Refactored Vixen scraper, moved XEmpire to generic Gamma scraper after site update, added BiPhoria.
This commit is contained in:
@@ -4,7 +4,7 @@
|
||||
const Promise = require('bluebird');
|
||||
const moment = require('moment');
|
||||
|
||||
const logger = require('../logger')(__filename);
|
||||
const qu = require('../utils/qu');
|
||||
const http = require('../utils/http');
|
||||
const slugify = require('../utils/slugify');
|
||||
|
||||
@@ -49,6 +49,26 @@ function getAvatarFallbacks(avatar) {
|
||||
.flat();
|
||||
}
|
||||
|
||||
function curateSources(sources, type = 'image/jpeg') {
|
||||
if (!sources) {
|
||||
return null;
|
||||
}
|
||||
|
||||
return sources
|
||||
.map((source) => ({
|
||||
src: source.src,
|
||||
width: source.width,
|
||||
height: source.height,
|
||||
type: source.type || type,
|
||||
expectType: {
|
||||
'binary/octet-stream': type,
|
||||
},
|
||||
}))
|
||||
.sort((resA, resB) => (resB.width * resB.height) - (resA.width * resA.height)) // number of pixels
|
||||
.sort((resA, resB) => Math.abs(1.8 - Number((resA.width / resA.height).toFixed(1))) // approximation to 16:9
|
||||
- Math.abs(1.8 - Number((resB.width / resB.height).toFixed(1))));
|
||||
}
|
||||
|
||||
async function getTrailer(scene, channel, url) {
|
||||
const res = await http.post(`${channel.url}/graphql`, {
|
||||
operationName: 'getToken',
|
||||
@@ -142,88 +162,27 @@ async function getTrailer(scene, channel, url) {
|
||||
return null;
|
||||
}
|
||||
|
||||
/*
|
||||
async function getPhotosLegacy(url) {
|
||||
const htmlRes = await http.get(url, {
|
||||
extract: {
|
||||
runScripts: 'dangerously',
|
||||
},
|
||||
});
|
||||
|
||||
try {
|
||||
const state = htmlRes?.window?.__APOLLO_STATE__;
|
||||
|
||||
if (!state) {
|
||||
return [];
|
||||
}
|
||||
|
||||
const key = Object.values(state?.ROOT_QUERY).find((query) => query?.__ref)?.__ref;
|
||||
const data = state[key];
|
||||
|
||||
if (!data) {
|
||||
return [];
|
||||
}
|
||||
|
||||
return data.carousel.slice(1).map((photo) => photo.main?.[0].src).filter(Boolean);
|
||||
} catch (error) {
|
||||
logger.warn(`Failed to retrieve Vixen images: ${error.message}`);
|
||||
return [];
|
||||
}
|
||||
}
|
||||
*/
|
||||
|
||||
async function getPhotos(url) {
|
||||
const htmlRes = await http.get(url, {
|
||||
parse: true,
|
||||
extract: {
|
||||
runScripts: 'dangerously',
|
||||
},
|
||||
});
|
||||
|
||||
try {
|
||||
const state = htmlRes?.window?.__APOLLO_STATE__;
|
||||
|
||||
console.log('state', state);
|
||||
|
||||
if (!state) {
|
||||
return [];
|
||||
}
|
||||
|
||||
const key = Object.values(state?.ROOT_QUERY).find((query) => query?.__ref)?.__ref;
|
||||
const data = state[key];
|
||||
|
||||
console.log('data', data);
|
||||
|
||||
if (!data) {
|
||||
return [];
|
||||
}
|
||||
|
||||
console.log(data.carousel);
|
||||
|
||||
return data.carousel.slice(1).map((photo) => photo.main?.[0].src).filter(Boolean);
|
||||
} catch (error) {
|
||||
logger.warn(`Failed to retrieve Vixen images: ${error.message}`);
|
||||
return [];
|
||||
}
|
||||
}
|
||||
|
||||
function scrapeAll(scenes, site, origin) {
|
||||
return scenes.map((scene) => {
|
||||
function scrapeAll(scenes, channel) {
|
||||
return scenes.map((data) => {
|
||||
const release = {};
|
||||
|
||||
release.title = scene.title;
|
||||
release.entryId = data.videoId;
|
||||
release.url = `${channel.url}/videos/${data.slug}`;
|
||||
release.title = data.title;
|
||||
|
||||
release.entryId = String(scene.newId);
|
||||
release.url = `${site?.url || origin}/videos${scene.targetUrl}`;
|
||||
release.date = qu.extractDate(data.releaseDate);
|
||||
release.actors = data.modelsSlugged.map((model) => ({
|
||||
name: model.name,
|
||||
url: `${channel.url}/models/${model.slugged}`,
|
||||
}));
|
||||
|
||||
release.date = moment.utc(scene.releaseDate).toDate();
|
||||
release.datePrecision = 'minute';
|
||||
release.poster = curateSources(data.images.listing);
|
||||
release.teaser = curateSources(data.previews.listing, 'video/mp4');
|
||||
|
||||
release.actors = scene.models;
|
||||
release.stars = Number(scene.textRating) / 2;
|
||||
release.stars = data.rating;
|
||||
|
||||
release.poster = getPosterFallbacks(scene.images.poster);
|
||||
release.teaser = getTeaserFallbacks(scene.previews.poster);
|
||||
console.log(data);
|
||||
console.log(release);
|
||||
|
||||
return release;
|
||||
});
|
||||
@@ -252,47 +211,47 @@ function scrapeUpcoming(scene, site) {
|
||||
|
||||
release.entryId = (release.poster[0] || release.teaser[0])?.src?.match(/\/(\d+)/)?.[1];
|
||||
|
||||
console.log('upcoming', scene);
|
||||
|
||||
return [release];
|
||||
}
|
||||
|
||||
async function scrapeScene(data, url, site, baseRelease, options) {
|
||||
const scene = data.video;
|
||||
|
||||
async function scrapeScene(data, url, channel, options) {
|
||||
const release = {
|
||||
url,
|
||||
title: scene.title,
|
||||
description: scene.description,
|
||||
actors: scene.models,
|
||||
director: scene.directorNames,
|
||||
duration: scene.runLength,
|
||||
stars: scene.totalRateVal,
|
||||
tags: scene.tags,
|
||||
entryId: data.video.videoId || data.video.newId,
|
||||
title: data.video.title,
|
||||
description: data.video.description,
|
||||
actors: data.video.models,
|
||||
director: data.video.directorNames,
|
||||
duration: qu.durationToSeconds(data.video.runLength),
|
||||
stars: data.video.rating,
|
||||
};
|
||||
|
||||
release.entryId = scene.newId;
|
||||
release.entryId = data.video.newId;
|
||||
release.date = qu.extractDate(data.video.releaseDate);
|
||||
|
||||
release.date = moment.utc(scene.releaseDate).toDate();
|
||||
release.productionDate = moment.utc(scene.shootDate).toDate();
|
||||
release.datePrecision = 'minute';
|
||||
release.actors = data.video.modelsSlugged.map((model) => ({
|
||||
name: model.name,
|
||||
url: `${channel.url}/models/${model.slugged}`,
|
||||
}));
|
||||
|
||||
release.actors = baseRelease?.actors || scene.models;
|
||||
release.poster = curateSources(data.video.images?.poster) || data.video.videoImage?.src;
|
||||
release.photos = data.galleryImages?.length > 0
|
||||
? data.galleryImages.map((image) => image.src)
|
||||
: data.video.carousel?.map((photo) => photo.main[0]?.src).filter(Boolean);
|
||||
|
||||
release.poster = getPosterFallbacks(scene.images.poster);
|
||||
if (options.includeTrailers) {
|
||||
const trailer = await getTrailer(data.video, channel, url);
|
||||
|
||||
// release.photos = data.pictureset.map(photo => photo.main[0]?.src).filter(Boolean);
|
||||
if (options.includePhotos) {
|
||||
release.photos = await getPhotos(url);
|
||||
if (trailer) {
|
||||
release.trailer = trailer;
|
||||
}
|
||||
}
|
||||
|
||||
release.teaser = getTeaserFallbacks(scene.previews.poster);
|
||||
release.qualities = data.video?.downloadResolutions.map((quality) => Number(quality.width)).filter(Boolean); // width property is actually the height
|
||||
|
||||
const trailer = await getTrailer(scene, site, url);
|
||||
if (trailer) release.trailer = trailer;
|
||||
|
||||
release.chapters = data.video.chapters?.video.map((chapter) => ({
|
||||
tags: [chapter.title],
|
||||
time: chapter.seconds,
|
||||
}));
|
||||
console.log(release);
|
||||
|
||||
return release;
|
||||
}
|
||||
@@ -346,13 +305,71 @@ async function scrapeProfile(data, origin, withReleases) {
|
||||
return profile;
|
||||
}
|
||||
|
||||
async function fetchLatestGraphql(channel, page = 1) {
|
||||
const query = `
|
||||
query($query: String!, $site: Site!) {
|
||||
searchVideos(input: {
|
||||
query: $query
|
||||
site: $site
|
||||
}) {
|
||||
edges {
|
||||
node {
|
||||
title
|
||||
slug
|
||||
description
|
||||
releaseDate
|
||||
categories {
|
||||
name
|
||||
}
|
||||
chapters {
|
||||
video {
|
||||
title
|
||||
seconds
|
||||
}
|
||||
}
|
||||
models {
|
||||
name
|
||||
}
|
||||
images {
|
||||
poster {
|
||||
...ImageInfo
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
fragment ImageInfo on Image {
|
||||
src
|
||||
highdpi {
|
||||
double
|
||||
}
|
||||
}
|
||||
`;
|
||||
|
||||
const variables = JSON.stringify({
|
||||
site: channel.slug.toUpperCase(),
|
||||
query: 'alone at last',
|
||||
});
|
||||
|
||||
const res = await http.get(`${channel.url}/graphql?query=${encodeURI(query)}&variables=${variables}`);
|
||||
|
||||
console.log(res.body);
|
||||
console.log(res.body.errors);
|
||||
console.log(res.body.data?.searchVideos?.edges.map((edge) => edge.node));
|
||||
}
|
||||
|
||||
async function fetchLatest(site, page = 1) {
|
||||
const url = `${site.url}/api/videos?page=${page}`;
|
||||
const res = await http.get(url);
|
||||
const url = `${site.url}/videos?page=${page}`;
|
||||
const res = await qu.get(url);
|
||||
|
||||
if (res.ok) {
|
||||
if (res.body.data.videos) {
|
||||
return scrapeAll(res.body.data.videos, site);
|
||||
const dataString = res.item.query.html('#__NEXT_DATA__');
|
||||
const data = dataString && JSON.parse(dataString);
|
||||
|
||||
if (data?.props.pageProps.edges) {
|
||||
return scrapeAll(data.props.pageProps.edges.map((edge) => edge.node), site);
|
||||
}
|
||||
|
||||
return [];
|
||||
@@ -376,22 +393,14 @@ async function fetchUpcoming(site) {
|
||||
return res.status;
|
||||
}
|
||||
|
||||
async function fetchScene(url, site, baseRelease, options) {
|
||||
const { origin, pathname } = new URL(url);
|
||||
const apiUrl = `${origin}/api/${pathname.split('/').slice(-1)[0]}`;
|
||||
|
||||
const res = await http.get(apiUrl, {
|
||||
extract: {
|
||||
runScripts: 'dangerously',
|
||||
},
|
||||
});
|
||||
async function fetchScene(url, channel, baseRelease, options) {
|
||||
const res = await qu.get(url);
|
||||
|
||||
if (res.ok) {
|
||||
if (res.body.data) {
|
||||
return scrapeScene(res.body.data, url, site, baseRelease, options);
|
||||
}
|
||||
const dataString = res.item.query.html('#__NEXT_DATA__');
|
||||
const data = dataString && JSON.parse(dataString);
|
||||
|
||||
return null;
|
||||
return scrapeScene(data.props.pageProps, url, channel, options);
|
||||
}
|
||||
|
||||
return res.status;
|
||||
@@ -415,6 +424,7 @@ async function fetchProfile({ name: actorName }, { site }, include) {
|
||||
}
|
||||
|
||||
module.exports = {
|
||||
// fetchLatest: fetchLatestGraphql,
|
||||
fetchLatest,
|
||||
fetchUpcoming,
|
||||
fetchScene,
|
||||
|
||||
Reference in New Issue
Block a user