From 100a35b4e8d7924fd1fabef27b0b987a45edba7f Mon Sep 17 00:00:00 2001
From: DebaucheryLibrarian <moonloop.adult@protonmail.com>
Date: Tue, 26 Oct 2021 23:42:32 +0200
Subject: [PATCH] Added before scene fetch method to prevent e.g. unnecessary
 session requests, moved scraper assignment to entity lookup. Removed channel
 URL hostname matching..

---
 src/deep.js              | 25 +++++++++++++++++++------
 src/entities.js          |  5 ++++-
 src/scrapers/mindgeek.js |  7 +++++--
 3 files changed, 28 insertions(+), 9 deletions(-)

diff --git a/src/deep.js b/src/deep.js
index d7d3d25d..e09bc5bd 100644
--- a/src/deep.js
+++ b/src/deep.js
@@ -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 },
 	);
 }
diff --git a/src/entities.js b/src/entities.js
index 787eb8db..f94f63f0 100644
--- a/src/entities.js
+++ b/src/entities.js
@@ -8,6 +8,7 @@ const argv = require('./argv');
 const knex = require('./knex');
 const { deleteScenes, deleteMovies } = require('./releases');
 const { flushOrphanedMedia } = require('./media');
+const { resolveScraper, resolveLayoutScraper } = require('./scrapers/resolve');
 
 function getRecursiveParent(entity) {
 	if (!entity) {
@@ -71,6 +72,9 @@ function curateEntity(entity, includeParameters = false) {
 		}, includeParameters));
 	}
 
+	const scraper = resolveScraper(curatedEntity);
+	curatedEntity.scraper = resolveLayoutScraper(entity, scraper);
+
 	return curatedEntity;
 }
 
@@ -187,7 +191,6 @@ async function fetchEntitiesBySlug(entitySlugs, sort = 'asc') {
 				array['parent'] as parent_path
 			FROM entities
 			WHERE slug = ANY(:entitySlugs)
-			OR substring(url from 'https%://%#"[a-z0-9-]+#".(com|net)%' for '#') = ANY(:entitySlugs)
 
 			UNION ALL
 
diff --git a/src/scrapers/mindgeek.js b/src/scrapers/mindgeek.js
index 6f7dbb92..36198d45 100644
--- a/src/scrapers/mindgeek.js
+++ b/src/scrapers/mindgeek.js
@@ -294,7 +294,7 @@ async function fetchScene(url, site, baseScene, options) {
 	}
 
 	const entryId = new URL(url).pathname.match(/\/(\d+)/)?.[1];
-	const { session, instanceToken } = await getSession(site, options.parameters);
+	const { session, instanceToken } = options.beforeFetchScene || await getSession(site, options.parameters);
 
 	const res = await http.get(`https://site-api.project1service.com/v2/releases/${entryId}`, {
 		session,
@@ -306,7 +306,9 @@ async function fetchScene(url, site, baseScene, options) {
 	});
 
 	if (res.status === 200 && res.body.result) {
-		return scrapeScene(res.body.result, url, site);
+		return {
+			scene: scrapeScene(res.body.result, url, site),
+		};
 	}
 
 	return null;
@@ -361,6 +363,7 @@ async function fetchProfile({ name: actorName, slug: actorSlug }, { entity, para
 }
 
 module.exports = {
+	beforeFetchScene: getSession,
 	scrapeLatestX,
 	fetchLatest,
 	fetchUpcoming,