Providing duplicate releases in predata. Using duplicates for filtering scenes without channel in Hush scraper.

This commit is contained in:
DebaucheryLibrarian
2020-08-23 02:43:10 +02:00
parent 3a5ea3dd9a
commit 278b74e78c
3 changed files with 34 additions and 22 deletions

View File

@@ -107,8 +107,6 @@ function scrapeAllT1(scenes, site, accNetworkReleases) {
// release.entryId = q('.img-div img', 'id')?.match(/set-target-(\d+)/)[1];
release.entryId = deriveEntryId(release);
console.log(site.name, accNetworkReleases.map(accRelease => accRelease.entryId), release.entryId, accNetworkReleases?.map(accRelease => accRelease.entryId).includes(release.entryId));
if (site.parameters?.accFilter && accNetworkReleases?.map(accRelease => accRelease.entryId).includes(release.entryId)) {
// filter out releases that were already scraped from a categorized site, requires sequential site scraping
return null;
@@ -360,7 +358,7 @@ function scrapeProfileTour({ el, qu }, site) {
return profile;
}
async function fetchLatest(site, page = 1, include, { accNetworkReleases }) {
async function fetchLatest(site, page = 1, include, { uniqueReleases, duplicateReleases }) {
const url = (site.parameters?.latest && util.format(site.parameters.latest, page))
|| (site.parameters?.t1 && `${site.url}/t1/categories/movies_${page}_d.html`)
|| `${site.url}/categories/movies_${page}_d.html`;
@@ -368,10 +366,10 @@ async function fetchLatest(site, page = 1, include, { accNetworkReleases }) {
const res = await geta(url, '.modelfeature, .item-video, .updateItem');
if (!res.ok) return res.status;
if (site.parameters?.t1) return scrapeAllT1(res.items, site, accNetworkReleases);
if (site.parameters?.tour) return scrapeAllTour(res.items, site, accNetworkReleases);
if (site.parameters?.t1) return scrapeAllT1(res.items, site, [...uniqueReleases, ...duplicateReleases]);
if (site.parameters?.tour) return scrapeAllTour(res.items, site);
return scrapeAll(res.items, site, accNetworkReleases);
return scrapeAll(res.items, site, uniqueReleases);
}
async function fetchScene(url, site, baseRelease, include, beforeFetchLatest) {