Fixed Team Skeet deep scrape and URL format, added duration.

This commit is contained in:
DebaucheryLibrarian
2026-10-21 02:30:33 +02:00
parent fef686f977
commit 131f744745
2 changed files with 102 additions and 26 deletions
+16 -19
View File
@@ -33,7 +33,8 @@ async function fetchTrailerUrl(videoId, entity) {
const token = res.data.token;
if (token) {
return `https://customer-bk3o9te23pydwwcb.cloudflarestream.com/${token}/manifest/video.mpd?parentOrigin=${entity.origin}`;
// ffmpeg fails to read the DASH manifest over a single connection, HLS works
return `https://customer-bk3o9te23pydwwcb.cloudflarestream.com/${token}/manifest/video.m3u8?parentOrigin=${entity.origin}`;
}
}
@@ -46,11 +47,12 @@ async function scrapeScene(scene, channel, _parameters, includeTrailers) {
// release.entryId = scene.id; // legacy
release.entryId = scene.itemId;
release.url = `${channel.type === 'network' || channel.parameters?.layout === 'organic' ? channel.url : channel.parent.url}/movies/${release.entryId}`;
release.url = `${channel.type === 'network' || channel.parameters?.layout === 'organic' ? channel.url : channel.parent.url}/movies/${scene.id || release.entryId}`; // numeric item ID URL 404s, id is the slug
release.title = scene.title;
release.description = scene.description;
release.date = unprint.extractDate(scene.publishedDate, 'YYYY-MM-DD');
release.duration = scene.d;
release.actors = scene.models?.map((model) => ({
name: model.modelName || model.name || model.title,
@@ -113,27 +115,22 @@ async function fetchScene(url, channel, baseScene, { parameters, includeTrailers
return baseScene;
}
const sceneSlug = new URL(url).pathname.match(/\/([\w-]+$)/)[1];
const sceneSlug = new URL(url).pathname.match(/\/([\w-]+)\/?$/)?.[1];
const res = await unprint.get(url, {
parser: {
runScripts: 'dangerously',
},
});
if (!sceneSlug) {
return null;
}
if (res.ok) {
const videos = res.context.window.__INITIAL_STATE__?.content?.videosContent;
// scene page no longer embeds state, look up the scene in the elastic API by its slug instead
// legacy URLs use the numeric item ID, which 404s on the site but can still be looked up
const isItemId = /^\d+$/.test(sceneSlug);
const res = await unprint.get(`https://tours-store.psmcdn.net/${parameters.endpoint}/_search?q=(${isItemId ? `itemId:${sceneSlug}` : `id:"${sceneSlug}"`} AND type:video)&size=1`);
res.context.window.fetch = () => {}; // suppress fetch missing error
if (res.ok && res.data?.hits) {
const scene = res.data.hits.hits.find(({ _source: hit }) => (isItemId ? String(hit.itemId) : hit.id) === sceneSlug)?._source;
if (!videos) {
return null;
}
const video = videos?.[sceneSlug] || Object.values(videos)[0];
if (video) {
return scrapeScene(video, channel, parameters, includeTrailers);
if (scene) {
return scrapeScene(scene, channel, parameters, includeTrailers);
}
return null;