improve subtitle scraping

src/runners/individualRunner.ts

@@ -9,7 +9,6 @@ import { NotFoundError } from '@/utils/errors';
 import { addOpenSubtitlesCaptions } from '@/utils/opensubtitles';
 import { requiresProxy, setupProxy } from '@/utils/proxy';
 import { isValidStream, validatePlayableStreams } from '@/utils/valid';
-import { addWyzieCaptions } from '@/utils/wyziesubs';
 
 export type IndividualSourceRunnerOptions = {
   features: FeatureMap;
@@ -95,36 +94,15 @@ export async function scrapeInvidualSource(
     // opensubtitles
     if (!ops.disableOpensubtitles) {
       for (const playableStream of playableStreams) {
-        // Try Wyzie subs first
-        if (ops.media.imdbId) {
-          playableStream.captions = await addWyzieCaptions(
-            playableStream.captions,
-            ops.media.tmdbId,
-            ops.media.imdbId,
-            ops.media.type === 'show' ? ops.media.season.number : undefined,
-            ops.media.type === 'show' ? ops.media.episode.number : undefined,
-          );
-
-          // Fall back to OpenSubtitles if no Wyzie subs found
-          if (!playableStream.captions.some((caption) => caption.wyziesubs)) {
-            const [imdbId, season, episode] = atob(ops.media.imdbId)
-              .split('.')
-              .map((x, i) => (i === 0 ? x : Number(x) || null));
-            const mediaInfo = {
-              ...ops,
-              media: {
-                type: season && episode ? 'show' : 'movie',
-                imdbId: imdbId?.toString() || '',
-                ...(season && episode ? { season: { number: season }, episode: { number: episode } } : {}),
-              } as ScrapeMedia,
-            };
-            playableStream.captions = await addOpenSubtitlesCaptions(
-              playableStream.captions,
-              mediaInfo,
-              ops.media.imdbId,
-            );
-          }
-        }
+        playableStream.captions = await addOpenSubtitlesCaptions(
+          playableStream.captions,
+          ops,
+          btoa(
+            `${ops.media.imdbId}${
+              ops.media.type === 'show' ? `.${ops.media.season.number}.${ops.media.episode.number}` : ''
+            }`,
+          ),
+        );
       }
     }
     output.stream = playableStreams;

src/utils/opensubtitles.ts

@@ -1,3 +1,4 @@
+/* eslint-disable no-console */
 import { ScrapeMedia } from '@/entrypoint/utils/media';
 import { Caption, labelToLanguageCode, removeDuplicatedLanguages } from '@/providers/captions';
 import { IndividualEmbedRunnerOptions } from '@/runners/individualRunner';
@@ -9,6 +10,14 @@ type CaptionOptions = (ProviderRunnerOptions | IndividualEmbedRunnerOptions) & {
   media?: ScrapeMedia;
 };
 
+const timeout = (ms: number, source: string) =>
+  new Promise<null>((resolve) => {
+    setTimeout(() => {
+      console.error(`${source} captions request timed out after ${ms}ms`);
+      resolve(null);
+    }, ms);
+  });
+
 export async function addOpenSubtitlesCaptions(
   captions: Caption[],
   ops: CaptionOptions,
@@ -20,59 +29,77 @@ export async function addOpenSubtitlesCaptions(
       .map((x, i) => (i === 0 ? x : Number(x) || null));
     if (!imdbId) return captions;
 
-    // First try Wyzie subs
-    const wyzieCaptions = await addWyzieCaptions(
-      [],
-      ops.media?.tmdbId?.toString() || '',
-      imdbId.toString(),
-      typeof season === 'number' ? season : undefined,
-      typeof episode === 'number' ? episode : undefined,
-    );
+    // Try Wyzie subs first. 2 second timeout
+    try {
+      const wyziePromise = addWyzieCaptions(
+        [],
+        ops.media?.tmdbId?.toString() || '',
+        imdbId.toString(),
+        typeof season === 'number' ? season : undefined,
+        typeof episode === 'number' ? episode : undefined,
+      );
 
-    // If we found Wyzie subs, return them as OpenSubtitles captions
-    if (wyzieCaptions.length > 0) {
-      return [
-        ...captions,
-        ...wyzieCaptions.map((caption) => ({
-          ...caption,
-          opensubtitles: true,
-        })),
-      ];
+      const wyzieCaptions = await Promise.race([wyziePromise, timeout(2000, 'Wyzie')]);
+
+      // If we found Wyzie subs, return them as OpenSubtitles captions
+      if (wyzieCaptions && wyzieCaptions.length > 0) {
+        return [
+          ...captions,
+          ...wyzieCaptions.map((caption) => ({
+            ...caption,
+            opensubtitles: true,
+          })),
+        ];
+      }
+    } catch (error) {
+      // Wyzie failed for a reason other than timeout
+      console.error('Wyzie subtitles fetch failed:', error);
     }
 
-    // Fall back to OpenSubtitles if no Wyzie subs found
-    const Res: {
-      LanguageName: string;
-      SubDownloadLink: string;
-      SubFormat: 'srt' | 'vtt';
-    }[] = await ops.proxiedFetcher(
-      `https://rest.opensubtitles.org/search/${
-        season && episode ? `episode-${episode}/` : ''
-      }imdbid-${(imdbId as string).slice(2)}${season && episode ? `/season-${season}` : ''}`,
-      {
-        headers: {
-          'X-User-Agent': 'VLSub 0.10.2',
+    // Fall back to OpenSubtitles with a 5 second timeout
+    try {
+      const openSubsPromise = ops.proxiedFetcher(
+        `https://rest.opensubtitles.org/search/${
+          season && episode ? `episode-${episode}/` : ''
+        }imdbid-${(imdbId as string).slice(2)}${season && episode ? `/season-${season}` : ''}`,
+        {
+          headers: {
+            'X-User-Agent': 'VLSub 0.10.2',
+          },
         },
-      },
-    );
+      );
 
-    const openSubtilesCaptions: Caption[] = [];
-    for (const caption of Res) {
-      const url = caption.SubDownloadLink.replace('.gz', '').replace('download/', 'download/subencoding-utf8/');
-      const language = labelToLanguageCode(caption.LanguageName);
-      if (!url || !language) continue;
-      else
-        openSubtilesCaptions.push({
-          id: url,
-          opensubtitles: true,
-          url,
-          type: caption.SubFormat || 'srt',
-          hasCorsRestrictions: false,
-          language,
-        });
+      const Res: {
+        LanguageName: string;
+        SubDownloadLink: string;
+        SubFormat: 'srt' | 'vtt';
+      }[] = await Promise.race([openSubsPromise, timeout(5000, 'OpenSubtitles')]);
+
+      if (!Res) return captions; // Timeout occurred
+
+      const openSubtilesCaptions: Caption[] = [];
+      for (const caption of Res) {
+        const url = caption.SubDownloadLink.replace('.gz', '').replace('download/', 'download/subencoding-utf8/');
+        const language = labelToLanguageCode(caption.LanguageName);
+        if (!url || !language) continue;
+        else
+          openSubtilesCaptions.push({
+            id: url,
+            opensubtitles: true,
+            url,
+            type: caption.SubFormat || 'srt',
+            hasCorsRestrictions: false,
+            language,
+          });
+      }
+      return [...captions, ...removeDuplicatedLanguages(openSubtilesCaptions)];
+    } catch (error) {
+      // OpenSubtitles failed for a reason other than timeout
+      console.error('OpenSubtitles fetch failed:', error);
+      return captions;
     }
-    return [...captions, ...removeDuplicatedLanguages(openSubtilesCaptions)];
-  } catch {
+  } catch (error) {
+    console.error('Error in addOpenSubtitlesCaptions:', error);
     return captions;
   }
 }