DebaucheryLibrarian
/
traxxx
2
1
Fork 1
Code Issues 37 Pull Requests 1 Releases Wiki Activity

Using date and title for Nubiles entry ID.

This commit is contained in:
DebaucheryLibrarian 2021-01-11 23:31:33 +01:00
parent e65828c729
commit ad73c11cb4
1 changed files with 41 additions and 38 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

79
src/scrapers/nubiles.js
View File

@ -1,6 +1,6 @@
'use strict'; 'use strict';
const { get, geta, ctxa } = require('../utils/q'); const qu = require('../utils/qu');
const slugify = require('../utils/slugify'); const slugify = require('../utils/slugify');
const { heightToCm } = require('../utils/convert'); const { heightToCm } = require('../utils/convert');
@ -10,112 +10,115 @@ const slugUrlMap = {
}; };
async function getPhotos(albumUrl) { async function getPhotos(albumUrl) {
const res = await geta(albumUrl, '.photo-thumb'); const res = await qu.getAll(albumUrl, '.photo-thumb');
return res.ok return res.ok
? res.items.map(({ q }) => q('source').srcset) ? res.items.map(({ query }) => query.q('source').srcset)
: []; : [];
} }
function scrapeAll(scenes, site, origin) { function scrapeAll(scenes, site, origin) {
return scenes.map(({ qu }) => { return scenes.map(({ query }) => {
const release = {}; const release = {};
release.title = qu.q('.title a', true); release.title = query.q('.title a', true);
const url = qu.url('.title a').split('?')[0]; const url = query.url('.title a').split('?')[0];
const channelUrl = qu.url('.site-link'); const channelUrl = query.url('.site-link');
if (/^http/.test(url)) { if (/^http/.test(url)) {
const { pathname } = new URL(url); const { pathname } = new URL(url);
release.entryId = pathname.split('/')[3]; // release.entryId = pathname.split('/')[3];
if (channelUrl) release.url = `${channelUrl}${pathname}`; if (channelUrl) release.url = `${channelUrl}${pathname}`;
else release.url = url; else release.url = url;
} else if (!/\/join/.test(url)) { } else if (!/\/join/.test(url)) {
release.entryId = url.split('/')[3]; // release.entryId = url.split('/')[3];
if (channelUrl) release.url = `${channelUrl}${url}`; if (channelUrl) release.url = `${channelUrl}${url}`;
else if (site?.url) release.url = `${site.url}${url}`; else if (site?.url) release.url = `${site.url}${url}`;
else if (origin) release.url = `${origin}${url}`; else if (origin) release.url = `${origin}${url}`;
} else { } else {
release.entryId = qu.q('a img', 'tube_tour_thumb_id'); // release.entryId = qu.q('a img', 'tube_tour_thumb_id');
} }
release.date = qu.date('.date', 'MMM D, YYYY'); release.date = query.date('.date', 'MMM D, YYYY');
release.actors = qu.all('.models a.model', true); release.actors = query.all('.models a.model', true);
const poster = qu.sourceSet('img', 'data-srcset')?.[0]; // no reliable entry ID between upcoming and released scenes
release.entryId = `${qu.formatDate(release.date, 'YYYY-MM-DD')}-${slugify(release.title)}`;
const poster = query.sourceSet('img', 'data-srcset')?.[0];
release.poster = poster && [ release.poster = poster && [
poster.replace('_640', '_1280'), poster.replace('_640', '_1280'),
poster, poster,
]; ];
release.stars = Number(qu.q('.rating', true)); release.stars = query.number('.rating');
release.likes = Number(qu.q('.likes', true)); release.likes = query.number('.likes');
return release; return release;
}); });
} }
async function scrapeScene({ qu }, url, site) { async function scrapeScene({ query }, url, site) {
const release = {}; const release = {};
const { origin, pathname } = new URL(url); const { origin, pathname } = new URL(url);
release.url = `${origin}${pathname}`; release.url = `${origin}${pathname}`;
release.entryId = new URL(url).pathname.split('/')[3]; release.entryId = new URL(url).pathname.split('/')[3];
release.title = qu.q('.content-pane-title h2', true); release.title = query.q('.content-pane-title h2', true);
release.description = qu.q('.content-pane-column div', true); release.description = query.q('.content-pane-column div', true);
release.date = qu.q('.date', 'MMM D, YYYY'); release.date = query.q('.date', 'MMM D, YYYY');
release.actors = qu.all('.content-pane-performers .model', true); release.actors = query.all('.content-pane-performers .model', true);
release.tags = qu.all('.categories a', true); release.tags = query.all('.categories a', true);
release.poster = qu.poster() || qu.img('.fake-video-player img'); release.poster = query.poster() || query.img('.fake-video-player img');
release.trailer = qu.all('source').map(source => ({ release.trailer = query.all('source').map(source => ({
src: source.src, src: source.src,
quality: Number(source.getAttribute('res')), quality: Number(source.getAttribute('res')),
})); }));
release.stars = Number(qu.q('.score', true)); release.stars = Number(query.q('.score', true));
release.likes = Number(qu.q('#likecount', true)); release.likes = Number(query.q('#likecount', true));
const albumLink = qu.url('.content-pane-related-links a[href*="gallery"]'); const albumLink = query.url('.content-pane-related-links a[href*="gallery"]');
if (albumLink) release.photos = await getPhotos(`${site.url}${albumLink}`); if (albumLink) release.photos = await getPhotos(`${site.url}${albumLink}`);
return release; return release;
} }
function scrapeProfile({ qu }, _actorName, origin) { function scrapeProfile({ query }, _actorName, origin) {
const profile = {}; const profile = {};
const keys = qu.all('.model-profile h5', true); const keys = query.all('.model-profile h5', true);
const values = qu.all('.model-profile h5 + p', true); const values = query.all('.model-profile h5 + p', true);
const bio = keys.reduce((acc, key, index) => ({ ...acc, [slugify(key, '_')]: values[index] }), {}); const bio = keys.reduce((acc, key, index) => ({ ...acc, [slugify(key, '_')]: values[index] }), {});
profile.age = Number(bio.age); profile.age = Number(bio.age);
profile.description = qu.q('.model-bio', true); profile.description = query.q('.model-bio', true);
profile.residencePlace = bio.location; profile.residencePlace = bio.location;
profile.height = heightToCm(bio.height); profile.height = heightToCm(bio.height);
[profile.bust, profile.waist, profile.hip] = bio.figure.split('-').map(v => Number(v) || v); [profile.bust, profile.waist, profile.hip] = bio.figure.split('-').map(v => Number(v) || v);
profile.avatar = qu.img('.model-profile img'); profile.avatar = query.img('.model-profile img');
const releases = qu.all('.content-grid-item').filter(el => /video\//.test(qu.url(el, '.img-wrapper a'))); // filter out photos const releases = query.all('.content-grid-item').filter(el => /video\//.test(query.url(el, '.img-wrapper a'))); // filter out photos
profile.releases = scrapeAll(ctxa(releases), null, origin); profile.releases = scrapeAll(query.initAll(releases), null, origin);
return profile; return profile;
} }
async function fetchLatest(site, page = 1) { async function fetchLatest(site, page = 1) {
const url = `${site.url}/video/gallery/${(page - 1) * 12}`; const url = `${site.url}/video/gallery/${(page - 1) * 12}`;
const res = await geta(url, '.content-grid-item'); const res = await qu.getAll(url, '.content-grid-item');
return res.ok ? scrapeAll(res.items, site) : res.status; return res.ok ? scrapeAll(res.items, site) : res.status;
} }
@ -123,7 +126,7 @@ async function fetchLatest(site, page = 1) {
async function fetchUpcoming(site) { async function fetchUpcoming(site) {
if (site.parameters?.upcoming) { if (site.parameters?.upcoming) {
const url = `${site.url}/video/upcoming`; const url = `${site.url}/video/upcoming`;
const res = await geta(url, '.content-grid-item'); const res = await qu.getAll(url, '.content-grid-item');
return res.ok ? scrapeAll(res.items, site) : res.status; return res.ok ? scrapeAll(res.items, site) : res.status;
} }
@ -132,7 +135,7 @@ async function fetchUpcoming(site) {
} }
async function fetchScene(url, site) { async function fetchScene(url, site) {
const res = await get(url); const res = await qu.get(url);
return res.ok ? scrapeScene(res.item, url, site) : res.status; return res.ok ? scrapeScene(res.item, url, site) : res.status;
} }
@ -142,7 +145,7 @@ async function fetchProfile({ name: actorName }, { site }) {
const origin = slugUrlMap[site.slug] || site.url; const origin = slugUrlMap[site.slug] || site.url;
const url = `${origin}/model/alpha/${firstLetter}`; const url = `${origin}/model/alpha/${firstLetter}`;
const resModels = await get(url); const resModels = await qu.get(url);
if (!resModels.ok) return resModels.status; if (!resModels.ok) return resModels.status;
@ -150,7 +153,7 @@ async function fetchProfile({ name: actorName }, { site }) {
if (modelPath) { if (modelPath) {
const modelUrl = `${origin}${modelPath}`; const modelUrl = `${origin}${modelPath}`;
const resModel = await get(modelUrl); const resModel = await qu.get(modelUrl);
return resModel.ok ? scrapeProfile(resModel.item, actorName, origin) : resModel.status; return resModel.ok ? scrapeProfile(resModel.item, actorName, origin) : resModel.status;
} }