1002 lines
31 KiB
JavaScript
1002 lines
31 KiB
JavaScript
'use strict';
|
|
|
|
const config = require('config');
|
|
const util = require('util');
|
|
const Promise = require('bluebird');
|
|
const moment = require('moment');
|
|
const blake2 = require('blake2');
|
|
const DOMPurify = require('dompurify');
|
|
const { JSDOM } = require('jsdom');
|
|
const omit = require('object.omit');
|
|
|
|
const { window } = new JSDOM('');
|
|
const domPurify = DOMPurify(window);
|
|
|
|
// const logger = require('./logger')(__filename);
|
|
const knex = require('./knex');
|
|
const scrapers = require('./scrapers/scrapers').actors;
|
|
|
|
const argv = require('./argv');
|
|
const include = require('./utils/argv-include')(argv);
|
|
const bulkInsert = require('./utils/bulk-insert');
|
|
const logger = require('./logger')(__filename);
|
|
|
|
const { toBaseReleases } = require('./deep');
|
|
const { associateAvatars, flushOrphanedMedia } = require('./media');
|
|
const { deleteScenes } = require('./releases');
|
|
|
|
const slugify = require('./utils/slugify');
|
|
const capitalize = require('./utils/capitalize');
|
|
const resolvePlace = require('./utils/resolve-place');
|
|
|
|
const hairColors = {
|
|
'jet-black': 'black',
|
|
'red-head': 'red',
|
|
'soft-black': 'black',
|
|
black: 'black',
|
|
blonde: 'blonde',
|
|
blondie: 'blonde',
|
|
brown: 'brown',
|
|
brunette: 'brown',
|
|
fair: 'blonde',
|
|
raven: 'black',
|
|
red: 'red',
|
|
redhead: 'red',
|
|
blue: 'blue',
|
|
green: 'green',
|
|
purple: 'purple',
|
|
pink: 'pink',
|
|
};
|
|
|
|
const eyeColors = {
|
|
blue: 'blue',
|
|
brown: 'brown',
|
|
dark: 'brown',
|
|
gray: 'gray',
|
|
green: 'green',
|
|
grey: 'gray',
|
|
hazel: 'hazel',
|
|
};
|
|
|
|
const ethnicities = {
|
|
'african american': 'black',
|
|
'african-american': 'black',
|
|
'native american': 'native american',
|
|
african: 'black',
|
|
aravic: 'arabic',
|
|
asian: 'asian',
|
|
black: 'black',
|
|
caucasian: 'white',
|
|
european: 'white',
|
|
hispanic: 'latin',
|
|
indian: 'indian',
|
|
japanese: 'japanese',
|
|
latin: 'latin',
|
|
latina: 'latina',
|
|
latino: 'latino',
|
|
white: 'white',
|
|
};
|
|
|
|
function getBoolean(value) {
|
|
if (typeof value === 'boolean') {
|
|
return value;
|
|
}
|
|
|
|
if (typeof value === 'string') {
|
|
if (/yes/i.test(value)) {
|
|
return true;
|
|
}
|
|
|
|
if (/no/i.test(value)) {
|
|
return true;
|
|
}
|
|
}
|
|
|
|
return null;
|
|
}
|
|
|
|
function getMostFrequent(items) {
|
|
const { mostFrequent } = items.reduce((acc, item) => {
|
|
if (item === undefined || item === null) {
|
|
return acc;
|
|
}
|
|
|
|
const slug = slugify(item);
|
|
|
|
acc.counts[slug] = (acc.counts[slug] || 0) + 1;
|
|
|
|
if (!acc.mostFrequent || acc.counts[slug] > acc.counts[slugify(acc.mostFrequent)]) {
|
|
acc.mostFrequent = item;
|
|
}
|
|
|
|
return acc;
|
|
}, {
|
|
counts: {},
|
|
mostFrequent: null,
|
|
});
|
|
|
|
return mostFrequent;
|
|
}
|
|
|
|
function getMostFrequentDate(dates) {
|
|
const year = getMostFrequent(dates.map(dateX => dateX.getFullYear()));
|
|
const month = getMostFrequent(dates.map(dateX => dateX.getMonth()));
|
|
const date = getMostFrequent(dates.map(dateX => dateX.getDate()));
|
|
|
|
if (year === null || month === null || date === null) {
|
|
return null;
|
|
}
|
|
|
|
return moment({ year, month, date }).toDate();
|
|
}
|
|
|
|
function getHighest(items) {
|
|
return items.reduce((prevItem, item) => (item > prevItem ? item : prevItem), null);
|
|
}
|
|
|
|
function getLongest(items) {
|
|
return items.sort((itemA, itemB) => itemB.length - itemA.length)[0] || null;
|
|
}
|
|
|
|
function getAverage(items) {
|
|
return Math.round(items.reduce((acc, item) => acc + item, 0) / items.length) || null;
|
|
}
|
|
|
|
function toBaseActors(actorsOrNames, release) {
|
|
if (!actorsOrNames) {
|
|
return [];
|
|
}
|
|
|
|
const baseActors = actorsOrNames
|
|
.filter(actorOrName => actorOrName && (typeof actorOrName === 'string' || actorOrName.name))
|
|
.map((actorOrName) => {
|
|
const [baseName, entryId] = (actorOrName.name || actorOrName).split(':');
|
|
|
|
const name = capitalize(baseName);
|
|
const slug = slugify(name);
|
|
|
|
const baseActor = {
|
|
name,
|
|
slug,
|
|
entryId: entryId || actorOrName.entryId || null,
|
|
entity: release?.entity?.parent || release?.entity || null,
|
|
hasProfile: !!actorOrName.name, // actor contains profile information
|
|
};
|
|
|
|
if (actorOrName.name) {
|
|
return {
|
|
...actorOrName,
|
|
...baseActor,
|
|
};
|
|
}
|
|
|
|
return baseActor;
|
|
});
|
|
|
|
return baseActors;
|
|
}
|
|
|
|
function curateActor(actor, withDetails = false, isProfile = false) {
|
|
if (!actor) {
|
|
return null;
|
|
}
|
|
|
|
const curatedActor = {
|
|
id: actor.id,
|
|
name: actor.name,
|
|
slug: actor.slug,
|
|
url: actor.url,
|
|
gender: actor.gender,
|
|
entityId: actor.entity_id,
|
|
aliasFor: actor.alias_for,
|
|
dateOfBirth: actor.date_of_birth,
|
|
age: actor.age,
|
|
birthCountry: actor.birth_country_alpha2,
|
|
...(withDetails && {
|
|
alias: actor.alias && {
|
|
id: actor.alias.id,
|
|
name: actor.alias.name,
|
|
slug: actor.slug,
|
|
gender: actor.alias.gender,
|
|
},
|
|
entity: actor.entity && {
|
|
id: actor.entity.id,
|
|
name: actor.entity.name,
|
|
slug: actor.entity.slug,
|
|
},
|
|
dateOfDeath: actor.date_of_death,
|
|
cup: actor.cup,
|
|
bust: actor.bust,
|
|
waist: actor.waist,
|
|
hip: actor.hip,
|
|
naturalBoobs: actor.natural_boobs,
|
|
penisLength: actor.penis_length,
|
|
penisGirth: actor.penis_girth,
|
|
circumcised: actor.circumcised,
|
|
height: actor.height,
|
|
weight: actor.weight,
|
|
eyes: actor.eyes,
|
|
hairColor: actor.hair_color,
|
|
hasTattoos: actor.has_tattoos,
|
|
hasPiercings: actor.has_piercings,
|
|
tattoos: actor.tattoos,
|
|
piercings: actor.piercings,
|
|
...(isProfile && { description: actor.description }),
|
|
placeOfBirth: actor.birth_country && {
|
|
country: {
|
|
alpha2: actor.birth_country.alpha2,
|
|
name: actor.birth_country.name,
|
|
alias: actor.birth_country.alias,
|
|
},
|
|
state: actor.birth_state,
|
|
city: actor.birth_city,
|
|
},
|
|
placeOfResidence: actor.residence_country && {
|
|
country: {
|
|
alpha2: actor.residence_country.alpha2,
|
|
name: actor.residence_country.name,
|
|
alias: actor.residence_country.alias,
|
|
},
|
|
state: actor.residence_state,
|
|
city: actor.residence_city,
|
|
},
|
|
avatar: actor.avatar && {
|
|
id: actor.avatar.id,
|
|
path: actor.avatar.path,
|
|
width: actor.avatar.width,
|
|
height: actor.avatar.height,
|
|
size: actor.avatar.size,
|
|
source: actor.avatar.source,
|
|
},
|
|
...(actor.profiles && { profiles: actor.profiles?.map(profile => curateActor(profile, true, true)) }),
|
|
}),
|
|
};
|
|
|
|
return curatedActor;
|
|
}
|
|
|
|
function curateActorEntry(baseActor, batchId) {
|
|
return {
|
|
name: baseActor.name,
|
|
slug: baseActor.slug,
|
|
entity_id: null,
|
|
entry_id: baseActor.entryId,
|
|
batch_id: batchId,
|
|
};
|
|
}
|
|
|
|
function curateActorEntries(baseActors, batchId) {
|
|
return baseActors.map(baseActor => curateActorEntry(baseActor, batchId));
|
|
}
|
|
|
|
function curateProfileEntry(profile) {
|
|
if (!profile.id) {
|
|
return null;
|
|
}
|
|
|
|
const curatedProfileEntry = {
|
|
...(profile.update !== false && { id: profile.update }),
|
|
actor_id: profile.id,
|
|
entity_id: profile.entity?.id || null,
|
|
date_of_birth: profile.dateOfBirth,
|
|
date_of_death: profile.dateOfDeath,
|
|
age: profile.age,
|
|
url: profile.url,
|
|
gender: profile.gender,
|
|
ethnicity: profile.ethnicity,
|
|
description: profile.description,
|
|
description_hash: profile.descriptionHash,
|
|
birth_city: profile.placeOfBirth?.city || null,
|
|
birth_state: profile.placeOfBirth?.state || null,
|
|
birth_country_alpha2: profile.placeOfBirth?.country || null,
|
|
residence_city: profile.placeOfResidence?.city || null,
|
|
residence_state: profile.placeOfResidence?.state || null,
|
|
residence_country_alpha2: profile.placeOfResidence?.country || null,
|
|
cup: profile.cup,
|
|
bust: profile.bust,
|
|
waist: profile.waist,
|
|
hip: profile.hip,
|
|
penis_length: profile.penisLength,
|
|
penis_girth: profile.penisGirth,
|
|
circumcised: profile.circumcised,
|
|
natural_boobs: profile.naturalBoobs,
|
|
height: profile.height,
|
|
weight: profile.weight,
|
|
hair_color: profile.hairColor,
|
|
eyes: profile.eyes,
|
|
has_tattoos: profile.hasTattoos,
|
|
has_piercings: profile.hasPiercings,
|
|
piercings: profile.piercings,
|
|
tattoos: profile.tattoos,
|
|
avatar_media_id: profile.avatarMediaId || null,
|
|
};
|
|
|
|
return curatedProfileEntry;
|
|
}
|
|
|
|
async function curateProfile(profile, actor) {
|
|
if (!profile) {
|
|
return null;
|
|
}
|
|
|
|
try {
|
|
const curatedProfile = {
|
|
id: profile.id,
|
|
name: profile.name,
|
|
avatar: profile.avatar,
|
|
scraper: profile.scraper,
|
|
entity: profile.entity,
|
|
update: profile.update,
|
|
};
|
|
|
|
curatedProfile.description = domPurify.sanitize(profile.description?.replace(/\s+/g, ' '), { ALLOWED_TAGS: [] }).trim() || null;
|
|
|
|
const hasher = curatedProfile.description && blake2
|
|
.createHash('blake2b', { digestLength: 24 })
|
|
.update(Buffer.from(slugify(curatedProfile.description)));
|
|
|
|
curatedProfile.descriptionHash = curatedProfile.description && hasher.digest('hex');
|
|
|
|
curatedProfile.nationality = profile.nationality?.trim() || null; // used to derive country when country not available
|
|
|
|
curatedProfile.ethnicity = ethnicities[profile.ethnicity?.trim().toLowerCase()] || null;
|
|
curatedProfile.hairColor = hairColors[(profile.hairColor || profile.hair)?.toLowerCase().replace('hair', '').trim()] || null;
|
|
curatedProfile.eyes = eyeColors[profile.eyes?.trim().toLowerCase()] || null;
|
|
|
|
curatedProfile.tattoos = profile.tattoos?.trim() || null;
|
|
curatedProfile.piercings = profile.piercings?.trim() || null;
|
|
|
|
curatedProfile.gender = (/female/i.test(profile.gender) && 'female')
|
|
|| (/shemale|trans/i.test(profile.gender) && 'transsexual')
|
|
|| (/male/i.test(profile.gender) && 'male')
|
|
|| null;
|
|
|
|
const dateOfBirth = profile.dateOfBirth || profile.birthdate;
|
|
|
|
curatedProfile.dateOfBirth = (!Number.isNaN(Number(dateOfBirth)) // possibly valid date
|
|
&& new Date() - dateOfBirth > 567648000000 // over 18
|
|
&& dateOfBirth)
|
|
|| null;
|
|
|
|
curatedProfile.dateOfDeath = Number.isNaN(Number(profile.dateOfDeath)) ? null : profile.dateOfDeath;
|
|
curatedProfile.age = Number(profile.age) || null;
|
|
|
|
curatedProfile.height = Number(profile.height) || profile.height?.match?.(/\d+/)?.[0] || null;
|
|
curatedProfile.weight = Number(profile.weight) || profile.weight?.match?.(/\d+/)?.[0] || null;
|
|
|
|
curatedProfile.cup = profile.cup || (typeof profile.bust === 'string' && profile.bust?.match?.(/[a-zA-Z]+/)?.[0]) || null;
|
|
curatedProfile.bust = Number(profile.bust) || profile.bust?.match?.(/\d+/)?.[0] || null;
|
|
curatedProfile.waist = Number(profile.waist) || profile.waist?.match?.(/\d+/)?.[0] || null;
|
|
curatedProfile.hip = Number(profile.hip) || profile.hip?.match?.(/\d+/)?.[0] || null;
|
|
curatedProfile.penisLength = Number(profile.penisLength) || profile.penisLength?.match?.(/\d+/)?.[0] || null;
|
|
curatedProfile.penisGirth = Number(profile.penisGirth) || profile.penisGirth?.match?.(/\d+/)?.[0] || null;
|
|
|
|
curatedProfile.circumcised = getBoolean(profile.circumcised);
|
|
curatedProfile.naturalBoobs = getBoolean(profile.naturalBoobs);
|
|
curatedProfile.hasTattoos = getBoolean(profile.hasTattoos);
|
|
curatedProfile.hasPiercings = getBoolean(profile.hasPiercings);
|
|
|
|
if (argv.resolvePlace) {
|
|
const [placeOfBirth, placeOfResidence] = await Promise.all([
|
|
resolvePlace(profile.birthPlace),
|
|
resolvePlace(profile.residencePlace),
|
|
]);
|
|
|
|
curatedProfile.placeOfBirth = placeOfBirth;
|
|
curatedProfile.placeOfResidence = placeOfResidence;
|
|
}
|
|
|
|
if (!curatedProfile.placeOfBirth && curatedProfile.nationality) {
|
|
const country = await knex('countries')
|
|
.where('nationality', 'ilike', `%${curatedProfile.nationality}%`)
|
|
.orWhere('alpha3', 'ilike', `%${curatedProfile.nationality}%`)
|
|
.orWhere('alpha2', 'ilike', `%${curatedProfile.nationality}%`)
|
|
.orderBy('priority', 'desc')
|
|
.first();
|
|
|
|
if (country) {
|
|
curatedProfile.placeOfBirth = {
|
|
country: country.alpha2,
|
|
};
|
|
}
|
|
}
|
|
|
|
curatedProfile.social = Array.isArray(profile.social)
|
|
? profile.social.map((social) => {
|
|
try {
|
|
const { href } = new URL(social);
|
|
return href;
|
|
} catch (error) {
|
|
logger.warn(`Profile scraper for '${profile.entity.name}' returned invalid social link: ${social}`);
|
|
return null;
|
|
}
|
|
}).filter(Boolean)
|
|
: [];
|
|
|
|
curatedProfile.scenes = toBaseReleases(profile.scenes || profile.releases, profile.entity, actor)
|
|
// attach actor to base scene, in case it was not scraped
|
|
.map((scene) => {
|
|
if (actor && !scene.actors?.find(sceneActor => slugify(sceneActor) === actor.slug || slugify(sceneActor.name) === actor.slug)) {
|
|
return {
|
|
...scene,
|
|
actors: [actor, ...(scene.actors || [])],
|
|
};
|
|
}
|
|
|
|
return scene;
|
|
});
|
|
|
|
if (profile.ethnicity && !curatedProfile.ethnicity) logger.warn(`Unrecognized ethnicity returned by '${profile.entity.name}' scraper: ${profile.ethnicity}`);
|
|
if ((profile.hairColor || profile.hair) && !curatedProfile.hairColor) logger.warn(`Unrecognized hair color returned by '${profile.entity.name}' scraper: ${profile.hairColor || profile.hair}`);
|
|
if (profile.eyes && !curatedProfile.eyes) logger.warn(`Unrecognized eye color returned by '${profile.entity.name}' scraper: ${profile.eyes}`);
|
|
|
|
return curatedProfile;
|
|
} catch (error) {
|
|
logger.error(`Failed to curate '${profile.name}': ${error.message}`);
|
|
|
|
return null;
|
|
}
|
|
}
|
|
|
|
async function fetchProfiles(actorIdsOrNames) {
|
|
return knex('actors_profiles')
|
|
.select(knex.raw('actors_profiles.*, row_to_json(actors) as actor, row_to_json(media) as avatar'))
|
|
.leftJoin('actors', 'actors.id', 'actors_profiles.actor_id')
|
|
.modify((query) => {
|
|
if (actorIdsOrNames) {
|
|
query
|
|
.whereIn('actor_id', actorIdsOrNames.filter(idOrName => typeof idOrName === 'number'))
|
|
.orWhere((builder) => {
|
|
builder
|
|
.whereIn('actors.name', actorIdsOrNames.filter(idOrName => typeof idOrName === 'string'))
|
|
.whereNull('actors.entity_id');
|
|
});
|
|
}
|
|
})
|
|
.leftJoin('media', 'actors_profiles.avatar_media_id', 'media.id');
|
|
}
|
|
|
|
async function interpolateProfiles(actorIdsOrNames) {
|
|
const profiles = await fetchProfiles(actorIdsOrNames);
|
|
|
|
const profilesByActorId = profiles.reduce((acc, profile) => ({
|
|
...acc,
|
|
[profile.actor_id]: [
|
|
...(acc[profile.actor_id] || []),
|
|
profile,
|
|
],
|
|
}), {});
|
|
|
|
logger.info(`Interpolating ${profiles.length} profiles from ${Object.keys(profilesByActorId).length} actors`);
|
|
|
|
const interpolatedProfiles = Object.entries(profilesByActorId).map(([actorId, actorProfiles]) => {
|
|
// group values from each profile
|
|
const valuesByProperty = actorProfiles.reduce((acc, profile) => Object
|
|
.entries(profile)
|
|
.reduce((profileAcc, [property, value]) => ({
|
|
...profileAcc,
|
|
[property]: [
|
|
...(acc[property] || []),
|
|
...(value === null ? [] : Array.from({ length: profile.priority }, () => value)), // multiply by priority, increasing the odds of being the most frequent value
|
|
],
|
|
}), {
|
|
// bundle location values so they can be assessed together, to ensure the most frequent city is in the most frequent state is in most frequent country
|
|
origin: [...acc.origin || [], {
|
|
...(profile.birth_country_alpha2 && { country: profile.birth_country_alpha2 }),
|
|
...(profile.birth_state && { state: profile.birth_state }),
|
|
...(profile.birth_city && { city: profile.birth_city }),
|
|
}].filter(location => Object.keys(location).length > 0),
|
|
residence: [...acc.residence || [], {
|
|
...(profile.residence_country_alpha2 && { country: profile.residence_country_alpha2 }),
|
|
...(profile.residence_state && { state: profile.residence_state }),
|
|
...(profile.residence_city && { city: profile.residence_city }),
|
|
}].filter(location => Object.keys(location).length > 0),
|
|
}), {});
|
|
|
|
const mostFrequentValues = [
|
|
'gender',
|
|
'ethnicity',
|
|
'cup',
|
|
'bust',
|
|
'waist',
|
|
'hip',
|
|
'penis_length',
|
|
'penis_girth',
|
|
'circumcised',
|
|
'natural_boobs',
|
|
'hair_color',
|
|
'eyes',
|
|
'has_tattoos',
|
|
'has_piercings',
|
|
].reduce((acc, property) => ({
|
|
...acc,
|
|
[property]: getMostFrequent(valuesByProperty[property]),
|
|
}), {});
|
|
|
|
const profile = {
|
|
id: actorId,
|
|
...mostFrequentValues,
|
|
};
|
|
|
|
profile.height = getMostFrequent(valuesByProperty.height.filter(height => height > 50 && height < 300)); // remove unlikely values
|
|
|
|
profile.date_of_birth = getMostFrequentDate(valuesByProperty.date_of_birth);
|
|
profile.date_of_death = getMostFrequentDate(valuesByProperty.date_of_death);
|
|
profile.age = getHighest(valuesByProperty.age);
|
|
|
|
// ensure most frequent country, city and state match up
|
|
profile.birth_country_alpha2 = getMostFrequent(valuesByProperty.origin.map(location => location.country));
|
|
const remainingOriginCountries = valuesByProperty.origin.filter(location => location.country === profile.birth_country_alpha2);
|
|
|
|
profile.birth_state = getMostFrequent(remainingOriginCountries.map(location => location.state));
|
|
const remainingOriginStates = remainingOriginCountries.filter(location => !profile.birth_state || location.state === profile.birth_state);
|
|
|
|
profile.birth_city = getMostFrequent(remainingOriginStates.map(location => location.city));
|
|
|
|
profile.residence_country_alpha2 = getMostFrequent(valuesByProperty.residence.map(location => location.country));
|
|
const remainingResidenceCountries = valuesByProperty.residence.filter(location => location.country === profile.residence_country_alpha2);
|
|
|
|
profile.residence_state = getMostFrequent(remainingResidenceCountries.map(location => location.state));
|
|
const remainingResidenceStates = remainingResidenceCountries.filter(location => !profile.residence_state || location.state === profile.residence_state);
|
|
|
|
profile.residence_city = getMostFrequent(remainingResidenceStates.map(location => location.city));
|
|
|
|
profile.weight = getAverage(valuesByProperty.weight);
|
|
|
|
profile.tattoos = getLongest(valuesByProperty.tattoos);
|
|
profile.piercings = getLongest(valuesByProperty.piercings);
|
|
|
|
profile.avatar_media_id = actorProfiles
|
|
.map(actorProfile => actorProfile.avatar)
|
|
.filter(avatar => avatar && (avatar.entropy === null || avatar.entropy > 6))
|
|
.sort((avatarA, avatarB) => avatarB.height - avatarA.height)[0]?.id || null;
|
|
|
|
return profile;
|
|
});
|
|
|
|
const transaction = await knex.transaction();
|
|
|
|
// clear existing interpolated data
|
|
const emptyProfile = Object
|
|
.keys(omit(curateProfileEntry({ id: 1 }), ['id', 'actor_id', 'entity_id', 'url', 'description_hash']))
|
|
.reduce((acc, key) => ({ ...acc, [key]: null }), {});
|
|
|
|
await knex('actors')
|
|
.modify((modifyBuilder) => {
|
|
if (actorIdsOrNames) {
|
|
modifyBuilder
|
|
.whereIn('id', actorIdsOrNames.filter(idOrName => typeof idOrName === 'number'))
|
|
.orWhere((whereBuilder) => {
|
|
whereBuilder
|
|
.whereIn('name', actorIdsOrNames.filter(idOrName => typeof idOrName === 'string'))
|
|
.whereNull('entity_id');
|
|
});
|
|
}
|
|
})
|
|
.update(emptyProfile)
|
|
.transacting(transaction);
|
|
|
|
// insert new interpolated data
|
|
const queries = interpolatedProfiles.map(profile => knex('actors')
|
|
.where('id', profile.id)
|
|
.update(profile)
|
|
.transacting(transaction));
|
|
|
|
await Promise.all(queries)
|
|
.then(transaction.commit)
|
|
.catch(transaction.rollback);
|
|
}
|
|
|
|
async function upsertProfiles(profiles) {
|
|
const newProfileEntries = profiles.filter(profile => !profile.update).map(profile => curateProfileEntry(profile)).filter(Boolean);
|
|
const updatingProfileEntries = profiles.filter(profile => profile.update).map(profile => curateProfileEntry(profile)).filter(Boolean);
|
|
|
|
if (newProfileEntries.length > 0) {
|
|
await bulkInsert('actors_profiles', newProfileEntries);
|
|
|
|
logger.info(`Saved ${newProfileEntries.length} actor profiles`);
|
|
}
|
|
|
|
if (argv.force && updatingProfileEntries.length > 0) {
|
|
const transaction = await knex.transaction();
|
|
const queries = updatingProfileEntries.map(profileEntry => knex('actors_profiles')
|
|
.where('id', profileEntry.id)
|
|
.update(profileEntry)
|
|
.returning(['id', 'actor_id'])
|
|
.transacting(transaction));
|
|
|
|
await Promise.all(queries)
|
|
.then(transaction.commit)
|
|
.catch(transaction.rollback);
|
|
|
|
logger.info(`Updated ${updatingProfileEntries.length} new actor profiles`);
|
|
}
|
|
}
|
|
|
|
async function scrapeProfiles(actor, sources, entitiesBySlug, existingProfilesByActorEntityId) {
|
|
const profiles = Promise.map(sources, async (source) => {
|
|
try {
|
|
// config may group sources to try until success
|
|
return await [].concat(source).reduce(async (outcome, scraperSlug) => outcome.catch(async () => {
|
|
try {
|
|
const entity = entitiesBySlug[scraperSlug] || null;
|
|
|
|
const scraper = scrapers[scraperSlug];
|
|
const layoutScraper = scraper?.[entity.parameters?.layout] || scraper;
|
|
|
|
const context = {
|
|
...entity,
|
|
// legacy
|
|
site: entity,
|
|
network: entity?.parent,
|
|
entity,
|
|
scraper: scraperSlug,
|
|
};
|
|
|
|
const label = context.entity?.name;
|
|
|
|
if (!layoutScraper?.fetchProfile) {
|
|
logger.warn(`No profile profile scraper available for ${scraperSlug}`);
|
|
throw new Error(`No profile profile scraper available for ${scraperSlug}`);
|
|
}
|
|
|
|
if (!context.entity) {
|
|
logger.warn(`No entity found for ${scraperSlug}`);
|
|
throw new Error(`No entity found for ${scraperSlug}`);
|
|
}
|
|
|
|
const existingProfile = existingProfilesByActorEntityId[actor.id]?.[context.entity?.id || null];
|
|
|
|
if (existingProfile && !argv.force) {
|
|
logger.verbose(`Found existing profile for '${actor.name}' on '${label}', use --force to scrape again`);
|
|
|
|
return null;
|
|
}
|
|
|
|
logger.verbose(`Searching profile for '${actor.name}' on '${label}'`);
|
|
|
|
const profile = await layoutScraper.fetchProfile(curateActor({
|
|
...existingProfile,
|
|
...actor,
|
|
}), context, include);
|
|
|
|
if (!profile || typeof profile === 'number') { // scraper returns HTTP code on request failure
|
|
logger.verbose(`Profile for '${actor.name}' not available on ${label}, scraper returned ${profile}`);
|
|
throw Object.assign(new Error(`Profile for '${actor.name}' not available on ${label}`), { code: 'PROFILE_NOT_AVAILABLE' });
|
|
}
|
|
|
|
logger.verbose(`Found profile for '${actor.name}' on '${label}'`);
|
|
|
|
return await curateProfile({
|
|
...actor,
|
|
...profile,
|
|
entity,
|
|
update: existingProfile?.id || false,
|
|
}, actor);
|
|
} catch (error) {
|
|
if (error.code !== 'PROFILE_NOT_AVAILABLE') {
|
|
logger.error(`Failed to fetch profile for '${actor.name}' from '${scraperSlug}': ${error.message}`);
|
|
}
|
|
|
|
// throw error to try next source
|
|
throw error;
|
|
}
|
|
}), Promise.reject(new Error()));
|
|
} catch (error) {
|
|
console.log(error);
|
|
|
|
if (error.code !== 'PROFILE_NOT_AVAILABLE') {
|
|
logger.error(`Failed to fetch profile for '${actor.name}': ${error.message}`);
|
|
}
|
|
}
|
|
|
|
return null;
|
|
});
|
|
|
|
return profiles.filter(Boolean);
|
|
}
|
|
|
|
async function getActorNames(actorNames) {
|
|
if (actorNames.length > 0) {
|
|
return actorNames;
|
|
}
|
|
|
|
const actorsWithoutProfiles = await knex.raw(`
|
|
SELECT actors.name
|
|
FROM actors
|
|
WHERE NOT EXISTS (
|
|
SELECT *
|
|
FROM actors_profiles
|
|
WHERE actors_profiles.actor_id = actors.id
|
|
AND actors_profiles.updated_at <= (?)
|
|
)
|
|
`, [argv.actorsUpdate || new Date()]);
|
|
|
|
return actorsWithoutProfiles.rows.map(actor => actor.name);
|
|
}
|
|
|
|
async function storeProfiles(profiles) {
|
|
const profilesWithAvatarIds = await associateAvatars(profiles);
|
|
const actorIds = Array.from(new Set(profiles.map(profile => profile.id)));
|
|
|
|
await upsertProfiles(profilesWithAvatarIds);
|
|
await interpolateProfiles(actorIds);
|
|
}
|
|
|
|
async function scrapeActors(argNames) {
|
|
const actorNames = await getActorNames(argNames);
|
|
const baseActors = toBaseActors(actorNames);
|
|
|
|
logger.info(`Scraping profiles for ${actorNames.length} actors`);
|
|
|
|
const sources = argv.profileSources || config.profiles || Object.keys(scrapers.actors);
|
|
const entitySlugs = sources.flat();
|
|
|
|
const [entities, existingActorEntries] = await Promise.all([
|
|
knex('entities')
|
|
.select(knex.raw('entities.*, row_to_json(parents) as parent, json_agg(children) as children'))
|
|
.whereIn('entities.slug', entitySlugs)
|
|
.whereIn('entities.type', ['network', 'channel'])
|
|
.leftJoin('entities as parents', 'parents.id', 'entities.parent_id')
|
|
.leftJoin('entities as children', 'children.parent_id', 'entities.id')
|
|
.orderBy('entities.type')
|
|
.groupBy('entities.id', 'parents.id'),
|
|
knex('actors')
|
|
.select(['id', 'name', 'slug', 'entry_id'])
|
|
.whereIn('slug', baseActors.map(baseActor => baseActor.slug))
|
|
.whereNull('alias_for'),
|
|
]);
|
|
|
|
const entitiesBySlug = entities.reduce((acc, entity) => ({ ...acc, [entity.slug]: acc[entity.slug] || entity }), {});
|
|
|
|
const existingActorEntriesBySlugAndEntryId = existingActorEntries.reduce((acc, actorEntry) => ({
|
|
...acc,
|
|
[actorEntry.slug]: {
|
|
...acc[actorEntry.slug],
|
|
[actorEntry.entryId || null]: actorEntry,
|
|
},
|
|
}), {});
|
|
|
|
const newBaseActors = baseActors.filter(baseActor => !existingActorEntriesBySlugAndEntryId[baseActor.slug]?.[baseActor.entryId]);
|
|
|
|
const [batchId] = newBaseActors.length > 0 ? await knex('batches').insert({ comment: null }).returning('id') : [null];
|
|
const curatedActorEntries = batchId && curateActorEntries(newBaseActors, batchId);
|
|
|
|
// TODO: associate entity when entry ID is provided
|
|
|
|
const newActorEntries = batchId && await bulkInsert('actors', curatedActorEntries);
|
|
|
|
const actors = existingActorEntries.concat(Array.isArray(newActorEntries) ? newActorEntries : []);
|
|
|
|
const existingProfiles = await knex('actors_profiles')
|
|
.select(knex.raw('actors_profiles.*, row_to_json(avatars) as avatar'))
|
|
.whereIn('actor_id', actors.map(actor => actor.id))
|
|
.leftJoin('media as avatars', 'avatars.id', 'actors_profiles.avatar_media_id');
|
|
|
|
const existingProfilesByActorEntityId = existingProfiles.reduce((acc, profile) => ({
|
|
...acc,
|
|
[profile.actor_id]: {
|
|
...acc[profile.actor_id],
|
|
[profile.entity_id]: profile,
|
|
},
|
|
}), {});
|
|
|
|
const profilesPerActor = await Promise.map(
|
|
actors,
|
|
async actor => scrapeProfiles(actor, sources, entitiesBySlug, existingProfilesByActorEntityId),
|
|
{ concurrency: 10 },
|
|
);
|
|
|
|
const profiles = profilesPerActor.flat().filter(Boolean);
|
|
|
|
logger.info(`Scraped ${profiles.length} profiles`);
|
|
|
|
if (argv.report) {
|
|
console.log(util.inspect(profiles, { depth: Infinity, colors: true }));
|
|
}
|
|
|
|
if (argv.save) {
|
|
await storeProfiles(profiles);
|
|
}
|
|
|
|
return profiles;
|
|
}
|
|
|
|
async function getOrCreateActors(baseActors, batchId) {
|
|
// WHERE IN causes stack depth error and performance issues with a large amount of values, no knex VALUES helper available
|
|
const actorValues = baseActors.map(actor => knex.raw('(:slug, :entityId)', { slug: actor.slug, entityId: actor.entity.id })).join(', ');
|
|
|
|
const existingActors = await knex
|
|
.select('actors.*')
|
|
.from(knex.raw(`actors, (VALUES ${actorValues}) AS base_actors (slug, entity_id)`))
|
|
.whereRaw('actors.slug = base_actors.slug AND actors.entity_id IS NULL')
|
|
.orWhereRaw('actors.slug = base_actors.slug AND actors.entity_id = base_actors.entity_id');
|
|
|
|
// const existingActorSlugs = new Set(existingActors.map(actor => actor.slug));
|
|
const existingActorSlugs = existingActors.reduce((acc, actor) => ({
|
|
...acc,
|
|
[actor.entity_id]: {
|
|
...acc[actor.entity_id],
|
|
[actor.slug]: true,
|
|
},
|
|
}), {});
|
|
|
|
const uniqueBaseActors = baseActors.filter(baseActor => !existingActorSlugs[baseActor.entity.id]?.[baseActor.slug] && !existingActorSlugs.null?.[baseActor.slug]);
|
|
|
|
const curatedActorEntries = curateActorEntries(uniqueBaseActors, batchId);
|
|
const newActors = await bulkInsert('actors', curatedActorEntries);
|
|
|
|
const newActorIdsByEntityIdAndSlug = newActors.reduce((acc, actor) => ({
|
|
...acc,
|
|
[actor.entity_id]: {
|
|
...acc[actor.entity_id],
|
|
[actor.slug]: actor.id,
|
|
},
|
|
}), {});
|
|
|
|
const newActorProfiles = await Promise.all(baseActors
|
|
.filter(actor => actor.hasProfile)
|
|
.map(actor => ({
|
|
...actor,
|
|
id: newActorIdsByEntityIdAndSlug[actor.entity?.id]?.[actor.slug] || newActorIdsByEntityIdAndSlug.null?.[actor.slug],
|
|
}))
|
|
.filter(actor => !!actor.id)
|
|
.map(actor => curateProfile(actor)));
|
|
|
|
await storeProfiles(newActorProfiles);
|
|
|
|
if (Array.isArray(newActors)) {
|
|
return newActors.concat(existingActors);
|
|
}
|
|
|
|
return existingActors;
|
|
}
|
|
|
|
async function associateActors(releases, batchId) {
|
|
const baseActorsByReleaseId = releases.reduce((acc, release) => {
|
|
if (release.actors) {
|
|
acc[release.id] = toBaseActors(release.actors, release);
|
|
}
|
|
|
|
return acc;
|
|
}, {});
|
|
|
|
const baseActors = Object.values(baseActorsByReleaseId).flat();
|
|
|
|
if (baseActors.length === 0) {
|
|
return [];
|
|
}
|
|
|
|
const baseActorsBySlug = baseActors.reduce((acc, baseActor) => ({
|
|
...acc,
|
|
[baseActor.slug]: baseActor,
|
|
}), {});
|
|
|
|
const uniqueBaseActors = Object.values(baseActorsBySlug);
|
|
|
|
const actors = await getOrCreateActors(uniqueBaseActors, batchId);
|
|
|
|
const actorIdsBySlug = actors.reduce((acc, actor) => ({
|
|
...acc,
|
|
[actor.slug]: actor.alias_for || actor.id,
|
|
}), {});
|
|
|
|
const releaseActorAssociations = Object.entries(baseActorsByReleaseId)
|
|
.map(([releaseId, releaseActors]) => releaseActors
|
|
.map(releaseActor => ({
|
|
release_id: releaseId,
|
|
actor_id: actorIdsBySlug[releaseActor.slug],
|
|
})))
|
|
.flat();
|
|
|
|
await bulkInsert('releases_actors', releaseActorAssociations, false);
|
|
|
|
logger.verbose(`Associated ${releaseActorAssociations.length} actors to ${releases.length} scenes`);
|
|
|
|
return actors;
|
|
}
|
|
|
|
async function fetchActor(actorId) {
|
|
const actor = await knex('actors')
|
|
.select(knex.raw(`
|
|
actors.*,
|
|
row_to_json(entities) as entity,
|
|
row_to_json(actor_alias) as alias,
|
|
row_to_json(birth_country) as birth_country,
|
|
row_to_json(residence_country) as residence_country,
|
|
row_to_json(media) as avatar,
|
|
json_agg(actors_profiles) as profiles
|
|
`))
|
|
.modify((queryBuilder) => {
|
|
if (Number.isNaN(Number(actorId))) {
|
|
queryBuilder.where('actors.slug', actorId);
|
|
return;
|
|
}
|
|
|
|
queryBuilder.where('actors.id', actorId);
|
|
})
|
|
.leftJoin('actors as actor_alias', 'actor_alias.id', 'actors.alias_for')
|
|
.leftJoin('actors_profiles', 'actors.id', 'actors_profiles.actor_id')
|
|
.leftJoin('entities', 'entities.id', 'actors.entity_id')
|
|
.leftJoin('countries as birth_country', 'birth_country.alpha2', 'actors.birth_country_alpha2')
|
|
.leftJoin('countries as residence_country', 'residence_country.alpha2', 'actors.residence_country_alpha2')
|
|
.leftJoin('media', 'media.id', 'actors.avatar_media_id')
|
|
.groupBy('actors.id', 'entities.id', 'actor_alias.id', 'birth_country.alpha2', 'residence_country.alpha2', 'media.id')
|
|
.first();
|
|
|
|
return curateActor(actor, true);
|
|
}
|
|
|
|
async function searchActors(query) {
|
|
const actors = await knex
|
|
.select('*')
|
|
.from(knex.raw('search_actors(?) as actors', [query]))
|
|
.limit(100);
|
|
|
|
return actors.map(actor => curateActor(actor));
|
|
}
|
|
|
|
async function flushProfiles(actorIdsOrNames) {
|
|
const profiles = await fetchProfiles(actorIdsOrNames);
|
|
const actorNames = Array.from(new Set(profiles.map(profile => profile.actor.name)));
|
|
|
|
const deleteCount = await knex('actors_profiles')
|
|
.whereIn('id', profiles.map(profile => profile.id))
|
|
.delete();
|
|
|
|
await interpolateProfiles(actorIdsOrNames);
|
|
await flushOrphanedMedia(); // don't flush until main avatar is detached by re-interpolating
|
|
|
|
if (actorNames.length > 20) {
|
|
logger.info(`Removed ${deleteCount} profiles for ${actorNames.length} actors`);
|
|
return;
|
|
}
|
|
|
|
if (deleteCount > 0) {
|
|
logger.info(`Removed ${deleteCount} profiles for ${actorNames.join(', ')}`);
|
|
return;
|
|
}
|
|
|
|
logger.info(`Removed ${deleteCount} profiles`);
|
|
}
|
|
|
|
async function flushActors(actorIdsOrNames) {
|
|
const actors = await knex('actors')
|
|
.whereIn('id', actorIdsOrNames.filter(idOrName => typeof idOrName === 'number'))
|
|
.orWhere((builder) => {
|
|
builder
|
|
.whereIn('name', actorIdsOrNames.filter(idOrName => typeof idOrName === 'string'))
|
|
.whereNull('entity_id');
|
|
});
|
|
|
|
const actorIds = actors.map(actor => actor.id);
|
|
|
|
const sceneIds = await knex('releases_actors')
|
|
.select('releases.id')
|
|
.whereIn('actor_id', actorIds)
|
|
.leftJoin('releases', 'releases.id', 'releases_actors.release_id')
|
|
.pluck('id');
|
|
|
|
const [deletedScenesCount, deletedActorsCount] = await Promise.all([
|
|
deleteScenes(sceneIds),
|
|
knex('actors')
|
|
.whereIn('id', actorIds)
|
|
.delete(),
|
|
]);
|
|
|
|
await flushOrphanedMedia();
|
|
|
|
logger.info(`Removed ${deletedActorsCount} actors with ${deletedScenesCount} scenes`);
|
|
}
|
|
|
|
module.exports = {
|
|
associateActors,
|
|
fetchActor,
|
|
flushActors,
|
|
flushProfiles,
|
|
interpolateProfiles,
|
|
scrapeActors,
|
|
searchActors,
|
|
toBaseActors,
|
|
};
|