Refactored media module. Returning 320p and 720p videos from MindGeek as teasers instead of trailers.

This commit is contained in:
ThePendulum 2020-02-19 04:41:53 +01:00
parent b9e617edfc
commit 97f5e49187
12 changed files with 389 additions and 193 deletions

View File

@ -116,7 +116,7 @@ export default {
.tiles { .tiles {
display: grid; display: grid;
grid-template-columns: repeat(auto-fit, minmax(10rem, .5fr)); grid-template-columns: repeat(auto-fit, 10rem);
grid-gap: 0 .5rem; grid-gap: 0 .5rem;
padding: 1rem; padding: 1rem;
flex-grow: 1; flex-grow: 1;
@ -127,10 +127,11 @@ export default {
justify-content: center; justify-content: center;
align-items: center; align-items: center;
padding: 0 1rem; padding: 0 1rem;
margin: 1rem 0 0 0; margin: 1rem 0;
} }
.genders { .genders {
flex-shrink: 0;
padding: 0 .5rem 0 0; padding: 0 .5rem 0 0;
border-right: solid 1px $shadow-hint; border-right: solid 1px $shadow-hint;
margin: 0 1rem 0 0; margin: 0 1rem 0 0;
@ -186,7 +187,7 @@ export default {
} }
@media(max-width: $breakpoint) { @media(max-width: $breakpoint) {
.actors { .tiles {
grid-template-columns: repeat(auto-fit, minmax(8rem, 1fr)); grid-template-columns: repeat(auto-fit, minmax(8rem, 1fr));
} }
} }

View File

@ -147,7 +147,19 @@ export default {
.tiles { .tiles {
display: grid; display: grid;
grid-template-columns: repeat(auto-fit, minmax(20rem, 1fr)); grid-template-columns: repeat(auto-fit, minmax(20rem, .25fr));
grid-gap: .5rem; grid-gap: .5rem;
} }
@media(max-width: $breakpoint3) {
.tiles {
grid-template-columns: repeat(auto-fit, minmax(20rem, .5fr));
}
}
@media(max-width: $breakpoint) {
.tiles {
grid-template-columns: repeat(auto-fit, minmax(20rem, 1fr));
}
}
</style> </style>

View File

@ -48,8 +48,10 @@
>{{ `(${formatDate(release.dateAdded, 'MMM D, YYYY')})` }}</a> >{{ `(${formatDate(release.dateAdded, 'MMM D, YYYY')})` }}</a>
</span> </span>
<router-link <a
:to="{ name: release.type || 'scene', params: { releaseId: release.id, releaseSlug: release.slug }, ...(referer && { hash: `#${referer}` }) }" :href="`/${release.type || 'scene'}/${release.id}/${release.slug}`"
target="_blank"
rel="noopener noreferrer"
class="link" class="link"
> >
<img <img
@ -71,12 +73,14 @@
:title="release.title" :title="release.title"
class="thumbnail" class="thumbnail"
>No thumbnail available</div> >No thumbnail available</div>
</router-link> </a>
</span> </span>
<div class="info"> <div class="info">
<router-link <a
:to="{ name: release.type || 'scene', params: { releaseId: release.id, releaseSlug: release.slug }, ...(referer && { hash: `#${referer}` }) }" :href="`/${release.type || 'scene'}/${release.id}/${release.slug}`"
target="_blank"
rel="noopener noreferrer"
class="row link" class="row link"
> >
<h3 <h3
@ -89,7 +93,7 @@
icon="film" icon="film"
/>{{ release.title }} />{{ release.title }}
</h3> </h3>
</router-link> </a>
<span class="row"> <span class="row">
<ul class="actors nolist"> <ul class="actors nolist">

View File

@ -98,7 +98,7 @@ module.exports = {
path: './media', path: './media',
thumbnailSize: 320, // width for 16:9 will be exactly 576px thumbnailSize: 320, // width for 16:9 will be exactly 576px
thumbnailQuality: 100, thumbnailQuality: 100,
trailerQuality: [480, 540], videoQuality: [480, 360, 320, 540, 720, 1080, 2160],
limit: 25, // max number of photos per release limit: 25, // max number of photos per release
}, },
titleSlugLength: 50, titleSlugLength: 50,

Binary file not shown.

After

Width:  |  Height:  |  Size: 998 KiB

Binary file not shown.

After

Width:  |  Height:  |  Size: 95 KiB

View File

@ -1,39 +1,39 @@
const upsert = require('../src/utils/upsert'); const upsert = require('../src/utils/upsert');
const tagPosters = Object.entries({ const tagPosters = [
'anal-creampie': [0, 'Gina Valentina and Jane Wilde in "A Very Special Anniversary" for Tushy'], ['anal-creampie', 0, 'Gina Valentina and Jane Wilde in "A Very Special Anniversary" for Tushy'],
'ass-to-mouth': ['poster', 'Alysa Gap and Logan in "Anal Buffet 4" for Evil Angel'], ['ass-to-mouth', 'poster', 'Alysa Gap and Logan in "Anal Buffet 4" for Evil Angel'],
'da-tp': [0, 'Natasha Teen in LegalPorno SZ2164'], ['da-tp', 0, 'Natasha Teen in LegalPorno SZ2164'],
'double-anal': [5, 'Riley Reid in "The Gangbang of Riley Reid" for Jules Jordan'], ['double-anal', 5, 'Riley Reid in "The Gangbang of Riley Reid" for Jules Jordan'],
'double-penetration': ['poster', 'Mia Malkova in "DP!" for HardX'], ['double-penetration', 'poster', 'Mia Malkova in "DP!" for HardX'],
'double-vaginal': ['poster', 'Riley Reid in "Pizza That Ass" for Reid My Lips'], ['double-vaginal', 'poster', 'Riley Reid in "Pizza That Ass" for Reid My Lips'],
'dv-tp': ['poster', 'Juelz Ventura in "Gangbanged 5" for Elegant Angel'], ['dv-tp', 'poster', 'Juelz Ventura in "Gangbanged 5" for Elegant Angel'],
'oral-creampie': [1, 'Keisha Grey in Brazzers House'], ['oral-creampie', 1, 'Keisha Grey in Brazzers House'],
'triple-anal': ['poster', 'Kristy Black in SZ1986 for LegalPorno'], ['triple-anal', 'poster', 'Kristy Black in SZ1986 for LegalPorno'],
airtight: [1, 'Jynx Maze in "Pump My Ass Full of Cum 3" for Jules Jordan'], ['airtight', 1, 'Jynx Maze in "Pump My Ass Full of Cum 3" for Jules Jordan'],
anal: ['poster', 'Jynx Maze in "Anal Buffet 6" for Evil Angel'], ['anal', 'poster', 'Jynx Maze in "Anal Buffet 6" for Evil Angel'],
asian: ['poster', 'Vina Sky in "Young and Glamorous 10" for Jules Jordan'], ['asian', 'poster', 'Vina Sky in "Young and Glamorous 10" for Jules Jordan'],
blowjob: [0, 'Adriana Chechik in "The Dinner Party" for Real Wife Stories (Brazzers)'], ['blowjob', 0, 'Adriana Chechik in "The Dinner Party" for Real Wife Stories (Brazzers)'],
blowbang: ['poster'], ['blowbang', 'poster'],
bukkake: ['poster'], ['bukkake', 'poster'],
caucasian: ['poster'], ['caucasian', 'poster'],
creampie: ['poster'], ['creampie', 'poster'],
ebony: [1, 'Sarah Banks for Brazzers'], ['ebony', 1, 'Sarah Banks for Brazzers'],
facial: ['poster'], ['facial', 'poster'],
facefucking: ['1', 'Carrie for Young Throats'], ['facefucking', '1', 'Carrie for Young Throats'],
gangbang: ['poster', 'Kristen Scott in "Interracial Gangbang!" for Jules Jordan'], ['gangbang', 'poster', 'Kristen Scott in "Interracial Gangbang!" for Jules Jordan'],
gaping: [0, 'McKenzee Miles in "Anal Buffet 4" for Evil Angel'], ['gaping', 1, 'Vina Sky in "Vina Sky Does Anal" for HardX'],
interracial: ['poster'], ['interracial', 'poster'],
latina: ['poster'], ['latina', 'poster'],
mff: ['poster'], ['mff', 'poster'],
mfm: ['poster'], ['mfm', 'poster'],
orgy: ['poster'], ['orgy', 'poster'],
schoolgirl: [1, 'Eliza Ibarra for Brazzers'], ['schoolgirl', 1, 'Eliza Ibarra for Brazzers'],
swallowing: ['poster'], ['swallowing', 'poster'],
tattoo: ['poster', 'Kali Roses in "Goes All In For Anal" for Hussie Pass'], ['tattoo', 'poster', 'Kali Roses in "Goes All In For Anal" for Hussie Pass'],
trainbang: ['poster', 'Kali Roses in "Passing Me Around" for Blacked'], ['trainbang', 'poster', 'Kali Roses in "Passing Me Around" for Blacked'],
}) ]
.map(([slug, [filename, comment]], index) => ({ .map(([slug, filename, comment], index) => ({
tagSlug: slug, tagSlug: slug,
path: `tags/${slug}/${filename}.jpeg`, path: `tags/${slug}/${filename}.jpeg`,
thumbnail: `tags/${slug}/${filename}_thumb.jpeg`, thumbnail: `tags/${slug}/${filename}_thumb.jpeg`,
@ -63,6 +63,7 @@ const tagPhotos = [
['gangbang', 1, 'Ginger Lynn in "Gangbang Mystique", a photoset shot by Suze Randall for Puritan No. 10, 1984. This photo pushed the boundaries of pornography at the time, as depicting a woman \'fully occupied\' was unheard of.'], ['gangbang', 1, 'Ginger Lynn in "Gangbang Mystique", a photoset shot by Suze Randall for Puritan No. 10, 1984. This photo pushed the boundaries of pornography at the time, as depicting a woman \'fully occupied\' was unheard of.'],
['gangbang', 2, 'Riley Reid\'s double anal in "The Gangbang of Riley Reid" for Jules Jordan'], ['gangbang', 2, 'Riley Reid\'s double anal in "The Gangbang of Riley Reid" for Jules Jordan'],
['gaping', 'poster', 'Paulina in "Anal Buffet 4" for Evil Angel'], ['gaping', 'poster', 'Paulina in "Anal Buffet 4" for Evil Angel'],
['gaping', 0, 'McKenzee Miles in "Anal Buffet 4" for Evil Angel'],
['trainbang', 0, 'Nicole Black in GIO971 for LegalPorno'], ['trainbang', 0, 'Nicole Black in GIO971 for LegalPorno'],
['triple-anal', 1, 'Natasha Teen in SZ2098 for LegalPorno'], ['triple-anal', 1, 'Natasha Teen in SZ2098 for LegalPorno'],
['triple-anal', 2, 'Kira Thorn in GIO1018 for LegalPorno'], ['triple-anal', 2, 'Kira Thorn in GIO1018 for LegalPorno'],

View File

@ -12,7 +12,8 @@ const scrapers = require('./scrapers/scrapers');
const whereOr = require('./utils/where-or'); const whereOr = require('./utils/where-or');
const resolvePlace = require('./utils/resolve-place'); const resolvePlace = require('./utils/resolve-place');
const slugify = require('./utils/slugify'); const slugify = require('./utils/slugify');
const { createMediaDirectory, storePhotos } = require('./media_legacy'); // const { createMediaDirectory, storePhotos } = require('./media_legacy');
const { storeMedia, associateMedia } = require('./media');
async function curateActor(actor) { async function curateActor(actor) {
const [aliases, avatar, photos, social] = await Promise.all([ const [aliases, avatar, photos, social] = await Promise.all([
@ -250,6 +251,17 @@ async function storeSocialLinks(urls, actorId) {
await knex('actors_social').insert(curatedSocialEntries); await knex('actors_social').insert(curatedSocialEntries);
} }
async function storeAvatars(avatars, actorId) {
if (!avatars || avatars.length === 0) {
return [];
}
const avatarsBySource = await storeMedia(avatars, 'actor', 'avatar');
await associateMedia({ [actorId]: avatars }, avatarsBySource, 'actor', 'photo', 'avatar');
return avatarsBySource;
}
async function storeActor(actor, scraped = false, scrapeSuccess = false) { async function storeActor(actor, scraped = false, scrapeSuccess = false) {
const curatedActor = curateActorEntry(actor, scraped, scrapeSuccess); const curatedActor = curateActorEntry(actor, scraped, scrapeSuccess);
@ -260,15 +272,7 @@ async function storeActor(actor, scraped = false, scrapeSuccess = false) {
await storeSocialLinks(actor.social, actorEntry.id); await storeSocialLinks(actor.social, actorEntry.id);
if (actor.avatars) { if (actor.avatars) {
await createMediaDirectory('actors', `${actorEntry.slug}/`); await storeAvatars(actor.avatars, actorEntry.id);
await storePhotos(actor.avatars, {
domain: 'actor',
role: 'photo',
primaryRole: 'avatar',
targetId: actorEntry.id,
subpath: `${actorEntry.slug}/`,
naming: 'timestamp',
}, actorEntry.name);
} }
logger.info(`Added new entry for actor '${actor.name}'`); logger.info(`Added new entry for actor '${actor.name}'`);
@ -421,19 +425,9 @@ async function scrapeActors(actorNames) {
if (argv.save) { if (argv.save) {
if (actorEntry && profile) { if (actorEntry && profile) {
await createMediaDirectory('actors', `${actorEntry.slug}/`);
await Promise.all([ await Promise.all([
updateActor(profile, true, true), updateActor(profile, true, true),
// storeAvatars(profile, actorEntry), storeAvatars(profile.avatars, actorEntry.id),
storePhotos(profile.avatars, {
domain: 'actor',
role: 'photo',
primaryRole: 'avatar',
targetId: actorEntry.id,
subpath: `${actorEntry.slug}/`,
naming: 'timestamp',
}, actorEntry.name),
]); ]);
return profile; return profile;

View File

@ -4,11 +4,14 @@ const config = require('config');
const Promise = require('bluebird'); const Promise = require('bluebird');
const bhttp = require('bhttp'); const bhttp = require('bhttp');
const mime = require('mime'); const mime = require('mime');
const fs = require('fs-extra');
const sharp = require('sharp'); const sharp = require('sharp');
const path = require('path');
const blake2 = require('blake2'); const blake2 = require('blake2');
const logger = require('./logger'); const logger = require('./logger')(__filename);
const knex = require('./knex'); const knex = require('./knex');
const { ex } = require('./utils/q');
function getHash(buffer) { function getHash(buffer) {
const hash = blake2.createHash('blake2b', { digestLength: 24 }); const hash = blake2.createHash('blake2b', { digestLength: 24 });
@ -17,6 +20,26 @@ function getHash(buffer) {
return hash.digest('hex'); return hash.digest('hex');
} }
async function createThumbnail(buffer) {
try {
const thumbnail = sharp(buffer)
.resize({
height: config.media.thumbnailSize,
withoutEnlargement: true,
})
.jpeg({
quality: config.media.thumbnailQuality,
})
.toBuffer();
return thumbnail;
} catch (error) {
logger.error(`Failed to create thumbnail: ${error.message}`);
}
return null;
}
function pluckItems(items, specifiedLimit) { function pluckItems(items, specifiedLimit) {
const limit = specifiedLimit || config.media.limit; const limit = specifiedLimit || config.media.limit;
@ -30,6 +53,13 @@ function pluckItems(items, specifiedLimit) {
return Array.from(new Set(plucked)).map(itemIndex => items[itemIndex - 1]); // remove duplicates, may happen when photo total and photo limit are close return Array.from(new Set(plucked)).map(itemIndex => items[itemIndex - 1]); // remove duplicates, may happen when photo total and photo limit are close
} }
function pickQuality(items) {
const itemsByQuality = items.reduce((acc, item) => ({ ...acc, [item.quality]: item }), {});
const item = config.media.videoQuality.reduce((acc, quality) => acc || itemsByQuality[quality], null);
return item || items[0];
}
async function getEntropy(buffer) { async function getEntropy(buffer) {
try { try {
const { entropy } = await sharp(buffer).stats(); const { entropy } = await sharp(buffer).stats();
@ -42,33 +72,58 @@ async function getEntropy(buffer) {
} }
} }
async function fetchItem(source, index, existingItemsBySource, attempt = 1) { async function extractItem(source) {
const res = await bhttp.get(source.src);
if (res.statusCode === 200) {
const { q } = ex(res.body.toString());
return source.extract(q);
}
return null;
}
async function fetchItem(source, index, existingItemsBySource, domain, role, attempt = 1, originalSource = null) {
try { try {
if (Array.isArray(source)) { if (Array.isArray(source)) {
if (source.every(sourceX => !!sourceX.quality)) {
// various video qualities provided
const selectedSource = pickQuality(source);
return fetchItem(selectedSource, index, existingItemsBySource, domain, role, attempt, originalSource);
}
// fallbacks provided // fallbacks provided
return source.reduce((outcome, sourceX) => outcome.catch(async () => { return source.reduce(
const item = await fetchItem(sourceX, index, existingItemsBySource); (outcome, sourceX) => outcome.catch(async () => fetchItem(sourceX, index, existingItemsBySource, domain, role, attempt, originalSource)),
Promise.reject(new Error()),
if (item) { );
return item;
}
throw new Error(`Item not available: ${source}`);
}), Promise.reject(new Error()));
} }
if (source.src && source.extract) {
// source links to page containing a (presumably) tokenized photo
const itemSource = await extractItem(source);
return fetchItem(itemSource, index, existingItemsBySource, domain, role, attempt, source);
}
if (existingItemsBySource[source]) { if (existingItemsBySource[source]) {
return existingItemsBySource[source]; return null;
} }
const res = await bhttp.get(source); logger.verbose(`Fetching media item from ${source.src || source}`);
const res = await bhttp.get(source.src || source);
if (res.statusCode === 200) { if (res.statusCode === 200) {
const { pathname } = new URL(source); const { pathname } = new URL(source.src || source);
const mimetype = mime.getType(pathname); const mimetype = mime.getType(pathname);
const extension = mime.getExtension(mimetype); const extension = mime.getExtension(mimetype);
const hash = getHash(res.body); const hash = getHash(res.body);
const entropy = await getEntropy(res.body); const entropy = /image/.test(mimetype) ? await getEntropy(res.body) : null;
logger.verbose(`Fetched media item from ${source.src || source}`);
return { return {
file: res.body, file: res.body,
@ -76,40 +131,193 @@ async function fetchItem(source, index, existingItemsBySource, attempt = 1) {
extension, extension,
hash, hash,
entropy, entropy,
source, quality: source.quality || null,
source: originalSource?.src || originalSource || source.src || source,
}; };
} }
throw new Error(`Response ${res.statusCode} not OK`); throw new Error(`Response ${res.statusCode} not OK`);
} catch (error) { } catch (error) {
if (attempt <= 3) { logger.warn(`Failed attempt ${attempt}/3 to fetch ${domain} ${role} ${index + 1} (${source.src || source}): ${error}`);
return fetchItem(source, index, existingItemsBySource, attempt + 1);
if (attempt < 3) {
await Promise.delay(5000);
return fetchItem(source, index, existingItemsBySource, domain, role, attempt + 1);
} }
throw new Error(`Failed to fetch media from ${source}: ${error}`); return null;
} }
} }
async function fetchItems(itemSources, existingItemsBySource) { async function fetchItems(itemSources, existingItemsBySource, domain, role) {
return Promise.map(itemSources, async (source, index) => fetchItem(source, index, existingItemsBySource)); return Promise.map(itemSources, async (source, index) => fetchItem(source, index, existingItemsBySource, domain, role), {
concurrency: 10,
}).filter(Boolean);
} }
async function storeReleaseMedia(releases, { async function saveItems(items, domain, role) {
type = 'poster', return Promise.map(items, async (item) => {
} = {}) { try {
const pluckedSources = releases.map(release => pluckItems(release[type])); const dir = item.hash.slice(0, 2);
const existingSourceItems = await knex('media').whereIn('source', pluckedSources.flat()); const subdir = item.hash.slice(2, 4);
const existingItemsBySource = existingSourceItems.reduce((acc, item) => ({ ...acc, [item.source]: item }), {}); const filename = item.quality
? `${item.hash.slice(4)}_${item.quality}.${item.extension}`
: `${item.hash.slice(4)}.${item.extension}`;
const fetchedItems = await fetchItems(pluckedSources, existingItemsBySource); const filedir = path.join(`${domain}s`, `${role}s`, dir, subdir);
const existingHashItems = await knex('media').whereIn('hash', fetchedItems.map(item => item.hash)); const filepath = path.join(filedir, filename);
const existingItemsByHash = existingHashItems.reduce((acc, item) => ({ ...acc, [item.hash]: item }), {});
const newItems = fetchedItems.filter(item => !existingItemsByHash[item.hash]); await fs.mkdir(path.join(config.media.path, filedir), { recursive: true });
await fs.writeFile(path.join(config.media.path, filepath), item.file);
console.log(fetchedItems, existingHashItems, existingItemsByHash, newItems); if (/image/.test(item.mimetype)) {
const thumbnail = await createThumbnail(item.file);
const thumbdir = path.join(`${domain}s`, `${role}s`, 'thumbs', dir, subdir);
const thumbpath = path.join(thumbdir, filename);
await fs.mkdir(path.join(config.media.path, thumbdir), { recursive: true });
await fs.writeFile(path.join(config.media.path, thumbpath), thumbnail);
logger.verbose(`Saved ${domain} ${role} with thumbnail to ${filepath}`);
return {
...item,
thumbnail,
filepath,
thumbpath,
};
}
logger.verbose(`Saved ${domain} ${role} to ${filepath}`);
return {
...item,
filepath,
};
} catch (error) {
logger.error(`Failed to store ${domain} ${role} from ${item.source}: ${error.message}`);
return null;
}
});
}
function curateItemEntries(items) {
return items.filter(Boolean).map((item, index) => ({
path: item.filepath,
thumbnail: item.thumbpath,
mime: item.mimetype,
hash: item.hash,
source: item.source,
entropy: item.entropy,
index,
}));
}
function groupItems(items) {
return items.reduce((acc, item) => ({
source: { ...acc.source, [item.source]: item },
hash: { ...acc.hash, [item.hash]: item },
}), {
source: {},
hash: {},
});
}
async function storeMedia(sources, domain, role) {
const presentSources = sources.filter(Boolean);
if (presentSources.length === 0) {
return {};
}
// find source duplicates that don't need to be re-downloaded or re-saved
const existingSourceItems = await knex('media').whereIn('source', presentSources.flat().map(source => source.src || source));
const { source: existingSourceItemsBySource, hash: existingSourceItemsByHash } = groupItems(existingSourceItems);
// download media items from new sources
const fetchedItems = await fetchItems(presentSources, existingSourceItemsBySource, domain, role);
const { hash: fetchedItemsByHash } = groupItems(fetchedItems);
// find hash duplicates that don't need to be re-saved
const uniqueFetchedItems = Object.values(fetchedItemsByHash);
const existingHashItems = await knex('media').whereIn('hash', uniqueFetchedItems.map(item => item.hash));
const { hash: existingHashItemsByHash } = groupItems(existingHashItems);
// save new items to disk
const newItems = uniqueFetchedItems.filter(item => !existingHashItemsByHash[item.hash]);
const savedItems = await saveItems(newItems, domain, role);
// store new items in database
const curatedItemEntries = curateItemEntries(savedItems);
const storedItems = await knex('media').insert(curatedItemEntries).returning('*');
const { hash: storedItemsByHash } = groupItems(Array.isArray(storedItems) ? storedItems : []);
// accumulate existing and new items by source to be mapped onto releases
const itemsByHash = { ...existingSourceItemsByHash, ...existingHashItemsByHash, ...storedItemsByHash };
const itemsBySource = {
...existingSourceItemsBySource,
...fetchedItems.reduce((acc, item) => ({ ...acc, [item.source]: itemsByHash[item.hash] }), {}),
};
logger.info(`Stored ${fetchedItems.length} new ${domain} ${role}s`);
return itemsBySource;
}
function extractPrimaryItem(associations, targetId, role, primaryRole, primaryItemsByTargetId) {
if (!primaryRole) {
return { [role]: associations, [primaryRole]: null };
}
if (primaryItemsByTargetId[targetId]) {
const remainingAssociations = associations.filter(association => association.media_id !== primaryItemsByTargetId[targetId].media_id);
return { [role]: remainingAssociations, [primaryRole]: null };
}
return {
[role]: associations.slice(1),
[primaryRole]: associations.slice(0, 1)[0],
};
}
function associateTargetMedia(targetId, sources, mediaBySource, domain, role, primaryRole, primaryItemsByTargetId) {
if (!sources) return { [role]: null, [primaryRole]: null };
const associations = sources
.filter(Boolean)
.map((source) => {
const mediaItem = Array.isArray(source)
? source.reduce((acc, sourceX) => acc || mediaBySource[sourceX.src || sourceX], null)
: mediaBySource[source.src || source];
return mediaItem && { [`${domain}_id`]: targetId, media_id: mediaItem.id };
})
.filter(Boolean);
logger.info(`Associating ${associations.length} ${role}s to ${domain} ${targetId}`);
return extractPrimaryItem(associations, targetId, role, primaryRole, primaryItemsByTargetId);
}
async function associateMedia(sourcesByTargetId, mediaBySource, domain, role, primaryRole) {
const primaryItems = primaryRole ? await knex(`${domain}s_${primaryRole}s`).whereIn(`${domain}_id`, Object.keys(sourcesByTargetId)) : [];
const primaryItemsByTargetId = primaryItems.reduce((acc, item) => ({ ...acc, [item[`${domain}_id`]]: item }), {});
const associationsPerTarget = await Promise.map(Object.entries(sourcesByTargetId), ([targetId, sources]) => associateTargetMedia(targetId, sources, mediaBySource, domain, role, primaryRole, primaryItemsByTargetId));
const associations = associationsPerTarget.map(association => association[role]).flat().filter(Boolean);
const primaryAssociations = associationsPerTarget.map(association => association[primaryRole]).filter(Boolean);
return Promise.all([
(associations.length > 0 && knex.raw(`${knex(`${domain}s_${role}s`).insert(associations).toString()} ON CONFLICT DO NOTHING`)),
(primaryAssociations.length > 0 && knex.raw(`${knex(`${domain}s_${primaryRole}s`).insert(primaryAssociations).toString()} ON CONFLICT DO NOTHING`)),
]);
} }
module.exports = { module.exports = {
storeReleaseMedia, pluckItems,
storeMedia,
associateMedia,
}; };

View File

@ -10,20 +10,11 @@ const argv = require('./argv');
const whereOr = require('./utils/where-or'); const whereOr = require('./utils/where-or');
const { associateTags } = require('./tags'); const { associateTags } = require('./tags');
const { associateActors, scrapeBasicActors } = require('./actors'); const { associateActors, scrapeBasicActors } = require('./actors');
/*
const { const {
createMediaDirectory, pluckItems,
storePhotos, storeMedia,
// storeReleasePhotos, associateMedia,
storeTrailer,
storeReleaseMedia,
} = require('./media'); } = require('./media');
*/
const {
createMediaDirectory,
storePhotos,
storeTrailer,
} = require('./media_legacy');
const { fetchSites, findSiteByUrl } = require('./sites'); const { fetchSites, findSiteByUrl } = require('./sites');
const slugify = require('./utils/slugify'); const slugify = require('./utils/slugify');
@ -337,55 +328,39 @@ function accumulateMovies(releases) {
} }
async function storeReleaseAssets(releases) { async function storeReleaseAssets(releases) {
// await storeReleasePhotos(releases); const releasePostersById = releases.reduce((acc, release) => ({ ...acc, [release.id]: [release.poster] }), {});
const releaseCoversById = releases.reduce((acc, release) => ({ ...acc, [release.id]: release.covers }), {});
const releaseTrailersById = releases.reduce((acc, release) => ({ ...acc, [release.id]: [release.trailer] }), {});
const releaseTeasersById = releases.reduce((acc, release) => ({ ...acc, [release.id]: [release.teaser] }), {});
const releasePhotosById = releases.reduce((acc, release) => ({
...acc,
[release.id]: pluckItems(release.photos),
}), {});
// return storeReleaseMedia(releases); const [posters, covers] = await Promise.all([
storeMedia(Object.values(releasePostersById).flat(), 'release', 'poster'),
storeMedia(Object.values(releaseCoversById).flat(), 'release', 'cover'),
]);
await Promise.map(releases, async (release) => { // ensure posters are available before fetching supplementary media
const subpath = `${release.site.network.slug}/${release.site.slug}/${release.id}/`; await Promise.all([
const identifier = `"${release.title}" (${release.id})`; associateMedia(releasePostersById, posters, 'release', 'poster'),
associateMedia(releaseCoversById, covers, 'release', 'cover'),
]);
try { const photos = await storeMedia(Object.values(releasePhotosById).flat(), 'release', 'photo');
await createMediaDirectory('releases', subpath); await associateMedia(releasePhotosById, photos, 'release', 'photo');
// don't use Promise.all to prevent concurrency issues with duplicate detection // videos take a long time, fetch last
if (release.poster) { const [trailers, teasers] = await Promise.all([
await storePhotos([release.poster], { storeMedia(Object.values(releaseTrailersById).flat(), 'release', 'trailer'),
role: 'poster', storeMedia(Object.values(releaseTeasersById).flat(), 'release', 'teaser'),
targetId: release.id, ]);
subpath,
}, identifier);
}
await storePhotos(release.photos, { await Promise.all([
targetId: release.id, associateMedia(releaseTrailersById, trailers, 'release', 'trailer'),
subpath, associateMedia(releaseTeasersById, teasers, 'release', 'teaser'),
primaryRole: release.poster ? null : 'poster', ]);
}, identifier);
await storePhotos(release.covers, {
role: 'cover',
targetId: release.id,
subpath,
}, identifier);
await storeTrailer(release.trailer, {
targetId: release.id,
subpath,
role: 'trailer',
}, identifier);
await storeTrailer(release.teaser, {
targetId: release.id,
subpath,
role: 'teaser',
}, identifier);
} catch (error) {
logger.error(error.message);
}
}, {
concurrency: 10,
});
} }
async function storeRelease(release) { async function storeRelease(release) {
@ -453,6 +428,8 @@ async function storeReleases(releases) {
concurrency: 10, concurrency: 10,
}).filter(release => release); }).filter(release => release);
logger.info(`Stored ${storedReleases.length} new releases`);
const actors = accumulateActors(storedReleases); const actors = accumulateActors(storedReleases);
const movies = accumulateMovies(storedReleases); const movies = accumulateMovies(storedReleases);

View File

@ -47,7 +47,7 @@ function scrapeLatest(html, site) {
const actors = title.split(/[,&]|\band\b/).map(actor => actor.trim()); const actors = title.split(/[,&]|\band\b/).map(actor => actor.trim());
const poster = `https:${element.querySelector('img').src}`; const poster = `https:${element.querySelector('img').src}`;
const trailer = sceneLinkElement.dataset.preview_clip_url; const teaser = sceneLinkElement.dataset.preview_clip_url;
return [ return [
...acc, ...acc,
@ -57,8 +57,8 @@ function scrapeLatest(html, site) {
title, title,
actors, actors,
poster, poster,
trailer: { teaser: {
src: trailer, src: teaser,
}, },
site, site,
}, },
@ -99,7 +99,7 @@ async function scrapeScene(html, url, site) {
const lastPhotosUrl = Array.from(document.querySelectorAll('.pagination a')).slice(-1)[0].href; const lastPhotosUrl = Array.from(document.querySelectorAll('.pagination a')).slice(-1)[0].href;
const photos = await getPhotos(`${origin}${pathname}${lastPhotosUrl}`, site, url); const photos = await getPhotos(`${origin}${pathname}${lastPhotosUrl}`, site, url);
const stars = Math.floor(Number(document.querySelector('span[itemprop="average"]').textContent) / 2); const stars = Math.floor(Number(document.querySelector('span[itemprop="average"]')?.textContent || document.querySelector('span[itemprop="ratingValue"]')?.textContent) / 2);
const tags = Array.from(document.querySelectorAll('.scene-details .categories a')).map(({ textContent }) => textContent); const tags = Array.from(document.querySelectorAll('.scene-details .categories a')).map(({ textContent }) => textContent);
return { return {

View File

@ -36,35 +36,33 @@ function scrapeLatestX(data, site) {
return null; return null;
} }
const { id: entryId, title, description } = data; const release = {
const hostname = site.parameters?.native ? site.url : site.network.url; entryId: data.id,
const url = `${hostname}/scene/${entryId}/`; title: data.title,
const date = new Date(data.dateReleased); description: data.description,
const actors = data.actors.map(actor => ({ name: actor.name, gender: actor.gender }));
const tags = data.tags.map(tag => tag.name);
const [poster, ...photos] = getThumbs(data);
const trailer = data.videos.mediabook && (data.videos.mediabook.files['720p'] || data.videos.mediabook.files['320p']);
const duration = data.videos.mediabook && data.videos.mediabook.length;
return {
url,
entryId,
title,
description,
actors,
tags,
duration,
poster,
photos,
trailer: trailer && {
src: trailer.urls.view,
quality: parseInt(trailer.format, 10),
},
date,
site,
}; };
const hostname = site.parameters?.native ? site.url : site.network.url;
release.url = `${hostname}/scene/${release.entryId}/`;
release.date = new Date(data.dateReleased);
release.actors = data.actors.map(actor => ({ name: actor.name, gender: actor.gender }));
release.tags = data.tags.map(tag => tag.name);
release.duration = data.videos.mediabook?.length;
[release.poster, ...release.photos] = getThumbs(data);
const teaserSources = data.videos.mediabook?.files;
if (teaserSources) {
release.teaser = Object.values(teaserSources).map(teaser => ({
src: teaser.urls.view,
quality: parseInt(teaser.format, 10),
}));
}
return release;
} }
async function scrapeLatest(items, site) { async function scrapeLatest(items, site) {
@ -89,12 +87,13 @@ function scrapeScene(data, url, _site, networkName) {
[release.poster, ...release.photos] = getThumbs(data); [release.poster, ...release.photos] = getThumbs(data);
const trailer = data.videos.mediabook && (data.videos.mediabook.files['720p'] || data.videos.mediabook.files['320p']); const teaserSources = data.videos.mediabook?.files;
if (trailer) {
release.trailer = { if (teaserSources) {
src: trailer.urls.view, release.teaser = Object.values(teaserSources).map(teaser => ({
quality: parseInt(trailer.format, 10), src: teaser.urls.view,
}; quality: parseInt(teaser.format, 10),
}));
} }
const siteName = data.collections[0]?.name || data.brand; const siteName = data.collections[0]?.name || data.brand;