feat(blacklist): Automatically add media with blacklisted tags to the blacklist

cypress/config/settings.cypress.json

@@ -19,6 +19,8 @@
     "discoverRegion": "",
     "streamingRegion": "",
     "originalLanguage": "",
+    "blacklistedTags": "",
+    "blacklistedTagsLimit": 50,
     "trustProxy": false,
     "mediaServerType": 1,
     "partialRequestsEnabled": true,

docs/using-jellyseerr/settings/general.md

@@ -62,6 +62,14 @@ Set the default display language for Jellyseerr. Users can override this setting
 
 These settings filter content shown on the "Discover" home page based on regional availability and original language, respectively. The Streaming Region filters the available streaming providers on the media page. Users can override these global settings by configuring these same options in their user settings.
 
+## Blacklist Content with Tags and Limit Content Blacklisted per Tag
+
+These settings blacklist any TV shows or movies that have one of the entered tags. The "Process Blacklisted Tags" job adds entries to the blacklist based on the configured blacklisted tags. If a blacklisted tag is removed, any media blacklisted under that tag will be removed from the blacklist when the "Process Blacklisted Tags" job runs.
+
+The limit setting determines how many pages per tag the job will process, with each page containing 20 entries. The job cycles through all 16 available discovery sort options, querying the defined number of pages to blacklist media that is most likely to appear at the top of each sort. Higher limits will create a more accurate blacklist, but will require more storage.
+
+Blacklisted tags are disabled until at least one tag is entered. These settings cannot be overridden in user settings.
+
 ## Hide Available Media
 
 When enabled, media which is already available will not appear on the "Discover" home page, or in the "Recommended" or "Similar" categories or other links on media detail pages.

overseerr-api.yml

@@ -4122,6 +4122,12 @@ paths:
             type: string
             nullable: true
             example: dune
+        - in: query
+          name: filter
+          schema:
+            type: string
+            enum: [all, manual, blacklistedTags]
+            default: manual
       responses:
         '200':
           description: Blacklisted items returned

server/api/themoviedb/index.ts

@@ -37,23 +37,26 @@ interface SingleSearchOptions extends SearchOptions {
   year?: number;
 }
 
-export type SortOptions =
-  | 'popularity.asc'
-  | 'popularity.desc'
-  | 'release_date.asc'
-  | 'release_date.desc'
-  | 'revenue.asc'
-  | 'revenue.desc'
-  | 'primary_release_date.asc'
-  | 'primary_release_date.desc'
-  | 'original_title.asc'
-  | 'original_title.desc'
-  | 'vote_average.asc'
-  | 'vote_average.desc'
-  | 'vote_count.asc'
-  | 'vote_count.desc'
-  | 'first_air_date.asc'
-  | 'first_air_date.desc';
+export const SortOptionsIterable = [
+  'popularity.desc',
+  'popularity.asc',
+  'release_date.desc',
+  'release_date.asc',
+  'revenue.desc',
+  'revenue.asc',
+  'primary_release_date.desc',
+  'primary_release_date.asc',
+  'original_title.asc',
+  'original_title.desc',
+  'vote_average.desc',
+  'vote_average.asc',
+  'vote_count.desc',
+  'vote_count.asc',
+  'first_air_date.desc',
+  'first_air_date.asc',
+] as const;
+
+export type SortOptions = (typeof SortOptionsIterable)[number];
 
 interface DiscoverMovieOptions {
   page?: number;

server/entity/Blacklist.ts

@@ -1,8 +1,9 @@
 import { MediaStatus, type MediaType } from '@server/constants/media';
-import { getRepository } from '@server/datasource';
+import dataSource from '@server/datasource';
 import Media from '@server/entity/Media';
 import { User } from '@server/entity/User';
 import type { BlacklistItem } from '@server/interfaces/api/blacklistInterfaces';
+import type { EntityManager } from 'typeorm';
 import {
   Column,
   CreateDateColumn,
@@ -35,42 +36,50 @@ export class Blacklist implements BlacklistItem {
   @ManyToOne(() => User, (user) => user.id, {
     eager: true,
   })
-  user: User;
+  user?: User;
 
   @OneToOne(() => Media, (media) => media.blacklist, {
     onDelete: 'CASCADE',
   })
   @JoinColumn()
   public media: Media;
 
+  @Column({ nullable: true, type: 'varchar' })
+  public blacklistedTags?: string;
+
   @CreateDateColumn()
   public createdAt: Date;
 
   constructor(init?: Partial<Blacklist>) {
     Object.assign(this, init);
   }
 
-  public static async addToBlacklist({
-    blacklistRequest,
-  }: {
-    blacklistRequest: {
-      mediaType: MediaType;
-      title?: ZodOptional<ZodString>['_output'];
-      tmdbId: ZodNumber['_output'];
-    };
-  }): Promise<void> {
+  public static async addToBlacklist(
+    {
+      blacklistRequest,
+    }: {
+      blacklistRequest: {
+        mediaType: MediaType;
+        title?: ZodOptional<ZodString>['_output'];
+        tmdbId: ZodNumber['_output'];
+        blacklistedTags?: string;
+      };
+    },
+    entityManager?: EntityManager
+  ): Promise<void> {
+    const em = entityManager ?? dataSource;
     const blacklist = new this({
       ...blacklistRequest,
     });
 
-    const mediaRepository = getRepository(Media);
+    const mediaRepository = em.getRepository(Media);
     let media = await mediaRepository.findOne({
       where: {
         tmdbId: blacklistRequest.tmdbId,
       },
     });
 
-    const blacklistRepository = getRepository(this);
+    const blacklistRepository = em.getRepository(this);
 
     await blacklistRepository.save(blacklist);
 

server/interfaces/api/blacklistInterfaces.ts

@@ -6,7 +6,8 @@ export interface BlacklistItem {
   mediaType: 'movie' | 'tv';
   title?: string;
   createdAt?: Date;
-  user: User;
+  user?: User;
+  blacklistedTags?: string;
 }
 
 export interface BlacklistResultsResponse extends PaginatedResponse {

server/job/blacklistedTagsProcessor.ts

@@ -0,0 +1,184 @@
+import type { SortOptions } from '@server/api/themoviedb';
+import { SortOptionsIterable } from '@server/api/themoviedb';
+import type {
+  TmdbSearchMovieResponse,
+  TmdbSearchTvResponse,
+} from '@server/api/themoviedb/interfaces';
+import { MediaType } from '@server/constants/media';
+import dataSource from '@server/datasource';
+import { Blacklist } from '@server/entity/Blacklist';
+import Media from '@server/entity/Media';
+import type {
+  RunnableScanner,
+  StatusBase,
+} from '@server/lib/scanners/baseScanner';
+import { getSettings } from '@server/lib/settings';
+import logger from '@server/logger';
+import { createTmdbWithRegionLanguage } from '@server/routes/discover';
+import type { EntityManager } from 'typeorm';
+
+const TMDB_API_DELAY_MS = 250;
+class AbortTransaction extends Error {}
+
+class BlacklistedTagProcessor implements RunnableScanner<StatusBase> {
+  private running = false;
+  private progress = 0;
+  private total = 0;
+
+  public async run() {
+    this.running = true;
+
+    try {
+      await dataSource.transaction(async (em) => {
+        await this.cleanBlacklist(em);
+        await this.createBlacklistEntries(em);
+      });
+    } catch (err) {
+      if (err instanceof AbortTransaction) {
+        logger.info('Aborting job: Process Blacklisted Tags', {
+          label: 'Jobs',
+        });
+      } else {
+        throw err;
+      }
+    } finally {
+      this.reset();
+    }
+  }
+
+  public status(): StatusBase {
+    return {
+      running: this.running,
+      progress: this.progress,
+      total: this.total,
+    };
+  }
+
+  public cancel() {
+    this.running = false;
+    this.progress = 0;
+    this.total = 0;
+  }
+
+  private reset() {
+    this.cancel();
+  }
+
+  private async createBlacklistEntries(em: EntityManager) {
+    const tmdb = createTmdbWithRegionLanguage();
+
+    const settings = getSettings();
+    const blacklistedTags = settings.main.blacklistedTags;
+    const blacklistedTagsArr = blacklistedTags.split(',');
+
+    const pageLimit = settings.main.blacklistedTagsLimit;
+
+    if (blacklistedTags.length === 0) {
+      return;
+    }
+
+    // The maximum number of queries we're expected to execute
+    this.total =
+      2 * blacklistedTagsArr.length * pageLimit * SortOptionsIterable.length;
+
+    for (const type of [MediaType.MOVIE, MediaType.TV]) {
+      const getDiscover =
+        type === MediaType.MOVIE ? tmdb.getDiscoverMovies : tmdb.getDiscoverTv;
+
+      // Iterate for each tag
+      for (const tag of blacklistedTagsArr) {
+        let queryMax = pageLimit * SortOptionsIterable.length;
+        let fixedSortMode = false; // Set to true when the page limit allows for getting every page of tag
+
+        for (let query = 0; query < queryMax; query++) {
+          const page: number = fixedSortMode
+            ? query + 1
+            : (query % pageLimit) + 1;
+          const sortBy: SortOptions | undefined = fixedSortMode
+            ? undefined
+            : SortOptionsIterable[query % SortOptionsIterable.length];
+
+          if (!this.running) {
+            throw new AbortTransaction();
+          }
+
+          const response = await getDiscover({
+            page,
+            sortBy,
+            keywords: tag,
+          });
+          await this.processResults(response, tag, type, em);
+          await new Promise((res) => setTimeout(res, TMDB_API_DELAY_MS));
+
+          this.progress++;
+          if (page === 1 && response.total_pages <= queryMax) {
+            // We will finish the tag with less queries than expected, move progress accordingly
+            this.progress += queryMax - response.total_pages;
+            fixedSortMode = true;
+            queryMax = response.total_pages;
+          }
+        }
+      }
+    }
+  }
+
+  private async processResults(
+    response: TmdbSearchMovieResponse | TmdbSearchTvResponse,
+    keywordId: string,
+    mediaType: MediaType,
+    em: EntityManager
+  ) {
+    const blacklistRepository = em.getRepository(Blacklist);
+
+    for (const entry of response.results) {
+      const blacklistEntry = await blacklistRepository.findOne({
+        where: { tmdbId: entry.id },
+      });
+
+      if (blacklistEntry) {
+        // Don't mark manual blacklists with tags
+        // If media wasn't previously blacklisted for this tag, add the tag to the media's blacklist
+        if (
+          blacklistEntry.blacklistedTags &&
+          !blacklistEntry.blacklistedTags.includes(`,${keywordId},`)
+        ) {
+          await blacklistRepository.update(blacklistEntry.id, {
+            blacklistedTags: `${blacklistEntry.blacklistedTags}${keywordId},`,
+          });
+        }
+      } else {
+        // Media wasn't previously blacklisted, add it to the blacklist
+        await Blacklist.addToBlacklist(
+          {
+            blacklistRequest: {
+              mediaType,
+              title: 'title' in entry ? entry.title : entry.name,
+              tmdbId: entry.id,
+              blacklistedTags: `,${keywordId},`,
+            },
+          },
+          em
+        );
+      }
+    }
+  }
+
+  private async cleanBlacklist(em: EntityManager) {
+    // Remove blacklist and media entries blacklisted by tags
+    const mediaRepository = em.getRepository(Media);
+    const mediaToRemove = await mediaRepository
+      .createQueryBuilder('media')
+      .innerJoinAndSelect(Blacklist, 'blist', 'blist.tmdbId = media.tmdbId')
+      .where(`blist.blacklistedTags IS NOT NULL`)
+      .getMany();
+
+    // Batch removes so the query doesn't get too large
+    for (let i = 0; i < mediaToRemove.length; i += 500) {
+      await mediaRepository.remove(mediaToRemove.slice(i, i + 500)); // This also deletes the blacklist entries via cascading
+    }
+  }
+}
+
+const blacklistedTagsProcessor = new BlacklistedTagProcessor();
+
+export default blacklistedTagsProcessor;

server/job/schedule.ts

@@ -1,4 +1,5 @@
 import { MediaServerType } from '@server/constants/server';
+import blacklistedTagsProcessor from '@server/job/blacklistedTagsProcessor';
 import availabilitySync from '@server/lib/availabilitySync';
 import downloadTracker from '@server/lib/downloadtracker';
 import ImageProxy from '@server/lib/imageproxy';
@@ -21,7 +22,7 @@ interface ScheduledJob {
   job: schedule.Job;
   name: string;
   type: 'process' | 'command';
-  interval: 'seconds' | 'minutes' | 'hours' | 'fixed';
+  interval: 'seconds' | 'minutes' | 'hours' | 'days' | 'fixed';
   cronSchedule: string;
   running?: () => boolean;
   cancelFn?: () => void;
@@ -237,5 +238,21 @@ export const startJobs = (): void => {
     }),
   });
 
+  scheduledJobs.push({
+    id: 'process-blacklisted-tags',
+    name: 'Process Blacklisted Tags',
+    type: 'process',
+    interval: 'days',
+    cronSchedule: jobs['process-blacklisted-tags'].schedule,
+    job: schedule.scheduleJob(jobs['process-blacklisted-tags'].schedule, () => {
+      logger.info('Starting scheduled job: Process Blacklisted Tags', {
+        label: 'Jobs',
+      });
+      blacklistedTagsProcessor.run();
+    }),
+    running: () => blacklistedTagsProcessor.status().running,
+    cancelFn: () => blacklistedTagsProcessor.cancel(),
+  });
+
   logger.info('Scheduled jobs loaded', { label: 'Jobs' });
 };