feat(youtube): add YouTube browser window and Whisper subtitle source (#273)

launcher/config/args-normalizer.test.ts

@@ -19,30 +19,16 @@ function withTempDir<T>(fn: (dir: string) => T): T {
   }
 }
 
-test('createDefaultArgs normalizes configured language codes and env thread override', () => {
-  const originalThreads = process.env.SUBMINER_WHISPER_THREADS;
-  process.env.SUBMINER_WHISPER_THREADS = '7';
+test('createDefaultArgs normalizes configured language codes', () => {
+  const parsed = createDefaultArgs({
+    primarySubLanguages: [' JA ', 'jpn', 'ja'],
+    secondarySubLanguages: ['en', 'ENG', ''],
+  });
 
-  try {
-    const parsed = createDefaultArgs({
-      primarySubLanguages: [' JA ', 'jpn', 'ja'],
-      secondarySubLanguages: ['en', 'ENG', ''],
-      whisperThreads: 2,
-    });
-
-    assert.deepEqual(parsed.youtubePrimarySubLangs, ['ja', 'jpn']);
-    assert.deepEqual(parsed.youtubeSecondarySubLangs, ['en', 'eng']);
-    assert.deepEqual(parsed.youtubeAudioLangs, ['ja', 'jpn', 'en', 'eng']);
-    assert.equal(parsed.whisperThreads, 7);
-    assert.equal(parsed.youtubeWhisperSourceLanguage, 'ja');
-    assert.equal(parsed.profile, '');
-  } finally {
-    if (originalThreads === undefined) {
-      delete process.env.SUBMINER_WHISPER_THREADS;
-    } else {
-      process.env.SUBMINER_WHISPER_THREADS = originalThreads;
-    }
-  }
+  assert.deepEqual(parsed.youtubePrimarySubLangs, ['ja', 'jpn']);
+  assert.deepEqual(parsed.youtubeSecondarySubLangs, ['en', 'eng']);
+  assert.deepEqual(parsed.youtubeAudioLangs, ['ja', 'jpn', 'en', 'eng']);
+  assert.equal(parsed.profile, '');
 });
 
 test('createDefaultArgs seeds mpv profile from launcher config', () => {
@@ -120,6 +106,7 @@ test('applyInvocationsToArgs maps config and jellyfin invocation state', () => {
       logLevel: 'warn',
     },
     settingsInvocation: null,
+    youtubeBrowserInvocation: null,
     mpvInvocation: null,
     appInvocation: null,
     dictionaryTriggered: false,
@@ -174,6 +161,7 @@ test('applyInvocationsToArgs maps settings invocation to settings window', () =>
     settingsInvocation: {
       logLevel: undefined,
     },
+    youtubeBrowserInvocation: null,
     mpvInvocation: null,
     appInvocation: null,
     dictionaryTriggered: false,
@@ -221,6 +209,7 @@ test('applyInvocationsToArgs fails when config invocation has no action', () =>
         action: undefined,
       },
       settingsInvocation: null,
+      youtubeBrowserInvocation: null,
       mpvInvocation: null,
       appInvocation: null,
       dictionaryTriggered: false,
@@ -266,6 +255,7 @@ test('applyInvocationsToArgs maps texthooker browser-open request', () => {
     jellyfinInvocation: null,
     configInvocation: null,
     settingsInvocation: null,
+    youtubeBrowserInvocation: null,
     mpvInvocation: null,
     appInvocation: null,
     dictionaryTriggered: false,

launcher/config/args-normalizer.ts

@@ -13,14 +13,8 @@ import {
   DEFAULT_JIMAKU_API_BASE_URL,
   DEFAULT_YOUTUBE_PRIMARY_SUB_LANGS,
   DEFAULT_YOUTUBE_SECONDARY_SUB_LANGS,
-  DEFAULT_YOUTUBE_SUBGEN_OUT_DIR,
 } from '../types.js';
-import {
-  inferWhisperLanguage,
-  isUrlTarget,
-  resolvePathMaybe,
-  uniqueNormalizedLangCodes,
-} from '../util.js';
+import { isUrlTarget, resolvePathMaybe, uniqueNormalizedLangCodes } from '../util.js';
 import type { CliInvocations } from './cli-parser-builder.js';
 
 function ensureTarget(target: string, parsed: Args): void {
@@ -131,18 +125,6 @@ export function createDefaultArgs(
     recursive: false,
     profile: mpvConfig.profile ?? '',
     startOverlay: false,
-    whisperBin: process.env.SUBMINER_WHISPER_BIN || launcherConfig.whisperBin || '',
-    whisperModel: process.env.SUBMINER_WHISPER_MODEL || launcherConfig.whisperModel || '',
-    whisperVadModel: process.env.SUBMINER_WHISPER_VAD_MODEL || launcherConfig.whisperVadModel || '',
-    whisperThreads: (() => {
-      const envValue = Number.parseInt(process.env.SUBMINER_WHISPER_THREADS || '', 10);
-      if (Number.isInteger(envValue) && envValue > 0) return envValue;
-      return launcherConfig.whisperThreads || 4;
-    })(),
-    youtubeSubgenOutDir: process.env.SUBMINER_YT_SUBGEN_OUT_DIR || DEFAULT_YOUTUBE_SUBGEN_OUT_DIR,
-    youtubeSubgenAudioFormat: process.env.SUBMINER_YT_SUBGEN_AUDIO_FORMAT || 'm4a',
-    youtubeSubgenKeepTemp: process.env.SUBMINER_YT_SUBGEN_KEEP_TEMP === '1',
-    youtubeFixWithAi: launcherConfig.fixWithAi === true,
     jimakuApiKey: process.env.SUBMINER_JIMAKU_API_KEY || '',
     jimakuApiKeyCommand: process.env.SUBMINER_JIMAKU_API_KEY_COMMAND || '',
     jimakuApiBaseUrl: process.env.SUBMINER_JIMAKU_API_BASE_URL || DEFAULT_JIMAKU_API_BASE_URL,
@@ -170,6 +152,7 @@ export function createDefaultArgs(
     version: false,
     update: false,
     settings: false,
+    youtubeBrowser: false,
     configPath: false,
     configShow: false,
     mpvIdle: false,
@@ -185,16 +168,6 @@ export function createDefaultArgs(
     youtubePrimarySubLangs: primarySubLangs,
     youtubeSecondarySubLangs: secondarySubLangs,
     youtubeAudioLangs,
-    youtubeWhisperSourceLanguage: inferWhisperLanguage(primarySubLangs, 'ja'),
-    aiConfig: {
-      enabled: launcherConfig.ai?.enabled,
-      apiKey: launcherConfig.ai?.apiKey,
-      apiKeyCommand: launcherConfig.ai?.apiKeyCommand,
-      baseUrl: launcherConfig.ai?.baseUrl,
-      model: launcherConfig.ai?.model,
-      systemPrompt: launcherConfig.ai?.systemPrompt,
-      requestTimeoutMs: launcherConfig.ai?.requestTimeoutMs,
-    },
     useTexthooker: true,
     autoStartOverlay: false,
     texthookerOnly: false,
@@ -363,6 +336,13 @@ export function applyInvocationsToArgs(parsed: Args, invocations: CliInvocations
     parsed.settings = true;
   }
 
+  if (invocations.youtubeBrowserInvocation) {
+    if (invocations.youtubeBrowserInvocation.logLevel) {
+      parsed.logLevel = parseLogLevel(invocations.youtubeBrowserInvocation.logLevel);
+    }
+    parsed.youtubeBrowser = true;
+  }
+
   if (invocations.mpvInvocation) {
     if (invocations.mpvInvocation.logLevel) {
       parsed.logLevel = parseLogLevel(invocations.mpvInvocation.logLevel);

launcher/config/cli-parser-builder.ts

@@ -29,6 +29,7 @@ export interface CliInvocations {
   jellyfinInvocation: JellyfinInvocation | null;
   configInvocation: CommandActionInvocation | null;
   settingsInvocation: CommandActionInvocation | null;
+  youtubeBrowserInvocation: CommandActionInvocation | null;
   mpvInvocation: CommandActionInvocation | null;
   appInvocation: { appArgs: string[] } | null;
   dictionaryTriggered: boolean;
@@ -121,6 +122,8 @@ function getTopLevelCommand(argv: string[]): { name: string; index: number } | n
     'doctor',
     'config',
     'settings',
+    'youtube',
+    'yt',
     'mpv',
     'logs',
     'dictionary',
@@ -175,6 +178,7 @@ export function parseCliPrograms(
   let jellyfinInvocation: JellyfinInvocation | null = null;
   let configInvocation: CommandActionInvocation | null = null;
   let settingsInvocation: CommandActionInvocation | null = null;
+  let youtubeBrowserInvocation: CommandActionInvocation | null = null;
   let mpvInvocation: CommandActionInvocation | null = null;
   let appInvocation: { appArgs: string[] } | null = null;
   let dictionaryTriggered = false;
@@ -512,6 +516,17 @@ export function parseCliPrograms(
       };
     });
 
+  commandProgram
+    .command('youtube')
+    .alias('yt')
+    .description('Open the YouTube browser (videos play in mpv)')
+    .option('--log-level <level>', 'Log level')
+    .action((options: Record<string, unknown>) => {
+      youtubeBrowserInvocation = {
+        logLevel: typeof options.logLevel === 'string' ? options.logLevel : undefined,
+      };
+    });
+
   commandProgram
     .command('mpv')
     .description('MPV helpers')
@@ -567,6 +582,7 @@ export function parseCliPrograms(
       jellyfinInvocation,
       configInvocation,
       settingsInvocation,
+      youtubeBrowserInvocation,
       mpvInvocation,
       appInvocation,
       dictionaryTriggered,

launcher/config/youtube-subgen-config.ts

@@ -1,5 +1,4 @@
 import type { LauncherYoutubeSubgenConfig } from '../types.js';
-import { mergeAiConfig } from '../../src/ai/config.js';
 
 function asStringArray(value: unknown): string[] | undefined {
   if (!Array.isArray(value)) return undefined;
@@ -9,11 +8,6 @@ function asStringArray(value: unknown): string[] | undefined {
 export function parseLauncherYoutubeSubgenConfig(
   root: Record<string, unknown>,
 ): LauncherYoutubeSubgenConfig {
-  const youtubeSubgenRaw = root.youtubeSubgen;
-  const youtubeSubgen =
-    youtubeSubgenRaw && typeof youtubeSubgenRaw === 'object'
-      ? (youtubeSubgenRaw as Record<string, unknown>)
-      : null;
   const youtubeRaw = root.youtube;
   const youtube =
     youtubeRaw && typeof youtubeRaw === 'object' ? (youtubeRaw as Record<string, unknown>) : null;
@@ -25,58 +19,11 @@ export function parseLauncherYoutubeSubgenConfig(
   const jimakuRaw = root.jimaku;
   const jimaku =
     jimakuRaw && typeof jimakuRaw === 'object' ? (jimakuRaw as Record<string, unknown>) : null;
-  const aiRaw = root.ai;
-  const ai = aiRaw && typeof aiRaw === 'object' ? (aiRaw as Record<string, unknown>) : null;
-  const youtubeAiRaw = youtubeSubgen?.ai;
-  const youtubeAi =
-    youtubeAiRaw && typeof youtubeAiRaw === 'object'
-      ? (youtubeAiRaw as Record<string, unknown>)
-      : null;
 
   const jimakuLanguagePreference = jimaku?.languagePreference;
   const jimakuMaxEntryResults = jimaku?.maxEntryResults;
 
   return {
-    whisperBin:
-      typeof youtubeSubgen?.whisperBin === 'string' ? youtubeSubgen.whisperBin : undefined,
-    whisperModel:
-      typeof youtubeSubgen?.whisperModel === 'string' ? youtubeSubgen.whisperModel : undefined,
-    whisperVadModel:
-      typeof youtubeSubgen?.whisperVadModel === 'string'
-        ? youtubeSubgen.whisperVadModel
-        : undefined,
-    whisperThreads:
-      typeof youtubeSubgen?.whisperThreads === 'number' &&
-      Number.isFinite(youtubeSubgen.whisperThreads) &&
-      youtubeSubgen.whisperThreads > 0
-        ? Math.floor(youtubeSubgen.whisperThreads)
-        : undefined,
-    fixWithAi: typeof youtubeSubgen?.fixWithAi === 'boolean' ? youtubeSubgen.fixWithAi : undefined,
-    ai: mergeAiConfig(
-      ai
-        ? {
-            enabled: typeof ai.enabled === 'boolean' ? ai.enabled : undefined,
-            apiKey: typeof ai.apiKey === 'string' ? ai.apiKey : undefined,
-            apiKeyCommand: typeof ai.apiKeyCommand === 'string' ? ai.apiKeyCommand : undefined,
-            baseUrl: typeof ai.baseUrl === 'string' ? ai.baseUrl : undefined,
-            model: typeof ai.model === 'string' ? ai.model : undefined,
-            systemPrompt: typeof ai.systemPrompt === 'string' ? ai.systemPrompt : undefined,
-            requestTimeoutMs:
-              typeof ai.requestTimeoutMs === 'number' &&
-              Number.isFinite(ai.requestTimeoutMs) &&
-              ai.requestTimeoutMs > 0
-                ? Math.floor(ai.requestTimeoutMs)
-                : undefined,
-          }
-        : undefined,
-      youtubeAi
-        ? {
-            model: typeof youtubeAi.model === 'string' ? youtubeAi.model : undefined,
-            systemPrompt:
-              typeof youtubeAi.systemPrompt === 'string' ? youtubeAi.systemPrompt : undefined,
-          }
-        : undefined,
-    ),
     primarySubLanguages: asStringArray(youtube?.primarySubLanguages),
     secondarySubLanguages: asStringArray(secondarySub?.secondarySubLanguages),
     jimakuApiKey: typeof jimaku?.apiKey === 'string' ? jimaku.apiKey : undefined,