Closing JSDOM window after deep scrape in an attempt to save memory. Reduced deep scrape concurrency to 5.

This commit is contained in:
DebaucheryLibrarian
2021-12-01 17:26:13 +01:00
parent 08f725a0b6
commit e29cbc9fea
6 changed files with 59 additions and 2 deletions

View File

@@ -9,6 +9,7 @@ const { fetchReleaseEntities, urlToSiteSlug } = require('./entities');
const logger = require('./logger')(__filename);
const qu = require('./utils/qu');
const getRecursiveParameters = require('./utils/get-recursive-parameters');
const windows = require('./utils/http-windows');
function toBaseReleases(baseReleasesOrUrls, entity = null) {
if (!baseReleasesOrUrls) {
@@ -116,10 +117,19 @@ async function scrapeRelease(baseRelease, entitiesBySlug, type = 'scene') {
parameters: getRecursiveParameters(entity),
};
logger.debug(`Memory usage before: ${process.memoryUsage.rss() / 1000000} MB (${baseRelease.url})`);
const rawScrapedRelease = type === 'scene'
? await fetchScene(layoutScraper, baseRelease.url, entity, baseRelease, options, null)
: await layoutScraper.fetchMovie(baseRelease.url, entity, baseRelease, options, null);
const pathname = new URL(baseRelease.url).pathname.replace(/\//g, '_');
windows.get(pathname)?.close();
windows.delete(pathname);
logger.debug(`Memory usage after: ${process.memoryUsage.rss() / 1000000} MB (${baseRelease.url})`);
const scrapedRelease = rawScrapedRelease?.scene || rawScrapedRelease;
if (!scrapedRelease || typeof scrapedRelease !== 'object' || Array.isArray(scrapedRelease)) {
@@ -186,7 +196,7 @@ async function scrapeReleases(baseReleases, entitiesBySlug, type) {
return Promise.map(
baseReleases,
async (baseRelease) => scrapeRelease(baseRelease, entitiesWithBeforeDataBySlug, type),
{ concurrency: 10 },
{ concurrency: 5 },
);
}