963 lines
29 KiB
JavaScript
Executable File
963 lines
29 KiB
JavaScript
Executable File
'use strict';
|
|
|
|
const Promise = require('bluebird');
|
|
const util = require('util');
|
|
const { JSDOM } = require('jsdom');
|
|
const moment = require('moment');
|
|
const format = require('template-format');
|
|
|
|
const logger = require('../logger')(__filename);
|
|
const qu = require('../utils/qu');
|
|
const http = require('../utils/http');
|
|
const slugify = require('../utils/slugify');
|
|
|
|
function getApiUrl(appId, apiKey) {
|
|
const userAgent = 'Algolia for vanilla JavaScript (lite) 3.27.0;instantsearch.js 2.7.4;JS Helper 2.26.0';
|
|
|
|
const apiUrl = `https://${appId.toLowerCase()}-dsn.algolia.net/1/indexes/*/queries?x-algolia-agent=${userAgent}&x-algolia-application-id=${appId}&x-algolia-api-key=${apiKey}`;
|
|
|
|
return {
|
|
appId,
|
|
apiKey,
|
|
userAgent,
|
|
apiUrl,
|
|
};
|
|
}
|
|
|
|
function getAvatarFallbacks(avatar) {
|
|
if (!avatar) {
|
|
return null;
|
|
}
|
|
|
|
return [
|
|
avatar.replace(/\d+x\d+/, '500x750'),
|
|
avatar.replace(/\d+x\d+/, '240x360'),
|
|
avatar.replace(/\d+x\d+/, '200x300'),
|
|
avatar,
|
|
];
|
|
}
|
|
|
|
async function fetchApiCredentials(referer, site) {
|
|
if (site?.parameters?.appId && site?.parameters?.apiKey) {
|
|
return getApiUrl(site.parameters.appId, site.parameters.apiKey);
|
|
}
|
|
|
|
const res = await http.get(referer);
|
|
const body = res.body.toString();
|
|
|
|
const apiLine = body.split('\n').find((bodyLine) => bodyLine.match('apiKey'));
|
|
|
|
if (!apiLine) {
|
|
throw new Error(`No Gamma API key found for ${referer}`);
|
|
}
|
|
|
|
const apiSerial = apiLine.slice(apiLine.indexOf('{'), apiLine.indexOf('};') + 1);
|
|
const apiData = JSON.parse(apiSerial);
|
|
|
|
const { applicationID: appId, apiKey } = apiData.api.algolia;
|
|
|
|
return getApiUrl(appId, apiKey);
|
|
}
|
|
|
|
function getAlbumUrl(albumPath, site) {
|
|
if (site.parameters?.photos) {
|
|
return /^http/.test(site.parameters.photos)
|
|
? `${site.parameters.photos}/${albumPath.split('/').slice(-2).join('/')}`
|
|
: `${site.url}${site.parameters.photos}/${albumPath.split('/').slice(-2).join('/')}`;
|
|
}
|
|
|
|
if (site.url && site.parameters?.photos !== false) {
|
|
return `${site.url}${albumPath}`;
|
|
}
|
|
|
|
return null;
|
|
}
|
|
|
|
async function fetchPhotos(url) {
|
|
const res = await qu.get(url);
|
|
|
|
return res.item;
|
|
}
|
|
|
|
function scrapePhotos({ query }, includeThumbnails = true) {
|
|
return query.all('.preview .imgLink, .pgFooterThumb a').map((linkEl) => {
|
|
const url = linkEl.href;
|
|
|
|
if (/\/join|\/createaccount/.test(url)) {
|
|
// URL links to join page instead of full photo, extract thumbnail
|
|
// /createaccount is used by e.g. Tricky Spa native site
|
|
const src = query.img(linkEl);
|
|
|
|
if (/previews\//.test(src)) {
|
|
// resource often serves full photo at a modifier URL anyway, add as primary source
|
|
const highRes = src
|
|
.replace('previews/', '')
|
|
.replace('_tb.jpg', '.jpg');
|
|
|
|
// keep original thumbnail as fallback in case full photo is not available
|
|
return [highRes, src];
|
|
}
|
|
|
|
if (!includeThumbnails) return null;
|
|
|
|
return src;
|
|
}
|
|
|
|
// URL links to full photo
|
|
return url;
|
|
}).filter(Boolean);
|
|
}
|
|
|
|
async function getPhotos(albumPath, site, includeThumbnails = true) {
|
|
const albumUrl = getAlbumUrl(albumPath, site);
|
|
|
|
if (!albumUrl) {
|
|
return [];
|
|
}
|
|
|
|
try {
|
|
const item = await fetchPhotos(albumUrl);
|
|
const photos = scrapePhotos(item, includeThumbnails);
|
|
|
|
const lastPage = item.query.url('.Gamma_Paginator a.last')?.match(/\d+$/)[0];
|
|
|
|
if (lastPage) {
|
|
const otherPages = Array.from({ length: Number(lastPage) }, (_value, index) => index + 1).slice(1);
|
|
|
|
const otherPhotos = await Promise.map(otherPages, async (page) => {
|
|
const pageItem = await fetchPhotos(`${albumUrl}/${page}`);
|
|
|
|
return scrapePhotos(pageItem, includeThumbnails);
|
|
}, {
|
|
concurrency: 2,
|
|
});
|
|
|
|
return photos.concat(otherPhotos.flat());
|
|
}
|
|
|
|
return photos;
|
|
} catch (error) {
|
|
logger.warn(`Failed to fetch ${site.name} photos from ${albumUrl}: ${error.message}`);
|
|
|
|
return [];
|
|
}
|
|
}
|
|
|
|
async function getFullPhotos(entryId, site) {
|
|
const res = await http.get(`${site.url}/media/signPhotoset/${entryId}`, {
|
|
headers: {
|
|
'X-Requested-With': 'XMLHttpRequest',
|
|
},
|
|
});
|
|
|
|
if (res.ok && typeof res.body === 'object') { // gives 200 OK even when redirected to signup page
|
|
return Object.values(res.body);
|
|
}
|
|
|
|
return [];
|
|
}
|
|
|
|
async function getThumbs(entryId, site, parameters) {
|
|
const referer = parameters?.referer || `${parameters?.networkReferer ? site.parent.url : site.url}/en/videos`;
|
|
const { apiUrl } = await fetchApiCredentials(referer, site);
|
|
|
|
const res = await http.post(apiUrl, {
|
|
requests: [
|
|
{
|
|
indexName: 'all_photosets',
|
|
params: `query=&page=0&facets=[]&tagFilters=&facetFilters=[["set_id:${entryId}"]]`,
|
|
},
|
|
],
|
|
}, {
|
|
headers: {
|
|
Referer: referer,
|
|
},
|
|
}, {
|
|
encodeJSON: true,
|
|
});
|
|
|
|
if (res.ok && res.body.results?.[0]?.hits[0]?.set_pictures) {
|
|
return res.body.results[0].hits[0].set_pictures.map((img) => img.thumb_path && ([
|
|
`https://images-fame.gammacdn.com/photo_set${img.thumb_path}`,
|
|
`https://images01-fame.gammacdn.com/photo_set${img.thumb_path}`,
|
|
`https://images02-fame.gammacdn.com/photo_set${img.thumb_path}`,
|
|
`https://images03-fame.gammacdn.com/photo_set${img.thumb_path}`,
|
|
`https://images04-fame.gammacdn.com/photo_set${img.thumb_path}`,
|
|
`https://images-evilangel.gammacdn.com/photo_set${img.thumb_path}`,
|
|
`https://transform.gammacdn.com/photo_set${img.thumb_path}`,
|
|
])).filter(Boolean);
|
|
}
|
|
|
|
return [];
|
|
}
|
|
|
|
async function getPhotosApi(entryId, site, parameters) {
|
|
const [photos, thumbs] = await Promise.all([
|
|
getFullPhotos(entryId, site, parameters),
|
|
getThumbs(entryId, site, parameters),
|
|
]);
|
|
|
|
return photos.concat(thumbs.slice(photos.length));
|
|
}
|
|
|
|
function getImageSources(source) {
|
|
return [
|
|
`https://images-fame.gammacdn.com/movies${source}`,
|
|
`https://images01-fame.gammacdn.com/movies${source}`,
|
|
`https://images02-fame.gammacdn.com/movies${source}`,
|
|
`https://images03-fame.gammacdn.com/movies${source}`,
|
|
`https://images04-fame.gammacdn.com/movies${source}`,
|
|
`https://images-evilangel.gammacdn.com/movies${source}`,
|
|
`https://transform.gammacdn.com/movies${source}`,
|
|
];
|
|
}
|
|
|
|
function curateTitle(title, channel) {
|
|
// some videos are redundantly prefixed with the name of the site, i.e. Bubblegum Dungeon, Forbidden Seductions and Lady Gonzo
|
|
return title.replace(new RegExp(`^\\s*${channel.name}\\s*[:|-]\\s`, 'i'), '');
|
|
}
|
|
|
|
async function scrapeApiReleases(json, site) {
|
|
return json.map((scene) => {
|
|
if (site.parameters?.extract && scene.sitename !== site.parameters.extract) {
|
|
return null;
|
|
}
|
|
|
|
if (site.parameters?.filterExclusive && scene.availableOnSite.length > 1) {
|
|
return null;
|
|
}
|
|
|
|
const release = {
|
|
entryId: scene.clip_id,
|
|
description: scene.description,
|
|
duration: scene.length,
|
|
likes: scene.ratings_up,
|
|
dislikes: scene.ratings_down,
|
|
};
|
|
|
|
release.title = curateTitle(scene.title, site);
|
|
release.path = `/${scene.url_title}/${release.entryId}`;
|
|
|
|
if (site.parameters?.scene) release.url = `${site.parameters.scene}${release.path}`;
|
|
else if (site.url && site.parameters?.scene !== false) release.url = `${site.url}/en/video${release.path}`;
|
|
|
|
release.date = moment.utc(scene.release_date, 'YYYY-MM-DD').toDate();
|
|
release.director = scene.directors[0]?.name || null;
|
|
|
|
release.actors = scene.actors.map((actor) => ({
|
|
entryId: actor.actor_id,
|
|
name: actor.name,
|
|
gender: actor.gender,
|
|
avatar: [
|
|
`https://images03-openlife.gammacdn.com/actors/${actor.actor_id}/${actor.actor_id}_500x750.jpg`,
|
|
`https://images03-openlife.gammacdn.com/actors/${actor.actor_id}/${actor.actor_id}_240x360.jpg`,
|
|
`https://images03-openlife.gammacdn.com/actors/${actor.actor_id}/${actor.actor_id}_200x300.jpg`,
|
|
],
|
|
}));
|
|
|
|
/* master categories include e.g. 'transgender' for non-trans Wicked scenes
|
|
release.tags = scene.master_categories
|
|
.concat(scene.categories?.map((category) => category.name))
|
|
.filter(Boolean); // some categories don't have a name
|
|
*/
|
|
|
|
release.tags = scene.categories?.map((category) => category.name).filter(Boolean); // some categories don't have a name
|
|
|
|
if (scene.availableOnSite.length > 1) {
|
|
release.comment = `Also available on ${scene.availableOnSite.filter((sisterSite) => sisterSite !== site.slug).join(', ')}`;
|
|
}
|
|
|
|
const posterPath = scene.pictures.resized || (scene.pictures.nsfw?.top && Object.values(scene.pictures.nsfw.top)[0]);
|
|
|
|
if (posterPath) {
|
|
release.poster = [
|
|
`https://images-evilangel.gammacdn.com/movies${posterPath}`,
|
|
`https://transform.gammacdn.com/movies${posterPath}`,
|
|
];
|
|
}
|
|
|
|
// release.movie = `${site.url}/en/movie/${scene.url_movie_title}/${scene.movie_id}`;
|
|
|
|
return release;
|
|
}).filter(Boolean);
|
|
}
|
|
|
|
function scrapeAll(scenes, site, networkUrl, hasTeaser = true) {
|
|
return scenes.map(({ query, el }) => {
|
|
const release = {};
|
|
|
|
release.url = query.url('.sceneTitle a, .tlcTitle a', 'href', { origin: networkUrl ? site.parent.url : site.url });
|
|
|
|
release.title = query.cnt('.sceneTitle a', 'tlcTitle a', 'title');
|
|
release.entryId = el.dataset.itemid;
|
|
|
|
release.date = query.date('.sceneDate, .tlcSpecsDate .tlcDetailsValue', ['MM-DD-YYYY', 'YYYY-MM-DD']);
|
|
release.actors = query.cnts('.sceneActors a, .tlcActors a', ' title');
|
|
|
|
[release.likes, release.dislikes] = query.all('.value').map((likeEl) => query.number(likeEl));
|
|
|
|
release.poster = query.img('.imgLink img, .tlcImageItem', 'data-original') || query.img('.imgLink img, .tlcImageItem');
|
|
|
|
if (hasTeaser) {
|
|
release.teaser = [
|
|
{ src: `https://videothumb.gammacdn.com/600x339/${release.entryId}.mp4` },
|
|
{ src: `https://videothumb.gammacdn.com/307x224/${release.entryId}.mp4` },
|
|
];
|
|
}
|
|
|
|
release.channel = query.el('.fromSite a', 'title')?.replace('.com', '');
|
|
|
|
return release;
|
|
});
|
|
}
|
|
|
|
async function scrapeScene({ query }, url, channel, baseRelease, mobileItem, options) {
|
|
const release = { query }; // used by XEmpire scraper to resolve channel-specific details
|
|
|
|
const json = query.html('script[type="application/ld+json"]');
|
|
const videoJson = query.htmls('script').find((script) => /ScenePlayerOptions/i.test(script));
|
|
|
|
const [data, data2] = json ? JSON.parse(json) : [];
|
|
const videoData = videoJson && JSON.parse(videoJson.slice(videoJson.indexOf('{'), videoJson.indexOf('};') + 1));
|
|
|
|
release.entryId = (baseRelease?.path || new URL(url).pathname).match(/\/(\d{2,})(\/|$)/)?.[1];
|
|
release.title = videoData?.playerOptions?.sceneInfos.sceneTitle || data?.name;
|
|
release.description = data?.description;
|
|
|
|
release.date = query.date('.updatedDate', ['MM-DD-YYYY', 'YYYY-MM-DD'])
|
|
|| qu.extractDate(data?.dateCreated, 'YYYY-MM-DD')
|
|
|| videoData?.playerOptions?.sceneInfos.sceneReleaseDate;
|
|
|
|
release.actors = (data?.actor || data2?.actor)?.map((actor) => ({
|
|
name: actor.name,
|
|
gender: actor.gender,
|
|
})) || [];
|
|
|
|
release.duration = qu.durationToSeconds(data.duration);
|
|
release.director = data?.director?.[0]?.name || data2?.director?.[0]?.name;
|
|
|
|
release.tags = data?.keywords?.split(', ') || data2?.keywords?.split(', ') || [];
|
|
release.stars = (data.aggregateRating.ratingValue / data.aggregateRating.bestRating) * 5 || null;
|
|
|
|
release.channel = slugify(data?.productionCompany?.name
|
|
|| query.el('.studioLink a, .siteLink a', 'title')
|
|
|| query.cnt('.siteNameSpan')?.toLowerCase().replace('.com', '')
|
|
|| query.meta('meta[name="twitter:domain"]')?.replace('.com', ''), '');
|
|
|
|
if (videoData?.picPreview && new URL(videoData.picPreview).pathname.length > 1) {
|
|
// sometimes links to just https://images02-fame.gammacdn.com/
|
|
const poster = new URL(videoData.picPreview);
|
|
|
|
release.poster = [
|
|
videoData.picPreview, // prefer original URL with width and height parameters, without may give a square crop on e.g. XEmpire
|
|
`${poster.origin}${poster.pathname}`,
|
|
];
|
|
}
|
|
|
|
const photoLink = query.url('.picturesItem a');
|
|
const mobilePhotos = mobileItem?.query.imgs('.preview-displayer a img') || [];
|
|
|
|
if (photoLink && options.includePhotos) {
|
|
const photos = await getPhotos(photoLink, channel, mobilePhotos.length < 3); // only get thumbnails when less than 3 mobile photos are available
|
|
|
|
if (photos.length < 7) {
|
|
release.photos = [...photos, ...mobilePhotos]; // probably only teaser photos available, supplement with mobile album
|
|
} else {
|
|
release.photos = photos;
|
|
}
|
|
} else {
|
|
release.photos = mobilePhotos;
|
|
}
|
|
|
|
const trailer = videoData && `${videoData.playerOptions.host}${videoData.url}`;
|
|
|
|
if (trailer) {
|
|
release.trailer = [
|
|
{
|
|
src: trailer.replace('hd', 'sm'),
|
|
quality: 240,
|
|
},
|
|
{
|
|
src: trailer.replace('hd', 'med'),
|
|
quality: 360,
|
|
},
|
|
{
|
|
src: trailer.replace('hd', 'big'),
|
|
quality: 480,
|
|
},
|
|
{
|
|
// probably 540p
|
|
src: trailer,
|
|
quality: parseInt(videoData.sizeOnLoad, 10),
|
|
},
|
|
{
|
|
src: trailer.replace('hd', '720p'),
|
|
quality: 720,
|
|
},
|
|
{
|
|
src: trailer.replace('hd', '1080p'),
|
|
quality: 1080,
|
|
},
|
|
{
|
|
src: trailer.replace('hd', '4k'),
|
|
quality: 2160,
|
|
},
|
|
];
|
|
}
|
|
|
|
const movieUrl = query.url('.dvdLink', 'href', { origin: channel.url });
|
|
|
|
if (movieUrl) {
|
|
release.movie = {
|
|
url: movieUrl,
|
|
title: query.el('.dvdLink', 'title'),
|
|
entryId: movieUrl.match(/\/(\d+)(\/|$)/)?.[1],
|
|
covers: [qu.imgs('.dvdLink img')],
|
|
};
|
|
}
|
|
|
|
return release;
|
|
}
|
|
|
|
async function scrapeReleaseApi(data, site, options, movieScenes) {
|
|
const release = {};
|
|
|
|
release.entryId = data.clip_id || data.movie_id;
|
|
release.title = curateTitle(data.title, site);
|
|
release.duration = data.length;
|
|
release.date = (data.date && new Date(data.date * 1000)) || qu.parseDate(data.release_date || data.last_modified, 'YYYY-MM-DD');
|
|
release.director = data.directors[0]?.name || null;
|
|
|
|
release.actors = data.actors.map((actor) => ({
|
|
entryId: actor.actor_id,
|
|
name: actor.name,
|
|
gender: actor.gender,
|
|
url: options.parameters?.actors
|
|
? format(options.parameters.actors, { id: actor.actor_id, slug: actor.url_name })
|
|
: qu.prefixUrl(`/en/pornstar/${actor.url_name}/${data.actor_id}`, site.url),
|
|
}));
|
|
|
|
release.tags = data.categories.map((category) => category.name);
|
|
|
|
if (data.pictures) {
|
|
release.poster = [
|
|
...(data.pictures['1920x1080'] ? getImageSources(data.pictures['1920x1080']) : []),
|
|
...(data.pictures.resized ? getImageSources(data.pictures.resized) : []),
|
|
...(data.pictures['960x544'] ? getImageSources(data.pictures['960x544']) : []),
|
|
];
|
|
}
|
|
|
|
if (data.photoset_id && options.includePhotos && options.parameters?.includePhotos !== false) {
|
|
release.photos = await getPhotosApi(data.photoset_id, site, options.parameters);
|
|
}
|
|
|
|
if (data.cover_path) {
|
|
release.covers = [
|
|
getImageSources(`${data.cover_path}_front_400x625.jpg?width=450&height=636&format=webp`),
|
|
getImageSources(`${data.cover_path}_back_400x625.jpg?width=450&height=636&format=webp`),
|
|
];
|
|
}
|
|
|
|
release.teaser = `https://videothumb.gammacdn.com/500x281/${release.entryId}.mp4`;
|
|
|
|
if (data.trailers) {
|
|
release.trailer = Object.entries(data.trailers).map(([quality, source]) => ({ src: source, quality }));
|
|
}
|
|
|
|
if (data.movie_id && !data.movie_path && options.parameters.sceneMovies !== false) {
|
|
release.movie = {
|
|
entryId: data.movie_id,
|
|
title: data.movie_title,
|
|
url: qu.prefixUrl(`${data.url_movie_title}/${data.movie_id}`, options.parameters.movie ? options.parameters.movie : `${site.url}/en/movie`),
|
|
};
|
|
}
|
|
|
|
if (movieScenes?.length > 0) {
|
|
release.scenes = await Promise.all(movieScenes.map((movieScene) => scrapeReleaseApi(movieScene, site, options)));
|
|
}
|
|
|
|
release.channel = data.sitename;
|
|
release.qualities = data.download_sizes;
|
|
|
|
return release;
|
|
}
|
|
|
|
async function fetchMovieTrailer(release) {
|
|
if (!release.entryId) {
|
|
return null;
|
|
}
|
|
|
|
const url = `https://www.evilangel.com/en/dvdtrailer/${release.entryId}`;
|
|
const res = await qu.get(url);
|
|
|
|
if (!res.ok) {
|
|
return null;
|
|
}
|
|
|
|
const trailerHost = res.html.match(/"host":\s*"(.*\.com)"/)?.[1].replace(/\\\//g, '/');
|
|
const trailerPath = res.html.match(/"url":\s*"(.*\.mp4)"/)?.[1].replace(/\\\//g, '/');
|
|
|
|
if (trailerHost && trailerPath) {
|
|
return qu.prefixUrl(trailerPath, trailerHost);
|
|
}
|
|
|
|
return null;
|
|
}
|
|
|
|
async function scrapeMovie({ query, el }, url, entity, baseRelease, options) {
|
|
const release = {};
|
|
|
|
const { dataLayer } = query.exec('//script[contains(text(), "dataLayer")]', ['dataLayer']);
|
|
const rawData = dataLayer?.[0]?.dvdDetails;
|
|
const data = rawData?.dvdId && rawData; // dvdDetails is mostly empty in some cache states
|
|
|
|
if (query.exists('.NotFound-Title')) {
|
|
return null;
|
|
}
|
|
|
|
release.entryId = new URL(url).pathname.match(/\/(\d+)(\/|$)/)?.[1];
|
|
|
|
release.covers = [
|
|
query.img('.frontCoverImg', 'href'),
|
|
query.img('.backCoverImg', 'href'),
|
|
];
|
|
|
|
release.description = query.cnt('.descriptionText');
|
|
release.date = qu.extractDate(data?.dvdReleaseDate) || query.date('.updatedOn', 'YYYY-MM-DD');
|
|
release.title = data?.dvdName || query.cnt('.dvdTitle');
|
|
release.director = query.el('.directedBy a', 'title');
|
|
|
|
release.actors = data?.dvdActors.map((actor) => ({ name: actor.actorName, entryId: actor.actorId }))
|
|
|| query.all('.actorCarousel a[href*="/pornstar"]').map((actorEl) => ({
|
|
entryId: query.url(actorEl, null).match(/\/(\d+)/)?.[1],
|
|
name: query.cnt(actorEl, 'span'),
|
|
href: query.url(actorEl, null, 'href', { origin: entity.url }),
|
|
avatar: getAvatarFallbacks(query.img(actorEl)),
|
|
}));
|
|
|
|
release.tags = query.cnts('.dvdCol a');
|
|
release.scenes = scrapeAll(qu.initAll(el, 'div[data-itemtype*=scene], li[data-itemtype*=scene]'), entity, entity.url);
|
|
|
|
if (options.includeTrailers) {
|
|
release.trailer = await fetchMovieTrailer(release);
|
|
}
|
|
|
|
return release;
|
|
}
|
|
|
|
function scrapeActorSearch(html, url, actorName) {
|
|
const { document } = new JSDOM(html).window;
|
|
const actorLink = document.querySelector(`a[title="${actorName}" i]`);
|
|
|
|
return actorLink ? actorLink.href : null;
|
|
}
|
|
|
|
async function fetchActorReleases(profileUrl, getActorReleasesUrl, page = 1, accReleases = [], context) {
|
|
const { origin, pathname } = new URL(profileUrl);
|
|
const profilePath = `/${pathname.split('/').slice(-2).join('/')}`;
|
|
|
|
const url = (context.parameters.actorScenes && format(context.parameters.actorScenes, { path: profilePath, page }))
|
|
|| getActorReleasesUrl?.(profilePath, page);
|
|
|
|
if (!url) {
|
|
return [];
|
|
}
|
|
|
|
const res = await qu.get(url);
|
|
|
|
if (!res.ok) {
|
|
return [];
|
|
}
|
|
|
|
const releases = scrapeAll(res.item.html, null, origin);
|
|
const nextPage = res.item.query.url('.Gamma_Paginator a.next');
|
|
|
|
if (nextPage) {
|
|
return fetchActorReleases(profileUrl, getActorReleasesUrl, page + 1, accReleases.concat(releases), context);
|
|
}
|
|
|
|
return accReleases.concat(releases);
|
|
}
|
|
|
|
async function scrapeProfile({ query }, url, actorName, _siteSlug, getActorReleasesUrl, withReleases, context) {
|
|
const avatar = query.img('img.actorPicture');
|
|
const hair = query.cnt('.actorProfile .attribute_hair_color');
|
|
const height = query.cnt('.actorProfile .attribute_height');
|
|
const weight = query.cnt('.actorProfile .attribute_weight');
|
|
const alias = query.cnt('.actorProfile .attribute_alternate_names');
|
|
const nationality = query.cnt('.actorProfile .attribute_home');
|
|
|
|
const profile = {
|
|
name: actorName,
|
|
};
|
|
|
|
if (avatar) {
|
|
// larger sizes usually available, provide fallbacks
|
|
const avatars = getAvatarFallbacks(avatar);
|
|
|
|
profile.avatar = avatars;
|
|
}
|
|
|
|
profile.description = query.cnt('.actorBio p:not(.bioTitle)');
|
|
|
|
if (hair) profile.hair = hair.split(':')[1].trim();
|
|
if (height) profile.height = Number(height.match(/\d+/)[0]);
|
|
if (weight) profile.weight = Number(weight.match(/\d+/)[0]);
|
|
if (alias) profile.aliases = alias.split(':')[1].trim().split(', ');
|
|
if (nationality) profile.nationality = nationality.split(':')[1].trim();
|
|
|
|
if ((getActorReleasesUrl || context.parameters.actorScenes) && withReleases) {
|
|
profile.releases = await fetchActorReleases(url, getActorReleasesUrl, 1, [], context);
|
|
}
|
|
|
|
return profile;
|
|
}
|
|
|
|
function scrapeApiProfile(data, releases, siteSlug) {
|
|
const profile = {};
|
|
|
|
if (data.male === 1) profile.gender = 'male';
|
|
if (data.female === 1) profile.gender = 'female';
|
|
if (data.shemale === 1 || data.trans === 1) profile.gender = 'transsexual';
|
|
|
|
if (data.description) profile.description = data.description.trim();
|
|
|
|
if (data.attributes.ethnicity) profile.ethnicity = data.attributes.ethnicity;
|
|
if (data.attributes.eye_color) profile.eyes = data.attributes.eye_color;
|
|
if (data.attributes.hair_color) profile.hair = data.attributes.hair_color;
|
|
|
|
const avatarPaths = Object.values(data.pictures).reverse();
|
|
if (avatarPaths.length > 0) profile.avatar = avatarPaths.map((avatarPath) => `https://images01-evilangel.gammacdn.com/actors${avatarPath}`);
|
|
|
|
if (releases) profile.releases = releases.map((release) => `https://${siteSlug}.com/en/video/${release.url_title}/${release.clip_id}`);
|
|
|
|
return profile;
|
|
}
|
|
|
|
async function fetchLatestApi(site, page = 1, options, preData, upcoming = false) {
|
|
const referer = options.parameters?.referer || `${options.parameters?.networkReferer ? site.parent.url : site.url}/en/videos`;
|
|
const { apiUrl } = await fetchApiCredentials(referer, site);
|
|
|
|
const res = await http.post(apiUrl, {
|
|
requests: [
|
|
{
|
|
indexName: 'all_scenes',
|
|
params: `query=&hitsPerPage=36&maxValuesPerFacet=100&page=${page - 1}&facetFilters=[["lesbian:"],["bisex:"],["shemale:"],["upcoming:${upcoming ? 1 : 0}"]]&filters=sitename:${site.slug}`, // OR channels.id:${site.slug}`,
|
|
},
|
|
],
|
|
}, {
|
|
headers: {
|
|
Referer: referer,
|
|
},
|
|
}, {
|
|
encodeJSON: true,
|
|
});
|
|
|
|
if (res.status === 200 && res.body.results?.[0]?.hits) {
|
|
return scrapeApiReleases(res.body.results[0].hits, site);
|
|
}
|
|
|
|
return res.status;
|
|
}
|
|
|
|
async function fetchUpcomingApi(site, page = 1, options, preData) {
|
|
return fetchLatestApi(site, page, options, preData, true);
|
|
}
|
|
|
|
async function fetchSceneApi(url, site, baseRelease, options) {
|
|
const referer = options.parameters?.referer || `${site.parameters?.networkReferer ? site.parent.url : site.url}/en/videos`;
|
|
const { apiUrl } = await fetchApiCredentials(referer, site);
|
|
|
|
const entryId = (baseRelease?.path || new URL(url).pathname).match(/\/(\d{2,})(\/|$)/)?.[1];
|
|
|
|
const res = await http.post(apiUrl, {
|
|
requests: [
|
|
{
|
|
indexName: 'all_scenes',
|
|
params: `query=&page=0&facets=[]&tagFilters=&facetFilters=[["clip_id:${entryId}"]]`,
|
|
},
|
|
{
|
|
indexName: 'all_scenes',
|
|
params: 'query=&page=0&hitsPerPage=1&attributesToRetrieve=[]&attributesToHighlight=[]&attributesToSnippet=[]&tagFilters=&analytics=false&clickAnalytics=false&facets=clip_id',
|
|
},
|
|
],
|
|
}, {
|
|
headers: {
|
|
Referer: referer,
|
|
},
|
|
}, {
|
|
encodeJSON: true,
|
|
});
|
|
|
|
if (res.status === 200 && res.body.results?.[0]?.hits.length > 0) {
|
|
return scrapeReleaseApi(res.body.results[0].hits[0], site, options);
|
|
}
|
|
|
|
if (res.status === 200) {
|
|
return null;
|
|
}
|
|
|
|
return res.status;
|
|
}
|
|
|
|
async function fetchMovieApi(url, site, baseRelease, options) {
|
|
const referer = options.parameters?.referer || `${site.parameters?.networkReferer ? site.parent.url : site.url}/en/movies`;
|
|
const { apiUrl } = await fetchApiCredentials(referer, site);
|
|
|
|
const entryId = (baseRelease?.path || new URL(url).pathname).match(/\/(\d{2,})(\/|$)/)?.[1];
|
|
|
|
const res = await http.post(apiUrl, {
|
|
requests: [
|
|
{
|
|
indexName: 'all_movies',
|
|
params: `query=&page=0&facets=[]&tagFilters=&facetFilters=[["movie_id:${entryId}"]]`,
|
|
},
|
|
{
|
|
indexName: 'all_scenes_latest_desc',
|
|
params: `query=&page=0&facets=[]&tagFilters=&facetFilters=[["movie_id:${entryId}"]]`,
|
|
},
|
|
/*
|
|
{
|
|
indexName: 'all_movies',
|
|
params: 'query=&page=0&hitsPerPage=1&attributesToRetrieve=[]&attributesToHighlight=[]&attributesToSnippet=[]&tagFilters=&analytics=false&clickAnalytics=false&facets=clip_id',
|
|
},
|
|
*/
|
|
],
|
|
}, {
|
|
headers: {
|
|
Referer: referer,
|
|
},
|
|
}, {
|
|
encodeJSON: true,
|
|
});
|
|
|
|
if (res.status === 200 && res.body.results?.[0]?.hits.length > 0) {
|
|
return scrapeReleaseApi(res.body.results[0].hits[0], site, options, res.body.results[1]?.hits);
|
|
}
|
|
|
|
if (res.status === 200) {
|
|
return null;
|
|
}
|
|
|
|
return res.status;
|
|
}
|
|
|
|
function getLatestUrl(site, page) {
|
|
if (site.parameters?.latest) {
|
|
if (/^http/.test(site.parameters.latest)) {
|
|
return /%d/.test(site.parameters.latest)
|
|
? util.format(site.parameters.latest, page)
|
|
: `${site.parameters.latest}${page}`;
|
|
}
|
|
|
|
return /%d/.test(site.parameters.latest)
|
|
? util.format(`${site.url}${site.parameters.latest}`, page)
|
|
: `${site.url}${site.parameters.latest}${page}`;
|
|
}
|
|
|
|
return `${site.url}/en/videos/AllCategories/0/${page}`;
|
|
}
|
|
|
|
function getUpcomingUrl(site) {
|
|
if (site.parameters?.upcoming) {
|
|
return /^http/.test(site.parameters.upcoming)
|
|
? `${site.parameters.upcoming}`
|
|
: `${site.url}${site.parameters.upcoming}`;
|
|
}
|
|
|
|
return `${site.url}/en/videos/AllCategories/0/1/upcoming`;
|
|
}
|
|
|
|
async function fetchLatest(site, page = 1) {
|
|
const url = getLatestUrl(site, page);
|
|
const res = await qu.getAll(url, 'li[data-itemtype=scene], div[data-itemtype*=scene]');
|
|
|
|
if (res.ok) {
|
|
return scrapeAll(res.items, site);
|
|
}
|
|
|
|
return res.status;
|
|
}
|
|
|
|
async function fetchUpcoming(site) {
|
|
const url = getUpcomingUrl(site);
|
|
const res = await qu.getAll(url, 'li[data-itemtype=scene], div[data-itemtype*=scene]');
|
|
|
|
if (res.ok) {
|
|
return scrapeAll(res.items, site, null, false);
|
|
}
|
|
|
|
return res.status;
|
|
}
|
|
|
|
function getDeepUrl(url, site, baseRelease, mobile) {
|
|
const filter = new Set(['en', 'video', 'scene', site.slug, site.parent.slug]);
|
|
const pathname = baseRelease?.path || new URL(url).pathname
|
|
.split('/')
|
|
.filter((component) => !filter.has(component))
|
|
.join('/'); // reduce to scene ID and title slug
|
|
|
|
const sceneId = baseRelease?.entryId || pathname.match(/\/(\d+)\//)?.[1];
|
|
|
|
if (mobile && /%d/.test(mobile)) {
|
|
return util.format(mobile, sceneId);
|
|
}
|
|
|
|
if (mobile && sceneId) {
|
|
return `${mobile}${pathname}`;
|
|
}
|
|
|
|
if (site.parameters?.deep) {
|
|
return `${site.parameters.deep}${pathname}`;
|
|
}
|
|
|
|
return url;
|
|
}
|
|
|
|
async function fetchScene(url, site, baseRelease, options) {
|
|
if (site.parameters?.deep === false) {
|
|
return baseRelease;
|
|
}
|
|
|
|
const deepUrl = getDeepUrl(url, site, baseRelease);
|
|
const mobileUrl = options.includePhotos && getDeepUrl(url, site, baseRelease, site.parameters?.mobile || site.parent?.parameters?.mobile);
|
|
|
|
if (deepUrl) {
|
|
const [res, mobileRes] = await Promise.all([
|
|
qu.get(deepUrl),
|
|
mobileUrl && qu.get(mobileUrl, null, {
|
|
headers: {
|
|
// don't redirect to main site
|
|
'user-agent': 'Mozilla/5.0 (Linux; Android 6.0; Nexus 5 Build/MRA58N) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/80.0.3987.122 Mobile Safari/537.36',
|
|
},
|
|
}),
|
|
]);
|
|
|
|
if (res.status === 200) {
|
|
const mobileItem = mobileRes?.status === 200 ? mobileRes.item : null;
|
|
const scene = await scrapeScene(res.item, url, site, baseRelease, mobileItem, options);
|
|
|
|
return { ...scene, deepUrl };
|
|
}
|
|
}
|
|
|
|
return null;
|
|
}
|
|
|
|
async function fetchActorScenes(actorName, apiUrl, siteSlug) {
|
|
const res = await http.post(apiUrl, {
|
|
requests: [
|
|
{
|
|
indexName: 'all_scenes',
|
|
params: `query=&filters=sitename:${siteSlug}&hitsPerPage=36&maxValuesPerFacet=100&page=0&facetFilters=[["lesbian:"],["bisex:"],["shemale:"],["actors.name:${actorName}"]]`,
|
|
},
|
|
],
|
|
}, {
|
|
headers: {
|
|
Referer: `https://www.${siteSlug}.com/en/videos`,
|
|
},
|
|
}, {
|
|
encodeJSON: true,
|
|
});
|
|
|
|
if (res.status === 200 && res.body.results[0].hits.length > 0) {
|
|
return res.body.results[0].hits;
|
|
}
|
|
|
|
return [];
|
|
}
|
|
|
|
async function fetchProfile({ name: actorName }, context, include, altSearchUrl, getActorReleasesUrl) {
|
|
const siteSlug = context.entity.slug || context.site?.slug || context.network?.slug;
|
|
|
|
const actorSlug = actorName.toLowerCase().replace(/\s+/, '+');
|
|
const searchUrl = altSearchUrl
|
|
? `https://www.${siteSlug}.com/en/search/${actorSlug}/1/actor`
|
|
: `https://www.${siteSlug}.com/en/search/${siteSlug}/actor/${actorSlug}`;
|
|
const searchRes = await http.get(searchUrl);
|
|
|
|
if (searchRes.status !== 200) {
|
|
return null;
|
|
}
|
|
|
|
const actorUrl = scrapeActorSearch(searchRes.body.toString(), searchUrl, actorName);
|
|
|
|
if (actorUrl) {
|
|
const url = `https://${siteSlug}.com${actorUrl}`;
|
|
const actorRes = await qu.get(url);
|
|
|
|
if (actorRes.status !== 200) {
|
|
return null;
|
|
}
|
|
|
|
return scrapeProfile(actorRes.item, url, actorName, siteSlug, getActorReleasesUrl, include.scenes, context);
|
|
}
|
|
|
|
return null;
|
|
}
|
|
|
|
async function fetchApiProfile({ name: actorName }, context, include) {
|
|
const siteSlug = context.entity.slug || context.site?.slug || context.network?.slug;
|
|
|
|
const actorSlug = encodeURI(actorName);
|
|
const referer = `https://www.${siteSlug}.com/en/search`;
|
|
|
|
const { apiUrl } = await fetchApiCredentials(referer);
|
|
|
|
const res = await http.post(apiUrl, {
|
|
requests: [
|
|
{
|
|
indexName: 'all_actors',
|
|
params: `query=${actorSlug}`,
|
|
},
|
|
],
|
|
}, {
|
|
headers: {
|
|
Referer: referer,
|
|
},
|
|
}, {
|
|
encodeJSON: true,
|
|
});
|
|
|
|
if (res.status === 200 && res.body.results[0].hits.length > 0) {
|
|
const actorData = res.body.results[0].hits.find((actor) => slugify(actor.name) === slugify(actorName));
|
|
|
|
if (actorData) {
|
|
const actorScenes = include.releases && await fetchActorScenes(actorData.name, apiUrl, siteSlug);
|
|
|
|
return scrapeApiProfile(actorData, actorScenes, siteSlug);
|
|
}
|
|
}
|
|
|
|
return null;
|
|
}
|
|
|
|
module.exports = {
|
|
fetchApiLatest: fetchLatestApi,
|
|
fetchApiProfile,
|
|
fetchApiUpcoming: fetchUpcomingApi,
|
|
fetchLatest,
|
|
fetchLatestApi,
|
|
fetchProfile,
|
|
fetchScene,
|
|
fetchSceneApi,
|
|
fetchUpcoming,
|
|
fetchUpcomingApi,
|
|
api: {
|
|
fetchLatest: fetchLatestApi,
|
|
fetchUpcoming: fetchUpcomingApi,
|
|
fetchProfile: fetchApiProfile,
|
|
// fetchScene,
|
|
fetchScene: fetchSceneApi,
|
|
// scrapeMovie,
|
|
fetchMovie: fetchMovieApi,
|
|
},
|
|
getPhotos,
|
|
scrapeApiProfile,
|
|
scrapeApiReleases,
|
|
scrapeProfile,
|
|
scrapeAll,
|
|
scrapeMovie,
|
|
scrapeScene,
|
|
};
|