Added fixed actor age. Added male profiles to Littlr Caprice Dreams scraper. Added various tag photos.

This commit is contained in:
DebaucheryLibrarian
2020-11-29 03:59:47 +01:00
parent 71c884fe48
commit 9a61d2305c
71 changed files with 123 additions and 33 deletions

View File

@@ -4,7 +4,11 @@ const qu = require('../utils/qu');
const slugify = require('../utils/slugify');
function matchChannel(release, channel) {
const series = channel.children || channel.parent.children;
const series = channel.children || channel.parent?.children;
if (!series) {
return null;
}
const serieNames = series.reduce((acc, serie) => ({
...acc,
@@ -26,7 +30,7 @@ function matchChannel(release, channel) {
if (serie) {
return {
slug: serie.slug,
channel: serie.slug,
title: release.title.replace(new RegExp(`(${serieName}|${serie.name}|${serie.slug})\\s*[-:/]+\\s*`, 'ig'), ''),
};
}
@@ -109,31 +113,34 @@ async function scrapeScene({ query }, url, channel, include) {
release.photos = await fetchPhotos(query.url('.vid_buttons a[href*="project/"]'));
}
release.trailer = {
src: query.video(),
type: query.video('source', 'type'),
quality: query.video('source', 'data-res'),
referer: url,
};
return {
...release,
...matchChannel(release, channel),
};
}
function scrapeProfile({ query }, url) {
const profile = {};
function scrapeProfile({ query, el }, { url, gender }, baseActor, entity) {
const profile = { url, gender };
const bio = query.cnts('div p').reduce((acc, item) => {
const [key, value] = item.split(/\s*:\s*/);
profile.age = query.number('div:nth-child(2) > p');
profile.birthPlace = query.cnt('div:nth-child(3) > p')?.match(/nationality[\s:]+(\w+)/i)?.[1];
return {
...acc,
[slugify(key, '_')]: value.trim(),
};
}, {});
profile.description = query.cnt('div:nth-child(4) > p');
profile.avatar = {
src: query.img('.model-page'),
referer: url,
};
console.log(bio);
console.log(profile);
profile.scenes = scrapeAll(qu.initAll(el, '.project_category-videos'), entity);
return profile;
}
@@ -160,12 +167,16 @@ async function fetchScene(url, channel, baseRelease, include) {
return res.status;
}
async function getActorUrl(baseActor) {
async function getActorUrl(baseActor, gender = 'female') {
if (baseActor.url) {
return baseActor.url;
}
const overviewRes = await qu.getAll('https://www.littlecaprice-dreams.com/pornstars', '.models');
const overviewUrl = gender === 'female'
? 'https://www.littlecaprice-dreams.com/pornstars/'
: 'https://www.littlecaprice-dreams.com/male-models-pornstars/';
const overviewRes = await qu.getAll(overviewUrl, '.models');
if (!overviewRes.ok) {
return overviewRes.status;
@@ -174,23 +185,36 @@ async function getActorUrl(baseActor) {
const actorItem = overviewRes.items.find(({ query }) => slugify(query.q('img', 'title')) === baseActor.slug);
if (!actorItem) {
if (gender === 'female') {
return getActorUrl(baseActor, 'male');
}
return null;
}
return actorItem.query.url('a');
const actorUrl = actorItem.query.url('a');
if (actorUrl) {
return {
url: actorUrl,
gender,
};
}
return null;
}
async function fetchProfile(baseActor, entity) {
async function fetchProfile(baseActor, { entity }) {
const actorUrl = await getActorUrl(baseActor);
if (!actorUrl) {
return null;
}
const actorRes = await qu.get(actorUrl, '#main-content');
const actorRes = await qu.get(actorUrl.url, '#main-content');
if (actorRes.ok) {
return scrapeProfile(actorRes.item, actorUrl, entity);
return scrapeProfile(actorRes.item, actorUrl, baseActor, entity);
}
return actorRes.status;