traxxx/src/scrapers/mariskax.js

'use strict';

const unprint = require('unprint');

const slugify = require('../utils/slugify');
const { convert } = require('../utils/convert');

function scrapeScene(data, channel) {
	const release = {};

	release.entryId = data.id;
	release.url = `${channel.origin}/scenes/${data.slug}`;

	release.title = data.title;
	release.description = data.description;

	release.date = unprint.extractDate(data.publish_date, 'YYYY/MM/DD hh:mm:ss');
	release.duration = unprint.extractDuration(data.videos_duration);

	release.actors = (data.models_thumbs || data.models_slugs)?.map((actor) => ({
		name: actor.name,
		url: actor.slug && `${channel.origin}/models/${actor.slug}`,
		avatar: actor.thumb,
	})) || data.models;

	release.tags = data.tags;

	release.poster = data.thumb || data.trailer_screencap;

	const posterPath = release.poster && new URL(release.poster).pathname.replace('//', '/');

	release.photos = data.extra_thumbnails?.filter((src) => !src.includes(posterPath));
	release.caps = data.thumbs;

	release.teaser = data.special_thumbnails;
	release.trailer = data.trailer_url;

	release.photoCount = data.photos_duration;
	release.channel = data.site?.toLowerCase();

	release.qualities = data.videos && Array.from(new Set(Object.values(data.videos).map((video) => video.height))).filter(Boolean);

	return release;
}

async function fetchLatest(channel, page) {
	const res = await unprint.get(`https://tour.mariskax.com/scenes?page=${page}`, {
		timeout: 30000, // slow site
	});

	if (res.ok) {
		const data = res.context.query.json('#__NEXT_DATA__')?.props.pageProps.contents.data;

		if (data) {
			return data.map((scene) => scrapeScene(scene, channel));
		}

		return null;
	}

	return res.status;
}

async function fetchScene(url, entity, baseRelease) {
	if (baseRelease.entryId) {
		// same as as deep data
		return baseRelease;
	}

	const res = await unprint.get(url, {
		timeout: 30000, // slow site
	});

	if (res.ok) {
		const data = res.context.query.json('#__NEXT_DATA__')?.props.pageProps.content;

		if (data) {
			return scrapeScene(data, entity);
		}

		return null;
	}

	return res.status;
}

function scrapeProfile(data) {
	const profile = {};

	const bio = Object.fromEntries(Object.entries(data).map(([key, value]) => [slugify(key, '_'), value]));

	profile.url = `https://tour.mariskax.com/models/${data.slug}`;
	profile.entryId = data.id;

	profile.gender = bio.gender;

	profile.dateOfBirth = bio.birthdate;
	profile.age = bio.age;
	profile.placeOfBirth = bio.born;

	profile.measurements = bio.measurements;
	profile.height = convert(bio.height, 'cm');
	profile.weight = convert(bio.weight, 'lb', 'kg');

	profile.hairColor = bio.hair_color;
	profile.eyes = bio.eye_color;

	profile.avatar = data.thumb;
	profile.socials = [bio.x && `https://x.com/${bio.x.replace('@', '')}`].filter(Boolean);

	return profile;
}

async function getActorUrl(actor, entity) {
	if (actor.url) {
		return { url: actor.url };
	}

	const res = await unprint.post(`${entity.origin}/api/search/${actor.name}`);

	if (res.ok) {
		const model = res.data.models.find((result) => slugify(result.name) === actor.slug);

		if (model?.slug) {
			return {
				url: `${entity.origin}/models/${model.slug}`,
				model,
			};
		}
	}

	return null;
}

async function fetchProfile(actor, entity) {
	const { url, model } = await getActorUrl(actor, entity);

	if (model) {
		// search data already contains everything except for age, but DOB is included
		return scrapeProfile(model);
	}

	if (url) {
		const res = await unprint.get(url, {
			parser: {
				runScripts: 'dangerously',
			},
		});

		if (res.ok) {
			const data = res.context.query.json('#__NEXT_DATA__')?.props.pageProps.model;

			if (data) {
				return scrapeProfile(data);
			}

			return null;
		}

		return res.status;
	}

	return null;
}

module.exports = {
	fetchLatest,
	scrapeScene,
	fetchScene,
	fetchProfile,
};