From 1aa876acd960fc91e66045ce11c7fbcb6017f535 Mon Sep 17 00:00:00 2001 From: DebaucheryLibrarian Date: Thu, 31 Oct 2024 03:01:13 +0100 Subject: [PATCH] Using common for place resolve. --- common | 2 +- src/actors.js | 29 ++++++++----- src/scrapers/bang.js | 1 + src/store-releases.js | 17 +++++++- src/utils/resolve-place.js | 85 -------------------------------------- 5 files changed, 36 insertions(+), 98 deletions(-) delete mode 100755 src/utils/resolve-place.js diff --git a/common b/common index c068c759..4316b69a 160000 --- a/common +++ b/common @@ -1 +1 @@ -Subproject commit c068c759b2bb62fad44c87226a780532b771479b +Subproject commit 4316b69a43daeaa7ddadb428ccc6cb3eb10f555e diff --git a/src/actors.js b/src/actors.js index fffed8fa..04d3dc5b 100755 --- a/src/actors.js +++ b/src/actors.js @@ -9,12 +9,14 @@ const DOMPurify = require('dompurify'); const { JSDOM } = require('jsdom'); const omit = require('object.omit'); const inquirer = require('inquirer'); +const unprint = require('unprint'); const { window } = new JSDOM(''); const domPurify = DOMPurify(window); // const logger = require('./logger')(__filename); const knex = require('./knex'); +const redis = require('./redis'); const scrapers = require('./scrapers/scrapers').actors; const argv = require('./argv'); @@ -29,13 +31,24 @@ const { fetchEntitiesBySlug } = require('./entities'); const { deleteScenes } = require('./releases'); const actorsCommon = import('../common/actors.mjs'); // eslint-disable-line import/extensions, import/no-relative-packages +const geoCommon = import('../common/geo.mjs'); // eslint-disable-line import/extensions, import/no-relative-packages const slugify = require('./utils/slugify'); const capitalize = require('./utils/capitalize'); -const resolvePlace = require('./utils/resolve-place'); const { resolveLayoutScraper } = require('./scrapers/resolve'); const getRecursiveParameters = require('./utils/get-recursive-parameters'); +const commonContext = { + argv, + knex, + redis, + logger, + moment, + slugify, + omit, + unprint, +}; + const hairColors = { 'jet-black': 'black', 'red-head': 'red', @@ -364,13 +377,7 @@ async function interpolateProfiles(actorIdsOrNames, refreshView) { const { interpolateProfiles: interpolateProfilesUtil } = await actorsCommon; try { - await interpolateProfilesUtil(actorIdsOrNames, { - knex, - logger, - moment, - slugify, - omit, - }, { refreshView }); + await interpolateProfilesUtil(actorIdsOrNames, commonContext, { refreshView }); } catch (error) { console.log(error); } @@ -381,6 +388,8 @@ async function curateProfile(profile, actor) { return null; } + const { resolvePlace } = await geoCommon; + try { const curatedProfile = { // id: profile.id, @@ -465,8 +474,8 @@ async function curateProfile(profile, actor) { if (argv.resolvePlace) { const [placeOfBirth, placeOfResidence] = await Promise.all([ - resolvePlace(profile.birthPlace), - resolvePlace(profile.residencePlace), + resolvePlace(profile.birthPlace, commonContext), + resolvePlace(profile.residencePlace, commonContext), ]); curatedProfile.placeOfBirth = placeOfBirth; diff --git a/src/scrapers/bang.js b/src/scrapers/bang.js index f87f0a83..b2e189b5 100755 --- a/src/scrapers/bang.js +++ b/src/scrapers/bang.js @@ -187,6 +187,7 @@ async function fetchLatest(channel, page = 1) { return res.status; } + async function fetchUpcoming(channel) { const url = `${channel.url}&by=date.desc&early-access=true`; const res = await unprint.get(url, { selectAll: '.video_container' }); diff --git a/src/store-releases.js b/src/store-releases.js index d5f6c6ce..64654c96 100755 --- a/src/store-releases.js +++ b/src/store-releases.js @@ -3,14 +3,15 @@ const config = require('config'); const Promise = require('bluebird'); const { decode } = require('html-entities'); +const unprint = require('unprint'); const argv = require('./argv'); const logger = require('./logger')(__filename); const knex = require('./knex'); const slugify = require('./utils/slugify'); const bulkInsert = require('./utils/bulk-insert'); -const resolvePlace = require('./utils/resolve-place'); const chunk = require('./utils/chunk'); +const redis = require('./redis'); const { formatDate } = require('./utils/qu'); const { associateActors, associateDirectors, scrapeActors, toBaseActors } = require('./actors'); const { associateReleaseTags } = require('./tags'); @@ -19,6 +20,17 @@ const { associateReleaseMedia } = require('./media'); const { updateSceneSearch, updateMovieSearch } = require('./update-search'); const { notify } = require('./alerts'); +const geoCommon = import('../common/geo.mjs'); // eslint-disable-line import/extensions, import/no-relative-packages + +const commonContext = { + argv, + knex, + redis, + logger, + slugify, + unprint, +}; + async function curateReleaseEntry(release, batchId, existingRelease, type = 'scene') { const slugBase = release.title || (release.actors?.length && `${release.entity.slug} ${release.actors.map((actor) => actor.name).join(' ')}`) @@ -64,7 +76,8 @@ async function curateReleaseEntry(release, batchId, existingRelease, type = 'sce curatedRelease.production_location = decode(release.productionLocation); if (argv.resolvePlace) { - const productionLocation = await resolvePlace(decode(release.productionLocation)); + const { resolvePlace } = await geoCommon; + const productionLocation = await resolvePlace(decode(release.productionLocation), commonContext); if (productionLocation) { curatedRelease.production_city = productionLocation.city; diff --git a/src/utils/resolve-place.js b/src/utils/resolve-place.js deleted file mode 100755 index 8bb25915..00000000 --- a/src/utils/resolve-place.js +++ /dev/null @@ -1,85 +0,0 @@ -'use strict'; - -const config = require('config'); - -const knex = require('../knex'); -const logger = require('../logger')(__filename); -const http = require('./http'); -const slugify = require('./slugify'); -const argv = require('../argv'); -const redis = require('../redis'); - -async function resolvePlace(query) { - if (!query) { - return null; - } - - const cacheKey = `place-${slugify(query)}`; - const cachedPlace = await redis.hGetAll(cacheKey); - - if (argv.placeCache !== false && await redis.exists(cacheKey)) { - await redis.expire(cacheKey, 3600 * 24 * 30); - - logger.debug(`Using cached place '${cacheKey}' for query '${query}': ${JSON.stringify(cachedPlace)}`); - - return cachedPlace; - } - - // query is a nationality, lookup would get weird results (British resolves to British, Northern Ireland) - const country = await knex('countries') - .where('nationality', 'ilike', `%${query}%`) - .orWhere('alpha3', 'ilike', `%${query}%`) - .orWhere('alpha2', 'ilike', `%${query}%`) - .orderBy('priority', 'desc') - .first(); - - if (country) { - return { - country: country.alpha2, - }; - } - - try { - // https://operations.osmfoundation.org/policies/nominatim/ - const res = await http.get(`https://nominatim.openstreetmap.org/search?q=${encodeURI(query)}&format=json&accept-language=en&addressdetails=1`, { - headers: { - 'User-Agent': config.location.userAgent, - }, - interval: 1000, - concurrency: 1, - }); - - const [item] = res.body; - - if (item && item.address) { - const rawPlace = item.address; - const place = {}; - - if (item.class === 'place' || item.class === 'boundary') { - const location = rawPlace[item.type] || rawPlace.city || rawPlace.place || rawPlace.town; - - if (location) { - place.place = location; - place.city = rawPlace.city || location; - } - } - - if (rawPlace.state) place.state = rawPlace.state; - if (rawPlace.country_code) place.country = rawPlace.country_code.toUpperCase(); - if (rawPlace.continent) place.continent = rawPlace.continent; - - logger.debug(`Resolved place '${query}' to ${JSON.stringify(place)}`); - - await redis.hSet(cacheKey, place); - await redis.expire(cacheKey, 3600 * 24 * 30); - - return place; - } - } catch (error) { - logger.error(`Failed to resolve place '${query}': ${error.message}`); - } - - return null; -} - -module.exports = resolvePlace;