fix: tag clean up query and add tests (#22633)

* fix delete empty tags query

* rewrite as a single statement

* create tag service medium test

* single tag exists, connected to one asset, and is not deleted

* do not delete parent tag if children have an asset

* hierarchical tag tests

* fix query to match 3 test

* remove transaction and format:fix

* remove transaction and format:fix

* simplify query, handle nested empty tag

* unused helper

---------

Co-authored-by: mertalev <101130780+mertalev@users.noreply.github.com>

server/src/repositories/tag.repository.ts

@@ -163,22 +163,22 @@ export class TagRepository {
   }
 
   async deleteEmptyTags() {
-    // TODO rewrite as a single statement
+    const result = await this.db
-    await this.db.transaction().execute(async (tx) => {
+      .deleteFrom('tag')
-      const result = await tx
+      .where(({ not, exists, selectFrom }) =>
-        .selectFrom('asset')
+        not(
-        .innerJoin('tag_asset', 'tag_asset.assetsId', 'asset.id')
+          exists(
-        .innerJoin('tag_closure', 'tag_closure.id_descendant', 'tag_asset.tagsId')
+            selectFrom('tag_closure')
-        .innerJoin('tag', 'tag.id', 'tag_closure.id_descendant')
+              .whereRef('tag.id', '=', 'tag_closure.id_ancestor')
-        .select((eb) => ['tag.id', eb.fn.count<number>('asset.id').as('count')])
+              .innerJoin('tag_asset', 'tag_closure.id_descendant', 'tag_asset.tagsId'),
-        .groupBy('tag.id')
+          ),
-        .execute();
+        ),
+      )
+      .executeTakeFirst();
 
-      const ids = result.filter(({ count }) => count === 0).map(({ id }) => id);
+    const deletedRows = Number(result.numDeletedRows);
-      if (ids.length > 0) {
+    if (deletedRows > 0) {
-        await this.db.deleteFrom('tag').where('id', 'in', ids).execute();
+      this.logger.log(`Deleted ${deletedRows} empty tags`);
-        this.logger.log(`Deleted ${ids.length} empty tags`);
     }
-    });
   }
 }

server/test/medium.factory.ts

@@ -39,6 +39,7 @@ import { StorageRepository } from 'src/repositories/storage.repository';
 import { SyncCheckpointRepository } from 'src/repositories/sync-checkpoint.repository';
 import { SyncRepository } from 'src/repositories/sync.repository';
 import { SystemMetadataRepository } from 'src/repositories/system-metadata.repository';
+import { TagRepository } from 'src/repositories/tag.repository';
 import { TelemetryRepository } from 'src/repositories/telemetry.repository';
 import { UserRepository } from 'src/repositories/user.repository';
 import { VersionHistoryRepository } from 'src/repositories/version-history.repository';
@@ -52,6 +53,8 @@ import { MemoryTable } from 'src/schema/tables/memory.table';
 import { PersonTable } from 'src/schema/tables/person.table';
 import { SessionTable } from 'src/schema/tables/session.table';
 import { StackTable } from 'src/schema/tables/stack.table';
+import { TagAssetTable } from 'src/schema/tables/tag-asset.table';
+import { TagTable } from 'src/schema/tables/tag.table';
 import { UserTable } from 'src/schema/tables/user.table';
 import { BASE_SERVICE_DEPENDENCIES, BaseService } from 'src/services/base.service';
 import { SyncService } from 'src/services/sync.service';
@@ -240,6 +243,18 @@ export class MediumTestContext<S extends BaseService = BaseService> {
       user,
     };
   }
+
+  async newTagAsset(tagBulkAssets: { tagIds: string[]; assetIds: string[] }) {
+    const tagsAssets: Insertable<TagAssetTable>[] = [];
+    for (const tagsId of tagBulkAssets.tagIds) {
+      for (const assetsId of tagBulkAssets.assetIds) {
+        tagsAssets.push({ tagsId, assetsId });
+      }
+    }
+
+    const result = await this.get(TagRepository).upsertAssetIds(tagsAssets);
+    return { tagsAssets, result };
+  }
 }
 
 export class SyncTestContext extends MediumTestContext<SyncService> {
@@ -318,6 +333,10 @@ const newRealRepository = <T>(key: ClassConstructor<T>, db: Kysely<DB>): T => {
       return new key(LoggingRepository.create());
     }
 
+    case TagRepository: {
+      return new key(db, LoggingRepository.create());
+    }
+
     case LoggingRepository as unknown as ClassConstructor<LoggingRepository>: {
       return new key() as unknown as T;
     }
@@ -345,7 +364,8 @@ const newMockRepository = <T>(key: ClassConstructor<T>) => {
     case SyncCheckpointRepository:
     case SystemMetadataRepository:
     case UserRepository:
-    case VersionHistoryRepository: {
+    case VersionHistoryRepository:
+    case TagRepository: {
       return automock(key);
     }
 
@@ -567,6 +587,23 @@ const memoryInsert = (memory: Partial<Insertable<MemoryTable>> = {}) => {
   return { ...defaults, ...memory, id };
 };
 
+const tagInsert = (tag: Partial<Insertable<TagTable>>) => {
+  const id = tag.id || newUuid();
+
+  const defaults: Insertable<TagTable> = {
+    id,
+    userId: '',
+    value: '',
+    createdAt: newDate(),
+    updatedAt: newDate(),
+    color: '',
+    parentId: null,
+    updateId: newUuid(),
+  };
+
+  return { ...defaults, ...tag, id };
+};
+
 class CustomWritable extends Writable {
   private data = '';
 
@@ -619,4 +656,5 @@ export const mediumFactory = {
   memoryInsert,
   loginDetails,
   loginResponse,
+  tagInsert,
 };

server/test/medium/specs/services/tag.service.spec.ts

@@ -0,0 +1,116 @@
+import { Kysely } from 'kysely';
+import { JobStatus } from 'src/enum';
+import { AccessRepository } from 'src/repositories/access.repository';
+import { LoggingRepository } from 'src/repositories/logging.repository';
+import { TagRepository } from 'src/repositories/tag.repository';
+import { DB } from 'src/schema';
+import { TagService } from 'src/services/tag.service';
+import { upsertTags } from 'src/utils/tag';
+import { newMediumService } from 'test/medium.factory';
+import { getKyselyDB } from 'test/utils';
+
+let defaultDatabase: Kysely<DB>;
+
+const setup = (db?: Kysely<DB>) => {
+  return newMediumService(TagService, {
+    database: db || defaultDatabase,
+    real: [TagRepository, AccessRepository],
+    mock: [LoggingRepository],
+  });
+};
+
+beforeAll(async () => {
+  defaultDatabase = await getKyselyDB();
+});
+
+describe(TagService.name, () => {
+  describe('deleteEmptyTags', () => {
+    it('single tag exists, not connected to any assets, and is deleted', async () => {
+      const { sut, ctx } = setup();
+      const { user } = await ctx.newUser();
+      const tagRepo = ctx.get(TagRepository);
+      const [tag] = await upsertTags(tagRepo, { userId: user.id, tags: ['tag-1'] });
+
+      await expect(tagRepo.getByValue(user.id, 'tag-1')).resolves.toEqual(expect.objectContaining({ id: tag.id }));
+      await expect(sut.handleTagCleanup()).resolves.toBe(JobStatus.Success);
+      await expect(tagRepo.getByValue(user.id, 'tag-1')).resolves.toBeUndefined();
+    });
+
+    it('single tag exists, connected to one asset, and is not deleted', async () => {
+      const { sut, ctx } = setup();
+      const { user } = await ctx.newUser();
+      const { asset } = await ctx.newAsset({ ownerId: user.id });
+      const tagRepo = ctx.get(TagRepository);
+      const [tag] = await upsertTags(tagRepo, { userId: user.id, tags: ['tag-1'] });
+
+      await ctx.newTagAsset({ tagIds: [tag.id], assetIds: [asset.id] });
+
+      await expect(tagRepo.getByValue(user.id, 'tag-1')).resolves.toEqual(expect.objectContaining({ id: tag.id }));
+      await expect(sut.handleTagCleanup()).resolves.toBe(JobStatus.Success);
+      await expect(tagRepo.getByValue(user.id, 'tag-1')).resolves.toEqual(expect.objectContaining({ id: tag.id }));
+    });
+
+    it('hierarchical tag exists, and the parent is connected to an asset, and the child is deleted', async () => {
+      const { sut, ctx } = setup();
+      const { user } = await ctx.newUser();
+      const { asset } = await ctx.newAsset({ ownerId: user.id });
+      const tagRepo = ctx.get(TagRepository);
+      const [parentTag, childTag] = await upsertTags(tagRepo, { userId: user.id, tags: ['parent', 'parent/child'] });
+
+      await ctx.newTagAsset({ tagIds: [parentTag.id], assetIds: [asset.id] });
+
+      await expect(tagRepo.getByValue(user.id, 'parent')).resolves.toEqual(
+        expect.objectContaining({ id: parentTag.id }),
+      );
+      await expect(tagRepo.getByValue(user.id, 'parent/child')).resolves.toEqual(
+        expect.objectContaining({ id: childTag.id }),
+      );
+      await expect(sut.handleTagCleanup()).resolves.toBe(JobStatus.Success);
+      await expect(tagRepo.getByValue(user.id, 'parent')).resolves.toEqual(
+        expect.objectContaining({ id: parentTag.id }),
+      );
+      await expect(tagRepo.getByValue(user.id, 'parent/child')).resolves.toBeUndefined();
+    });
+
+    it('hierarchical tag exists, and only the child is connected to an asset, and nothing is deleted', async () => {
+      const { sut, ctx } = setup();
+      const { user } = await ctx.newUser();
+      const { asset } = await ctx.newAsset({ ownerId: user.id });
+      const tagRepo = ctx.get(TagRepository);
+      const [parentTag, childTag] = await upsertTags(tagRepo, { userId: user.id, tags: ['parent', 'parent/child'] });
+
+      await ctx.newTagAsset({ tagIds: [childTag.id], assetIds: [asset.id] });
+
+      await expect(tagRepo.getByValue(user.id, 'parent')).resolves.toEqual(
+        expect.objectContaining({ id: parentTag.id }),
+      );
+      await expect(tagRepo.getByValue(user.id, 'parent/child')).resolves.toEqual(
+        expect.objectContaining({ id: childTag.id }),
+      );
+      await expect(sut.handleTagCleanup()).resolves.toBe(JobStatus.Success);
+      await expect(tagRepo.getByValue(user.id, 'parent')).resolves.toEqual(
+        expect.objectContaining({ id: parentTag.id }),
+      );
+      await expect(tagRepo.getByValue(user.id, 'parent/child')).resolves.toEqual(
+        expect.objectContaining({ id: childTag.id }),
+      );
+    });
+
+    it('hierarchical tag exists, and neither parent nor child is connected to an asset, and both are deleted', async () => {
+      const { sut, ctx } = setup();
+      const { user } = await ctx.newUser();
+      const tagRepo = ctx.get(TagRepository);
+      const [parentTag, childTag] = await upsertTags(tagRepo, { userId: user.id, tags: ['parent', 'parent/child'] });
+
+      await expect(tagRepo.getByValue(user.id, 'parent')).resolves.toEqual(
+        expect.objectContaining({ id: parentTag.id }),
+      );
+      await expect(tagRepo.getByValue(user.id, 'parent/child')).resolves.toEqual(
+        expect.objectContaining({ id: childTag.id }),
+      );
+      await expect(sut.handleTagCleanup()).resolves.toBe(JobStatus.Success);
+      await expect(tagRepo.getByValue(user.id, 'parent/child')).resolves.toBeUndefined();
+      await expect(tagRepo.getByValue(user.id, 'parent')).resolves.toBeUndefined();
+    });
+  });
+});