\bpgs\b/.test(text)
+    || /\bhdmv\b/.test(text)
+    || /\bsup\b/.test(text)
+    || /\bvobsub\b/.test(text)
+    || /\bdvd[-_\s]?sub/.test(text)
+    || /\bdvb[-_\s]?sub/.test(text)
+  );
+}
+
+function isLikelySdhSubtitleTrack(track) {
+  const text = collectSubtitleText(track);
+  if (!text) {
+    return false;
+  }
+  return (
+    /\bsdh\b/.test(text)
+    || /\bcc\b/.test(text)
+    || /\bhoh\b/.test(text)
+    || /\bcaptions?\b/.test(text)
+    || /hard[-\s]?of[-\s]?hearing/.test(text)
+    || /hearing\s+impaired/.test(text)
+    || /(?:gehörlos|hoergeschaedigt|hörgeschädigt)/.test(text)
+  );
+}
+
+function isLikelyForcedSubtitleTrack(track) {
+  const text = collectSubtitleText(track);
+  if (!text) {
+    return false;
+  }
+  if (isLikelySdhSubtitleTrack(track) || /\bnot forced\b/.test(text)) {
+    return false;
+  }
+  return (
+    /\bforced(?:\s+only)?\b/.test(text)
+    || /nur\s+erzwungen/.test(text)
+    || /\berzwungen\b/.test(text)
+  );
+}
+
+function compareSubtitleTracksByForcedHeuristic(a, b) {
+  const aSdh = Number(Boolean(a?.sdhLikely));
+  const bSdh = Number(Boolean(b?.sdhLikely));
+  if (aSdh !== bSdh) {
+    return aSdh - bSdh;
+  }
+
+  const aDefault = Number(Boolean(a?.defaultFlag));
+  const bDefault = Number(Boolean(b?.defaultFlag));
+  if (aDefault !== bDefault) {
+    return aDefault - bDefault;
+  }
+
+  const aEventKnown = Number.isFinite(a?.eventCount) ? 0 : 1;
+  const bEventKnown = Number.isFinite(b?.eventCount) ? 0 : 1;
+  if (aEventKnown !== bEventKnown) {
+    return aEventKnown - bEventKnown;
+  }
+  if (aEventKnown === 0 && a.eventCount !== b.eventCount) {
+    return a.eventCount - b.eventCount;
+  }
+
+  const aSizeKnown = Number.isFinite(a?.streamSizeBytes) ? 0 : 1;
+  const bSizeKnown = Number.isFinite(b?.streamSizeBytes) ? 0 : 1;
+  if (aSizeKnown !== bSizeKnown) {
+    return aSizeKnown - bSizeKnown;
+  }
+  if (aSizeKnown === 0 && a.streamSizeBytes !== b.streamSizeBytes) {
+    return a.streamSizeBytes - b.streamSizeBytes;
+  }
+
+  const aTrackId = Number.isFinite(a?.id) && a.id > 0 ? a.id : Number.MAX_SAFE_INTEGER;
+  const bTrackId = Number.isFinite(b?.id) && b.id > 0 ? b.id : Number.MAX_SAFE_INTEGER;
+  if (aTrackId !== bTrackId) {
+    return aTrackId - bTrackId;
+  }
+  return a.originalIndex - b.originalIndex;
+}
+
+function compareSubtitleTracksForDedup(a, b) {
+  const aSdh = Number(Boolean(a?.sdhLikely));
+  const bSdh = Number(Boolean(b?.sdhLikely));
+  if (aSdh !== bSdh) {
+    return aSdh - bSdh;
+  }
+
+  const defaultDiff = Number(Boolean(b?.defaultFlag)) - Number(Boolean(a?.defaultFlag));
+  if (defaultDiff !== 0) {
+    return defaultDiff;
+  }
+  const confidenceDiff = subtitleConfidenceScore(b?.sourceConfidence) - subtitleConfidenceScore(a?.sourceConfidence);
+  if (confidenceDiff !== 0) {
+    return confidenceDiff;
+  }
+  const aTrackId = Number.isFinite(a?.id) && a.id > 0 ? a.id : Number.MAX_SAFE_INTEGER;
+  const bTrackId = Number.isFinite(b?.id) && b.id > 0 ? b.id : Number.MAX_SAFE_INTEGER;
+  if (aTrackId !== bTrackId) {
+    return aTrackId - bTrackId;
+  }
+  return a.originalIndex - b.originalIndex;
+}
+
+function isHeuristicForcedSubtitleCandidate(languageEntries, candidate) {
+  if (!candidate) {
+    return false;
+  }
+  if (!isLikelyBitmapSubtitleFormat(candidate)) {
+    return false;
+  }
+  const comparable = (Array.isArray(languageEntries) ? languageEntries : [])
+    .filter((entry) => entry !== candidate && !entry.sdhLikely && isLikelyBitmapSubtitleFormat(entry));
+  if (comparable.length === 0) {
+    return false;
+  }
+
+  const candidateEventCount = Number(candidate?.eventCount);
+  const comparableEventCounts = comparable
+    .map((entry) => Number(entry?.eventCount))
+    .filter((value) => Number.isFinite(value) && value > 0);
+  const maxComparableEventCount = comparableEventCounts.length > 0
+    ? Math.max(...comparableEventCounts)
+    : null;
+  const hasEventSignal = Number.isFinite(candidateEventCount)
+    && candidateEventCount >= 0
+    && Number.isFinite(maxComparableEventCount)
+    && maxComparableEventCount > 0
+    && candidateEventCount <= FORCED_SUBTITLE_MAX_EVENT_COUNT
+    && (candidateEventCount / maxComparableEventCount) <= FORCED_SUBTITLE_EVENT_RATIO_THRESHOLD
+    && (maxComparableEventCount - candidateEventCount) >= FORCED_SUBTITLE_MIN_EVENT_GAP;
+
+  const candidateStreamSize = Number(candidate?.streamSizeBytes);
+  const comparableSizes = comparable
+    .map((entry) => Number(entry?.streamSizeBytes))
+    .filter((value) => Number.isFinite(value) && value > 0);
+  const maxComparableSize = comparableSizes.length > 0
+    ? Math.max(...comparableSizes)
+    : null;
+  const hasSizeSignal = Number.isFinite(candidateStreamSize)
+    && candidateStreamSize > 0
+    && Number.isFinite(maxComparableSize)
+    && maxComparableSize > 0
+    && (candidateStreamSize / maxComparableSize) <= FORCED_SUBTITLE_SIZE_RATIO_THRESHOLD
+    && (maxComparableSize - candidateStreamSize) >= FORCED_SUBTITLE_MIN_SIZE_GAP_BYTES;
+
+  const signalCount = Number(hasEventSignal) + Number(hasSizeSignal);
+  if (signalCount === 0) {
+    return false;
+  }
+
+  if (candidate.defaultFlag && signalCount < 2) {
+    return false;
+  }
+  return true;
+}
+
+function annotateSubtitleTracks(subtitleTracks) {
+  const tracks = Array.isArray(subtitleTracks) ? subtitleTracks : [];
+  if (tracks.length === 0) {
+    return [];
+  }
+
+  const entries = tracks.map((track, index) => ({
+    ...track,
+    language: normalizeLanguage(track?.language || track?.languageLabel || 'und'),
+    defaultFlag: Boolean(track?.defaultFlag),
+    forcedFlag: parseSubtitleForcedFlag(track),
+    forcedTrack: false,
+    sourceConfidence: null,
+    confidenceSource: 'heuristic',
+    duplicate: false,
+    selected: false,
+    subtitleType: 'full',
+    forcedAvailable: false,
+    forcedSourceTrackIds: [],
+    sdhLikely: (parseSubtitleSdhFlag(track) === true) || Boolean(track?.sdhLikely) || isLikelySdhSubtitleTrack(track),
+    originalIndex: index
+  }));
+
+  const byLanguage = new Map();
+  for (const entry of entries) {
+    if (!byLanguage.has(entry.language)) {
+      byLanguage.set(entry.language, []);
+    }
+    byLanguage.get(entry.language).push(entry);
+  }
+
+  for (const languageEntries of byLanguage.values()) {
+    for (const entry of languageEntries) {
+      const forcedByFlag = entry.forcedFlag === true;
+      const forcedBlockedByFlag = entry.forcedFlag === false;
+      const forcedByTitle = !entry.sdhLikely && !forcedBlockedByFlag && isLikelyForcedSubtitleTrack(entry);
+
+      if (forcedByFlag) {
+        entry.forcedTrack = true;
+        entry.sourceConfidence = 'high';
+        entry.confidenceSource = 'explicit_flag';
+      } else if (forcedByTitle) {
+        entry.forcedTrack = true;
+        entry.sourceConfidence = 'medium';
+        entry.confidenceSource = 'title';
+      }
+    }
+
+    if (!languageEntries.some((entry) => entry.forcedTrack)) {
+      const candidates = languageEntries
+        .filter((entry) => !entry.sdhLikely && entry.forcedFlag !== false)
+        .sort(compareSubtitleTracksByForcedHeuristic);
+      const forcedCandidate = candidates[0] || null;
+      if (forcedCandidate && isHeuristicForcedSubtitleCandidate(languageEntries, forcedCandidate)) {
+        forcedCandidate.forcedTrack = true;
+        forcedCandidate.sourceConfidence = 'low';
+        forcedCandidate.confidenceSource = 'heuristic';
+      }
+    }
+
+    for (const entry of languageEntries) {
+      if (!entry.forcedTrack) {
+        continue;
+      }
+      entry.sourceConfidence = normalizeSubtitleConfidence(entry.sourceConfidence || 'low');
+    }
+
+    const forcedEntries = languageEntries.filter((entry) => entry.forcedTrack);
+    const fullEntries = languageEntries.filter((entry) => !entry.forcedTrack && !entry.sdhLikely);
+    const sdhEntries = languageEntries.filter((entry) => !entry.forcedTrack && entry.sdhLikely);
+    const forcedWinner = forcedEntries.length > 0 ? [...forcedEntries].sort(compareSubtitleTracksForDedup)[0] : null;
+    const fullWinner = fullEntries.length > 0 ? [...fullEntries].sort(compareSubtitleTracksForDedup)[0] : null;
+    const sdhWinner = sdhEntries.length > 0 ? [...sdhEntries].sort(compareSubtitleTracksForDedup)[0] : null;
+    const forcedSourceTrackIds = forcedEntries
+      .map((entry) => Number(entry?.sourceTrackId ?? entry?.id))
+      .filter((value) => Number.isFinite(value) && value > 0)
+      .map((value) => Math.trunc(value));
+    const forcedAvailable = Boolean(forcedWinner);
+
+    for (const entry of languageEntries) {
+      const typeWinner = entry.forcedTrack
+        ? forcedWinner
+        : (entry.sdhLikely ? sdhWinner : fullWinner);
+      entry.duplicate = Boolean(typeWinner && typeWinner !== entry);
+      if (entry.forcedTrack) {
+        entry.selected = Boolean(typeWinner && typeWinner === entry && !entry.duplicate);
+      } else if (entry.sdhLikely) {
+        entry.selected = Boolean(!fullWinner && typeWinner && typeWinner === entry && !entry.duplicate);
+      } else {
+        entry.selected = Boolean(typeWinner && typeWinner === entry && !entry.duplicate);
+      }
+      entry.subtitleType = entry.forcedTrack ? 'forced' : 'full';
+      entry.forcedAvailable = forcedAvailable;
+      entry.forcedSourceTrackIds = forcedSourceTrackIds;
+      const fullHasForced = !entry.forcedTrack && Boolean(
+        entry.fullHasForced
+        ?? entry.subtitleFullHasForced
+        ?? entry.hasForcedVariant
+      );
+      const explicitForcedOnly = entry.isForcedOnly ?? entry.forcedOnly ?? entry.subtitlePreviewForcedOnly;
+      entry.isForcedOnly = typeof explicitForcedOnly === 'boolean'
+        ? explicitForcedOnly
+        : (entry.forcedTrack && !fullHasForced);
+      entry.fullHasForced = fullHasForced;
+    }
+  }
+
+  return entries
+    .sort((a, b) => a.originalIndex - b.originalIndex)
+    .map((entry) => {
+      const { originalIndex, ...rest } = entry;
+      return rest;
+    });
+}
+
 function parseEncoderList(raw) {
   return String(raw || '')
     .split(',')
@@ -638,7 +1115,10 @@ function buildTrackSelectors(settings, presetProfile) {
 
 function selectTrackIds(tracks, selector, trackType) {
   const available = Array.isArray(tracks) ? tracks : [];
-  if (available.length === 0) {
+  const selectable = trackType === 'subtitle'
+    ? available.filter((track) => !Boolean(track?.duplicate))
+    : available;
+  if (selectable.length === 0) {
     return [];
   }
 
@@ -648,13 +1128,13 @@ function selectTrackIds(tracks, selector, trackType) {
 
   if (selector.mode === 'all') {
     if (selector.firstOnly) {
-      return [available[0].id];
+      return [selectable[0].id];
     }
-    return available.map((track) => track.id);
+    return selectable.map((track) => track.id);
   }
 
   if (selector.mode === 'explicit') {
-    const explicit = available
+    const explicit = selectable
       .filter((track) => selector.explicitIds.includes(track.id))
       .map((track) => track.id);
     if (selector.firstOnly) {
@@ -664,7 +1144,7 @@ function selectTrackIds(tracks, selector, trackType) {
   }
 
   if (selector.mode === 'language') {
-    const matches = available.filter((track) => selector.languages.includes(track.language));
+    const matches = selectable.filter((track) => selector.languages.includes(track.language));
     if (selector.firstOnly) {
       return matches.length > 0 ? [matches[0].id] : [];
     }
@@ -672,11 +1152,11 @@ function selectTrackIds(tracks, selector, trackType) {
   }
 
   if (selector.mode === 'first') {
-    return [available[0].id];
+    return [selectable[0].id];
   }
 
   if (trackType === 'audio') {
-    return [available[0].id];
+    return [selectable[0].id];
   }
 
   return [];
@@ -916,21 +1396,45 @@ function buildMediainfoReview({
     const normalizedSubtitle = title.subtitleTracks.map((track) => {
       const selectedByRule = selectedSubtitleIds.includes(track.id);
       const subtitleFlags = computeSubtitleFlags(track.id, selectedSubtitleIds, trackSelectors.subtitle);
+      const inferredForced = Boolean(
+        track?.forcedTrack
+        || String(track?.subtitleType || '').trim().toLowerCase() === 'forced'
+      );
+      const inferredForcedOnly = Boolean(
+        track?.isForcedOnly
+        ?? track?.forcedOnly
+        ?? track?.subtitlePreviewForcedOnly
+        ?? inferredForced
+      );
+      const inferredDefault = Boolean(track?.defaultFlag);
+      const subtitlePreviewFlags = [];
+      if (subtitleFlags.burned) {
+        subtitlePreviewFlags.push('burned');
+      }
+      if (subtitleFlags.forced || inferredForced) {
+        subtitlePreviewFlags.push('forced');
+      }
+      if (subtitleFlags.forcedOnly || inferredForcedOnly) {
+        subtitlePreviewFlags.push('forced-only');
+      }
+      if (subtitleFlags.default || inferredDefault) {
+        subtitlePreviewFlags.push('default');
+      }
       const subtitlePreviewSummary = !selectedByRule
         ? 'Nicht übernommen'
-        : (subtitleFlags.flags.length > 0
-          ? `Übernehmen (${subtitleFlags.flags.join(', ')})`
+        : (subtitlePreviewFlags.length > 0
+          ? `Übernehmen (${subtitlePreviewFlags.join(', ')})`
           : 'Übernehmen');
 
       return {
         ...track,
         selectedByRule,
         subtitlePreviewSummary,
-        subtitlePreviewFlags: subtitleFlags.flags,
+        subtitlePreviewFlags: selectedByRule ? subtitlePreviewFlags : [],
         subtitlePreviewBurnIn: subtitleFlags.burned,
-        subtitlePreviewForced: subtitleFlags.forced,
-        subtitlePreviewForcedOnly: subtitleFlags.forcedOnly,
-        subtitlePreviewDefaultTrack: subtitleFlags.default
+        subtitlePreviewForced: selectedByRule ? (subtitleFlags.forced || inferredForced) : false,
+        subtitlePreviewForcedOnly: selectedByRule ? (subtitleFlags.forcedOnly || inferredForcedOnly) : false,
+        subtitlePreviewDefaultTrack: selectedByRule ? (subtitleFlags.default || inferredDefault) : false
       };
     });