remove external subtitle scraping

This commit is contained in:
Pas 2025-08-01 14:54:29 -06:00
parent 6bea2a2ee5
commit 970fb7e598
6 changed files with 2 additions and 275 deletions

View file

@ -94,8 +94,7 @@
"nanoid": "^3.3.8",
"node-fetch": "^3.3.2",
"set-cookie-parser": "^2.7.1",
"unpacker": "^1.0.1",
"wyzie-lib": "^2.2.5"
"unpacker": "^1.0.1"
},
"packageManager": "pnpm@9.14.4"
}

View file

@ -44,9 +44,6 @@ importers:
unpacker:
specifier: ^1.0.1
version: 1.0.1
wyzie-lib:
specifier: ^2.2.5
version: 2.2.5
devDependencies:
'@nabla/vite-plugin-eslint':
specifier: ^2.0.5
@ -2539,9 +2536,6 @@ packages:
utf-8-validate:
optional: true
wyzie-lib@2.2.5:
resolution: {integrity: sha512-hOvAqS0VJhqTi2TRFkQ1YSEvX5ltrKC+ZWGiGDZ2Qn6b2t5Ov5DJILFCO2W/1ScK+BKOKHen7tRidatj1W62oA==}
y18n@5.0.8:
resolution: {integrity: sha512-0pfFzegeDWJHJIAmTLRP2DwHjdF5s7jo9tuztdQxAhINCdvS+3nGINqPd00AphqJR/0LhANUS6/+7SCb98YOfA==}
engines: {node: '>=10'}
@ -5248,8 +5242,6 @@ snapshots:
ws@8.18.0: {}
wyzie-lib@2.2.5: {}
y18n@5.0.8: {}
yallist@4.0.0: {}

View file

@ -6,7 +6,6 @@ import { EmbedOutput, SourcererOutput } from '@/providers/base';
import { ProviderList } from '@/providers/get';
import { ScrapeContext } from '@/utils/context';
import { NotFoundError } from '@/utils/errors';
import { addOpenSubtitlesCaptions } from '@/utils/opensubtitles';
import { requiresProxy, setupProxy } from '@/utils/proxy';
import { isValidStream, validatePlayableStreams } from '@/utils/valid';
@ -18,7 +17,6 @@ export type IndividualSourceRunnerOptions = {
id: string;
events?: IndividualScraperEvents;
proxyStreams?: boolean; // temporary
disableOpensubtitles?: boolean;
};
export async function scrapeInvidualSource(
@ -74,15 +72,6 @@ export async function scrapeInvidualSource(
return true;
});
// opensubtitles
if (!ops.disableOpensubtitles)
for (const embed of output.embeds)
embed.url = `${embed.url}${btoa('MEDIA=')}${btoa(
`${ops.media.imdbId}${
ops.media.type === 'show' ? `.${ops.media.season.number}.${ops.media.episode.number}` : ''
}`,
)}`;
if ((!output.stream || output.stream.length === 0) && output.embeds.length === 0)
throw new NotFoundError('No streams found');
@ -91,20 +80,6 @@ export async function scrapeInvidualSource(
const playableStreams = await validatePlayableStreams(output.stream, ops, sourceScraper.id);
if (playableStreams.length === 0) throw new NotFoundError('No playable streams found');
// opensubtitles
if (!ops.disableOpensubtitles) {
for (const playableStream of playableStreams) {
playableStream.captions = await addOpenSubtitlesCaptions(
playableStream.captions,
ops,
btoa(
`${ops.media.imdbId}${
ops.media.type === 'show' ? `.${ops.media.season.number}.${ops.media.episode.number}` : ''
}`,
),
);
}
}
output.stream = playableStreams;
}
return output;
@ -118,7 +93,6 @@ export type IndividualEmbedRunnerOptions = {
id: string;
events?: IndividualScraperEvents;
proxyStreams?: boolean; // temporary
disableOpensubtitles?: boolean;
};
export async function scrapeIndividualEmbed(
@ -128,9 +102,7 @@ export async function scrapeIndividualEmbed(
const embedScraper = list.embeds.find((v) => ops.id === v.id);
if (!embedScraper) throw new Error('Embed with ID not found');
let url = ops.url;
let media;
if (ops.url.includes(btoa('MEDIA='))) [url, media] = url.split(btoa('MEDIA='));
const url = ops.url;
const output = await embedScraper.scrape({
fetcher: ops.fetcher,
@ -157,22 +129,6 @@ export async function scrapeIndividualEmbed(
const playableStreams = await validatePlayableStreams(output.stream, ops, embedScraper.id);
if (playableStreams.length === 0) throw new NotFoundError('No playable streams found');
if (media && !ops.disableOpensubtitles) {
const [imdbId, season, episode] = atob(media)
.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,
};
for (const playableStream of playableStreams)
playableStream.captions = await addOpenSubtitlesCaptions(playableStream.captions, mediaInfo, media);
}
output.stream = playableStreams;
return output;

View file

@ -8,7 +8,6 @@ import { Stream } from '@/providers/streams';
import { ScrapeContext } from '@/utils/context';
import { NotFoundError } from '@/utils/errors';
import { reorderOnIdList } from '@/utils/list';
import { addOpenSubtitlesCaptions } from '@/utils/opensubtitles';
import { requiresProxy, setupProxy } from '@/utils/proxy';
import { isValidStream, validatePlayableStream } from '@/utils/valid';
@ -38,7 +37,6 @@ export type ProviderRunnerOptions = {
events?: FullScraperEvents;
media: ScrapeMedia;
proxyStreams?: boolean; // temporary
disableOpensubtitles?: boolean;
};
export async function runAllProviders(list: ProviderList, ops: ProviderRunnerOptions): Promise<RunOutput | null> {
@ -115,21 +113,6 @@ export async function runAllProviders(list: ProviderList, ops: ProviderRunnerOpt
const playableStream = await validatePlayableStream(output.stream[0], ops, source.id);
if (!playableStream) throw new NotFoundError('No streams found');
// opensubtitles
if (!ops.disableOpensubtitles) {
if (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}` : ''
}`,
),
);
}
}
return {
sourceId: source.id,
stream: playableStream,
@ -181,20 +164,6 @@ export async function runAllProviders(list: ProviderList, ops: ProviderRunnerOpt
const playableStream = await validatePlayableStream(embedOutput.stream[0], ops, embed.embedId);
if (!playableStream) throw new NotFoundError('No streams found');
// opensubtitles
if (!ops.disableOpensubtitles) {
if (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}` : ''
}`,
),
);
}
}
embedOutput.stream = [playableStream];
} catch (error) {
const updateParams: UpdateEvent = {

View file

@ -1,110 +0,0 @@
/* eslint-disable no-console */
import { ScrapeMedia } from '@/entrypoint/utils/media';
import { Caption, labelToLanguageCode } from '@/providers/captions';
import { IndividualEmbedRunnerOptions } from '@/runners/individualRunner';
import { ProviderRunnerOptions } from '@/runners/runner';
import { addWyzieCaptions } from './wyziesubs';
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 ${ms}ms`);
resolve(null);
}, ms);
});
export async function addOpenSubtitlesCaptions(
captions: Caption[],
ops: CaptionOptions,
media: string,
): Promise<Caption[]> {
try {
const [imdbId, season, episode] = atob(media)
.split('.')
.map((x, i) => (i === 0 ? x : Number(x) || null));
if (!imdbId) return captions;
const allCaptions = [...captions];
// Fetch Wyzie captions with 2 second timeout
const wyziePromise = addWyzieCaptions(
[],
ops.media?.tmdbId?.toString() || '',
imdbId.toString(),
typeof season === 'number' ? season : undefined,
typeof episode === 'number' ? episode : undefined,
)
.then((wyzieCaptions) => {
if (wyzieCaptions && wyzieCaptions.length > 0) {
return wyzieCaptions.map((caption) => ({
...caption,
opensubtitles: true,
}));
}
return [];
})
.catch((error) => {
console.error('Wyzie subtitles fetch failed:', error);
return [];
});
// Fetch OpenSubtitles captions with 5 second timeout
const openSubsPromise = ops
.proxiedFetcher(
`https://rest.opensubtitles.org/search/${
season && episode ? `episode-${episode}/` : ''
}imdbid-${(imdbId as string).slice(2)}${season && episode ? `/season-${season}` : ''}`,
{
headers: {
'X-User-Agent': 'VLSub 0.10.2',
},
},
)
.then((Res) => {
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 openSubtilesCaptions;
})
.catch((error) => {
console.error('OpenSubtitles fetch failed:', error);
return [];
});
// Wait for both promises with their respective timeouts
const [wyzieCaptions, openSubsCaptions] = await Promise.all([
Promise.race([wyziePromise, timeout(2000, 'Wyzie')]),
Promise.race([openSubsPromise, timeout(5000, 'OpenSubtitles')]),
]);
// Debug logging
// console.log('Wyzie captions found:', wyzieCaptions?.length || 0);
// console.log('OpenSubtitles captions found:', openSubsCaptions?.length || 0);
// Add any successful captions to our result
if (wyzieCaptions) allCaptions.push(...wyzieCaptions);
if (openSubsCaptions) allCaptions.push(...openSubsCaptions);
return allCaptions;
} catch (error) {
console.error('Error in addOpenSubtitlesCaptions:', error);
return captions;
}
}

View file

@ -1,79 +0,0 @@
/* eslint-disable no-console */
import { type SubtitleData, searchSubtitles } from 'wyzie-lib';
import { Caption } from '@/providers/captions';
// function isSubdlUrl(url: string) {
// return url.endsWith('.subdl');
// }
// export function filterSubtitles(list: Caption[]) {
// const selected: Record<string, Caption> = {};
// for (const sub of list) {
// const existing = selected[sub.language];
// if (!existing) {
// selected[sub.language] = sub;
// continue;
// }
// const existingIsSubdl = isSubdlUrl(existing.url);
// const currentIsSubdl = isSubdlUrl(sub.url);
// if (existingIsSubdl && !currentIsSubdl) {
// selected[sub.language] = sub;
// }
// }
// return Object.values(selected);
// }
export async function addWyzieCaptions(
captions: Caption[],
tmdbId: string | number,
imdbId: string,
season?: number,
episode?: number,
): Promise<Caption[]> {
try {
const searchParams: any = {
encoding: 'utf-8',
source: 'all',
imdb_id: imdbId,
};
if (tmdbId && !imdbId) {
searchParams.tmdb_id = typeof tmdbId === 'string' ? parseInt(tmdbId, 10) : tmdbId;
}
if (season && episode) {
searchParams.season = season;
searchParams.episode = episode;
}
console.log('Searching Wyzie subtitles with params:', searchParams);
const wyzieSubtitles: SubtitleData[] = await searchSubtitles(searchParams);
// console.log('Found Wyzie subtitles:', wyzieSubtitles);
const wyzieCaptions: Caption[] = wyzieSubtitles.map((subtitle) => ({
id: subtitle.id,
url: subtitle.url,
type: subtitle.format === 'srt' || subtitle.format === 'vtt' ? subtitle.format : 'srt',
hasCorsRestrictions: false,
language: subtitle.language,
// Additional metadata from Wyzie
flagUrl: subtitle.flagUrl,
display: subtitle.display,
media: subtitle.media,
isHearingImpaired: subtitle.isHearingImpaired,
source: typeof subtitle.source === 'number' ? subtitle.source.toString() : subtitle.source,
encoding: subtitle.encoding,
}));
return [...captions, ...wyzieCaptions];
} catch (error) {
console.error('Error fetching Wyzie subtitles:', error);
return captions;
}
}