Updated AnalVids.

config/default.js

@@ -375,7 +375,7 @@ module.exports = {
 		thumbnailQuality: 100,
 		lazySize: 90,
 		lazyQuality: 90,
-		trailerQuality: [480, 540, 360, 720, 960, 1080, 320, 1440, 1600, 1920, 2160, 270, 240, 180],
+		trailerQuality: [540, 720, 480, 360, 960, 1080, 320, 1440, 1600, 1920, 2160, 270, 240, 180],
 		limit: 25, // max number of photos per release
 		attempts: 2,
 		fetchStreams: true,

package-lock.json

@@ -81,7 +81,7 @@
                 "tunnel": "0.0.6",
                 "ua-parser-js": "^1.0.32",
                 "undici": "^4.13.0",
-                "unprint": "^0.10.3",
+                "unprint": "^0.10.7",
                 "url-pattern": "^1.0.3",
                 "v-tooltip": "^2.0.3",
                 "video.js": "^7.11.4",
@@ -17626,9 +17626,9 @@
             }
         },
         "node_modules/unprint": {
-            "version": "0.10.3",
-            "resolved": "https://registry.npmjs.org/unprint/-/unprint-0.10.3.tgz",
-            "integrity": "sha512-ui8BbBo4JmKR++w50rSUFyg8X6l9EAbLRpATxdjxyS7yYevjcGMEt3HT0nrBG2JXDMkLwWZ+WoOaz3qC5stSxQ==",
+            "version": "0.10.7",
+            "resolved": "https://registry.npmjs.org/unprint/-/unprint-0.10.7.tgz",
+            "integrity": "sha512-AuWdWvxkNrFSa8jGE3HIUJ7Pwvp4GR5kJ6eYu5zO8Devizc2D7qifvQv1ksLa9ZenKRjmsgoFVVHOAroGFNE3g==",
             "dependencies": {
                 "axios": "^0.27.2",
                 "bottleneck": "^2.19.5",
@@ -32536,9 +32536,9 @@
             "integrity": "sha1-sr9O6FFKrmFltIF4KdIbLvSZBOw="
         },
         "unprint": {
-            "version": "0.10.3",
-            "resolved": "https://registry.npmjs.org/unprint/-/unprint-0.10.3.tgz",
-            "integrity": "sha512-ui8BbBo4JmKR++w50rSUFyg8X6l9EAbLRpATxdjxyS7yYevjcGMEt3HT0nrBG2JXDMkLwWZ+WoOaz3qC5stSxQ==",
+            "version": "0.10.7",
+            "resolved": "https://registry.npmjs.org/unprint/-/unprint-0.10.7.tgz",
+            "integrity": "sha512-AuWdWvxkNrFSa8jGE3HIUJ7Pwvp4GR5kJ6eYu5zO8Devizc2D7qifvQv1ksLa9ZenKRjmsgoFVVHOAroGFNE3g==",
             "requires": {
                 "axios": "^0.27.2",
                 "bottleneck": "^2.19.5",

package.json

@@ -140,7 +140,7 @@
         "tunnel": "0.0.6",
         "ua-parser-js": "^1.0.32",
         "undici": "^4.13.0",
-        "unprint": "^0.10.3",
+        "unprint": "^0.10.7",
         "url-pattern": "^1.0.3",
         "v-tooltip": "^2.0.3",
         "video.js": "^7.11.4",

src/scrapers/analvids.js

@@ -1,8 +1,6 @@
 'use strict';
 
-const { JSDOM } = require('jsdom');
-const cheerio = require('cheerio');
-const moment = require('moment');
+const unprint = require('unprint');
 
 const http = require('../utils/http');
 const slugify = require('../utils/slugify');
@@ -17,187 +15,132 @@ function extractTitle(originalTitle) {
 	return { shootId, title };
 }
 
-function getPoster(posterElement, sceneId) {
-	const posterStyle = posterElement.attr('style');
+function scrapeAll(scenes, channel) {
+	return scenes.map(({ query }) => {
+		const release = {};
 
-	if (posterStyle) {
-		return posterStyle.slice(posterStyle.indexOf('(') + 1, -1);
+		release.url = query.url('.card-scene__view > a');
+		release.entryId = query.dataset(null, 'content') || new URL(release.url).pathname.match(/watch\/(\d+)/)?.[1];
+
+		release.title = query.content('.card-scene__text');
+		release.shootId = extractTitle(release.title).shootId;
+
+		release.date = query.date('.label--time:nth-child(2)', 'YYYY-MM-DD');
+		release.duration = query.duration('.label--time:first-child');
+
+		release.poster = query.img('.card-scene__view img', { attribute: 'data-src' });
+		const caps = query.json('.card-scene__view > a', { attribute: 'data-casting' })?.map((timestamp) => `${channel.url}/casting/${release.entryId}/${timestamp}`);
+
+		if (caps && release.poster) {
+			release.caps = caps;
+		} else if (caps) {
+			[release.poster, ...release.caps] = caps;
 		}
 
-	const posterRange = posterElement.attr('data-casting');
-	const posterRangeData = posterRange ? JSON.parse(posterRange) : null;
-	const posterTimeRange = posterRangeData[Math.floor(Math.random() * posterRangeData.length)];
+		release.teaser = query.video('.card-scene__view > a', { attribute: 'data-preview' });
 
-	if (!posterTimeRange) {
-		return null;
-	}
-
-	if (typeof posterTimeRange === 'number') {
-		// poster time is already a single time value
-		return `https://analvids.com/casting/${sceneId}/${posterTimeRange}`;
-	}
-
-	const [max, min] = posterTimeRange.split('-');
-	const posterTime = Math.floor(Math.random() * (Number(max) - Number(min) + 1) + Number(min));
-
-	return `https://analvids.com/casting/${sceneId}/${posterTime}`;
-}
-
-function scrapeAll(html) {
-	const $ = cheerio.load(html, { normalizeWhitespace: true });
-	const scenesElements = $('.thumbnails > div').toArray();
-
-	return scenesElements.map((element) => {
-		const sceneLinkElement = $(element).find('.thumbnail-title a');
-		const url = sceneLinkElement.attr('href');
-
-		const originalTitle = sceneLinkElement.text().trim(); // title attribute breaks when they use \\ escaping
-		const { shootId, title } = extractTitle(originalTitle);
-		const entryId = new URL(url).pathname.split('/')[2];
-
-		const date = moment.utc($(element).attr('release'), 'YYYY/MM/DD').toDate();
-
-		const sceneId = $(element).attr('data-content');
-		const posterElement = $(element).find('.thumbnail-avatar');
-
-		const poster = getPoster(posterElement, sceneId);
-
-		return {
-			url,
-			shootId,
-			entryId,
-			title,
-			date,
-			poster,
-		};
+		return release;
 	});
 }
 
-async function scrapeScene(html, url, site, useGallery) {
-	const $ = cheerio.load(html, { normalizeWhitespace: true });
-	const playerObject = $('script:contains("new WatchPage")').html();
-	const playerData = playerObject && playerObject.slice(playerObject.indexOf('{"swf":'), playerObject.lastIndexOf('},') + 1);
-	const data = playerData && JSON.parse(playerData);
+function scrapeScene({ query }, { url }) {
+	const release = {};
 
-	const release = { url };
+	release.entryId = new URL(url).pathname.match(/watch\/(\d+)/)?.[1];
 
-	const originalTitle = $('h1.watchpage-title').text().trim();
-	const { shootId, title } = extractTitle(originalTitle);
+	const featuring = query.content('.watch__title .watch__featuring_models');
 
-	release.shootId = shootId;
-	release.entryId = new URL(url).pathname.split('/')[2];
+	release.title = query.content('.watch__title').replace(featuring, '');
+	release.description = query.content('.text-mob-more');
+	release.shootId = extractTitle(release.title).shootId;
 
-	release.title = title;
-	release.date = moment.utc($('span[title="Release date"] a').text(), 'YYYY-MM-DD').toDate();
+	release.date = query.date('.bi-calendar3', 'YYYY-MM-DD');
+	release.duration = query.duration('.bi-clock');
 
-	const [actorsElement, tagsElement, descriptionElement] = $('.scene-description__row').toArray();
-
-	release.description = $('meta[name="description"]')?.attr('content')?.trim()
-        || (descriptionElement && $(descriptionElement).find('dd').text().trim());
-
-	release.actors = $(actorsElement)
-		.find('a[href*="com/model"]')
-		.map((actorIndex, actorElement) => $(actorElement).text()).toArray();
-
-	release.duration = moment.duration($('span[title="Runtime"]').text().trim()).asSeconds();
-	release.tags = $(tagsElement).find('a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
-
-	const photos = useGallery
-		? $('.gallery a img').map((photoIndex, photoElement) => $(photoElement).attr('src')).toArray()
-		: $('.screenshots img').map((photoIndex, photoElement) => $(photoElement).attr('src')).toArray();
-
-	release.photos = photos.map((source) => {
-		// source without parameters sometimes serves larger preview photo
-		const { origin, pathname } = new URL(source);
-
-		return `${origin}${pathname}`;
-	});
-
-	const posterStyle = $('#player').attr('style');
-	const poster = posterStyle.slice(posterStyle.indexOf('(') + 1, -1);
-
-	release.poster = poster || release.photos.slice(Math.floor(release.photos.length / 3) * -1); // poster unavailable, try last 1/3rd of high res photos as fallback
-
-	if (data) {
-		const qualityMap = {
-			web: 240,
-			vga: 480,
-			hd: 720,
-			'1080p': 1080,
-		};
-
-		release.trailer = data.clip.qualities.map((trailer) => ({
-			src: trailer.src,
-			type: trailer.type,
-			quality: qualityMap[trailer.quality] || trailer.quality,
+	release.actors = query.all('.watch__title a[href*="/model"]').map((el) => ({
+		name: unprint.query.content(el),
+		url: unprint.query.url(el, null),
 	}));
-	}
 
-	const studioName = $('.watchpage-studioname').first().text().trim();
-	release.studio = slugify(studioName, '', { removePunctuation: true });
+	release.tags = query.contents('.genres-list a[href*="/genre"]');
+
+	release.studio = slugify(query.content('.genres-list a[href*="/studios"]'), '', { removePunctuation: true });
+
+	release.poster = query.img('.watch__video video', { attribute: 'data-poster' });
+	release.photoCount = query.number('.bi-eye');
+
+	release.trailer = query.all('.watch__video source').map((el) => ({
+		src: unprint.query.video(el, null, { attribute: 'src' }),
+		quality: unprint.query.number(el, null, { attribute: 'size' }),
+	}));
 
 	return release;
 }
 
-async function scrapeProfile(html, _url, actorName) {
-	const { document } = new JSDOM(html).window;
+function scrapeProfile({ query }, url, channel) {
+	const profile = { url };
 
-	const profile = {
-		name: actorName,
-	};
+	profile.nationality = query.content('.model__info a[href*="/nationality"]');
+	profile.age = query.number('//td[contains(text(), "Age")]/following-sibling::td');
 
-	const avatarEl = document.querySelector('.model--avatar img[src^="http"]');
-	const entries = Array.from(document.querySelectorAll('.model--description tr'), (el) => el.textContent.replace(/\n/g, '').split(':'));
+	profile.avatar = query.img('.model__left img');
 
-	const bio = entries
-		.filter((entry) => entry.length === 2) // ignore entries without ':' (About section, see Blanche Bradburry)
-		.reduce((acc, [key, value]) => ({ ...acc, [key.trim()]: value.trim() }), {});
-
-	profile.birthPlace = bio.Nationality;
-
-	if (bio.Age) profile.age = bio.Age;
-	if (avatarEl) profile.avatar = avatarEl.src;
-
-	profile.releases = scrapeAll(html);
+	profile.scenes = scrapeAll(unprint.initAll(query.all('.card-scene')), channel);
 
 	return profile;
 }
 
-async function fetchLatest(site, page = 1) {
-	const res = await http.get(`${site.url}/new-videos/${page}`);
+async function fetchLatest(channel, page) {
+	const res = await unprint.get(`${channel.url}/new-videos/${page}`, { selectAll: '.card-scene' });
 
-	return scrapeAll(res.body.toString(), site);
+	if (res.ok) {
+		return scrapeAll(res.context, channel);
 	}
 
-async function fetchScene(url, site) {
-	const useGallery = true;
-
-	// TODO: fall back on screenshots when gallery is not available
-	const res = useGallery
-		? await http.get(`${url}/gallery#gallery`)
-		: await http.get(`${url}/screenshots#screenshots`);
-
-	return scrapeScene(res.body.toString(), url, site, useGallery);
+	return res.status;
 }
 
-async function fetchProfile({ name: actorName }) {
-	const res = await http.get(`https://www.analvids.com/api/autocomplete/search?q=${actorName.replace(' ', '+')}`);
-	const data = res.body;
+async function getActorUrl(actor, channel) {
+	if (actor.url) {
+		return actor.url;
+	}
 
-	const result = data.terms.find((item) => item.type === 'model');
+	const searchRes = await http.get(`${channel.url}/api/autocomplete/search?q=${slugify(actor.name, '+')}`);
+
+	if (!searchRes.ok) {
+		return searchRes.status;
+	}
+
+	const result = searchRes.body.terms.find((item) => item.type === 'model');
 
 	if (result) {
-		const bioRes = await http.get(result.url);
-		const html = bioRes.body.toString();
-
-		return scrapeProfile(html, result.url, actorName);
+		return result.url;
 	}
 
 	return null;
 }
 
+async function fetchProfile(actor, { channel }) {
+	const actorUrl = await getActorUrl(actor, channel);
+
+	if (typeof actorUrl !== 'string') {
+		return actorUrl;
+	}
+
+	const bioRes = await unprint.get(actorUrl);
+
+	if (bioRes.ok) {
+		return scrapeProfile(bioRes.context, actorUrl, channel);
+	}
+
+	return bioRes.status;
+}
+
 module.exports = {
 	fetchLatest,
+	scrapeScene: {
+		scraper: scrapeScene,
+		unprint: true,
+	},
 	fetchProfile,
-	fetchScene,
 };

src/scrapers/analvids_legacy.js

@@ -0,0 +1,203 @@
+'use strict';
+
+const { JSDOM } = require('jsdom');
+const cheerio = require('cheerio');
+const moment = require('moment');
+
+const http = require('../utils/http');
+const slugify = require('../utils/slugify');
+
+function extractTitle(originalTitle) {
+	const titleComponents = originalTitle.split(' ');
+	// const sceneIdMatch = titleComponents.slice(-1)[0].match(/(AB|AF|GP|SZ|IV|GIO|RS|TW|MA|FM|SAL|NR|AA|GL|BZ|FS|KS|OTS|NF|NT|AX|RV|CM|BTG)\d+/); // detect studio prefixes
+	const sceneIdMatch = titleComponents.slice(-1)[0].match(/\w+\d+\s*$/); // detect studio prefixes
+	const shootId = sceneIdMatch ? sceneIdMatch[0] : null;
+	const title = sceneIdMatch ? titleComponents.slice(0, -1).join(' ') : originalTitle;
+
+	return { shootId, title };
+}
+
+function getPoster(posterElement, sceneId) {
+	const posterStyle = posterElement.attr('style');
+
+	if (posterStyle) {
+		return posterStyle.slice(posterStyle.indexOf('(') + 1, -1);
+	}
+
+	const posterRange = posterElement.attr('data-casting');
+	const posterRangeData = posterRange ? JSON.parse(posterRange) : null;
+	const posterTimeRange = posterRangeData[Math.floor(Math.random() * posterRangeData.length)];
+
+	if (!posterTimeRange) {
+		return null;
+	}
+
+	if (typeof posterTimeRange === 'number') {
+		// poster time is already a single time value
+		return `https://analvids.com/casting/${sceneId}/${posterTimeRange}`;
+	}
+
+	const [max, min] = posterTimeRange.split('-');
+	const posterTime = Math.floor(Math.random() * (Number(max) - Number(min) + 1) + Number(min));
+
+	return `https://analvids.com/casting/${sceneId}/${posterTime}`;
+}
+
+function scrapeAll(html) {
+	const $ = cheerio.load(html, { normalizeWhitespace: true });
+	const scenesElements = $('.thumbnails > div').toArray();
+
+	return scenesElements.map((element) => {
+		const sceneLinkElement = $(element).find('.thumbnail-title a');
+		const url = sceneLinkElement.attr('href');
+
+		const originalTitle = sceneLinkElement.text().trim(); // title attribute breaks when they use \\ escaping
+		const { shootId, title } = extractTitle(originalTitle);
+		const entryId = new URL(url).pathname.split('/')[2];
+
+		const date = moment.utc($(element).attr('release'), 'YYYY/MM/DD').toDate();
+
+		const sceneId = $(element).attr('data-content');
+		const posterElement = $(element).find('.thumbnail-avatar');
+
+		const poster = getPoster(posterElement, sceneId);
+
+		return {
+			url,
+			shootId,
+			entryId,
+			title,
+			date,
+			poster,
+		};
+	});
+}
+
+async function scrapeScene(html, url, site, useGallery) {
+	const $ = cheerio.load(html, { normalizeWhitespace: true });
+	const playerObject = $('script:contains("new WatchPage")').html();
+	const playerData = playerObject && playerObject.slice(playerObject.indexOf('{"swf":'), playerObject.lastIndexOf('},') + 1);
+	const data = playerData && JSON.parse(playerData);
+
+	const release = { url };
+
+	const originalTitle = $('h1.watchpage-title').text().trim();
+	const { shootId, title } = extractTitle(originalTitle);
+
+	release.shootId = shootId;
+	release.entryId = new URL(url).pathname.split('/')[2];
+
+	release.title = title;
+	release.date = moment.utc($('span[title="Release date"] a').text(), 'YYYY-MM-DD').toDate();
+
+	const [actorsElement, tagsElement, descriptionElement] = $('.scene-description__row').toArray();
+
+	release.description = $('meta[name="description"]')?.attr('content')?.trim()
+        || (descriptionElement && $(descriptionElement).find('dd').text().trim());
+
+	release.actors = $(actorsElement)
+		.find('a[href*="com/model"]')
+		.map((actorIndex, actorElement) => $(actorElement).text()).toArray();
+
+	release.duration = moment.duration($('span[title="Runtime"]').text().trim()).asSeconds();
+	release.tags = $(tagsElement).find('a').map((tagIndex, tagElement) => $(tagElement).text()).toArray();
+
+	const photos = useGallery
+		? $('.gallery a img').map((photoIndex, photoElement) => $(photoElement).attr('src')).toArray()
+		: $('.screenshots img').map((photoIndex, photoElement) => $(photoElement).attr('src')).toArray();
+
+	release.photos = photos.map((source) => {
+		// source without parameters sometimes serves larger preview photo
+		const { origin, pathname } = new URL(source);
+
+		return `${origin}${pathname}`;
+	});
+
+	const posterStyle = $('#player').attr('style');
+	const poster = posterStyle.slice(posterStyle.indexOf('(') + 1, -1);
+
+	release.poster = poster || release.photos.slice(Math.floor(release.photos.length / 3) * -1); // poster unavailable, try last 1/3rd of high res photos as fallback
+
+	if (data) {
+		const qualityMap = {
+			web: 240,
+			vga: 480,
+			hd: 720,
+			'1080p': 1080,
+		};
+
+		release.trailer = data.clip.qualities.map((trailer) => ({
+			src: trailer.src,
+			type: trailer.type,
+			quality: qualityMap[trailer.quality] || trailer.quality,
+		}));
+	}
+
+	const studioName = $('.watchpage-studioname').first().text().trim();
+	release.studio = slugify(studioName, '', { removePunctuation: true });
+
+	return release;
+}
+
+async function scrapeProfile(html, _url, actorName) {
+	const { document } = new JSDOM(html).window;
+
+	const profile = {
+		name: actorName,
+	};
+
+	const avatarEl = document.querySelector('.model--avatar img[src^="http"]');
+	const entries = Array.from(document.querySelectorAll('.model--description tr'), (el) => el.textContent.replace(/\n/g, '').split(':'));
+
+	const bio = entries
+		.filter((entry) => entry.length === 2) // ignore entries without ':' (About section, see Blanche Bradburry)
+		.reduce((acc, [key, value]) => ({ ...acc, [key.trim()]: value.trim() }), {});
+
+	profile.birthPlace = bio.Nationality;
+
+	if (bio.Age) profile.age = bio.Age;
+	if (avatarEl) profile.avatar = avatarEl.src;
+
+	profile.releases = scrapeAll(html);
+
+	return profile;
+}
+
+async function fetchLatest(site, page = 1) {
+	const res = await http.get(`${site.url}/new-videos/${page}`);
+
+	return scrapeAll(res.body.toString(), site);
+}
+
+async function fetchScene(url, site) {
+	const useGallery = true;
+
+	// TODO: fall back on screenshots when gallery is not available
+	const res = useGallery
+		? await http.get(`${url}/gallery#gallery`)
+		: await http.get(`${url}/screenshots#screenshots`);
+
+	return scrapeScene(res.body.toString(), url, site, useGallery);
+}
+
+async function fetchProfile({ name: actorName }) {
+	const res = await http.get(`https://www.analvids.com/api/autocomplete/search?q=${actorName.replace(' ', '+')}`);
+	const data = res.body;
+
+	const result = data.terms.find((item) => item.type === 'model');
+
+	if (result) {
+		const bioRes = await http.get(result.url);
+		const html = bioRes.body.toString();
+
+		return scrapeProfile(html, result.url, actorName);
+	}
+
+	return null;
+}
+
+module.exports = {
+	fetchLatest,
+	fetchProfile,
+	fetchScene,
+};