traxxx/src/scrapers/template.js

92 lines
2.0 KiB
JavaScript
Executable File

'use strict';
const unprint = require('unprint');
const slugify = require('../utils/slugify');
function scrapeAll(scenes) {
return scenes.map(({ query }) => {
const release = {};
release.url = query.url('.title a');
release.entryId = new URL(release.url).pathname.match(/\/scene\/(\d+)/)[1];
release.title = query.content('.title a');
release.date = query.date('.date', 'MMM DD, YYYY');
release.duration = query.duration('.duration');
release.actors = query.all('.models a.model').map((actorEl) => ({
name: unprint.query.content(actorEl),
url: unprint.query.url(actorEl, null),
}));
release.poster = query.img('img.poster');
release.teaser = query.video('.teaser video');
console.log(release);
return release;
});
}
function scrapeScene({ query }, { url }) {
const release = {};
release.entryId = new URL(url).pathname.match(/\/scene\/(\d+)/)[1];
release.title = query.content('h3.title');
release.description = query.content('p.description');
release.date = query.date('.date', 'MMMM D, YYYY');
release.duration = query.duration('.duration');
[release.poster, ...release.photos] = query.imgs('.preview-thumb');
release.trailer = query.video('.trailer video');
console.log(release);
return release;
}
function scrapeProfile({ query }) {
const profile = {};
profile.description = query.content('.bio-text');
profile.birthPlace = query.content('.birth-place span');
profile.avatar = query.img('.actor-photo img');
console.log(profile);
return profile;
}
async function fetchLatest(channel, page = 1) {
const url = `${channel.url}/${page}`;
const res = await unprint.get(url, { selectAll: '.scene' });
if (res.ok) {
return scrapeAll(res.context, channel);
}
return res.status;
}
async function fetchProfile({ name: actorName }, entity) {
const url = `${entity.url}/actors/${slugify(actorName, '_')}`;
const res = await unprint.get(url);
if (res.ok) {
return scrapeProfile(res.context, entity);
}
return res.status;
}
module.exports = {
fetchLatest,
fetchProfile,
scrapeScene,
};