Chunked duplicate check to prevent postgres stack depth errors.

This commit is contained in:
DebaucheryLibrarian
2022-01-10 02:17:17 +01:00
parent 43a0bc8a2c
commit 506971b44b
3 changed files with 33 additions and 22 deletions

View File

@@ -1,6 +1,7 @@
'use strict';
const config = require('config');
const Promise = require('bluebird');
const argv = require('./argv');
const logger = require('./logger')(__filename);
@@ -8,6 +9,7 @@ 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 { formatDate } = require('./utils/qu');
const { associateActors, associateDirectors, scrapeActors, toBaseActors } = require('./actors');
const { associateReleaseTags } = require('./tags');
@@ -192,13 +194,16 @@ function filterInternalDuplicateReleases(releases) {
async function filterDuplicateReleases(releases) {
const internalUniqueReleases = filterInternalDuplicateReleases(releases);
const internalUniqueReleaseChunks = chunk(internalUniqueReleases);
const duplicateReleaseEntries = await knex('releases')
.whereIn(['entry_id', 'entity_id'], internalUniqueReleases.map((release) => [release.entryId, release.entity.id]))
.orWhereIn(['entry_id', 'entity_id'], internalUniqueReleases
const duplicateReleaseEntryChunks = await Promise.map(internalUniqueReleaseChunks, async (internalUniqueReleasesChunk) => knex('releases')
.whereIn(['entry_id', 'entity_id'], internalUniqueReleasesChunk.map((release) => [release.entryId, release.entity.id]))
.orWhereIn(['entry_id', 'entity_id'], internalUniqueReleasesChunk
// scene IDs shared across network, mark as duplicate so scene can be updated with channel if only available on release day (i.e. Perv City)
.filter((release) => release.entity.parent?.parameters?.networkEntryIds)
.map((release) => [release.entryId, release.entity.parent.id]));
.map((release) => [release.entryId, release.entity.parent.id])), { concurrency: 10 });
const duplicateReleaseEntries = duplicateReleaseEntryChunks.flat();
const duplicateReleasesByEntityIdAndEntryId = duplicateReleaseEntries.reduce((acc, release) => {
if (!acc[release.entity_id]) acc[release.entity_id] = {};