forked from DebaucheryLibrarian/traxxx
Changed sites from argument query to group by network.
This commit is contained in:
112
src/entities.js
112
src/entities.js
@@ -14,16 +14,10 @@ function curateEntity(entity, includeParameters = false) {
|
||||
url: entity.url,
|
||||
description: entity.description,
|
||||
slug: entity.slug,
|
||||
independent: !!entity.parameters && entity.parameters.independent,
|
||||
type: entity.type,
|
||||
parameters: includeParameters ? entity.parameters : null,
|
||||
network: {
|
||||
id: entity.network_id,
|
||||
name: entity.network_name,
|
||||
description: entity.network_description,
|
||||
slug: entity.network_slug,
|
||||
url: entity.network_url,
|
||||
parameters: includeParameters ? entity.network_parameters : null,
|
||||
},
|
||||
parent: entity.parent,
|
||||
children: (entity.children || []).map(child => curateEntity(child)),
|
||||
};
|
||||
|
||||
return curatedEntity;
|
||||
@@ -33,56 +27,59 @@ async function curateEntities(entities, includeParameters) {
|
||||
return Promise.all(entities.map(async entity => curateEntity(entity, includeParameters)));
|
||||
}
|
||||
|
||||
async function findSiteByUrl(url) {
|
||||
const { origin, hostname, pathname } = new URL(url);
|
||||
// const domain = hostname.replace(/www.|tour./, '');
|
||||
const dirUrl = `${origin}${pathname.split('/').slice(0, 2).join('/')}`; // allow for sites on URI directory
|
||||
|
||||
const site = await knex('sites')
|
||||
.leftJoin('networks', 'sites.network_id', 'networks.id')
|
||||
.select(
|
||||
'sites.*',
|
||||
'networks.name as network_name', 'networks.slug as network_slug', 'networks.url as network_url', 'networks.description as network_description', 'networks.parameters as network_parameters',
|
||||
)
|
||||
.where('sites.url', url)
|
||||
.orWhere('sites.url', origin)
|
||||
.orWhere('sites.url', origin.replace(/www\.|tour\./, ''))
|
||||
.orWhere('sites.url', `https://www.${hostname}`)
|
||||
.orWhere('sites.url', `http://www.${hostname}`)
|
||||
.orWhere('sites.url', dirUrl)
|
||||
// .orWhere('sites.url', 'like', `%${domain}`)
|
||||
.first();
|
||||
|
||||
if (site) {
|
||||
const curatedSite = curateSite(site, true, false);
|
||||
|
||||
return curatedSite;
|
||||
}
|
||||
|
||||
return null;
|
||||
}
|
||||
|
||||
async function fetchEntitiesFromArgv() {
|
||||
async function fetchSitesFromArgv() {
|
||||
const rawEntities = await knex.raw(`
|
||||
WITH RECURSIVE temp AS (
|
||||
SELECT id, parent_id, name, slug, type FROM entities WHERE slug IN (?)
|
||||
UNION
|
||||
SELECT entities.id, entities.parent_id, entities.name, entities.slug, entities.type FROM entities
|
||||
INNER JOIN temp ON temp.id = entities.parent_id
|
||||
) SELECT * FROM temp;
|
||||
`, argv.sites || argv.networks || argv.entities);
|
||||
|
||||
console.log(rawEntities.rows);
|
||||
SELECT
|
||||
id, parent_id, name, slug, type, url, description, parameters
|
||||
FROM
|
||||
entities
|
||||
WHERE
|
||||
slug = ANY(?) AND entities.type = 1
|
||||
UNION ALL
|
||||
SELECT
|
||||
entities.id, entities.parent_id, entities.name, entities.slug, entities.type, entities.url, entities.description, entities.parameters
|
||||
FROM
|
||||
entities
|
||||
INNER JOIN
|
||||
temp ON temp.id = entities.parent_id
|
||||
)
|
||||
SELECT
|
||||
entities.*, row_to_json(parents) as parent, json_agg(temp) as children
|
||||
FROM
|
||||
temp
|
||||
LEFT JOIN
|
||||
entities ON entities.id = temp.parent_id
|
||||
LEFT JOIN
|
||||
entities AS parents ON parents.id = entities.parent_id
|
||||
WHERE
|
||||
temp.type = 2
|
||||
GROUP BY
|
||||
temp.parent_id, entities.id, entities.name, parents.id
|
||||
UNION ALL
|
||||
SELECT
|
||||
entities.*, row_to_json(parents) as parent, json_build_array(row_to_json(children))
|
||||
FROM
|
||||
entities AS children
|
||||
LEFT JOIN
|
||||
entities ON entities.id = children.parent_id
|
||||
LEFT JOIN
|
||||
entities AS parents ON parents.id = entities.parent_id
|
||||
WHERE
|
||||
children.slug = ANY(?) AND children.type = 2
|
||||
GROUP BY
|
||||
entities.id, parents.id, children.id;
|
||||
`, [argv.networks || [], argv.sites || []]);
|
||||
|
||||
const curatedEntities = await curateEntities(rawEntities.rows, true);
|
||||
logger.info(`Found ${curatedEntities.length} entities in database`);
|
||||
|
||||
console.log(curatedEntities);
|
||||
console.log(rawEntities.rows);
|
||||
|
||||
return curatedEntities;
|
||||
}
|
||||
|
||||
async function fetchEntitiesFromConfig() {
|
||||
async function fetchSitesFromConfig() {
|
||||
const rawSites = await knex('entities')
|
||||
.select('entities.*')
|
||||
.leftJoin('entities as entities_parents', 'entities_parents.id', 'entities.id')
|
||||
@@ -95,18 +92,18 @@ async function fetchEntitiesFromConfig() {
|
||||
builder.whereIn('entities.slug', config.exclude || []);
|
||||
});
|
||||
|
||||
const curatedSites = await curateSites(rawSites, true);
|
||||
logger.info(`Found ${curatedSites.length} sites in database`);
|
||||
const curatedSites = await curateEntities(rawSites, true);
|
||||
logger.info(`Found ${curatedSites.length} entities in database`);
|
||||
|
||||
return curatedSites;
|
||||
}
|
||||
|
||||
async function fetchIncludedEntities() {
|
||||
if (argv.networks || argv.sites) {
|
||||
return fetchEntitiesFromArgv();
|
||||
return fetchSitesFromArgv();
|
||||
}
|
||||
|
||||
return fetchEntitiesFromConfig();
|
||||
return fetchSitesFromConfig();
|
||||
}
|
||||
|
||||
async function fetchSites(queryObject) {
|
||||
@@ -119,7 +116,7 @@ async function fetchSites(queryObject) {
|
||||
.leftJoin('networks', 'sites.network_id', 'networks.id')
|
||||
.limit(100);
|
||||
|
||||
return curateSites(sites);
|
||||
return curateEntities(sites);
|
||||
}
|
||||
|
||||
async function fetchSitesFromReleases() {
|
||||
@@ -129,7 +126,7 @@ async function fetchSitesFromReleases() {
|
||||
.groupBy('sites.id')
|
||||
.limit(100);
|
||||
|
||||
return curateSites(sites);
|
||||
return curateEntities(sites);
|
||||
}
|
||||
|
||||
module.exports = {
|
||||
@@ -137,8 +134,7 @@ module.exports = {
|
||||
curateEntities,
|
||||
fetchIncludedEntities,
|
||||
fetchSites,
|
||||
fetchEntitiesFromConfig,
|
||||
fetchEntitiesFromArgv,
|
||||
fetchSitesFromConfig,
|
||||
fetchSitesFromArgv,
|
||||
fetchSitesFromReleases,
|
||||
findSiteByUrl,
|
||||
};
|
||||
|
||||
@@ -108,7 +108,7 @@ function scrapeAllT1(scenes, site, accSiteReleases) {
|
||||
release.entryId = deriveEntryId(release);
|
||||
|
||||
if (site.parameters?.accFilter && accSiteReleases?.map(accRelease => accRelease.entryId).includes(release.entryId)) {
|
||||
// filter out releases that were already scraped from a categorized site
|
||||
// filter out releases that were already scraped from a categorized site, requires sequential site scraping
|
||||
return null;
|
||||
}
|
||||
|
||||
|
||||
@@ -8,7 +8,7 @@ const logger = require('./logger')(__filename);
|
||||
const knex = require('./knex');
|
||||
const include = require('./utils/argv-include')(argv);
|
||||
const scrapers = require('./scrapers/scrapers');
|
||||
const { fetchEntitiesFromArgv, fetchEntitiesFromConfig } = require('./entities');
|
||||
const { fetchSitesFromArgv, fetchSitesFromConfig } = require('./entities');
|
||||
|
||||
const afterDate = (() => {
|
||||
if (/\d{2,4}-\d{2}-\d{2,4}/.test(argv.after)) {
|
||||
@@ -219,8 +219,10 @@ async function scrapeNetworkParallel(network) {
|
||||
|
||||
async function fetchUpdates() {
|
||||
const includedNetworks = argv.sites || argv.networks || argv.from
|
||||
? await fetchEntitiesFromArgv()
|
||||
: await fetchEntitiesFromConfig();
|
||||
? await fetchSitesFromArgv()
|
||||
: await fetchSitesFromConfig();
|
||||
|
||||
// console.log('included', includedNetworks);
|
||||
|
||||
const scrapedNetworks = await Promise.map(
|
||||
includedNetworks,
|
||||
|
||||
Reference in New Issue
Block a user