refactor: queue in batches (#30698)

This commit is contained in:
Jason Rasmussen 2026-08-11 10:01:42 -04:00 committed by GitHub
parent d1662fb2b6
commit ee525f159b
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
12 changed files with 136 additions and 292 deletions

View file

@ -1,7 +1,6 @@
import { BadRequestException, Injectable } from '@nestjs/common'; import { BadRequestException, Injectable } from '@nestjs/common';
import _ from 'lodash'; import _ from 'lodash';
import { DateTime, Duration } from 'luxon'; import { DateTime, Duration } from 'luxon';
import { JOBS_ASSET_PAGINATION_SIZE } from 'src/constants';
import { AssetFile } from 'src/database'; import { AssetFile } from 'src/database';
import { OnJob } from 'src/decorators'; import { OnJob } from 'src/decorators';
import { AssetResponseDto, SanitizedAssetResponseDto, mapAsset } from 'src/dtos/asset-response.dto'; import { AssetResponseDto, SanitizedAssetResponseDto, mapAsset } from 'src/dtos/asset-response.dto';
@ -46,7 +45,7 @@ import {
} from 'src/utils/asset.util'; } from 'src/utils/asset.util';
import { updateLockedColumns } from 'src/utils/database'; import { updateLockedColumns } from 'src/utils/database';
import { extractTimeZone } from 'src/utils/date'; import { extractTimeZone } from 'src/utils/date';
import { findOrFail } from 'src/utils/misc'; import { batched, findOrFail } from 'src/utils/misc';
import { transformOcrBoundingBox } from 'src/utils/transform'; import { transformOcrBoundingBox } from 'src/utils/transform';
@Injectable() @Injectable()
@ -279,31 +278,12 @@ export class AssetService extends BaseService {
.minus(Duration.fromObject({ days: trashedDays })) .minus(Duration.fromObject({ days: trashedDays }))
.toJSDate(); .toJSDate();
let chunk: Array<{ id: string; isOffline: boolean }> = []; for await (const assets of batched(this.assetJobRepository.streamForDeletedJob(trashedBefore))) {
const queueChunk = async () => {
if (chunk.length === 0) {
return;
}
await this.jobRepository.queueAll( await this.jobRepository.queueAll(
chunk.map(({ id, isOffline }) => ({ assets.map(({ id, isOffline }) => ({ name: JobName.AssetDelete, data: { id, deleteOnDisk: !isOffline } })),
name: JobName.AssetDelete,
data: { id, deleteOnDisk: !isOffline },
})),
); );
chunk = [];
};
const assets = this.assetJobRepository.streamForDeletedJob(trashedBefore);
for await (const asset of assets) {
chunk.push(asset);
if (chunk.length >= JOBS_ASSET_PAGINATION_SIZE) {
await queueChunk();
}
} }
await queueChunk();
return JobStatus.Success; return JobStatus.Success;
} }

View file

@ -1,5 +1,4 @@
import { Injectable } from '@nestjs/common'; import { Injectable } from '@nestjs/common';
import { JOBS_ASSET_PAGINATION_SIZE } from 'src/constants';
import { OnJob } from 'src/decorators'; import { OnJob } from 'src/decorators';
import { BulkIdErrorReason, BulkIdResponseDto, BulkIdsDto } from 'src/dtos/asset-ids.response.dto'; import { BulkIdErrorReason, BulkIdResponseDto, BulkIdsDto } from 'src/dtos/asset-ids.response.dto';
import { MapAsset, mapAsset } from 'src/dtos/asset-response.dto'; import { MapAsset, mapAsset } from 'src/dtos/asset-response.dto';
@ -8,9 +7,9 @@ import { DuplicateResolveDto, DuplicateResolveGroupDto, DuplicateResponseDto } f
import { AssetStatus, AssetVisibility, JobName, JobStatus, Permission, QueueName } from 'src/enum'; import { AssetStatus, AssetVisibility, JobName, JobStatus, Permission, QueueName } from 'src/enum';
import { AssetDuplicateResult } from 'src/repositories/search.repository'; import { AssetDuplicateResult } from 'src/repositories/search.repository';
import { BaseService } from 'src/services/base.service'; import { BaseService } from 'src/services/base.service';
import { JobItem, JobOf } from 'src/types'; import { JobOf } from 'src/types';
import { suggestDuplicateKeepAssetIds } from 'src/utils/duplicate'; import { suggestDuplicateKeepAssetIds } from 'src/utils/duplicate';
import { isDuplicateDetectionEnabled } from 'src/utils/misc'; import { batched, isDuplicateDetectionEnabled } from 'src/utils/misc';
type ResolveRequest = { type ResolveRequest = {
assetUpdate: { assetUpdate: {
@ -307,22 +306,12 @@ export class DuplicateService extends BaseService {
return JobStatus.Skipped; return JobStatus.Skipped;
} }
let jobs: JobItem[] = []; for await (const assets of batched(this.assetJobRepository.streamForSearchDuplicates(force))) {
const queueAll = async () => { await this.jobRepository.queueAll(
await this.jobRepository.queueAll(jobs); assets.map((asset) => ({ name: JobName.AssetDetectDuplicates, data: { id: asset.id } })),
jobs = []; );
};
const assets = this.assetJobRepository.streamForSearchDuplicates(force);
for await (const asset of assets) {
jobs.push({ name: JobName.AssetDetectDuplicates, data: { id: asset.id } });
if (jobs.length >= JOBS_ASSET_PAGINATION_SIZE) {
await queueAll();
}
} }
await queueAll();
return JobStatus.Success; return JobStatus.Success;
} }

View file

@ -35,7 +35,7 @@ import {
IIntegrityUntrackedFilesJob, IIntegrityUntrackedFilesJob,
} from 'src/types'; } from 'src/types';
import { ImmichFileResponse } from 'src/utils/file'; import { ImmichFileResponse } from 'src/utils/file';
import { handlePromiseError } from 'src/utils/misc'; import { batched, handlePromiseError } from 'src/utils/misc';
/** /**
* Untracked Files: * Untracked Files:
@ -201,7 +201,7 @@ export class IntegrityService extends BaseService {
const reports = this.integrityRepository.streamIntegrityReportsWithAssetChecksum(IntegrityReport.UntrackedFile); const reports = this.integrityRepository.streamIntegrityReportsWithAssetChecksum(IntegrityReport.UntrackedFile);
let total = 0; let total = 0;
for await (const batchReports of chunk(reports, JOBS_LIBRARY_PAGINATION_SIZE)) { for await (const batchReports of batched(reports, JOBS_LIBRARY_PAGINATION_SIZE)) {
await this.jobRepository.queue({ await this.jobRepository.queue({
name: JobName.IntegrityUntrackedFilesRefresh, name: JobName.IntegrityUntrackedFilesRefresh,
data: { data: {
@ -338,7 +338,7 @@ export class IntegrityService extends BaseService {
const reports = this.integrityRepository.streamIntegrityReportsWithAssetChecksum(IntegrityReport.MissingFile); const reports = this.integrityRepository.streamIntegrityReportsWithAssetChecksum(IntegrityReport.MissingFile);
let total = 0; let total = 0;
for await (const batchReports of chunk(reports, JOBS_LIBRARY_PAGINATION_SIZE)) { for await (const batchReports of batched(reports, JOBS_LIBRARY_PAGINATION_SIZE)) {
await this.jobRepository.queue({ await this.jobRepository.queue({
name: JobName.IntegrityMissingFilesRefresh, name: JobName.IntegrityMissingFilesRefresh,
data: { data: {
@ -365,7 +365,7 @@ export class IntegrityService extends BaseService {
const assetPaths = this.integrityRepository.streamAssetPathsForMissingFiles(); const assetPaths = this.integrityRepository.streamAssetPathsForMissingFiles();
let total = 0; let total = 0;
for await (const batchPaths of chunk(assetPaths, JOBS_LIBRARY_PAGINATION_SIZE)) { for await (const batchPaths of batched(assetPaths, JOBS_LIBRARY_PAGINATION_SIZE)) {
await this.jobRepository.queue({ await this.jobRepository.queue({
name: JobName.IntegrityMissingFiles, name: JobName.IntegrityMissingFiles,
data: { data: {
@ -450,7 +450,7 @@ export class IntegrityService extends BaseService {
const reports = this.integrityRepository.streamIntegrityReportsWithAssetChecksum(IntegrityReport.ChecksumFail); const reports = this.integrityRepository.streamIntegrityReportsWithAssetChecksum(IntegrityReport.ChecksumFail);
let total = 0; let total = 0;
for await (const batchReports of chunk(reports, JOBS_LIBRARY_PAGINATION_SIZE)) { for await (const batchReports of batched(reports, JOBS_LIBRARY_PAGINATION_SIZE)) {
await this.jobRepository.queue({ await this.jobRepository.queue({
name: JobName.IntegrityChecksumFilesRefresh, name: JobName.IntegrityChecksumFilesRefresh,
data: { data: {
@ -656,7 +656,7 @@ export class IntegrityService extends BaseService {
for (const property of properties) { for (const property of properties) {
const reports = this.integrityRepository.streamIntegrityReportsByProperty(property, type); const reports = this.integrityRepository.streamIntegrityReportsByProperty(property, type);
for await (const batch of chunk(reports, JOBS_LIBRARY_PAGINATION_SIZE)) { for await (const batch of batched(reports, JOBS_LIBRARY_PAGINATION_SIZE)) {
await this.jobRepository.queue({ await this.jobRepository.queue({
name: JobName.IntegrityDeleteReports, name: JobName.IntegrityDeleteReports,
data: { data: {
@ -705,19 +705,3 @@ export class IntegrityService extends BaseService {
return JobStatus.Success; return JobStatus.Success;
} }
} }
async function* chunk<T>(generator: AsyncIterableIterator<T>, n: number) {
let chunk: T[] = [];
for await (const item of generator) {
chunk.push(item);
if (chunk.length === n) {
yield chunk;
chunk = [];
}
}
if (chunk.length > 0) {
yield chunk;
}
}

View file

@ -34,7 +34,7 @@ import { AssetTable } from 'src/schema/tables/asset.table';
import { BaseService } from 'src/services/base.service'; import { BaseService } from 'src/services/base.service';
import { JobOf } from 'src/types'; import { JobOf } from 'src/types';
import { mimeTypes } from 'src/utils/mime-types'; import { mimeTypes } from 'src/utils/mime-types';
import { findOrFail, handlePromiseError } from 'src/utils/misc'; import { batched, findOrFail, handlePromiseError } from 'src/utils/misc';
@Injectable() @Injectable()
export class LibraryService extends BaseService { export class LibraryService extends BaseService {
@ -375,35 +375,20 @@ export class LibraryService extends BaseService {
await this.assetRepository.updateByLibraryId(libraryId, { deletedAt: new Date() }); await this.assetRepository.updateByLibraryId(libraryId, { deletedAt: new Date() });
let isAssetsFound = false;
let chunk: string[] = [];
const queueChunk = async () => {
if (chunk.length === 0) {
return;
}
isAssetsFound = true;
this.logger.debug(`Queueing deletion of ${chunk.length} asset(s) in library ${libraryId}`);
await this.jobRepository.queueAll(
chunk.map((id) => ({ name: JobName.AssetDelete, data: { id, deleteOnDisk: false } })),
);
chunk = [];
};
this.logger.debug(`Will delete all assets in library ${libraryId}`); this.logger.debug(`Will delete all assets in library ${libraryId}`);
const assets = this.libraryRepository.streamAssetIds(libraryId); let hasAssets = false;
for await (const asset of assets) { for await (const assets of batched(
chunk.push(asset.id); this.libraryRepository.streamAssetIds(libraryId),
JOBS_LIBRARY_PAGINATION_SIZE,
if (chunk.length >= JOBS_LIBRARY_PAGINATION_SIZE) { )) {
await queueChunk(); this.logger.debug(`Queueing deletion of ${assets.length} asset(s) in library ${libraryId}`);
} await this.jobRepository.queueAll(
assets.map((asset) => ({ name: JobName.AssetDelete, data: { id: asset.id, deleteOnDisk: false } })),
);
hasAssets = true;
} }
await queueChunk(); if (!hasAssets) {
if (!isAssetsFound) {
this.logger.log(`Deleting library ${libraryId}`); this.logger.log(`Deleting library ${libraryId}`);
await this.libraryRepository.delete(libraryId); await this.libraryRepository.delete(libraryId);
} }
@ -746,15 +731,12 @@ export class LibraryService extends BaseService {
return JobStatus.Success; return JobStatus.Success;
} }
let chunk: string[] = []; this.logger.log(`Scanning library ${library.id} for assets missing from disk...`);
let count = 0; let count = 0;
const existingAssets = this.libraryRepository.streamAssetIds(library.id);
const queueChunk = async () => { for await (const assets of batched(existingAssets, JOBS_LIBRARY_PAGINATION_SIZE)) {
if (chunk.length === 0) { count += assets.length;
return;
}
count += chunk.length;
await this.jobRepository.queue({ await this.jobRepository.queue({
name: JobName.LibrarySyncAssets, name: JobName.LibrarySyncAssets,
@ -762,32 +744,19 @@ export class LibraryService extends BaseService {
libraryId: library.id, libraryId: library.id,
importPaths: library.importPaths, importPaths: library.importPaths,
exclusionPatterns: library.exclusionPatterns, exclusionPatterns: library.exclusionPatterns,
assetIds: chunk.map((id) => id), assetIds: assets.map(({ id }) => id),
progressCounter: count, progressCounter: count,
totalAssets: assetCount, totalAssets: assetCount,
}, },
}); });
chunk = [];
const completePercentage = ((100 * count) / assetCount).toFixed(1); const completePercentage = ((100 * count) / assetCount).toFixed(1);
this.logger.log( this.logger.log(
`Queued check of ${count} of ${assetCount} (${completePercentage} %) existing asset(s) so far in library ${library.id}`, `Queued check of ${count} of ${assetCount} (${completePercentage} %) existing asset(s) so far in library ${library.id}`,
); );
};
this.logger.log(`Scanning library ${library.id} for assets missing from disk...`);
const existingAssets = this.libraryRepository.streamAssetIds(library.id);
for await (const asset of existingAssets) {
chunk.push(asset.id);
if (chunk.length === JOBS_LIBRARY_PAGINATION_SIZE) {
await queueChunk();
}
} }
await queueChunk();
this.logger.log(`Finished queuing ${count} asset check(s) for library ${library.id}`); this.logger.log(`Finished queuing ${count} asset check(s) for library ${library.id}`);
return JobStatus.Success; return JobStatus.Success;

View file

@ -207,8 +207,7 @@ describe(MediaService.name, () => {
await sut.handleQueueGenerateThumbnails({ force: false }); await sut.handleQueueGenerateThumbnails({ force: false });
expect(mocks.assetJob.streamForThumbnailJob).toHaveBeenCalledWith({ force: false, fullsizeEnabled: false }); expect(mocks.assetJob.streamForThumbnailJob).toHaveBeenCalledWith({ force: false, fullsizeEnabled: false });
expect(mocks.job.queueAll).toHaveBeenCalledWith([]); expect(mocks.job.queueAll).toHaveBeenCalledTimes(1);
expect(mocks.person.getAll).toHaveBeenCalledWith({ thumbnailPath: '' }); expect(mocks.person.getAll).toHaveBeenCalledWith({ thumbnailPath: '' });
}); });
@ -237,7 +236,10 @@ describe(MediaService.name, () => {
await sut.handleQueueGenerateThumbnails({ force: false }); await sut.handleQueueGenerateThumbnails({ force: false });
expect(mocks.assetJob.streamForThumbnailJob).toHaveBeenCalledWith({ force: false, fullsizeEnabled: false }); expect(mocks.assetJob.streamForThumbnailJob).toHaveBeenCalledWith({ force: false, fullsizeEnabled: false });
expect(mocks.job.queueAll).toHaveBeenCalledWith([]); expect(mocks.job.queueAll).toHaveBeenCalledTimes(1);
expect(mocks.job.queueAll).toHaveBeenCalledWith([
{ name: JobName.AssetEditThumbnailGeneration, data: { id: asset.id } },
]);
expect(mocks.person.getAll).toHaveBeenCalledWith({ thumbnailPath: '' }); expect(mocks.person.getAll).toHaveBeenCalledWith({ thumbnailPath: '' });
}); });

View file

@ -1,6 +1,6 @@
import { Injectable } from '@nestjs/common'; import { Injectable } from '@nestjs/common';
import { SystemConfig } from 'src/config'; import { SystemConfig } from 'src/config';
import { FACE_THUMBNAIL_SIZE, JOBS_ASSET_PAGINATION_SIZE } from 'src/constants'; import { FACE_THUMBNAIL_SIZE } from 'src/constants';
import { ImagePathOptions, StorageCore, ThumbnailPathEntity } from 'src/cores/storage.core'; import { ImagePathOptions, StorageCore, ThumbnailPathEntity } from 'src/cores/storage.core';
import { AssetFile } from 'src/database'; import { AssetFile } from 'src/database';
import { OnEvent, OnJob } from 'src/decorators'; import { OnEvent, OnJob } from 'src/decorators';
@ -43,7 +43,7 @@ import { getAssetFile, getDimensions } from 'src/utils/asset.util';
import { checkFaceVisibility, checkOcrVisibility } from 'src/utils/editor'; import { checkFaceVisibility, checkOcrVisibility } from 'src/utils/editor';
import { BaseConfig, ThumbnailConfig } from 'src/utils/media'; import { BaseConfig, ThumbnailConfig } from 'src/utils/media';
import { mimeTypes } from 'src/utils/mime-types'; import { mimeTypes } from 'src/utils/mime-types';
import { clamp } from 'src/utils/misc'; import { batched, clamp } from 'src/utils/misc';
import { getOutputDimensions } from 'src/utils/transform'; import { getOutputDimensions } from 'src/utils/transform';
interface UpsertFileOptions { interface UpsertFileOptions {
@ -69,52 +69,42 @@ export class MediaService extends BaseService {
@OnJob({ name: JobName.AssetGenerateThumbnailsQueueAll, queue: QueueName.ThumbnailGeneration }) @OnJob({ name: JobName.AssetGenerateThumbnailsQueueAll, queue: QueueName.ThumbnailGeneration })
async handleQueueGenerateThumbnails({ force }: JobOf<JobName.AssetGenerateThumbnailsQueueAll>): Promise<JobStatus> { async handleQueueGenerateThumbnails({ force }: JobOf<JobName.AssetGenerateThumbnailsQueueAll>): Promise<JobStatus> {
const config = await this.getConfig({ withCache: true }); const config = await this.getConfig({ withCache: true });
let jobs: JobItem[] = [];
const queueAll = async () => {
await this.jobRepository.queueAll(jobs);
jobs = [];
};
const isFullsizeEnabled = config.image.fullsize.enabled; const isFullsizeEnabled = config.image.fullsize.enabled;
for await (const asset of this.assetJobRepository.streamForThumbnailJob({ for await (const assets of batched(
force, this.assetJobRepository.streamForThumbnailJob({ force, fullsizeEnabled: isFullsizeEnabled }),
fullsizeEnabled: isFullsizeEnabled, )) {
})) { const jobs: JobItem[] = [];
if (force || !asset.isEdited) { for (const asset of assets) {
jobs.push({ name: JobName.AssetGenerateThumbnails, data: { id: asset.id } }); if (force || !asset.isEdited) {
} jobs.push({ name: JobName.AssetGenerateThumbnails, data: { id: asset.id } });
if (asset.isEdited) {
jobs.push({ name: JobName.AssetEditThumbnailGeneration, data: { id: asset.id } });
}
if (jobs.length >= JOBS_ASSET_PAGINATION_SIZE) {
await queueAll();
}
}
await queueAll();
const people = this.personRepository.getAll(force ? undefined : { thumbnailPath: '' });
for await (const person of people) {
if (!person.faceAssetId) {
const face = await this.personRepository.getRandomFace(person.id);
if (!face) {
continue;
} }
await this.personRepository.update({ id: person.id, faceAssetId: face.id }); if (asset.isEdited) {
jobs.push({ name: JobName.AssetEditThumbnailGeneration, data: { id: asset.id } });
}
} }
jobs.push({ name: JobName.PersonGenerateThumbnail, data: { id: person.id } }); await this.jobRepository.queueAll(jobs);
if (jobs.length >= JOBS_ASSET_PAGINATION_SIZE) {
await queueAll();
}
} }
await queueAll(); for await (const people of batched(this.personRepository.getAll(force ? undefined : { thumbnailPath: '' }))) {
const jobs: JobItem[] = [];
for (const person of people) {
if (!person.faceAssetId) {
const face = await this.personRepository.getRandomFace(person.id);
if (!face) {
continue;
}
await this.personRepository.update({ id: person.id, faceAssetId: face.id });
}
jobs.push({ name: JobName.PersonGenerateThumbnail, data: { id: person.id } });
}
await this.jobRepository.queueAll(jobs);
}
return JobStatus.Success; return JobStatus.Success;
} }
@ -127,30 +117,18 @@ export class MediaService extends BaseService {
await this.storageCore.removeEmptyDirs(StorageFolder.EncodedVideo); await this.storageCore.removeEmptyDirs(StorageFolder.EncodedVideo);
} }
let jobs: JobItem[] = []; for await (const assets of batched(this.assetJobRepository.streamForMigrationJob())) {
const assets = this.assetJobRepository.streamForMigrationJob(); await this.jobRepository.queueAll(
for await (const asset of assets) { assets.map((asset) => ({ name: JobName.AssetFileMigration, data: { id: asset.id } })),
jobs.push({ name: JobName.AssetFileMigration, data: { id: asset.id } }); );
if (jobs.length >= JOBS_ASSET_PAGINATION_SIZE) {
await this.jobRepository.queueAll(jobs);
jobs = [];
}
} }
await this.jobRepository.queueAll(jobs); for await (const people of batched(this.personRepository.getAll())) {
jobs = []; await this.jobRepository.queueAll(
people.map((person) => ({ name: JobName.PersonFileMigration, data: { id: person.id } })),
for await (const person of this.personRepository.getAll()) { );
jobs.push({ name: JobName.PersonFileMigration, data: { id: person.id } });
if (jobs.length === JOBS_ASSET_PAGINATION_SIZE) {
await this.jobRepository.queueAll(jobs);
jobs = [];
}
} }
await this.jobRepository.queueAll(jobs);
return JobStatus.Success; return JobStatus.Success;
} }
@ -551,18 +529,12 @@ export class MediaService extends BaseService {
async handleQueueVideoConversion(job: JobOf<JobName.AssetEncodeVideoQueueAll>): Promise<JobStatus> { async handleQueueVideoConversion(job: JobOf<JobName.AssetEncodeVideoQueueAll>): Promise<JobStatus> {
const { force } = job; const { force } = job;
let queue: { name: JobName.AssetEncodeVideo; data: { id: string } }[] = []; for await (const assets of batched(this.assetJobRepository.streamForVideoConversion(force))) {
for await (const asset of this.assetJobRepository.streamForVideoConversion(force)) { await this.jobRepository.queueAll(
queue.push({ name: JobName.AssetEncodeVideo, data: { id: asset.id } }); assets.map((asset) => ({ name: JobName.AssetEncodeVideo, data: { id: asset.id } })),
);
if (queue.length >= JOBS_ASSET_PAGINATION_SIZE) {
await this.jobRepository.queueAll(queue);
queue = [];
}
} }
await this.jobRepository.queueAll(queue);
return JobStatus.Success; return JobStatus.Success;
} }

View file

@ -6,7 +6,7 @@ import { DateTime, Duration } from 'luxon';
import { Stats } from 'node:fs'; import { Stats } from 'node:fs';
import { constants } from 'node:fs/promises'; import { constants } from 'node:fs/promises';
import { join, parse } from 'node:path'; import { join, parse } from 'node:path';
import { JOBS_ASSET_PAGINATION_SIZE } from 'src/constants';
import { StorageCore } from 'src/cores/storage.core'; import { StorageCore } from 'src/cores/storage.core';
import { Asset, AssetFile } from 'src/database'; import { Asset, AssetFile } from 'src/database';
import { OnEvent, OnJob } from 'src/decorators'; import { OnEvent, OnJob } from 'src/decorators';
@ -30,12 +30,12 @@ import { AssetExifTable } from 'src/schema/tables/asset-exif.table';
import { AssetFaceTable } from 'src/schema/tables/asset-face.table'; import { AssetFaceTable } from 'src/schema/tables/asset-face.table';
import { PersonTable } from 'src/schema/tables/person.table'; import { PersonTable } from 'src/schema/tables/person.table';
import { BaseService } from 'src/services/base.service'; import { BaseService } from 'src/services/base.service';
import { JobItem, JobOf } from 'src/types'; import { JobOf } from 'src/types';
import { getAssetFiles } from 'src/utils/asset.util'; import { getAssetFiles } from 'src/utils/asset.util';
import { isAssetChecksumConstraint } from 'src/utils/database'; import { isAssetChecksumConstraint } from 'src/utils/database';
import { mergeTimeZone } from 'src/utils/date'; import { mergeTimeZone } from 'src/utils/date';
import { mimeTypes } from 'src/utils/mime-types'; import { mimeTypes } from 'src/utils/mime-types';
import { isFaceImportEnabled } from 'src/utils/misc'; import { batched, isFaceImportEnabled } from 'src/utils/misc';
import { upsertTags } from 'src/utils/tag'; import { upsertTags } from 'src/utils/tag';
import { Tasks } from 'src/utils/tasks'; import { Tasks } from 'src/utils/tasks';
@ -218,17 +218,12 @@ export class MetadataService extends BaseService {
async handleQueueMetadataExtraction(job: JobOf<JobName.AssetExtractMetadataQueueAll>): Promise<JobStatus> { async handleQueueMetadataExtraction(job: JobOf<JobName.AssetExtractMetadataQueueAll>): Promise<JobStatus> {
const { force } = job; const { force } = job;
let queue: { name: JobName.AssetExtractMetadata; data: { id: string } }[] = []; for await (const assets of batched(this.assetJobRepository.streamForMetadataExtraction(force))) {
for await (const asset of this.assetJobRepository.streamForMetadataExtraction(force)) { await this.jobRepository.queueAll(
queue.push({ name: JobName.AssetExtractMetadata, data: { id: asset.id } }); assets.map((asset) => ({ name: JobName.AssetExtractMetadata, data: { id: asset.id } })),
);
if (queue.length >= JOBS_ASSET_PAGINATION_SIZE) {
await this.jobRepository.queueAll(queue);
queue = [];
}
} }
await this.jobRepository.queueAll(queue);
return JobStatus.Success; return JobStatus.Success;
} }
@ -417,22 +412,12 @@ export class MetadataService extends BaseService {
@OnJob({ name: JobName.SidecarQueueAll, queue: QueueName.Sidecar }) @OnJob({ name: JobName.SidecarQueueAll, queue: QueueName.Sidecar })
async handleQueueSidecar({ force }: JobOf<JobName.SidecarQueueAll>): Promise<JobStatus> { async handleQueueSidecar({ force }: JobOf<JobName.SidecarQueueAll>): Promise<JobStatus> {
let jobs: JobItem[] = []; for await (const assets of batched(this.assetJobRepository.streamForSidecar(force))) {
const queueAll = async () => { await this.jobRepository.queueAll(
await this.jobRepository.queueAll(jobs); assets.map((asset) => ({ name: JobName.SidecarCheck, data: { id: asset.id } })),
jobs = []; );
};
const assets = this.assetJobRepository.streamForSidecar(force);
for await (const asset of assets) {
jobs.push({ name: JobName.SidecarCheck, data: { id: asset.id } });
if (jobs.length >= JOBS_ASSET_PAGINATION_SIZE) {
await queueAll();
}
} }
await queueAll();
return JobStatus.Success; return JobStatus.Success;
} }

View file

@ -1,12 +1,12 @@
import { Injectable } from '@nestjs/common'; import { Injectable } from '@nestjs/common';
import { JOBS_ASSET_PAGINATION_SIZE } from 'src/constants';
import { OnJob } from 'src/decorators'; import { OnJob } from 'src/decorators';
import { AssetVisibility, JobName, JobStatus, QueueName } from 'src/enum'; import { AssetVisibility, JobName, JobStatus, QueueName } from 'src/enum';
import { OCR } from 'src/repositories/machine-learning.repository'; import { OCR } from 'src/repositories/machine-learning.repository';
import { BaseService } from 'src/services/base.service'; import { BaseService } from 'src/services/base.service';
import { JobItem, JobOf } from 'src/types'; import { JobOf } from 'src/types';
import { tokenizeForSearch } from 'src/utils/database'; import { tokenizeForSearch } from 'src/utils/database';
import { isOcrEnabled } from 'src/utils/misc'; import { batched, isOcrEnabled } from 'src/utils/misc';
@Injectable() @Injectable()
export class OcrService extends BaseService { export class OcrService extends BaseService {
@ -21,19 +21,10 @@ export class OcrService extends BaseService {
await this.ocrRepository.deleteAll(); await this.ocrRepository.deleteAll();
} }
let jobs: JobItem[] = []; for await (const assets of batched(this.assetJobRepository.streamForOcrJob(force))) {
const assets = this.assetJobRepository.streamForOcrJob(force); await this.jobRepository.queueAll(assets.map((asset) => ({ name: JobName.Ocr, data: { id: asset.id } })));
for await (const asset of assets) {
jobs.push({ name: JobName.Ocr, data: { id: asset.id } });
if (jobs.length >= JOBS_ASSET_PAGINATION_SIZE) {
await this.jobRepository.queueAll(jobs);
jobs = [];
}
} }
await this.jobRepository.queueAll(jobs);
return JobStatus.Success; return JobStatus.Success;
} }

View file

@ -1,6 +1,5 @@
import { BadRequestException, Injectable, NotFoundException } from '@nestjs/common'; import { BadRequestException, Injectable, NotFoundException } from '@nestjs/common';
import { Insertable, Updateable } from 'kysely'; import { Insertable, Updateable } from 'kysely';
import { JOBS_ASSET_PAGINATION_SIZE } from 'src/constants';
import { Person } from 'src/database'; import { Person } from 'src/database';
import { Chunked, OnJob } from 'src/decorators'; import { Chunked, OnJob } from 'src/decorators';
import { BulkIdErrorReason, BulkIdResponseDto, BulkIdsDto } from 'src/dtos/asset-ids.response.dto'; import { BulkIdErrorReason, BulkIdResponseDto, BulkIdsDto } from 'src/dtos/asset-ids.response.dto';
@ -43,7 +42,7 @@ import { JobItem, JobOf } from 'src/types';
import { getDimensions } from 'src/utils/asset.util'; import { getDimensions } from 'src/utils/asset.util';
import { ImmichFileResponse } from 'src/utils/file'; import { ImmichFileResponse } from 'src/utils/file';
import { mimeTypes } from 'src/utils/mime-types'; import { mimeTypes } from 'src/utils/mime-types';
import { findOrFail, isFacialRecognitionEnabled } from 'src/utils/misc'; import { batched, findOrFail, isFacialRecognitionEnabled } from 'src/utils/misc';
import { Point, transformPoints } from 'src/utils/transform'; import { Point, transformPoints } from 'src/utils/transform';
@Injectable() @Injectable()
@ -277,19 +276,12 @@ export class PersonService extends BaseService {
await this.personRepository.vacuum({ reindexVectors: true }); await this.personRepository.vacuum({ reindexVectors: true });
} }
let jobs: JobItem[] = []; for await (const assets of batched(this.assetJobRepository.streamForDetectFacesJob(force))) {
const assets = this.assetJobRepository.streamForDetectFacesJob(force); await this.jobRepository.queueAll(
for await (const asset of assets) { assets.map((asset) => ({ name: JobName.AssetDetectFaces, data: { id: asset.id } })),
jobs.push({ name: JobName.AssetDetectFaces, data: { id: asset.id } }); );
if (jobs.length >= JOBS_ASSET_PAGINATION_SIZE) {
await this.jobRepository.queueAll(jobs);
jobs = [];
}
} }
await this.jobRepository.queueAll(jobs);
if (force === undefined) { if (force === undefined) {
await this.jobRepository.queue({ name: JobName.PersonCleanup }); await this.jobRepository.queue({ name: JobName.PersonCleanup });
} }
@ -435,22 +427,16 @@ export class PersonService extends BaseService {
await this.databaseRepository.prewarm(VectorIndex.Face); await this.databaseRepository.prewarm(VectorIndex.Face);
const lastRun = new Date().toISOString(); const lastRun = new Date().toISOString();
const facePagination = this.personRepository.getAllFaces(
const faces = this.personRepository.getAllFaces(
force ? undefined : { personId: null, sourceType: SourceType.MachineLearning }, force ? undefined : { personId: null, sourceType: SourceType.MachineLearning },
); );
for await (const batch of batched(faces)) {
let jobs: { name: JobName.FacialRecognition; data: { id: string; deferred: false } }[] = []; await this.jobRepository.queueAll(
for await (const face of facePagination) { batch.map((face) => ({ name: JobName.FacialRecognition, data: { id: face.id, deferred: false } })),
jobs.push({ name: JobName.FacialRecognition, data: { id: face.id, deferred: false } }); );
if (jobs.length === JOBS_ASSET_PAGINATION_SIZE) {
await this.jobRepository.queueAll(jobs);
jobs = [];
}
} }
await this.jobRepository.queueAll(jobs);
await this.systemMetadataRepository.set(SystemMetadataKey.FacialRecognitionState, { lastRun }); await this.systemMetadataRepository.set(SystemMetadataKey.FacialRecognitionState, { lastRun });
return JobStatus.Success; return JobStatus.Success;

View file

@ -1,12 +1,12 @@
import { Injectable } from '@nestjs/common'; import { Injectable } from '@nestjs/common';
import { SystemConfig } from 'src/config'; import { SystemConfig } from 'src/config';
import { JOBS_ASSET_PAGINATION_SIZE } from 'src/constants';
import { OnEvent, OnJob } from 'src/decorators'; import { OnEvent, OnJob } from 'src/decorators';
import { AssetVisibility, DatabaseLock, ImmichWorker, JobName, JobStatus, QueueName } from 'src/enum'; import { AssetVisibility, DatabaseLock, ImmichWorker, JobName, JobStatus, QueueName } from 'src/enum';
import { ArgOf } from 'src/repositories/event.repository'; import { ArgOf } from 'src/repositories/event.repository';
import { BaseService } from 'src/services/base.service'; import { BaseService } from 'src/services/base.service';
import { JobItem, JobOf } from 'src/types'; import { JobOf } from 'src/types';
import { getCLIPModelInfo, isSmartSearchEnabled } from 'src/utils/misc'; import { batched, getCLIPModelInfo, isSmartSearchEnabled } from 'src/utils/misc';
@Injectable() @Injectable()
export class SmartInfoService extends BaseService { export class SmartInfoService extends BaseService {
@ -77,18 +77,10 @@ export class SmartInfoService extends BaseService {
await this.databaseRepository.setDimensionSize(dimSize); await this.databaseRepository.setDimensionSize(dimSize);
} }
let queue: JobItem[] = []; for await (const assets of batched(this.assetJobRepository.streamForEncodeClip(force))) {
const assets = this.assetJobRepository.streamForEncodeClip(force); await this.jobRepository.queueAll(assets.map((asset) => ({ name: JobName.SmartSearch, data: { id: asset.id } })));
for await (const asset of assets) {
queue.push({ name: JobName.SmartSearch, data: { id: asset.id } });
if (queue.length >= JOBS_ASSET_PAGINATION_SIZE) {
await this.jobRepository.queueAll(queue);
queue = [];
}
} }
await this.jobRepository.queueAll(queue);
return JobStatus.Success; return JobStatus.Success;
} }

View file

@ -1,11 +1,11 @@
import { Injectable } from '@nestjs/common'; import { Injectable } from '@nestjs/common';
import { JOBS_ASSET_PAGINATION_SIZE } from 'src/constants';
import { OnEvent, OnJob } from 'src/decorators'; import { OnEvent, OnJob } from 'src/decorators';
import { BulkIdsDto } from 'src/dtos/asset-ids.response.dto'; import { BulkIdsDto } from 'src/dtos/asset-ids.response.dto';
import { AuthDto } from 'src/dtos/auth.dto'; import { AuthDto } from 'src/dtos/auth.dto';
import { TrashResponseDto } from 'src/dtos/trash.dto'; import { TrashResponseDto } from 'src/dtos/trash.dto';
import { JobName, JobStatus, Permission, QueueName } from 'src/enum'; import { JobName, JobStatus, Permission, QueueName } from 'src/enum';
import { BaseService } from 'src/services/base.service'; import { BaseService } from 'src/services/base.service';
import { batched } from 'src/utils/misc';
@Injectable() @Injectable()
export class TrashService extends BaseService { export class TrashService extends BaseService {
@ -47,39 +47,16 @@ export class TrashService extends BaseService {
@OnJob({ name: JobName.AssetEmptyTrash, queue: QueueName.BackgroundTask }) @OnJob({ name: JobName.AssetEmptyTrash, queue: QueueName.BackgroundTask })
async handleEmptyTrash() { async handleEmptyTrash() {
const assets = this.trashRepository.getDeletedIds();
let count = 0; let count = 0;
const batch: string[] = []; for await (const assets of batched(this.trashRepository.getDeletedIds())) {
for await (const { id } of assets) { await this.jobRepository.queueAll(
batch.push(id); assets.map(({ id }) => ({ name: JobName.AssetDelete, data: { id, deleteOnDisk: true } })),
);
if (batch.length === JOBS_ASSET_PAGINATION_SIZE) { count += assets.length;
await this.handleBatch(batch);
count += batch.length;
batch.length = 0;
}
} }
await this.handleBatch(batch);
count += batch.length;
batch.length = 0;
this.logger.log(`Queued ${count} asset(s) for deletion from the trash`); this.logger.log(`Queued ${count} asset(s) for deletion from the trash`);
return JobStatus.Success; return JobStatus.Success;
} }
private async handleBatch(ids: string[]) {
this.logger.debug(`Queueing ${ids.length} asset(s) for deletion from the trash`);
await this.jobRepository.queueAll(
ids.map((assetId) => ({
name: JobName.AssetDelete,
data: {
id: assetId,
deleteOnDisk: true,
},
})),
);
}
} }

View file

@ -14,7 +14,7 @@ import path from 'node:path';
import picomatch from 'picomatch'; import picomatch from 'picomatch';
import parse from 'picomatch/lib/parse'; import parse from 'picomatch/lib/parse';
import { SystemConfig } from 'src/config'; import { SystemConfig } from 'src/config';
import { CLIP_MODEL_INFO, endpointTags, serverVersion } from 'src/constants'; import { CLIP_MODEL_INFO, JOBS_ASSET_PAGINATION_SIZE, endpointTags, serverVersion } from 'src/constants';
import { extraModels } from 'src/decorators'; import { extraModels } from 'src/decorators';
import { ApiCustomExtension, ImmichCookie, ImmichHeader, MetadataKey } from 'src/enum'; import { ApiCustomExtension, ImmichCookie, ImmichHeader, MetadataKey } from 'src/enum';
import { LoggingRepository } from 'src/repositories/logging.repository'; import { LoggingRepository } from 'src/repositories/logging.repository';
@ -120,6 +120,23 @@ export const findOrFail = async <T>(find: () => Promise<T>, entity: string): Pro
return value; return value;
}; };
export async function* batched<T>(items: AsyncIterable<T>, size = JOBS_ASSET_PAGINATION_SIZE): AsyncGenerator<T[]> {
let batch: T[] = [];
for await (const item of items) {
batch.push(item);
if (batch.length >= size) {
yield batch;
batch = [];
}
}
if (batch.length > 0) {
yield batch;
}
}
export interface OpenGraphTags { export interface OpenGraphTags {
title: string; title: string;
description: string; description: string;