feat(server): library refresh go brrr (#14456)

* feat: brr

---------
Co-authored-by: mertalev <101130780+mertalev@users.noreply.github.com>
This commit is contained in:
Jonathan Jogenfors
2025-03-06 16:00:18 +01:00
committed by GitHub
parent bc61497461
commit 3af26ee94a
15 changed files with 855 additions and 531 deletions

View File

@@ -469,8 +469,9 @@ export enum JobName {
// library management
LIBRARY_QUEUE_SYNC_FILES = 'library-queue-sync-files',
LIBRARY_QUEUE_SYNC_ASSETS = 'library-queue-sync-assets',
LIBRARY_SYNC_FILE = 'library-sync-file',
LIBRARY_SYNC_ASSET = 'library-sync-asset',
LIBRARY_SYNC_FILES = 'library-sync-files',
LIBRARY_SYNC_ASSETS = 'library-sync-assets',
LIBRARY_ASSET_REMOVAL = 'handle-library-file-deletion',
LIBRARY_DELETE = 'library-delete',
LIBRARY_QUEUE_SCAN_ALL = 'library-queue-scan-all',
LIBRARY_QUEUE_CLEANUP = 'library-queue-cleanup',

View File

@@ -1,5 +1,5 @@
import { Injectable } from '@nestjs/common';
import { Insertable, Kysely, Updateable, sql } from 'kysely';
import { Insertable, Kysely, UpdateResult, Updateable, sql } from 'kysely';
import { isEmpty, isUndefined, omitBy } from 'lodash';
import { InjectKysely } from 'nestjs-kysely';
import { ASSET_FILE_CONFLICT_KEYS, EXIF_CONFLICT_KEYS, JOB_STATUS_CONFLICT_KEYS } from 'src/constants';
@@ -24,7 +24,8 @@ import {
} from 'src/entities/asset.entity';
import { AssetFileType, AssetOrder, AssetStatus, AssetType } from 'src/enum';
import { AssetSearchOptions, SearchExploreItem, SearchExploreItemSet } from 'src/repositories/search.repository';
import { anyUuid, asUuid, mapUpsertColumns } from 'src/utils/database';
import { anyUuid, asUuid, mapUpsertColumns, unnest } from 'src/utils/database';
import { globToSqlPattern } from 'src/utils/misc';
import { Paginated, PaginationOptions, paginationHelper } from 'src/utils/pagination';
export type AssetStats = Record<AssetType, number>;
@@ -191,6 +192,10 @@ export class AssetRepository {
.executeTakeFirst() as any as Promise<AssetEntityPlaceholder>;
}
createAll(assets: Insertable<Assets>[]): Promise<AssetEntity[]> {
return this.db.insertInto('assets').values(assets).returningAll().execute() as any as Promise<AssetEntity[]>;
}
@GenerateSql({ params: [DummyValue.UUID, { day: 1, month: 1 }] })
getByDayOfYear(ownerIds: string[], { day, month }: MonthDay) {
return this.db
@@ -384,6 +389,17 @@ export class AssetRepository {
return paginationHelper(items as any as AssetEntity[], pagination.take);
}
async getAllInLibrary(pagination: PaginationOptions, libraryId: string): Paginated<AssetEntity> {
const builder = this.db
.selectFrom('assets')
.select('id')
.where('libraryId', '=', asUuid(libraryId))
.limit(pagination.take + 1)
.offset(pagination.skip ?? 0);
const items = await builder.execute();
return paginationHelper(items as any as AssetEntity[], pagination.take);
}
/**
* Get assets by device's Id on the database
* @param ownerId
@@ -470,6 +486,10 @@ export class AssetRepository {
await this.db.updateTable('assets').set(options).where('id', '=', anyUuid(ids)).execute();
}
async updateByLibraryId(libraryId: string, options: Updateable<Assets>): Promise<void> {
await this.db.updateTable('assets').set(options).where('libraryId', '=', asUuid(libraryId)).execute();
}
@GenerateSql({
params: [{ targetDuplicateId: DummyValue.UUID, duplicateIds: [DummyValue.UUID], assetIds: [DummyValue.UUID] }],
})
@@ -939,4 +959,64 @@ export class AssetRepository {
)
.execute();
}
@GenerateSql({
params: [{ libraryId: DummyValue.UUID, importPaths: [DummyValue.STRING], exclusionPatterns: [DummyValue.STRING] }],
})
async detectOfflineExternalAssets(
libraryId: string,
importPaths: string[],
exclusionPatterns: string[],
): Promise<UpdateResult> {
const paths = importPaths.map((importPath) => `${importPath}%`);
const exclusions = exclusionPatterns.map((pattern) => globToSqlPattern(pattern));
return this.db
.updateTable('assets')
.set({
isOffline: true,
deletedAt: new Date(),
})
.where('isOffline', '=', false)
.where('isExternal', '=', true)
.where('libraryId', '=', asUuid(libraryId))
.where((eb) =>
eb.or([eb('originalPath', 'not like', paths.join('|')), eb('originalPath', 'like', exclusions.join('|'))]),
)
.executeTakeFirstOrThrow();
}
@GenerateSql({
params: [{ libraryId: DummyValue.UUID, paths: [DummyValue.STRING] }],
})
async filterNewExternalAssetPaths(libraryId: string, paths: string[]): Promise<string[]> {
const result = await this.db
.selectFrom(unnest(paths).as('path'))
.select('path')
.where((eb) =>
eb.not(
eb.exists(
this.db
.selectFrom('assets')
.select('originalPath')
.whereRef('assets.originalPath', '=', eb.ref('path'))
.where('libraryId', '=', asUuid(libraryId))
.where('isExternal', '=', true),
),
),
)
.execute();
return result.map((row) => row.path as string);
}
async getLibraryAssetCount(libraryId: string): Promise<number> {
const { count } = await this.db
.selectFrom('assets')
.select((eb) => eb.fn.countAll().as('count'))
.where('libraryId', '=', asUuid(libraryId))
.executeTakeFirstOrThrow();
return Number(count);
}
}

View File

@@ -26,6 +26,13 @@ const userColumns = [
'users.profileChangedAt',
] as const;
export enum AssetSyncResult {
DO_NOTHING,
UPDATE,
OFFLINE,
CHECK_OFFLINE,
}
const withOwner = (eb: ExpressionBuilder<DB, 'libraries'>) => {
return jsonObjectFrom(eb.selectFrom('users').whereRef('users.id', '=', 'libraries.ownerId').select(userColumns)).as(
'owner',

View File

@@ -6,7 +6,7 @@ import { mapLibrary } from 'src/dtos/library.dto';
import { UserEntity } from 'src/entities/user.entity';
import { AssetType, ImmichWorker, JobName, JobStatus } from 'src/enum';
import { LibraryService } from 'src/services/library.service';
import { ILibraryAssetJob, ILibraryFileJob } from 'src/types';
import { ILibraryBulkIdsJob, ILibraryFileJob } from 'src/types';
import { assetStub } from 'test/fixtures/asset.stub';
import { authStub } from 'test/fixtures/auth.stub';
import { libraryStub } from 'test/fixtures/library.stub';
@@ -22,10 +22,11 @@ async function* mockWalk() {
describe(LibraryService.name, () => {
let sut: LibraryService;
let mocks: ServiceMocks;
beforeEach(() => {
({ sut, mocks } = newTestService(LibraryService));
({ sut, mocks } = newTestService(LibraryService, {}));
mocks.database.tryLock.mockResolvedValue(true);
mocks.config.getWorker.mockReturnValue(ImmichWorker.MICROSERVICES);
@@ -152,21 +153,74 @@ describe(LibraryService.name, () => {
describe('handleQueueSyncFiles', () => {
it('should queue refresh of a new asset', async () => {
mocks.library.get.mockResolvedValue(libraryStub.externalLibrary1);
mocks.library.get.mockResolvedValue(libraryStub.externalLibraryWithImportPaths1);
mocks.storage.walk.mockImplementation(mockWalk);
mocks.storage.stat.mockResolvedValue({ isDirectory: () => true } as Stats);
mocks.storage.checkFileExists.mockResolvedValue(true);
mocks.asset.filterNewExternalAssetPaths.mockResolvedValue(['/data/user1/photo.jpg']);
await sut.handleQueueSyncFiles({ id: libraryStub.externalLibrary1.id });
await sut.handleQueueSyncFiles({ id: libraryStub.externalLibraryWithImportPaths1.id });
expect(mocks.job.queueAll).toHaveBeenCalledWith([
{
name: JobName.LIBRARY_SYNC_FILE,
data: {
id: libraryStub.externalLibrary1.id,
ownerId: libraryStub.externalLibrary1.owner.id,
assetPath: '/data/user1/photo.jpg',
},
expect(mocks.job.queue).toHaveBeenCalledWith({
name: JobName.LIBRARY_SYNC_FILES,
data: {
libraryId: libraryStub.externalLibraryWithImportPaths1.id,
paths: ['/data/user1/photo.jpg'],
progressCounter: 1,
},
]);
});
});
it("should fail when library can't be found", async () => {
await expect(sut.handleQueueSyncFiles({ id: libraryStub.externalLibraryWithImportPaths1.id })).resolves.toBe(
JobStatus.SKIPPED,
);
});
it('should ignore import paths that do not exist', async () => {
mocks.storage.stat.mockImplementation((path): Promise<Stats> => {
if (path === libraryStub.externalLibraryWithImportPaths1.importPaths[0]) {
const error = { code: 'ENOENT' } as any;
throw error;
}
return Promise.resolve({
isDirectory: () => true,
} as Stats);
});
mocks.storage.checkFileExists.mockResolvedValue(true);
mocks.library.get.mockResolvedValue(libraryStub.externalLibraryWithImportPaths1);
await sut.handleQueueSyncFiles({ id: libraryStub.externalLibraryWithImportPaths1.id });
expect(mocks.storage.walk).toHaveBeenCalledWith({
pathsToCrawl: [libraryStub.externalLibraryWithImportPaths1.importPaths[1]],
exclusionPatterns: [],
includeHidden: false,
take: JOBS_LIBRARY_PAGINATION_SIZE,
});
});
});
describe('handleQueueSyncFiles', () => {
it('should queue refresh of a new asset', async () => {
mocks.library.get.mockResolvedValue(libraryStub.externalLibraryWithImportPaths1);
mocks.storage.walk.mockImplementation(mockWalk);
mocks.storage.stat.mockResolvedValue({ isDirectory: () => true } as Stats);
mocks.storage.checkFileExists.mockResolvedValue(true);
mocks.asset.filterNewExternalAssetPaths.mockResolvedValue(['/data/user1/photo.jpg']);
await sut.handleQueueSyncFiles({ id: libraryStub.externalLibraryWithImportPaths1.id });
expect(mocks.job.queue).toHaveBeenCalledWith({
name: JobName.LIBRARY_SYNC_FILES,
data: {
libraryId: libraryStub.externalLibraryWithImportPaths1.id,
paths: ['/data/user1/photo.jpg'],
progressCounter: 1,
},
});
});
it("should fail when library can't be found", async () => {
@@ -199,142 +253,226 @@ describe(LibraryService.name, () => {
});
});
describe('handleQueueRemoveDeleted', () => {
it('should queue online check of existing assets', async () => {
describe('handleQueueSyncAssets', () => {
it('should call the offline check', async () => {
mocks.library.get.mockResolvedValue(libraryStub.externalLibrary1);
mocks.storage.walk.mockImplementation(async function* generator() {});
mocks.asset.getAll.mockResolvedValue({ items: [assetStub.external], hasNextPage: false });
mocks.asset.getLibraryAssetCount.mockResolvedValue(1);
mocks.asset.detectOfflineExternalAssets.mockResolvedValue({ numUpdatedRows: BigInt(1) });
await sut.handleQueueSyncAssets({ id: libraryStub.externalLibrary1.id });
const response = await sut.handleQueueSyncAssets({ id: libraryStub.externalLibrary1.id });
expect(mocks.job.queueAll).toHaveBeenCalledWith([
{
name: JobName.LIBRARY_SYNC_ASSET,
data: {
id: assetStub.external.id,
importPaths: libraryStub.externalLibrary1.importPaths,
exclusionPatterns: [],
},
},
]);
expect(response).toBe(JobStatus.SUCCESS);
expect(mocks.asset.detectOfflineExternalAssets).toHaveBeenCalledWith(
libraryStub.externalLibrary1.id,
libraryStub.externalLibrary1.importPaths,
libraryStub.externalLibrary1.exclusionPatterns,
);
});
it("should fail when library can't be found", async () => {
it('should skip an empty library', async () => {
mocks.library.get.mockResolvedValue(libraryStub.externalLibrary1);
mocks.storage.walk.mockImplementation(async function* generator() {});
mocks.asset.getAll.mockResolvedValue({ items: [assetStub.external], hasNextPage: false });
mocks.asset.getLibraryAssetCount.mockResolvedValue(0);
mocks.asset.detectOfflineExternalAssets.mockResolvedValue({ numUpdatedRows: BigInt(1) });
const response = await sut.handleQueueSyncAssets({ id: libraryStub.externalLibrary1.id });
expect(response).toBe(JobStatus.SUCCESS);
expect(mocks.asset.detectOfflineExternalAssets).not.toHaveBeenCalled();
});
it('should queue asset sync', async () => {
mocks.library.get.mockResolvedValue(libraryStub.externalLibraryWithImportPaths1);
mocks.storage.walk.mockImplementation(async function* generator() {});
mocks.asset.getAll.mockResolvedValue({ items: [assetStub.external], hasNextPage: false });
mocks.asset.getLibraryAssetCount.mockResolvedValue(1);
mocks.asset.detectOfflineExternalAssets.mockResolvedValue({ numUpdatedRows: BigInt(0) });
mocks.asset.getAllInLibrary.mockResolvedValue({ items: [assetStub.external], hasNextPage: false });
const response = await sut.handleQueueSyncAssets({ id: libraryStub.externalLibraryWithImportPaths1.id });
expect(mocks.job.queue).toBeCalledWith({
name: JobName.LIBRARY_SYNC_ASSETS,
data: {
libraryId: libraryStub.externalLibraryWithImportPaths1.id,
importPaths: libraryStub.externalLibraryWithImportPaths1.importPaths,
exclusionPatterns: libraryStub.externalLibraryWithImportPaths1.exclusionPatterns,
assetIds: [assetStub.external.id],
progressCounter: 1,
totalAssets: 1,
},
});
expect(response).toBe(JobStatus.SUCCESS);
expect(mocks.asset.detectOfflineExternalAssets).toHaveBeenCalledWith(
libraryStub.externalLibraryWithImportPaths1.id,
libraryStub.externalLibraryWithImportPaths1.importPaths,
libraryStub.externalLibraryWithImportPaths1.exclusionPatterns,
);
});
it("should fail if library can't be found", async () => {
await expect(sut.handleQueueSyncAssets({ id: libraryStub.externalLibrary1.id })).resolves.toBe(JobStatus.SKIPPED);
});
});
describe('handleSyncAsset', () => {
it('should skip missing assets', async () => {
const mockAssetJob: ILibraryAssetJob = {
id: assetStub.external.id,
importPaths: ['/'],
exclusionPatterns: [],
};
await expect(sut.handleSyncAsset(mockAssetJob)).resolves.toBe(JobStatus.SKIPPED);
expect(mocks.asset.remove).not.toHaveBeenCalled();
});
describe('handleSyncAssets', () => {
it('should offline assets no longer on disk', async () => {
const mockAssetJob: ILibraryAssetJob = {
id: assetStub.external.id,
const mockAssetJob: ILibraryBulkIdsJob = {
assetIds: [assetStub.external.id],
libraryId: libraryStub.externalLibrary1.id,
importPaths: ['/'],
exclusionPatterns: [],
totalAssets: 1,
progressCounter: 0,
};
mocks.asset.getById.mockResolvedValue(assetStub.external);
mocks.asset.getByIds.mockResolvedValue([assetStub.external]);
mocks.storage.stat.mockRejectedValue(new Error('ENOENT, no such file or directory'));
await expect(sut.handleSyncAsset(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
await expect(sut.handleSyncAssets(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
expect(mocks.asset.updateAll).toHaveBeenCalledWith([assetStub.external.id], {
isOffline: true,
deletedAt: expect.any(Date),
deletedAt: expect.anything(),
});
});
it('should offline assets matching an exclusion pattern', async () => {
const mockAssetJob: ILibraryAssetJob = {
id: assetStub.external.id,
importPaths: ['/'],
exclusionPatterns: ['**/user1/**'],
};
mocks.asset.getById.mockResolvedValue(assetStub.external);
await expect(sut.handleSyncAsset(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
expect(mocks.asset.updateAll).toHaveBeenCalledWith([assetStub.external.id], {
isOffline: true,
deletedAt: expect.any(Date),
});
});
it('should set assets outside of import paths as offline', async () => {
const mockAssetJob: ILibraryAssetJob = {
id: assetStub.external.id,
it('should set assets deleted from disk as offline', async () => {
const mockAssetJob: ILibraryBulkIdsJob = {
assetIds: [assetStub.external.id],
libraryId: libraryStub.externalLibrary1.id,
importPaths: ['/data/user2'],
exclusionPatterns: [],
totalAssets: 1,
progressCounter: 0,
};
mocks.asset.getById.mockResolvedValue(assetStub.external);
mocks.storage.checkFileExists.mockResolvedValue(true);
mocks.asset.getByIds.mockResolvedValue([assetStub.external]);
mocks.storage.stat.mockRejectedValue(new Error('Could not read file'));
await expect(sut.handleSyncAsset(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
await expect(sut.handleSyncAssets(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
expect(mocks.asset.updateAll).toHaveBeenCalledWith([assetStub.external.id], {
isOffline: true,
deletedAt: expect.any(Date),
deletedAt: expect.anything(),
});
});
it('should do nothing with online assets', async () => {
const mockAssetJob: ILibraryAssetJob = {
id: assetStub.external.id,
importPaths: ['/'],
it('should do nothing with offline assets deleted from disk', async () => {
const mockAssetJob: ILibraryBulkIdsJob = {
assetIds: [assetStub.trashedOffline.id],
libraryId: libraryStub.externalLibrary1.id,
importPaths: ['/data/user2'],
exclusionPatterns: [],
totalAssets: 1,
progressCounter: 0,
};
mocks.asset.getById.mockResolvedValue(assetStub.external);
mocks.storage.stat.mockResolvedValue({ mtime: assetStub.external.fileModifiedAt } as Stats);
mocks.asset.getByIds.mockResolvedValue([assetStub.trashedOffline]);
mocks.storage.stat.mockRejectedValue(new Error('Could not read file'));
await expect(sut.handleSyncAsset(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
await expect(sut.handleSyncAssets(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
expect(mocks.asset.updateAll).not.toHaveBeenCalled();
});
it('should un-trash an asset previously marked as offline', async () => {
const mockAssetJob: ILibraryAssetJob = {
id: assetStub.external.id,
importPaths: ['/'],
const mockAssetJob: ILibraryBulkIdsJob = {
assetIds: [assetStub.trashedOffline.id],
libraryId: libraryStub.externalLibrary1.id,
importPaths: ['/original/'],
exclusionPatterns: [],
totalAssets: 1,
progressCounter: 0,
};
mocks.asset.getById.mockResolvedValue(assetStub.trashedOffline);
mocks.storage.stat.mockResolvedValue({ mtime: assetStub.trashedOffline.fileModifiedAt } as Stats);
mocks.asset.getByIds.mockResolvedValue([assetStub.trashedOffline]);
mocks.storage.stat.mockResolvedValue({ mtime: assetStub.external.fileModifiedAt } as Stats);
await expect(sut.handleSyncAsset(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
await expect(sut.handleSyncAssets(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
expect(mocks.asset.updateAll).toHaveBeenCalledWith([assetStub.trashedOffline.id], {
deletedAt: null,
fileModifiedAt: assetStub.trashedOffline.fileModifiedAt,
expect(mocks.asset.updateAll).toHaveBeenCalledWith([assetStub.external.id], {
isOffline: false,
originalFileName: 'path.jpg',
deletedAt: null,
});
});
it('should not touch fileCreatedAt when un-trashing an asset previously marked as offline', async () => {
const mockAssetJob: ILibraryAssetJob = {
id: assetStub.external.id,
importPaths: ['/'],
exclusionPatterns: [],
it('should do nothing with offline asset if covered by exclusion pattern', async () => {
const mockAssetJob: ILibraryBulkIdsJob = {
assetIds: [assetStub.trashedOffline.id],
libraryId: libraryStub.externalLibrary1.id,
importPaths: ['/original/'],
exclusionPatterns: ['**/path.jpg'],
totalAssets: 1,
progressCounter: 0,
};
mocks.asset.getById.mockResolvedValue(assetStub.trashedOffline);
mocks.asset.getByIds.mockResolvedValue([assetStub.trashedOffline]);
mocks.storage.stat.mockResolvedValue({ mtime: assetStub.external.fileModifiedAt } as Stats);
await expect(sut.handleSyncAssets(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
expect(mocks.asset.updateAll).not.toHaveBeenCalled();
expect(mocks.job.queueAll).not.toHaveBeenCalled();
});
it('should do nothing with offline asset if not in import path', async () => {
const mockAssetJob: ILibraryBulkIdsJob = {
assetIds: [assetStub.trashedOffline.id],
libraryId: libraryStub.externalLibrary1.id,
importPaths: ['/import/'],
exclusionPatterns: [],
totalAssets: 1,
progressCounter: 0,
};
mocks.asset.getByIds.mockResolvedValue([assetStub.trashedOffline]);
mocks.storage.stat.mockResolvedValue({ mtime: assetStub.external.fileModifiedAt } as Stats);
await expect(sut.handleSyncAssets(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
expect(mocks.asset.updateAll).not.toHaveBeenCalled();
expect(mocks.job.queueAll).not.toHaveBeenCalled();
});
it('should do nothing with unchanged online assets', async () => {
const mockAssetJob: ILibraryBulkIdsJob = {
assetIds: [assetStub.external.id],
libraryId: libraryStub.externalLibrary1.id,
importPaths: ['/'],
exclusionPatterns: [],
totalAssets: 1,
progressCounter: 0,
};
mocks.asset.getByIds.mockResolvedValue([assetStub.external]);
mocks.storage.stat.mockResolvedValue({ mtime: assetStub.external.fileModifiedAt } as Stats);
await expect(sut.handleSyncAssets(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
expect(mocks.asset.updateAll).not.toHaveBeenCalled();
});
it('should not touch fileCreatedAt when un-trashing an asset previously marked as offline', async () => {
const mockAssetJob: ILibraryBulkIdsJob = {
assetIds: [assetStub.trashedOffline.id],
libraryId: libraryStub.externalLibrary1.id,
importPaths: ['/'],
exclusionPatterns: [],
totalAssets: 1,
progressCounter: 0,
};
mocks.asset.getByIds.mockResolvedValue([assetStub.trashedOffline]);
mocks.storage.stat.mockResolvedValue({ mtime: assetStub.trashedOffline.fileModifiedAt } as Stats);
await expect(sut.handleSyncAsset(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
await expect(sut.handleSyncAssets(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
expect(mocks.asset.updateAll).toHaveBeenCalledWith(
[assetStub.trashedOffline.id],
@@ -343,30 +481,41 @@ describe(LibraryService.name, () => {
}),
);
});
});
it('should update file when mtime has changed', async () => {
const mockAssetJob: ILibraryAssetJob = {
id: assetStub.external.id,
importPaths: ['/'],
exclusionPatterns: [],
};
it('should update with online assets that have changed', async () => {
const mockAssetJob: ILibraryBulkIdsJob = {
assetIds: [assetStub.external.id],
libraryId: libraryStub.externalLibrary1.id,
importPaths: ['/'],
exclusionPatterns: [],
totalAssets: 1,
progressCounter: 0,
};
const newMTime = new Date();
mocks.asset.getById.mockResolvedValue(assetStub.external);
mocks.storage.stat.mockResolvedValue({ mtime: newMTime } as Stats);
if (assetStub.external.fileModifiedAt == null) {
throw new Error('fileModifiedAt is null');
}
await expect(sut.handleSyncAsset(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
const mtime = new Date(assetStub.external.fileModifiedAt.getDate() + 1);
expect(mocks.asset.updateAll).toHaveBeenCalledWith([assetStub.external.id], {
fileModifiedAt: newMTime,
isOffline: false,
originalFileName: 'photo.jpg',
deletedAt: null,
mocks.asset.getByIds.mockResolvedValue([assetStub.external]);
mocks.storage.stat.mockResolvedValue({ mtime } as Stats);
await expect(sut.handleSyncAssets(mockAssetJob)).resolves.toBe(JobStatus.SUCCESS);
expect(mocks.job.queueAll).toHaveBeenCalledWith([
{
name: JobName.SIDECAR_DISCOVERY,
data: {
id: assetStub.external.id,
source: 'upload',
},
},
]);
});
});
describe('handleSyncFile', () => {
describe('handleSyncFiles', () => {
let mockUser: UserEntity;
beforeEach(() => {
@@ -381,187 +530,57 @@ describe(LibraryService.name, () => {
it('should import a new asset', async () => {
const mockLibraryJob: ILibraryFileJob = {
id: libraryStub.externalLibrary1.id,
ownerId: mockUser.id,
assetPath: '/data/user1/photo.jpg',
libraryId: libraryStub.externalLibrary1.id,
paths: ['/data/user1/photo.jpg'],
};
mocks.asset.create.mockResolvedValue(assetStub.image);
mocks.asset.createAll.mockResolvedValue([assetStub.image]);
mocks.library.get.mockResolvedValue(libraryStub.externalLibrary1);
await expect(sut.handleSyncFile(mockLibraryJob)).resolves.toBe(JobStatus.SUCCESS);
await expect(sut.handleSyncFiles(mockLibraryJob)).resolves.toBe(JobStatus.SUCCESS);
expect(mocks.asset.create.mock.calls).toEqual([
expect(mocks.asset.createAll.mock.calls).toEqual([
[
{
ownerId: mockUser.id,
libraryId: libraryStub.externalLibrary1.id,
checksum: expect.any(Buffer),
originalPath: '/data/user1/photo.jpg',
deviceAssetId: expect.any(String),
deviceId: 'Library Import',
fileCreatedAt: expect.any(Date),
fileModifiedAt: expect.any(Date),
localDateTime: expect.any(Date),
type: AssetType.IMAGE,
originalFileName: 'photo.jpg',
isExternal: true,
},
[
expect.objectContaining({
ownerId: mockUser.id,
libraryId: libraryStub.externalLibrary1.id,
originalPath: '/data/user1/photo.jpg',
deviceId: 'Library Import',
type: AssetType.IMAGE,
originalFileName: 'photo.jpg',
isExternal: true,
}),
],
],
]);
expect(mocks.job.queue.mock.calls).toEqual([
expect(mocks.job.queueAll.mock.calls).toEqual([
[
{
name: JobName.SIDECAR_DISCOVERY,
data: {
id: assetStub.image.id,
source: 'upload',
[
{
name: JobName.SIDECAR_DISCOVERY,
data: {
id: assetStub.image.id,
source: 'upload',
},
},
},
],
]);
});
it('should import a new video', async () => {
const mockLibraryJob: ILibraryFileJob = {
id: libraryStub.externalLibrary1.id,
ownerId: mockUser.id,
assetPath: '/data/user1/video.mp4',
};
mocks.asset.create.mockResolvedValue(assetStub.video);
mocks.library.get.mockResolvedValue(libraryStub.externalLibrary1);
await expect(sut.handleSyncFile(mockLibraryJob)).resolves.toBe(JobStatus.SUCCESS);
expect(mocks.asset.create.mock.calls).toEqual([
[
{
ownerId: mockUser.id,
libraryId: libraryStub.externalLibrary1.id,
checksum: expect.any(Buffer),
originalPath: '/data/user1/video.mp4',
deviceAssetId: expect.any(String),
deviceId: 'Library Import',
fileCreatedAt: expect.any(Date),
fileModifiedAt: expect.any(Date),
localDateTime: expect.any(Date),
type: AssetType.VIDEO,
originalFileName: 'video.mp4',
isExternal: true,
},
],
]);
expect(mocks.job.queue.mock.calls).toEqual([
[
{
name: JobName.SIDECAR_DISCOVERY,
data: {
id: assetStub.image.id,
source: 'upload',
},
},
],
],
]);
});
it('should not import an asset to a soft deleted library', async () => {
const mockLibraryJob: ILibraryFileJob = {
id: libraryStub.externalLibrary1.id,
ownerId: mockUser.id,
assetPath: '/data/user1/photo.jpg',
libraryId: libraryStub.externalLibrary1.id,
paths: ['/data/user1/photo.jpg'],
};
mocks.asset.create.mockResolvedValue(assetStub.image);
mocks.library.get.mockResolvedValue({ ...libraryStub.externalLibrary1, deletedAt: new Date() });
await expect(sut.handleSyncFile(mockLibraryJob)).resolves.toBe(JobStatus.FAILED);
await expect(sut.handleSyncFiles(mockLibraryJob)).resolves.toBe(JobStatus.FAILED);
expect(mocks.asset.create.mock.calls).toEqual([]);
});
it('should not refresh a file whose mtime matches existing asset', async () => {
const mockLibraryJob: ILibraryFileJob = {
id: libraryStub.externalLibrary1.id,
ownerId: mockUser.id,
assetPath: assetStub.hasFileExtension.originalPath,
};
mocks.storage.stat.mockResolvedValue({
size: 100,
mtime: assetStub.hasFileExtension.fileModifiedAt,
ctime: new Date('2023-01-01'),
} as Stats);
mocks.asset.getByLibraryIdAndOriginalPath.mockResolvedValue(assetStub.hasFileExtension);
await expect(sut.handleSyncFile(mockLibraryJob)).resolves.toBe(JobStatus.SKIPPED);
expect(mocks.job.queue).not.toHaveBeenCalled();
expect(mocks.job.queueAll).not.toHaveBeenCalled();
});
it('should skip existing asset', async () => {
const mockLibraryJob: ILibraryFileJob = {
id: libraryStub.externalLibrary1.id,
ownerId: mockUser.id,
assetPath: '/data/user1/photo.jpg',
};
mocks.asset.getByLibraryIdAndOriginalPath.mockResolvedValue(assetStub.image);
await expect(sut.handleSyncFile(mockLibraryJob)).resolves.toBe(JobStatus.SKIPPED);
});
it('should not refresh an asset trashed by user', async () => {
const mockLibraryJob: ILibraryFileJob = {
id: libraryStub.externalLibrary1.id,
ownerId: mockUser.id,
assetPath: assetStub.hasFileExtension.originalPath,
};
mocks.asset.getByLibraryIdAndOriginalPath.mockResolvedValue(assetStub.trashed);
await expect(sut.handleSyncFile(mockLibraryJob)).resolves.toBe(JobStatus.SKIPPED);
expect(mocks.job.queue).not.toHaveBeenCalled();
expect(mocks.job.queueAll).not.toHaveBeenCalled();
});
it('should fail when the file could not be read', async () => {
mocks.storage.stat.mockRejectedValue(new Error('Could not read file'));
const mockLibraryJob: ILibraryFileJob = {
id: libraryStub.externalLibrary1.id,
ownerId: userStub.admin.id,
assetPath: '/data/user1/photo.jpg',
};
mocks.asset.create.mockResolvedValue(assetStub.image);
await expect(sut.handleSyncFile(mockLibraryJob)).resolves.toBe(JobStatus.FAILED);
expect(mocks.library.get).not.toHaveBeenCalled();
expect(mocks.asset.create).not.toHaveBeenCalled();
});
it('should skip if the file could not be found', async () => {
const error = new Error('File not found') as any;
error.code = 'ENOENT';
mocks.storage.stat.mockRejectedValue(error);
const mockLibraryJob: ILibraryFileJob = {
id: libraryStub.externalLibrary1.id,
ownerId: userStub.admin.id,
assetPath: '/data/user1/photo.jpg',
};
mocks.asset.create.mockResolvedValue(assetStub.image);
await expect(sut.handleSyncFile(mockLibraryJob)).resolves.toBe(JobStatus.SKIPPED);
expect(mocks.library.get).not.toHaveBeenCalled();
expect(mocks.asset.create).not.toHaveBeenCalled();
expect(mocks.asset.createAll.mock.calls).toEqual([]);
});
});
@@ -641,10 +660,6 @@ describe(LibraryService.name, () => {
expect(mocks.library.getStatistics).toHaveBeenCalledWith(libraryStub.externalLibrary1.id);
});
it('should throw an error if the library could not be found', async () => {
await expect(sut.getStatistics('foo')).rejects.toBeInstanceOf(BadRequestException);
});
});
describe('create', () => {
@@ -898,19 +913,13 @@ describe(LibraryService.name, () => {
await sut.watchAll();
expect(mocks.job.queueAll).toHaveBeenCalledWith([
{
name: JobName.LIBRARY_SYNC_FILE,
data: {
id: libraryStub.externalLibraryWithImportPaths1.id,
assetPath: '/foo/photo.jpg',
ownerId: libraryStub.externalLibraryWithImportPaths1.owner.id,
},
expect(mocks.job.queue).toHaveBeenCalledWith({
name: JobName.LIBRARY_SYNC_FILES,
data: {
libraryId: libraryStub.externalLibraryWithImportPaths1.id,
paths: ['/foo/photo.jpg'],
},
]);
expect(mocks.job.queueAll).toHaveBeenCalledWith([
{ name: JobName.LIBRARY_SYNC_ASSET, data: expect.objectContaining({ id: assetStub.image.id }) },
]);
});
});
it('should handle a file change event', async () => {
@@ -923,19 +932,13 @@ describe(LibraryService.name, () => {
await sut.watchAll();
expect(mocks.job.queueAll).toHaveBeenCalledWith([
{
name: JobName.LIBRARY_SYNC_FILE,
data: {
id: libraryStub.externalLibraryWithImportPaths1.id,
assetPath: '/foo/photo.jpg',
ownerId: libraryStub.externalLibraryWithImportPaths1.owner.id,
},
expect(mocks.job.queue).toHaveBeenCalledWith({
name: JobName.LIBRARY_SYNC_FILES,
data: {
libraryId: libraryStub.externalLibraryWithImportPaths1.id,
paths: ['/foo/photo.jpg'],
},
]);
expect(mocks.job.queueAll).toHaveBeenCalledWith([
{ name: JobName.LIBRARY_SYNC_ASSET, data: expect.objectContaining({ id: assetStub.image.id }) },
]);
});
});
it('should handle a file unlink event', async () => {
@@ -943,14 +946,18 @@ describe(LibraryService.name, () => {
mocks.library.getAll.mockResolvedValue([libraryStub.externalLibraryWithImportPaths1]);
mocks.asset.getByLibraryIdAndOriginalPath.mockResolvedValue(assetStub.image);
mocks.storage.watch.mockImplementation(
makeMockWatcher({ items: [{ event: 'unlink', value: '/foo/photo.jpg' }] }),
makeMockWatcher({ items: [{ event: 'unlink', value: assetStub.image.originalPath }] }),
);
await sut.watchAll();
expect(mocks.job.queueAll).toHaveBeenCalledWith([
{ name: JobName.LIBRARY_SYNC_ASSET, data: expect.objectContaining({ id: assetStub.image.id }) },
]);
expect(mocks.job.queue).toHaveBeenCalledWith({
name: JobName.LIBRARY_ASSET_REMOVAL,
data: {
libraryId: libraryStub.externalLibraryWithImportPaths1.id,
paths: [assetStub.image.originalPath],
},
});
});
it('should handle an error event', async () => {
@@ -966,10 +973,10 @@ describe(LibraryService.name, () => {
await expect(sut.watchAll()).resolves.toBeUndefined();
});
it('should ignore unknown extensions', async () => {
it('should not import a file with unknown extension', async () => {
mocks.library.get.mockResolvedValue(libraryStub.externalLibraryWithImportPaths1);
mocks.library.getAll.mockResolvedValue([libraryStub.externalLibraryWithImportPaths1]);
mocks.storage.watch.mockImplementation(makeMockWatcher({ items: [{ event: 'add', value: '/foo/photo.jpg' }] }));
mocks.storage.watch.mockImplementation(makeMockWatcher({ items: [{ event: 'add', value: '/foo/photo.xyz' }] }));
await sut.watchAll();
@@ -1100,27 +1107,6 @@ describe(LibraryService.name, () => {
});
});
describe('handleQueueAssetOfflineCheck', () => {
it('should queue removal jobs', async () => {
mocks.library.get.mockResolvedValue(libraryStub.externalLibrary1);
mocks.asset.getAll.mockResolvedValue({ items: [assetStub.image1], hasNextPage: false });
mocks.asset.getById.mockResolvedValue(assetStub.image1);
await expect(sut.handleQueueSyncAssets({ id: libraryStub.externalLibrary1.id })).resolves.toBe(JobStatus.SUCCESS);
expect(mocks.job.queueAll).toHaveBeenCalledWith([
{
name: JobName.LIBRARY_SYNC_ASSET,
data: {
id: assetStub.image1.id,
importPaths: libraryStub.externalLibrary1.importPaths,
exclusionPatterns: libraryStub.externalLibrary1.exclusionPatterns,
},
},
]);
});
});
describe('validate', () => {
it('should not require import paths', async () => {
await expect(sut.validate('library-id', {})).resolves.toEqual({ importPaths: [] });

View File

@@ -1,5 +1,6 @@
import { BadRequestException, Injectable } from '@nestjs/common';
import { R_OK } from 'node:constants';
import { Stats } from 'node:fs';
import path, { basename, isAbsolute, parse } from 'node:path';
import picomatch from 'picomatch';
import { JOBS_LIBRARY_PAGINATION_SIZE } from 'src/constants';
@@ -16,9 +17,9 @@ import {
ValidateLibraryResponseDto,
} from 'src/dtos/library.dto';
import { AssetEntity } from 'src/entities/asset.entity';
import { LibraryEntity } from 'src/entities/library.entity';
import { AssetType, DatabaseLock, ImmichWorker, JobName, JobStatus, QueueName } from 'src/enum';
import { AssetStatus, AssetType, DatabaseLock, ImmichWorker, JobName, JobStatus, QueueName } from 'src/enum';
import { ArgOf } from 'src/repositories/event.repository';
import { AssetSyncResult } from 'src/repositories/library.repository';
import { BaseService } from 'src/services/base.service';
import { JobOf } from 'src/types';
import { mimeTypes } from 'src/utils/mime-types';
@@ -98,6 +99,26 @@ export class LibraryService extends BaseService {
let _resolve: () => void;
const ready$ = new Promise<void>((resolve) => (_resolve = resolve));
const handler = async (event: string, path: string) => {
if (matcher(path)) {
this.logger.debug(`File ${event} event received for ${path} in library ${library.id}}`);
await this.jobRepository.queue({
name: JobName.LIBRARY_SYNC_FILES,
data: { libraryId: library.id, paths: [path] },
});
} else {
this.logger.verbose(`Ignoring file ${event} event for ${path} in library ${library.id}`);
}
};
const deletionHandler = async (path: string) => {
this.logger.debug(`File unlink event received for ${path} in library ${library.id}}`);
await this.jobRepository.queue({
name: JobName.LIBRARY_ASSET_REMOVAL,
data: { libraryId: library.id, paths: [path] },
});
};
this.watchers[id] = this.storageRepository.watch(
library.importPaths,
{
@@ -107,43 +128,13 @@ export class LibraryService extends BaseService {
{
onReady: () => _resolve(),
onAdd: (path) => {
const handler = async () => {
this.logger.debug(`File add event received for ${path} in library ${library.id}}`);
if (matcher(path)) {
const asset = await this.assetRepository.getByLibraryIdAndOriginalPath(library.id, path);
if (asset) {
await this.syncAssets(library, [asset.id]);
}
if (matcher(path)) {
await this.syncFiles(library, [path]);
}
}
};
return handlePromiseError(handler(), this.logger);
return handlePromiseError(handler('add', path), this.logger);
},
onChange: (path) => {
const handler = async () => {
this.logger.debug(`Detected file change for ${path} in library ${library.id}`);
const asset = await this.assetRepository.getByLibraryIdAndOriginalPath(library.id, path);
if (asset) {
await this.syncAssets(library, [asset.id]);
}
if (matcher(path)) {
// Note: if the changed file was not previously imported, it will be imported now.
await this.syncFiles(library, [path]);
}
};
return handlePromiseError(handler(), this.logger);
return handlePromiseError(handler('change', path), this.logger);
},
onUnlink: (path) => {
const handler = async () => {
this.logger.debug(`Detected deleted file at ${path} in library ${library.id}`);
const asset = await this.assetRepository.getByLibraryIdAndOriginalPath(library.id, path);
if (asset) {
await this.syncAssets(library, [asset.id]);
}
};
return handlePromiseError(handler(), this.logger);
return handlePromiseError(deletionHandler(path), this.logger);
},
onError: (error) => {
this.logger.error(`Library watcher for library ${library.id} encountered error: ${error}`);
@@ -234,26 +225,38 @@ export class LibraryService extends BaseService {
return mapLibrary(library);
}
private async syncFiles({ id, ownerId }: LibraryEntity, assetPaths: string[]) {
await this.jobRepository.queueAll(
assetPaths.map((assetPath) => ({
name: JobName.LIBRARY_SYNC_FILE,
data: {
id,
assetPath,
ownerId,
},
})),
);
}
@OnJob({ name: JobName.LIBRARY_SYNC_FILES, queue: QueueName.LIBRARY })
async handleSyncFiles(job: JobOf<JobName.LIBRARY_SYNC_FILES>): Promise<JobStatus> {
const library = await this.libraryRepository.get(job.libraryId);
// We need to check if the library still exists as it could have been deleted after the scan was queued
if (!library) {
this.logger.debug(`Library ${job.libraryId} not found, skipping file import`);
return JobStatus.FAILED;
} else if (library.deletedAt) {
this.logger.debug(`Library ${job.libraryId} is deleted, won't import assets into it`);
return JobStatus.FAILED;
}
private async syncAssets({ importPaths, exclusionPatterns }: LibraryEntity, assetIds: string[]) {
await this.jobRepository.queueAll(
assetIds.map((assetId) => ({
name: JobName.LIBRARY_SYNC_ASSET,
data: { id: assetId, importPaths, exclusionPatterns },
})),
);
const assetImports = job.paths.map((assetPath) => this.processEntity(assetPath, library.ownerId, job.libraryId));
const assetIds: string[] = [];
for (let i = 0; i < assetImports.length; i += 5000) {
// Chunk the imports to avoid the postgres limit of max parameters at once
const chunk = assetImports.slice(i, i + 5000);
await this.assetRepository.createAll(chunk).then((assets) => assetIds.push(...assets.map((asset) => asset.id)));
}
const progressMessage =
job.progressCounter && job.totalAssets
? `(${job.progressCounter} of ${job.totalAssets})`
: `(${job.progressCounter} done so far)`;
this.logger.log(`Imported ${assetIds.length} ${progressMessage} file(s) into library ${job.libraryId}`);
await this.queuePostSyncJobs(assetIds);
return JobStatus.SUCCESS;
}
private async validateImportPath(importPath: string): Promise<ValidateLibraryImportPathResponseDto> {
@@ -336,6 +339,8 @@ export class LibraryService extends BaseService {
async handleDeleteLibrary(job: JobOf<JobName.LIBRARY_DELETE>): Promise<JobStatus> {
const libraryId = job.id;
await this.assetRepository.updateByLibraryId(libraryId, { deletedAt: new Date() });
const assetPagination = usePagination(JOBS_LIBRARY_PAGINATION_SIZE, (pagination) =>
this.assetRepository.getAll(pagination, { libraryId, withDeleted: true }),
);
@@ -367,84 +372,52 @@ export class LibraryService extends BaseService {
return JobStatus.SUCCESS;
}
@OnJob({ name: JobName.LIBRARY_SYNC_FILE, queue: QueueName.LIBRARY })
async handleSyncFile(job: JobOf<JobName.LIBRARY_SYNC_FILE>): Promise<JobStatus> {
// Only needs to handle new assets
const assetPath = path.normalize(job.assetPath);
private processEntity(filePath: string, ownerId: string, libraryId: string) {
const assetPath = path.normalize(filePath);
let asset = await this.assetRepository.getByLibraryIdAndOriginalPath(job.id, assetPath);
if (asset) {
return JobStatus.SKIPPED;
}
let stat;
try {
stat = await this.storageRepository.stat(assetPath);
} catch (error: any) {
if (error.code === 'ENOENT') {
this.logger.error(`File not found: ${assetPath}`);
return JobStatus.SKIPPED;
}
this.logger.error(`Error reading file: ${assetPath}. Error: ${error}`);
return JobStatus.FAILED;
}
this.logger.log(`Importing new library asset: ${assetPath}`);
const library = await this.libraryRepository.get(job.id, true);
if (!library || library.deletedAt) {
this.logger.error('Cannot import asset into deleted library');
return JobStatus.FAILED;
}
// TODO: device asset id is deprecated, remove it
const deviceAssetId = `${basename(assetPath)}`.replaceAll(/\s+/g, '');
const pathHash = this.cryptoRepository.hashSha1(`path:${assetPath}`);
const assetType = mimeTypes.isVideo(assetPath) ? AssetType.VIDEO : AssetType.IMAGE;
const mtime = stat.mtime;
asset = await this.assetRepository.create({
ownerId: job.ownerId,
libraryId: job.id,
checksum: pathHash,
return {
ownerId,
libraryId,
checksum: this.cryptoRepository.hashSha1(`path:${assetPath}`),
originalPath: assetPath,
deviceAssetId,
fileCreatedAt: null,
fileModifiedAt: null,
localDateTime: null,
// TODO: device asset id is deprecated, remove it
deviceAssetId: `${basename(assetPath)}`.replaceAll(/\s+/g, ''),
deviceId: 'Library Import',
fileCreatedAt: mtime,
fileModifiedAt: mtime,
localDateTime: mtime,
type: assetType,
type: mimeTypes.isVideo(assetPath) ? AssetType.VIDEO : AssetType.IMAGE,
originalFileName: parse(assetPath).base,
isExternal: true,
});
await this.queuePostSyncJobs(asset);
return JobStatus.SUCCESS;
livePhotoVideoId: null,
};
}
async queuePostSyncJobs(asset: AssetEntity) {
this.logger.debug(`Queueing metadata extraction for: ${asset.originalPath}`);
async queuePostSyncJobs(assetIds: string[]) {
this.logger.debug(`Queuing sidecar discovery for ${assetIds.length} asset(s)`);
// We queue a sidecar discovery which, in turn, queues metadata extraction
await this.jobRepository.queue({
name: JobName.SIDECAR_DISCOVERY,
data: { id: asset.id, source: 'upload' },
});
await this.jobRepository.queueAll(
assetIds.map((assetId) => ({
name: JobName.SIDECAR_DISCOVERY,
data: { id: assetId, source: 'upload' },
})),
);
}
async queueScan(id: string) {
await this.findOrFail(id);
this.logger.log(`Starting to scan library ${id}`);
await this.jobRepository.queue({
name: JobName.LIBRARY_QUEUE_SYNC_FILES,
data: {
id,
},
});
await this.jobRepository.queue({ name: JobName.LIBRARY_QUEUE_SYNC_ASSETS, data: { id } });
}
@@ -454,11 +427,12 @@ export class LibraryService extends BaseService {
@OnJob({ name: JobName.LIBRARY_QUEUE_SCAN_ALL, queue: QueueName.LIBRARY })
async handleQueueScanAll(): Promise<JobStatus> {
this.logger.log(`Refreshing all external libraries`);
this.logger.log(`Initiating scan of all external libraries...`);
await this.jobRepository.queue({ name: JobName.LIBRARY_QUEUE_CLEANUP, data: {} });
const libraries = await this.libraryRepository.getAll(true);
await this.jobRepository.queueAll(
libraries.map((library) => ({
name: JobName.LIBRARY_QUEUE_SYNC_FILES,
@@ -475,64 +449,141 @@ export class LibraryService extends BaseService {
},
})),
);
return JobStatus.SUCCESS;
}
@OnJob({ name: JobName.LIBRARY_SYNC_ASSET, queue: QueueName.LIBRARY })
async handleSyncAsset(job: JobOf<JobName.LIBRARY_SYNC_ASSET>): Promise<JobStatus> {
const asset = await this.assetRepository.getById(job.id);
if (!asset) {
return JobStatus.SKIPPED;
}
@OnJob({ name: JobName.LIBRARY_SYNC_ASSETS, queue: QueueName.LIBRARY })
async handleSyncAssets(job: JobOf<JobName.LIBRARY_SYNC_ASSETS>): Promise<JobStatus> {
const assets = await this.assetRepository.getByIds(job.assetIds);
const markOffline = async (explanation: string) => {
if (!asset.isOffline) {
this.logger.debug(`${explanation}, removing: ${asset.originalPath}`);
await this.assetRepository.updateAll([asset.id], { isOffline: true, deletedAt: new Date() });
const assetIdsToOffline: string[] = [];
const trashedAssetIdsToOffline: string[] = [];
const assetIdsToOnline: string[] = [];
const trashedAssetIdsToOnline: string[] = [];
const assetIdsToUpdate: string[] = [];
this.logger.debug(`Checking batch of ${assets.length} existing asset(s) in library ${job.libraryId}`);
const stats = await Promise.all(
assets.map((asset) => this.storageRepository.stat(asset.originalPath).catch(() => null)),
);
for (let i = 0; i < assets.length; i++) {
const asset = assets[i];
const stat = stats[i];
const action = this.checkExistingAsset(asset, stat);
switch (action) {
case AssetSyncResult.OFFLINE: {
if (asset.status === AssetStatus.TRASHED) {
trashedAssetIdsToOffline.push(asset.id);
} else {
assetIdsToOffline.push(asset.id);
}
break;
}
case AssetSyncResult.UPDATE: {
assetIdsToUpdate.push(asset.id);
break;
}
case AssetSyncResult.CHECK_OFFLINE: {
const isInImportPath = job.importPaths.find((path) => asset.originalPath.startsWith(path));
if (!isInImportPath) {
this.logger.verbose(
`Offline asset ${asset.originalPath} is still not in any import path, keeping offline in library ${job.libraryId}`,
);
break;
}
const isExcluded = job.exclusionPatterns.some((pattern) => picomatch.isMatch(asset.originalPath, pattern));
if (!isExcluded) {
this.logger.debug(`Offline asset ${asset.originalPath} is now online in library ${job.libraryId}`);
if (asset.status === AssetStatus.TRASHED) {
trashedAssetIdsToOnline.push(asset.id);
} else {
assetIdsToOnline.push(asset.id);
}
break;
}
this.logger.verbose(
`Offline asset ${asset.originalPath} is in an import path but still covered by exclusion pattern, keeping offline in library ${job.libraryId}`,
);
break;
}
}
};
const isInPath = job.importPaths.find((path) => asset.originalPath.startsWith(path));
if (!isInPath) {
await markOffline('Asset is no longer in an import path');
return JobStatus.SUCCESS;
}
const isExcluded = job.exclusionPatterns.some((pattern) => picomatch.isMatch(asset.originalPath, pattern));
if (isExcluded) {
await markOffline('Asset is covered by an exclusion pattern');
return JobStatus.SUCCESS;
const promises = [];
if (assetIdsToOffline.length > 0) {
promises.push(this.assetRepository.updateAll(assetIdsToOffline, { isOffline: true, deletedAt: new Date() }));
}
let stat;
try {
stat = await this.storageRepository.stat(asset.originalPath);
} catch {
await markOffline('Asset is no longer on disk or is inaccessible because of permissions');
return JobStatus.SUCCESS;
if (trashedAssetIdsToOffline.length > 0) {
promises.push(this.assetRepository.updateAll(trashedAssetIdsToOffline, { isOffline: true }));
}
const mtime = stat.mtime;
const isAssetModified = !asset.fileModifiedAt || mtime.toISOString() !== asset.fileModifiedAt.toISOString();
if (asset.isOffline || isAssetModified) {
this.logger.debug(`Asset was offline or modified, updating asset record ${asset.originalPath}`);
//TODO: When we have asset status, we need to leave deletedAt as is when status is trashed
await this.assetRepository.updateAll([asset.id], {
isOffline: false,
deletedAt: null,
fileModifiedAt: mtime,
originalFileName: parse(asset.originalPath).base,
});
if (assetIdsToOnline.length > 0) {
promises.push(this.assetRepository.updateAll(assetIdsToOnline, { isOffline: false, deletedAt: null }));
}
if (isAssetModified) {
this.logger.debug(`Asset was modified, queuing metadata extraction for: ${asset.originalPath}`);
await this.queuePostSyncJobs(asset);
if (trashedAssetIdsToOnline.length > 0) {
promises.push(this.assetRepository.updateAll(trashedAssetIdsToOnline, { isOffline: false }));
}
if (assetIdsToUpdate.length > 0) {
promises.push(this.queuePostSyncJobs(assetIdsToUpdate));
}
await Promise.all(promises);
const remainingCount = assets.length - assetIdsToOffline.length - assetIdsToUpdate.length - assetIdsToOnline.length;
const cumulativePercentage = ((100 * job.progressCounter) / job.totalAssets).toFixed(1);
this.logger.log(
`Checked existing asset(s): ${assetIdsToOffline.length + trashedAssetIdsToOffline.length} offlined, ${assetIdsToOnline.length + trashedAssetIdsToOnline.length} onlined, ${assetIdsToUpdate.length} updated, ${remainingCount} unchanged of current batch of ${assets.length} (Total progress: ${job.progressCounter} of ${job.totalAssets}, ${cumulativePercentage} %) in library ${job.libraryId}.`,
);
return JobStatus.SUCCESS;
}
private checkExistingAsset(asset: AssetEntity, stat: Stats | null): AssetSyncResult {
if (!stat) {
// File not found on disk or permission error
if (asset.isOffline) {
this.logger.verbose(
`Asset ${asset.originalPath} is still not accessible, keeping offline in library ${asset.libraryId}`,
);
return AssetSyncResult.DO_NOTHING;
}
this.logger.debug(
`Asset ${asset.originalPath} is no longer on disk or is inaccessible because of permissions, marking offline in library ${asset.libraryId}`,
);
return AssetSyncResult.OFFLINE;
}
if (asset.isOffline && asset.status !== AssetStatus.DELETED) {
// Only perform the expensive check if the asset is offline
return AssetSyncResult.CHECK_OFFLINE;
}
if (
!asset.fileCreatedAt ||
!asset.localDateTime ||
!asset.fileModifiedAt ||
stat.mtime.valueOf() !== asset.fileModifiedAt.valueOf()
) {
this.logger.verbose(`Asset ${asset.originalPath} needs metadata extraction in library ${asset.libraryId}`);
return AssetSyncResult.UPDATE;
}
return AssetSyncResult.DO_NOTHING;
}
@OnJob({ name: JobName.LIBRARY_QUEUE_SYNC_FILES, queue: QueueName.LIBRARY })
async handleQueueSyncFiles(job: JobOf<JobName.LIBRARY_QUEUE_SYNC_FILES>): Promise<JobStatus> {
const library = await this.libraryRepository.get(job.id);
@@ -541,7 +592,7 @@ export class LibraryService extends BaseService {
return JobStatus.SKIPPED;
}
this.logger.log(`Refreshing library ${library.id} for new assets`);
this.logger.debug(`Validating import paths for library ${library.id}...`);
const validImportPaths: string[] = [];
@@ -556,35 +607,67 @@ export class LibraryService extends BaseService {
if (validImportPaths.length === 0) {
this.logger.warn(`No valid import paths found for library ${library.id}`);
return JobStatus.SKIPPED;
}
const assetsOnDisk = this.storageRepository.walk({
const pathsOnDisk = this.storageRepository.walk({
pathsToCrawl: validImportPaths,
includeHidden: false,
exclusionPatterns: library.exclusionPatterns,
take: JOBS_LIBRARY_PAGINATION_SIZE,
});
let count = 0;
let importCount = 0;
let crawlCount = 0;
for await (const assetBatch of assetsOnDisk) {
count += assetBatch.length;
this.logger.debug(`Discovered ${count} asset(s) on disk for library ${library.id}...`);
await this.syncFiles(library, assetBatch);
this.logger.verbose(`Queued scan of ${assetBatch.length} crawled asset(s) in library ${library.id}...`);
this.logger.log(`Starting disk crawl of ${validImportPaths.length} import path(s) for library ${library.id}...`);
for await (const pathBatch of pathsOnDisk) {
crawlCount += pathBatch.length;
const paths = await this.assetRepository.filterNewExternalAssetPaths(library.id, pathBatch);
if (paths.length > 0) {
importCount += paths.length;
await this.jobRepository.queue({
name: JobName.LIBRARY_SYNC_FILES,
data: {
libraryId: library.id,
paths,
progressCounter: crawlCount,
},
});
}
this.logger.log(
`Crawled ${crawlCount} file(s) so far: ${paths.length} of current batch of ${pathBatch.length} will be imported to library ${library.id}...`,
);
}
if (count > 0) {
this.logger.debug(`Finished queueing scan of ${count} assets on disk for library ${library.id}`);
} else if (validImportPaths.length > 0) {
this.logger.debug(`No non-excluded assets found in any import path for library ${library.id}`);
}
this.logger.log(
`Finished disk crawl, ${crawlCount} file(s) found on disk and queued ${importCount} file(s) for import into ${library.id}`,
);
await this.libraryRepository.update(job.id, { refreshedAt: new Date() });
return JobStatus.SUCCESS;
}
@OnJob({ name: JobName.LIBRARY_ASSET_REMOVAL, queue: QueueName.LIBRARY })
async handleAssetRemoval(job: JobOf<JobName.LIBRARY_ASSET_REMOVAL>): Promise<JobStatus> {
// This is only for handling file unlink events via the file watcher
this.logger.verbose(`Deleting asset(s) ${job.paths} from library ${job.libraryId}`);
for (const assetPath of job.paths) {
const asset = await this.assetRepository.getByLibraryIdAndOriginalPath(job.libraryId, assetPath);
if (asset) {
await this.assetRepository.remove(asset);
}
}
return JobStatus.SUCCESS;
}
@OnJob({ name: JobName.LIBRARY_QUEUE_SYNC_ASSETS, queue: QueueName.LIBRARY })
async handleQueueSyncAssets(job: JobOf<JobName.LIBRARY_QUEUE_SYNC_ASSETS>): Promise<JobStatus> {
const library = await this.libraryRepository.get(job.id);
@@ -592,27 +675,68 @@ export class LibraryService extends BaseService {
return JobStatus.SKIPPED;
}
this.logger.log(`Scanning library ${library.id} for removed assets`);
const assetCount = await this.assetRepository.getLibraryAssetCount(job.id);
const onlineAssets = usePagination(JOBS_LIBRARY_PAGINATION_SIZE, (pagination) =>
this.assetRepository.getAll(pagination, { libraryId: job.id, withDeleted: true }),
);
let assetCount = 0;
for await (const assets of onlineAssets) {
assetCount += assets.length;
this.logger.debug(`Discovered ${assetCount} asset(s) in library ${library.id}...`);
await this.jobRepository.queueAll(
assets.map((asset) => ({
name: JobName.LIBRARY_SYNC_ASSET,
data: { id: asset.id, importPaths: library.importPaths, exclusionPatterns: library.exclusionPatterns },
})),
);
this.logger.debug(`Queued check of ${assets.length} asset(s) in library ${library.id}...`);
if (!assetCount) {
this.logger.log(`Library ${library.id} is empty, no need to check assets`);
return JobStatus.SUCCESS;
}
if (assetCount) {
this.logger.log(`Finished queueing check of ${assetCount} assets for library ${library.id}`);
this.logger.log(
`Checking ${assetCount} asset(s) against import paths and exclusion patterns in library ${library.id}...`,
);
const offlineResult = await this.assetRepository.detectOfflineExternalAssets(
library.id,
library.importPaths,
library.exclusionPatterns,
);
const affectedAssetCount = Number(offlineResult.numUpdatedRows);
this.logger.log(
`${affectedAssetCount} asset(s) out of ${assetCount} were offlined due to import paths and/or exclusion pattern(s) in library ${library.id}`,
);
if (affectedAssetCount === assetCount) {
return JobStatus.SUCCESS;
}
this.logger.log(`Scanning library ${library.id} for assets missing from disk...`);
const existingAssets = usePagination(JOBS_LIBRARY_PAGINATION_SIZE, (pagination) =>
this.assetRepository.getAllInLibrary(pagination, job.id),
);
let currentAssetCount = 0;
for await (const assets of existingAssets) {
if (assets.length === 0) {
throw new BadRequestException(`Failed to get assets for library ${job.id}`);
}
currentAssetCount += assets.length;
await this.jobRepository.queue({
name: JobName.LIBRARY_SYNC_ASSETS,
data: {
libraryId: library.id,
importPaths: library.importPaths,
exclusionPatterns: library.exclusionPatterns,
assetIds: assets.map(({ id }) => id),
progressCounter: currentAssetCount,
totalAssets: assetCount,
},
});
const completePercentage = ((100 * currentAssetCount) / assetCount).toFixed(1);
this.logger.log(
`Queued check of ${currentAssetCount} of ${assetCount} (${completePercentage} %) existing asset(s) so far in library ${library.id}`,
);
}
if (currentAssetCount) {
this.logger.log(`Finished queuing ${currentAssetCount} asset check(s) for library ${library.id}`);
}
return JobStatus.SUCCESS;

View File

@@ -662,10 +662,10 @@ export class MetadataService extends BaseService {
let dateTimeOriginal = dateTime?.toDate();
let localDateTime = dateTime?.toDateTime().setZone('UTC', { keepLocalTime: true }).toJSDate();
if (!localDateTime || !dateTimeOriginal) {
this.logger.debug(
`No exif date time found, falling back on earliest of file creation and modification for assset ${asset.id}: ${asset.originalPath}`,
);
const earliestDate = this.earliestDate(asset.fileModifiedAt, asset.fileCreatedAt);
this.logger.debug(
`No exif date time found, falling back on ${earliestDate.toISOString()}, earliest of file creation and modification for assset ${asset.id}: ${asset.originalPath}`,
);
dateTimeOriginal = earliestDate;
localDateTime = earliestDate;
}

View File

@@ -208,17 +208,23 @@ export interface IAssetDeleteJob extends IEntityJob {
deleteOnDisk: boolean;
}
export interface ILibraryFileJob extends IEntityJob {
ownerId: string;
assetPath: string;
export interface ILibraryFileJob {
libraryId: string;
paths: string[];
progressCounter?: number;
totalAssets?: number;
}
export interface ILibraryAssetJob extends IEntityJob {
export interface ILibraryBulkIdsJob {
libraryId: string;
importPaths: string[];
exclusionPatterns: string[];
assetIds: string[];
progressCounter: number;
totalAssets: number;
}
export interface IBulkEntityJob extends IBaseJob {
export interface IBulkEntityJob {
ids: string[];
}
@@ -354,10 +360,11 @@ export type JobItem =
| { name: JobName.ASSET_DELETION_CHECK; data?: IBaseJob }
// Library Management
| { name: JobName.LIBRARY_SYNC_FILE; data: ILibraryFileJob }
| { name: JobName.LIBRARY_SYNC_FILES; data: ILibraryFileJob }
| { name: JobName.LIBRARY_QUEUE_SYNC_FILES; data: IEntityJob }
| { name: JobName.LIBRARY_QUEUE_SYNC_ASSETS; data: IEntityJob }
| { name: JobName.LIBRARY_SYNC_ASSET; data: ILibraryAssetJob }
| { name: JobName.LIBRARY_SYNC_ASSETS; data: ILibraryBulkIdsJob }
| { name: JobName.LIBRARY_ASSET_REMOVAL; data: ILibraryFileJob }
| { name: JobName.LIBRARY_DELETE; data: IEntityJob }
| { name: JobName.LIBRARY_QUEUE_SCAN_ALL; data?: IBaseJob }
| { name: JobName.LIBRARY_QUEUE_CLEANUP; data: IBaseJob }

View File

@@ -44,6 +44,8 @@ export const anyUuid = (ids: string[]) => sql<string>`any(${`{${ids}}`}::uuid[])
export const asVector = (embedding: number[]) => sql<string>`${`[${embedding}]`}::vector`;
export const unnest = (array: string[]) => sql<Record<string, string>>`unnest(array[${sql.join(array)}]::text[])`;
/**
* Mainly for type debugging to make VS Code display a more useful tooltip.
* Source: https://stackoverflow.com/a/69288824

View File

@@ -1,4 +1,4 @@
import { getKeysDeep, unsetDeep } from 'src/utils/misc';
import { getKeysDeep, globToSqlPattern, unsetDeep } from 'src/utils/misc';
import { describe, expect, it } from 'vitest';
describe('getKeysDeep', () => {
@@ -51,3 +51,19 @@ describe('unsetDeep', () => {
expect(unsetDeep({ foo: 'bar', nested: { enabled: true } }, 'nested.enabled')).toEqual({ foo: 'bar' });
});
});
describe('globToSqlPattern', () => {
const testCases = [
['**/Raw/**', '%/Raw/%'],
['**/abc/*.tif', '%/abc/%.tif'],
['**/*.tif', '%/%.tif'],
['**/*.jp?', '%/%.jp_'],
['**/@eaDir/**', '%/@eaDir/%'],
['**/._*', `%/._%`],
['/absolute/path/**', `/absolute/path/%`],
];
it.each(testCases)('should convert %s to %s', (input, expected) => {
expect(globToSqlPattern(input)).toEqual(expected);
});
});

View File

@@ -10,6 +10,8 @@ import { ReferenceObject, SchemaObject } from '@nestjs/swagger/dist/interfaces/o
import _ from 'lodash';
import { writeFileSync } from 'node:fs';
import path from 'node:path';
import picomatch from 'picomatch';
import parse from 'picomatch/lib/parse';
import { SystemConfig } from 'src/config';
import { CLIP_MODEL_INFO, serverVersion } from 'src/constants';
import { extraSyncModels } from 'src/dtos/sync.dto';
@@ -268,3 +270,35 @@ export const useSwagger = (app: INestApplication, { write }: { write: boolean })
writeFileSync(outputPath, JSON.stringify(patchOpenAPI(specification), null, 2), { encoding: 'utf8' });
}
};
const convertTokenToSqlPattern = (token: parse.Token): string => {
switch (token.type) {
case 'slash': {
return '/';
}
case 'text': {
return token.value;
}
case 'globstar':
case 'star': {
return '%';
}
case 'underscore': {
return String.raw`\_`;
}
case 'qmark': {
return '_';
}
case 'dot': {
return '.';
}
default: {
return '';
}
}
};
export const globToSqlPattern = (glob: string) => {
const tokens = picomatch.parse(glob).tokens;
return tokens.map((token) => convertTokenToSqlPattern(token)).join('');
};