Refactored Hitzefrei to use unprint.
This commit is contained in:
parent
ba2fd254ff
commit
d380c64f7c
|
|
@ -426,7 +426,7 @@ const networks = [
|
|||
{
|
||||
slug: 'hitzefrei',
|
||||
name: 'Hitzefrei',
|
||||
url: 'http://www.hitzefrei.com',
|
||||
url: 'http://tour.hitzefrei.com',
|
||||
},
|
||||
{
|
||||
slug: 'hussiepass',
|
||||
|
|
|
|||
|
|
@ -1,9 +1,8 @@
|
|||
'use strict';
|
||||
|
||||
const qu = require('../utils/qu');
|
||||
const http = require('../utils/http');
|
||||
const { lbsToKg, feetInchesToCm } = require('../utils/convert');
|
||||
const slugify = require('../utils/slugify');
|
||||
const unprint = require('unprint');
|
||||
|
||||
const { convert } = require('../utils/convert');
|
||||
|
||||
function scrapeAll(scenes) {
|
||||
return scenes.map(({ query }) => {
|
||||
|
|
@ -12,39 +11,50 @@ function scrapeAll(scenes) {
|
|||
release.url = query.url('.content-title a');
|
||||
release.entryId = new URL(release.url).pathname.match(/\/view\/(\d+)/)[1];
|
||||
|
||||
release.title = query.cnt('.content-title a');
|
||||
release.title = query.content('.content-title a');
|
||||
release.date = query.date('.content-date strong', 'DD/MM/YYYY');
|
||||
release.duration = query.dur('.content-date');
|
||||
release.duration = query.duration('.content-date');
|
||||
|
||||
release.actors = query.cnts('.content-models a');
|
||||
release.actors = query.all('.content-models a').map((actorEl) => ({
|
||||
name: unprint.query.content(actorEl),
|
||||
url: unprint.query.url(actorEl, null),
|
||||
}));
|
||||
|
||||
release.poster = query.img('.content-thumbnail img, .large-thumbnail img') || query.poster('.content-thumbnail video, .large-thumbnail video');
|
||||
release.photos = query.imgs('.small-thumbnails img'); // actor page only
|
||||
|
||||
const teaser = query.video('.vid-hover source');
|
||||
release.teaser = { src: teaser };
|
||||
|
||||
release.channel = slugify(query.cnt('.content-site a'), '');
|
||||
release.teaser = query.video('.vid-hover source');
|
||||
|
||||
return release;
|
||||
});
|
||||
}
|
||||
|
||||
async function fetchLatest(channel, page = 1) {
|
||||
const res = await unprint.get(`https://tour.hitzefrei.com/videos?site=${channel.parameters.siteId}&page=${page}`, { selectAll: '.hitem' });
|
||||
|
||||
if (res.ok) {
|
||||
return scrapeAll(res.context, channel);
|
||||
}
|
||||
|
||||
return res.status;
|
||||
}
|
||||
|
||||
function scrapeScene({ query }, url) {
|
||||
const release = {};
|
||||
|
||||
release.entryId = new URL(url).pathname.match(/\/view\/(\d+)/)[1];
|
||||
|
||||
release.title = query.cnt('.content-title');
|
||||
release.description = query.cnt('.content-description p');
|
||||
release.title = query.content('.content-title');
|
||||
release.description = query.content('.content-description p');
|
||||
|
||||
release.date = query.date('.content-metas span:nth-child(4)', 'DD/MM/YYYY');
|
||||
release.duration = query.dur('.content-metas span:nth-child(2)');
|
||||
release.duration = query.duration('.content-metas span:nth-child(2)');
|
||||
release.likes = query.number('.content-metas span:nth-child(6)');
|
||||
|
||||
release.actors = query.all('.model-thumb img').map((el) => ({
|
||||
name: query.q(el, null, 'alt'),
|
||||
avatar: query.img(el, null, 'src'),
|
||||
release.actors = query.all('.model-thumb a').map((actorEl) => ({
|
||||
name: unprint.query.attribute(actorEl, 'img', 'alt'),
|
||||
url: unprint.query.url(actorEl, null),
|
||||
avatar: unprint.query.img(actorEl),
|
||||
}));
|
||||
|
||||
release.poster = query.poster('.content-video video');
|
||||
|
|
@ -54,48 +64,50 @@ function scrapeScene({ query }, url) {
|
|||
photo.replace('/full', '/thumbs'),
|
||||
]);
|
||||
|
||||
const trailer = query.video('.content-video source');
|
||||
release.trailer = { src: trailer };
|
||||
release.trailer = query.video('.content-video source');
|
||||
|
||||
return release;
|
||||
}
|
||||
|
||||
async function fetchScene(url) {
|
||||
const res = await unprint.get(url, { select: '#content-details' });
|
||||
|
||||
if (res.ok) {
|
||||
return scrapeScene(res.context, url);
|
||||
}
|
||||
|
||||
return res.status;
|
||||
}
|
||||
|
||||
async function fetchActorScenes({ query }, accReleases = []) {
|
||||
const releases = scrapeAll(qu.initAll(query.all('.container-large-video-thumb')));
|
||||
const releases = scrapeAll(unprint.initAll(query.all('.container-large-video-thumb')));
|
||||
const nextPage = query.url('.pagination li:nth-last-child(2) a');
|
||||
|
||||
if (nextPage) {
|
||||
const res = await qu.get(nextPage);
|
||||
const res = await unprint.get(nextPage);
|
||||
|
||||
if (res.ok) {
|
||||
return fetchActorScenes(res.item, accReleases.concat(releases));
|
||||
return fetchActorScenes(res.context, accReleases.concat(releases));
|
||||
}
|
||||
}
|
||||
|
||||
return accReleases.concat(releases);
|
||||
}
|
||||
|
||||
async function scrapeProfile({ query }, include) {
|
||||
async function scrapeProfile({ query }, _entity, include) {
|
||||
const profile = {};
|
||||
|
||||
/*
|
||||
const bio = query.all('.model-stats-info div div').reduce((acc, el) => ({
|
||||
...acc,
|
||||
[slugify(query.cnt(el, '.stat-label'), '_')]: query.cnt(el, '.stat-value'),
|
||||
}), {});
|
||||
*/
|
||||
|
||||
profile.dateOfBirth = query.date('.col-birtdate .stat-value, .col-birthdate .stat-value', 'YYYY-MM-DD'); // sic
|
||||
profile.birthPlace = query.cnt('.col-birth .stat-value');
|
||||
profile.birthPlace = query.content('.col-birth .stat-value');
|
||||
|
||||
[profile.bust, profile.waist, profile.hip] = query.cnt('.col-measurements .stat-value').split('-').map(Number);
|
||||
profile.height = feetInchesToCm(query.cnt('.col-height .stat-value'));
|
||||
profile.weight = lbsToKg(query.number('.col-weight .stat-value'));
|
||||
profile.measurements = query.content('.col-measurements .stat-value');
|
||||
profile.height = convert(query.content('.col-height .stat-value'), 'cm');
|
||||
profile.weight = convert(query.number('.col-weight .stat-value'), 'lb', 'kg');
|
||||
|
||||
profile.eyes = query.cnt('.col-eyes .stat-value');
|
||||
profile.hair = query.cnt('.col-hair .stat-value');
|
||||
profile.eyes = query.content('.col-eyes .stat-value');
|
||||
profile.hair = query.content('.col-hair .stat-value');
|
||||
|
||||
profile.description = query.cnt('.model-profile .model-profile');
|
||||
profile.description = query.content('.model-profile .model-profile');
|
||||
profile.avatar = query.img('.model-thumbnail img');
|
||||
|
||||
if (include.releases) {
|
||||
|
|
@ -105,28 +117,12 @@ async function scrapeProfile({ query }, include) {
|
|||
return profile;
|
||||
}
|
||||
|
||||
async function fetchLatest(channel, page = 1) {
|
||||
const res = await qu.getAll(`https://tour.hitzefrei.com/videos?site=${channel.parameters.siteId}&page=${page}`, '.hitem');
|
||||
|
||||
if (res.ok) {
|
||||
return scrapeAll(res.items, channel);
|
||||
async function getActorUrl(baseActor) {
|
||||
if (baseActor.url) {
|
||||
return baseActor.url;
|
||||
}
|
||||
|
||||
return res.status;
|
||||
}
|
||||
|
||||
async function fetchScene(url) {
|
||||
const res = await qu.get(url, '#content-details');
|
||||
|
||||
if (res.ok) {
|
||||
return scrapeScene(res.item, url);
|
||||
}
|
||||
|
||||
return res.status;
|
||||
}
|
||||
|
||||
async function fetchProfile(baseActor, entity, include) {
|
||||
const searchRes = await http.post('https://tour.hitzefrei.com/search-preview', {
|
||||
const res = await unprint.post('https://tour.hitzefrei.com/search-preview', {
|
||||
q: baseActor.name,
|
||||
}, {
|
||||
headers: {
|
||||
|
|
@ -134,23 +130,31 @@ async function fetchProfile(baseActor, entity, include) {
|
|||
},
|
||||
});
|
||||
|
||||
if (searchRes.ok) {
|
||||
const actor = searchRes.body.find((result) => result.type === 'model' && result.title === baseActor.name);
|
||||
if (res.ok) {
|
||||
const actor = res.data.find((result) => result.type === 'model' && result.title === baseActor.name);
|
||||
|
||||
if (actor) {
|
||||
const actorRes = await qu.get(actor.url);
|
||||
|
||||
if (actorRes.ok) {
|
||||
return scrapeProfile(actorRes.item, include);
|
||||
}
|
||||
|
||||
return actorRes.status;
|
||||
return actor.url;
|
||||
}
|
||||
|
||||
return null;
|
||||
}
|
||||
|
||||
return searchRes.status;
|
||||
return null;
|
||||
}
|
||||
|
||||
async function fetchProfile(baseActor, entity, include) {
|
||||
const actorUrl = await getActorUrl(baseActor);
|
||||
|
||||
if (actorUrl) {
|
||||
const res = await unprint.get(actorUrl);
|
||||
|
||||
if (res.ok) {
|
||||
return scrapeProfile(res.context, entity, include);
|
||||
}
|
||||
|
||||
return res.status;
|
||||
}
|
||||
|
||||
return null;
|
||||
}
|
||||
|
||||
module.exports = {
|
||||
|
|
|
|||
|
|
@ -183,6 +183,7 @@ const actors = [
|
|||
{ entity: 'score', name: 'Vanessa Blue', fields: ['avatar', 'gender', 'placeOfResidence', 'ethnicity', 'height', 'weight', 'measurements', 'hairColor', 'dateOfBirth'] },
|
||||
{ entity: 'pierrewoodman', name: 'Abby Lee Brazil', fields: ['avatar', 'nationality'] },
|
||||
{ entity: 'dorcelclub', name: 'Clea Gaultier', fields: ['avatar'] },
|
||||
{ entity: 'hitzefrei', name: 'Jolee Love', fields: ['avatar', 'dateOfBirth', 'birthPlace', 'measurements', 'height', 'weight', 'eyes', 'hair', 'description'] },
|
||||
];
|
||||
|
||||
const actorScrapers = scrapers.actors;
|
||||
|
|
|
|||
Loading…
Reference in New Issue