Improved duplicate filtering, now also applying to upcoming updates. Updated Gamma fetchLatest method parameters. Added shortcut for SFW-mode.

This commit is contained in:
ThePendulum 2020-03-28 04:37:04 +01:00
parent 95d115b585
commit 6d9f96c5d5
10 changed files with 112 additions and 40 deletions

View File

@ -9,6 +9,7 @@
<img <img
v-if="tag.poster" v-if="tag.poster"
:src="sfw ? `/img/${tag.poster.sfw.thumbnail}` : `/img/${tag.poster.thumbnail}`" :src="sfw ? `/img/${tag.poster.sfw.thumbnail}` : `/img/${tag.poster.thumbnail}`"
:title="sfw ? tag.poster.sfw.comment : tag.poster.comment"
:alt="tag.name" :alt="tag.name"
class="poster" class="poster"
> >

View File

@ -5,6 +5,8 @@ import VTooltip from 'v-tooltip';
import router from './router'; import router from './router';
import initStore from './store'; import initStore from './store';
import initUiObservers from './ui/observers';
import '../css/style.scss'; import '../css/style.scss';
import Container from '../components/container/container.vue'; import Container from '../components/container/container.vue';
@ -13,6 +15,8 @@ import Icon from '../components/icon/icon.vue';
function init() { function init() {
const store = initStore(router); const store = initStore(router);
initUiObservers(store, router);
if (window.env.sfw) { if (window.env.sfw) {
store.dispatch('setSfw', true); store.dispatch('setSfw', true);
} }

View File

@ -122,8 +122,10 @@ function initTagsActions(store, _router) {
poster: tagsPosterByTagId { poster: tagsPosterByTagId {
media { media {
thumbnail thumbnail
comment
sfw: sfwMedia { sfw: sfwMedia {
thumbnail thumbnail
comment
} }
} }
} }

30
assets/js/ui/observers.js Normal file
View File

@ -0,0 +1,30 @@
function initUiObservers(store, _router) {
let keyHistory = [];
document.addEventListener('keypress', (event) => {
if (event.target.tagName === 'INPUT') {
return;
}
keyHistory = keyHistory
.concat({
key: event.key,
time: new Date().getTime(),
})
.slice(-3);
if (keyHistory.length === 3
&& keyHistory.every(input => input.key === 's')
&& new Date().getTime() - keyHistory[0].time < 1000
) {
store.dispatch('setSfw', false);
return;
}
if (event.key === 's') {
store.dispatch('setSfw', true);
}
});
}
export default initUiObservers;

View File

@ -15,7 +15,7 @@ const tagPosters = [
['brunette', 0, 'Nicole Black in GIO971 for LegalPorno'], ['brunette', 0, 'Nicole Black in GIO971 for LegalPorno'],
['bukkake', 'poster', 'Mia Malkova in "Facialized 2" for HardX'], ['bukkake', 'poster', 'Mia Malkova in "Facialized 2" for HardX'],
['caucasian', 0, 'Remy Lacroix for HardX'], ['caucasian', 0, 'Remy Lacroix for HardX'],
['creampie', 'poster'], ['creampie', 'poster', 'ALina Lopez in "Making Yourself Unforgettable" for Blacked'],
['da-tp', 0, 'Natasha Teen in LegalPorno SZ2164'], ['da-tp', 0, 'Natasha Teen in LegalPorno SZ2164'],
['deepthroat', 0, 'Chanel Grey in "Deepthroating Is Fun" for Throated'], ['deepthroat', 0, 'Chanel Grey in "Deepthroating Is Fun" for Throated'],
['double-anal', 7, 'Adriana Chechik in "DP Masters 6" for Jules Jordan'], ['double-anal', 7, 'Adriana Chechik in "DP Masters 6" for Jules Jordan'],
@ -34,9 +34,9 @@ const tagPosters = [
['maid', 0, 'Whitney Wright in "Dredd Up Your Ass 2" for Jules Jordan'], ['maid', 0, 'Whitney Wright in "Dredd Up Your Ass 2" for Jules Jordan'],
['milf', 0, 'Olivia Austin in "Dredd 3" for Jules Jordan'], ['milf', 0, 'Olivia Austin in "Dredd 3" for Jules Jordan'],
['mff', 0, 'Madison Ivy, Adriana Chechik and Keiran Lee in "Day With A Pornstar" for Brazzers'], ['mff', 0, 'Madison Ivy, Adriana Chechik and Keiran Lee in "Day With A Pornstar" for Brazzers'],
['mfm', 5, 'Vina Sky, Jules Jordan and Steve Holmes in "Slut Puppies 15" for Jules Jordan'], ['mfm', 5, 'Vina Sky in "Slut Puppies 15" for Jules Jordan'],
['cum-in-mouth', 1, 'Keisha Grey in Brazzers House'], ['cum-in-mouth', 1, 'Keisha Grey in Brazzers House'],
['orgy', 1, 'Megan Rain (DP), Morgan Lee (anal), Jessa Rhodes, Melissa Moore, Kimmy Granger, Mick Blue and Markus Dupree in "Orgy Masters 8" for Jules Jordan'], ['orgy', 1, 'Megan Rain (DP), Morgan Lee (anal), Jessa Rhodes, Melissa Moore and Kimmy Granger in "Orgy Masters 8" for Jules Jordan'],
['pussy-eating', 0, 'Elena Kosha and Ivy Wolfe in "Bare" for Jules Jordan'], ['pussy-eating', 0, 'Elena Kosha and Ivy Wolfe in "Bare" for Jules Jordan'],
['redhead', 0, 'Penny Pax in "The Submission of Emma Marx: Boundaries" for New Sensations'], ['redhead', 0, 'Penny Pax in "The Submission of Emma Marx: Boundaries" for New Sensations'],
['schoolgirl', 1, 'Eliza Ibarra for Brazzers'], ['schoolgirl', 1, 'Eliza Ibarra for Brazzers'],
@ -96,7 +96,7 @@ const tagPhotos = [
['gaping', 2, 'Alex Grey in "DP Masters 5" for Jules Jordan'], ['gaping', 2, 'Alex Grey in "DP Masters 5" for Jules Jordan'],
// ['mfm', 0, 'Vina Sky in "Jules Jordan\'s Three Ways" for Jules Jordan'], // ['mfm', 0, 'Vina Sky in "Jules Jordan\'s Three Ways" for Jules Jordan'],
['mfm', 1, 'Jynx Maze in "Don\'t Make Me Beg 4" for Evil Angel'], ['mfm', 1, 'Jynx Maze in "Don\'t Make Me Beg 4" for Evil Angel'],
['orgy', 'poster', 'Zoey Mornoe (DP), Jillian Janson (sex), Frida Sante, Katerina Kay, Natasha Starr, Mick Blue and Ramon Nomar in "Orgy Masters 6" for Jules Jordan'], ['orgy', 'poster', 'Zoey Mornoe (DP), Jillian Janson (sex), Frida Sante, Katerina Kay and Natasha Starr in "Orgy Masters 6" for Jules Jordan'],
['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

@ -112,6 +112,8 @@ async function scrapeRelease(baseRelease, sites, type = 'scene') {
} }
try { try {
logger.verbose(`Fetching ${type} ${baseRelease.url}`);
const scrapedRelease = type === 'scene' const scrapedRelease = type === 'scene'
? await scraper.fetchScene(baseRelease.url, site, baseRelease, null, include) ? await scraper.fetchScene(baseRelease.url, site, baseRelease, null, include)
: await scraper.fetchMovie(baseRelease.url, site, baseRelease, null, include); : await scraper.fetchMovie(baseRelease.url, site, baseRelease, null, include);

View File

@ -136,7 +136,7 @@ async function scrapeApiReleases(json, site) {
]; ];
} }
release.movie = `${site.url}/en/movie/${scene.url_movie_title}/${scene.movie_id}`; // release.movie = `${site.url}/en/movie/${scene.url_movie_title}/${scene.movie_id}`;
return release; return release;
}).filter(Boolean); }).filter(Boolean);
@ -408,7 +408,7 @@ async function fetchApiCredentials(referer, site) {
return getApiUrl(appId, apiKey); return getApiUrl(appId, apiKey);
} }
async function fetchApiLatest(site, page = 1, upcoming = false) { async function fetchApiLatest(site, page = 1, preData, include, upcoming = false) {
const referer = site.parameters?.referer || `${site.parameters?.networkReferer ? site.network.url : site.url}/en/videos`; const referer = site.parameters?.referer || `${site.parameters?.networkReferer ? site.network.url : site.url}/en/videos`;
const { apiUrl } = await fetchApiCredentials(referer, site); const { apiUrl } = await fetchApiCredentials(referer, site);
@ -433,8 +433,8 @@ async function fetchApiLatest(site, page = 1, upcoming = false) {
return []; return [];
} }
async function fetchApiUpcoming(site) { async function fetchApiUpcoming(site, page = 1, preData, include) {
return fetchApiLatest(site, 1, true); return fetchApiLatest(site, page, preData, include, true);
} }
function getLatestUrl(site, page) { function getLatestUrl(site, page) {

View File

@ -110,9 +110,27 @@ function attachReleaseIds(releases, storedReleases) {
return releasesWithId; return releasesWithId;
} }
async function extractUniqueReleases(releases) { function filterInternalDuplicateReleases(releases) {
const releasesBySiteIdAndEntryId = releases.reduce((acc, release) => {
if (!acc[release.site.id]) {
acc[release.site.id] = {};
}
acc[release.site.id][release.entryId] = release;
return acc;
}, {});
return Object.values(releasesBySiteIdAndEntryId)
.map(siteReleases => Object.values(siteReleases))
.flat();
}
async function filterDuplicateReleases(releases) {
const internalUniqueReleases = filterInternalDuplicateReleases(releases);
const duplicateReleaseEntries = await knex('releases') const duplicateReleaseEntries = await knex('releases')
.whereIn(['entry_id', 'site_id'], releases.map(release => [release.entryId, release.site.id])); .whereIn(['entry_id', 'site_id'], internalUniqueReleases.map(release => [release.entryId, release.site.id]));
const duplicateReleasesBySiteIdAndEntryId = duplicateReleaseEntries.reduce((acc, release) => { const duplicateReleasesBySiteIdAndEntryId = duplicateReleaseEntries.reduce((acc, release) => {
if (!acc[release.site_id]) acc[release.site_id] = {}; if (!acc[release.site_id]) acc[release.site_id] = {};
@ -121,8 +139,8 @@ async function extractUniqueReleases(releases) {
return acc; return acc;
}, {}); }, {});
const duplicateReleases = releases.filter(release => duplicateReleasesBySiteIdAndEntryId[release.site.id]?.[release.entryId]); const duplicateReleases = internalUniqueReleases.filter(release => duplicateReleasesBySiteIdAndEntryId[release.site.id]?.[release.entryId]);
const uniqueReleases = releases.filter(release => !duplicateReleasesBySiteIdAndEntryId[release.site.id]?.[release.entryId]); const uniqueReleases = internalUniqueReleases.filter(release => !duplicateReleasesBySiteIdAndEntryId[release.site.id]?.[release.entryId]);
return { return {
uniqueReleases, uniqueReleases,
@ -138,21 +156,25 @@ async function storeReleases(releases) {
const releasesWithStudios = await attachStudios(releasesWithSites); const releasesWithStudios = await attachStudios(releasesWithSites);
// uniqueness is site ID + entry ID, filter uniques after adding sites // uniqueness is site ID + entry ID, filter uniques after adding sites
const { uniqueReleases, duplicateReleaseEntries } = await extractUniqueReleases(releasesWithStudios); const { uniqueReleases, duplicateReleases, duplicateReleaseEntries } = await filterDuplicateReleases(releasesWithStudios);
const curatedNewReleaseEntries = uniqueReleases.map(release => curateReleaseEntry(release, batchId)); const curatedNewReleaseEntries = uniqueReleases.map(release => curateReleaseEntry(release, batchId));
// console.log(curatedNewReleaseEntries);
const storedReleases = await knex('releases').insert(curatedNewReleaseEntries).returning('*'); const storedReleases = await knex('releases').insert(curatedNewReleaseEntries).returning('*');
// TODO: update duplicate releases // TODO: update duplicate releases
const storedReleaseEntries = Array.isArray(storedReleases) ? storedReleases : []; const storedReleaseEntries = Array.isArray(storedReleases) ? storedReleases : [];
const releasesWithId = attachReleaseIds(releases, [].concat(storedReleaseEntries, duplicateReleaseEntries)); const releasesWithId = attachReleaseIds([].concat(uniqueReleases, duplicateReleases), [].concat(storedReleaseEntries, duplicateReleaseEntries));
await Promise.all([ await Promise.all([
associateTags(releasesWithId), associateTags(releasesWithId),
associateActors(releasesWithId), associateActors(releasesWithId),
]); ]);
logger.info(`Stored ${storedReleaseEntries.length} releases`);
return releasesWithId; return releasesWithId;
} }

View File

@ -54,7 +54,7 @@ function buildReleaseTagAssociations(releases, tagIdsBySlug, siteTagIdsBySiteId)
? releaseTags // obsolete scraper returned pre-matched tags ? releaseTags // obsolete scraper returned pre-matched tags
: releaseTags.map(tag => tagIdsBySlug[slugify(tag)]); : releaseTags.map(tag => tagIdsBySlug[slugify(tag)]);
return [...new Set( const tags = [...new Set(
// filter duplicates and empties // filter duplicates and empties
releaseTagIds releaseTagIds
.concat(siteTagIds) .concat(siteTagIds)
@ -64,14 +64,17 @@ function buildReleaseTagAssociations(releases, tagIdsBySlug, siteTagIdsBySiteId)
release_id: release.id, release_id: release.id,
tag_id: tagId, tag_id: tagId,
})); }));
return tags;
}) })
.flat(); .flat();
return tagAssociations; return tagAssociations;
} }
async function extractUniqueAssociations(tagAssociations) { async function filterUniqueAssociations(tagAssociations) {
const duplicateAssociations = await knex('releases_tags').whereIn(['release_id', 'tag_id'], tagAssociations.map(association => [association.release_id, association.tag_id])); const duplicateAssociations = await knex('releases_tags')
.whereIn(['release_id', 'tag_id'], tagAssociations.map(association => [association.release_id, association.tag_id]));
const duplicateAssociationsByReleaseIdAndTagId = duplicateAssociations.reduce((acc, association) => { const duplicateAssociationsByReleaseIdAndTagId = duplicateAssociations.reduce((acc, association) => {
if (!acc[association.release_id]) { if (!acc[association.release_id]) {
@ -94,7 +97,7 @@ async function associateTags(releases) {
const siteTagIdsBySiteId = await getSiteTags(releases); const siteTagIdsBySiteId = await getSiteTags(releases);
const tagAssociations = buildReleaseTagAssociations(releases, tagIdsBySlug, siteTagIdsBySiteId); const tagAssociations = buildReleaseTagAssociations(releases, tagIdsBySlug, siteTagIdsBySiteId);
const uniqueAssociations = await extractUniqueAssociations(tagAssociations); const uniqueAssociations = await filterUniqueAssociations(tagAssociations);
await knex('releases_tags').insert(uniqueAssociations); await knex('releases_tags').insert(uniqueAssociations);
} }

View File

@ -25,7 +25,7 @@ const afterDate = (() => {
.toDate(); .toDate();
})(); })();
async function extractUniqueReleases(latestReleases, accReleases) { async function filterUniqueReleases(latestReleases, accReleases) {
const latestReleaseIdentifiers = latestReleases const latestReleaseIdentifiers = latestReleases
.map(release => [release.site.id, release.entryId]); .map(release => [release.site.id, release.entryId]);
@ -75,13 +75,11 @@ function needNextPage(uniqueReleases, pageAccReleases) {
return pageAccReleases.length <= argv.nullDateLimit; return pageAccReleases.length <= argv.nullDateLimit;
} }
async function scrapeLatestReleases(scraper, site, preData) { async function scrapeReleases(scraper, site, preData, upcoming = false) {
if (!scraper.fetchLatest) {
return [];
}
const scrapePage = async (page = 1, accReleases = []) => { const scrapePage = async (page = 1, accReleases = []) => {
const latestReleases = await scraper.fetchLatest(site, page, preData, include); const latestReleases = upcoming
? await scraper.fetchUpcoming(site, page, preData, include)
: await scraper.fetchLatest(site, page, preData, include);
if (!Array.isArray(latestReleases)) { if (!Array.isArray(latestReleases)) {
// scraper is unable to fetch the releases and returned a HTTP code or null // scraper is unable to fetch the releases and returned a HTTP code or null
@ -98,11 +96,11 @@ async function scrapeLatestReleases(scraper, site, preData) {
const uniqueReleases = argv.redownload const uniqueReleases = argv.redownload
? latestReleasesWithSite ? latestReleasesWithSite
: await extractUniqueReleases(latestReleasesWithSite, accReleases); : await filterUniqueReleases(latestReleasesWithSite, accReleases);
const pageAccReleases = accReleases.concat(uniqueReleases); const pageAccReleases = accReleases.concat(uniqueReleases);
logger.verbose(`Scraped '${site.name}' (${site.network.name}) page ${page}, found ${uniqueReleases.length} unique releases`); logger.verbose(`Scraped '${site.name}' (${site.network.name}) ${upcoming ? 'upcoming' : 'latest'} page ${page}, found ${uniqueReleases.length} unique updates`);
if (needNextPage(uniqueReleases, pageAccReleases)) { if (needNextPage(uniqueReleases, pageAccReleases)) {
return scrapePage(page + 1, pageAccReleases); return scrapePage(page + 1, pageAccReleases);
@ -111,37 +109,45 @@ async function scrapeLatestReleases(scraper, site, preData) {
return pageAccReleases; return pageAccReleases;
}; };
const releases = await scrapePage(1, []); const rawReleases = await scrapePage(1, []);
const releases = upcoming
? rawReleases.map(rawRelease => ({ ...rawRelease, upcoming: true }))
: rawReleases;
if (argv.last) { if (argv.last) {
return releases.slice(0, argv.last); return releases.slice(0, argv.last);
} }
if (releases.every(release => release.date)) { if (releases.every(release => release.date)) {
return releases return releases.filter(release => moment(release.date).isAfter(afterDate));
.filter(release => moment(release.date).isAfter(afterDate));
} }
return releases.slice(0, argv.nullDateLimit); return releases.slice(0, argv.nullDateLimit);
} }
async function scrapeLatestReleases(scraper, site, preData) {
if (!scraper.fetchLatest) {
return [];
}
try {
return await scrapeReleases(scraper, site, preData, false);
} catch (error) {
logger.warn(`Failed to scrape latest updates for '${site.slug}' (${site.network.slug})`);
}
return [];
}
async function scrapeUpcomingReleases(scraper, site, preData) { async function scrapeUpcomingReleases(scraper, site, preData) {
if (!scraper.fetchUpcoming) { if (!scraper.fetchUpcoming) {
return []; return [];
} }
try { try {
const upcomingReleases = await scraper.fetchUpcoming(site, 1, preData, include); return await scrapeReleases(scraper, site, preData, true);
if (upcomingReleases) {
return upcomingReleases.map(release => ({
...release,
site,
upcoming: true,
}));
}
} catch (error) { } catch (error) {
logger.warn(`Failed to scrape upcoming releases for '${site.slug}' (${site.network.slug})`); logger.warn(`Failed to scrape upcoming updates for '${site.slug}' (${site.network.slug})`);
} }
return []; return [];
@ -157,6 +163,8 @@ async function scrapeSiteReleases(scraper, site, preData) {
: [], : [],
]); ]);
logger.info(`Fetching ${latestReleases.length} latest and ${upcomingReleases.length} upcoming updates for '${site.name}' (${site.network.name})`);
return [...latestReleases, ...upcomingReleases]; return [...latestReleases, ...upcomingReleases];
} }