Added before scene fetch method to prevent e.g. unnecessary session requests, moved scraper assignment to entity lookup. Removed channel URL hostname matching..

This commit is contained in:
DebaucheryLibrarian
2021-10-26 23:42:32 +02:00
parent 6c5d4389fe
commit 100a35b4e8
3 changed files with 28 additions and 9 deletions

View File

@@ -5,7 +5,6 @@ const { mergeAdvanced: merge } = require('object-merge-advanced');
const argv = require('./argv');
const include = require('./utils/argv-include')(argv);
const { resolveScraper, resolveLayoutScraper } = require('./scrapers/resolve');
const { fetchReleaseEntities, urlToSiteSlug } = require('./entities');
const logger = require('./logger')(__filename);
const qu = require('./utils/qu');
@@ -96,10 +95,9 @@ async function scrapeRelease(baseRelease, entitiesBySlug, type = 'scene') {
};
}
const scraper = resolveScraper(entity);
const layoutScraper = resolveLayoutScraper(entity, scraper);
const layoutScraper = entity.scraper;
if (!layoutScraper) {
if (!entity.scraper) {
logger.warn(`Could not find scraper for ${baseRelease.url}`);
return baseRelease;
}
@@ -114,13 +112,16 @@ async function scrapeRelease(baseRelease, entitiesBySlug, type = 'scene') {
const options = {
...include,
beforeFetchScene: entity.preData,
parameters: getRecursiveParameters(entity),
};
const scrapedRelease = type === 'scene'
const rawScrapedRelease = type === 'scene'
? await fetchScene(layoutScraper, baseRelease.url, entity, baseRelease, options, null)
: await layoutScraper.fetchMovie(baseRelease.url, entity, baseRelease, options, null);
const scrapedRelease = rawScrapedRelease?.scene || rawScrapedRelease;
if (!scrapedRelease || typeof scrapedRelease !== 'object' || Array.isArray(scrapedRelease)) {
// scraper is unable to fetch the releases and returned a HTTP code or null
throw new Error(`Scraper returned '${scrapedRelease}' when fetching latest from '${entity.name}' (${entity.parent?.name})`);
@@ -170,9 +171,21 @@ async function scrapeRelease(baseRelease, entitiesBySlug, type = 'scene') {
}
async function scrapeReleases(baseReleases, entitiesBySlug, type) {
const entitiesWithBeforeDataEntries = await Promise.all(Object.entries(entitiesBySlug).map(async ([slug, entity]) => {
if (entity.scraper?.beforeFetchScene) {
const preData = await entity.scraper.beforeFetchScene(entity);
return [slug, { ...entity, preData }];
}
return null;
}));
const entitiesWithBeforeDataBySlug = Object.fromEntries(entitiesWithBeforeDataEntries.filter(Boolean));
return Promise.map(
baseReleases,
async baseRelease => scrapeRelease(baseRelease, entitiesBySlug, type),
async baseRelease => scrapeRelease(baseRelease, entitiesWithBeforeDataBySlug, type),
{ concurrency: 10 },
);
}