traxxx/src/scrapers/aziani.js

210 lines
6.0 KiB
JavaScript
Executable File

'use strict';
const unprint = require('unprint');
const { decode } = require('html-entities');
const slugify = require('../utils/slugify');
const { feetInchesToCm, femaleFeetUsToEu } = require('../utils/convert');
function scrapeScene(data, channel, parameters) {
const release = {};
release.entryId = data.cms_set_id;
release.url = `${parameters.scene || channel.url}/video/${data.cms_set_id}`;
release.title = data.name;
release.description = data.description;
release.date = unprint.extractDate(data.added_nice, 'YYYY-MM-DD');
release.duration = Number(data.lengths.total);
release.actors = data.data_types.find((dataType) => dataType.data_type === 'Models' || dataType.data_type === 'Talent')?.data_values.map((actor) => ({
name: actor.name,
url: `${channel.url}/model/${actor.cms_data_value_id}?models=${encodeURI(actor.name)}`, // slug does not work unless it's also the ID
}));
release.directors = data.data_types.find((dataType) => dataType.data_type === 'Videographers')?.data_values.map((director) => ({
name: director.name,
// url: `${channel.url}/model/${director.slug}?models=${director.name}`,
}));
release.tags = data.data_types
.filter((dataType) => dataType.data_type === 'Tags' || dataType.data_type === 'Category')
.flatMap((tags) => tags.data_values.map((tag) => tag.name));
const poster = data.preview_formatted.thumb;
const teaser = data.preview_formatted.clip?.[0];
const trailer = data.preview_formatted.trailer?.formats?.[0]?.content?.[0];
if (poster) {
release.poster = Object.keys(poster)
.filter((key) => poster[key].length > 0)
.toSorted((keyA, keyB) => keyB.split('-')[0] - keyA.split('-')[0])
.map((key) => unprint.prefixUrl(`${poster[key][0].fileuri}?${poster[key][0].signature}`, parameters.cdn));
}
release.caps = data.content
?.filter((item) => item.content_type === 'image' && item.content?.[0])
.map((item) => [
item.content[0],
...(item.thumb
? Object.keys(item.thumb).toSorted((keyA, keyB) => keyB.split('-')[0] - keyA.split('-')[0]).map((key) => item.thumb[key][0]).filter(Boolean)
: []),
])
.map((sources) => sources.map((source) => ({
src: unprint.prefixUrl(`${source.fileuri}?${source.signature}`, parameters.cdn),
expectType: {
'application/octet-stream': 'image/jpeg',
},
})));
if (teaser && teaser.fileuri !== trailer?.fileuri) {
release.teaser = unprint.prefixUrl(`${teaser.fileuri}?${teaser.signature}`, parameters.cdn);
}
if (trailer) {
release.trailer = unprint.prefixUrl(`${trailer.fileuri}?${trailer.signature}`, parameters.cdn);
}
release.channel = slugify(data.data_types.find((dataType) => dataType.data_type === 'Series')?.data_values[0]?.name, '');
return release;
}
async function fetchLatest(channel, page = 1, { parameters }) {
const query = new URLSearchParams({
cms_area_id: parameters.areaId,
cms_block_id: parameters.blockId,
count: 100,
start: (page - 1) * 100,
orderby: 'published_desc',
content_type: 'video',
// unsure what this does
status: 'enabled',
cms_set_ids: undefined,
data_types: 1,
content_count: 1,
data_type_search: parameters.seriesId && JSON.stringify({ 7: parameters.seriesId.toString() }), // doesn't seem relevant
}).toString();
const url = `https://azianistudios.com/tour_api.php/content/sets?${query}`;
const res = await unprint.get(url, {
headers: {
Referer: channel.url,
'x-nats-cms-area-id': parameters.areaId,
},
});
if (res.ok && res.data.success) {
return res.data.sets.map((data) => scrapeScene(data, channel, parameters));
}
return res.status;
}
async function fetchScene(url, entity, _baseRelease, { parameters }) {
const entryId = new URL(url).pathname.match(/\/video\/(\w+)/)[1];
const query = new URLSearchParams({
cms_set_ids: entryId,
cms_area_id: parameters.areaId,
cms_block_id: parameters.blockId,
content: 1,
orderby: 'published_desc',
content_type: 'video',
// unsure what this does
data_types: 1,
content_count: 1,
}).toString();
const apiUrl = `https://azianistudios.com/tour_api.php/content/sets?${query}`;
const res = await unprint.get(apiUrl, {
headers: {
Referer: entity.url,
'x-nats-cms-area-id': parameters.areaId,
},
});
if (res.ok && res.data.success) {
return scrapeScene(res.data.sets[0], entity, parameters);
}
return res.status;
}
function scrapeProfile(data, entity, parameters) {
const profile = {};
const bio = Object.fromEntries(Object.values(data.data_detail_values).map((detail) => [
slugify(detail.name, '_'),
decode(detail.value || detail.content_formatted || detail.content),
]));
profile.url = `${entity.url}/model/${data.cms_data_value_id}`;
profile.entryId = data.cms_data_value_id;
profile.description = data.description;
profile.gender = bio.gender?.toLowerCase();
profile.age = bio.age;
profile.dateOfBirth = unprint.extractDate(`${bio.born} 0`, 'MMMM Do YYYY', { match: /\w+ \d+\w{2} \d{1,4}/ });
profile.measurements = bio.measurements;
profile.height = feetInchesToCm(bio.height);
profile.foot = femaleFeetUsToEu(bio.foot_size);
profile.hairColor = bio.hair_color;
profile.eyeColor = bio.eye_color;
const avatar = bio.thumbnail?.image[0];
if (avatar) {
profile.avatar = {
src: `${parameters.cdn}${avatar.fileuri}?${avatar.signature}`,
expectType: {
'application/octet-stream': 'image/jpeg',
},
};
}
return profile;
}
async function fetchProfile({ url }, { entity, parameters }) {
if (!url) {
// no easy search option
return null;
}
const actorId = new URL(url).pathname.match(/model\/(\d+)/)[1];
const query = new URLSearchParams({
cms_data_value_ids: actorId,
cms_block_id: entity.parameters.modelBlockId,
cms_data_type_id: 4,
}).toString();
const apiUrl = `https://azianistudios.com/tour_api.php/content/data-values?${query}`;
const res = await unprint.get(apiUrl, {
headers: {
Referer: entity.url,
'x-nats-cms-area-id': entity.parameters.areaId,
},
});
if (res.ok && res.data.success) {
return scrapeProfile(res.data.data_values[0], entity, parameters);
}
return res.status;
}
module.exports = {
fetchLatest,
fetchProfile,
fetchScene,
};