smarter "clean remote files"

this should (optionally) skip in-use files, and files that have been
seen recently

packages/backend/src/core/QueueService.ts

@@ -684,8 +684,11 @@ export class QueueService {
 	}
 
 	@bindThis
-	public createCleanRemoteFilesJob() {
-		return this.objectStorageQueue.add('cleanRemoteFiles', {}, {
+	public createCleanRemoteFilesJob(olderThanSeconds: number = 0, keepFilesInUse: boolean = false) {
+		return this.objectStorageQueue.add('cleanRemoteFiles', {
+			keepFilesInUse,
+			olderThanSeconds,
+		}, {
 			removeOnComplete: {
 				age: 3600 * 24 * 7, // keep up to 7 days
 				count: 30,

packages/backend/src/queue/processors/CleanRemoteFilesProcessorService.ts

@@ -4,14 +4,17 @@
  */
 
 import { Inject, Injectable } from '@nestjs/common';
-import { IsNull, MoreThan, Not } from 'typeorm';
+import { IsNull, MoreThan, Not, Brackets } from 'typeorm';
 import { DI } from '@/di-symbols.js';
 import type { MiDriveFile, DriveFilesRepository } from '@/models/_.js';
+import { MiUser } from '@/models/_.js';
 import type Logger from '@/logger.js';
 import { DriveService } from '@/core/DriveService.js';
 import { bindThis } from '@/decorators.js';
 import { QueueLoggerService } from '../QueueLoggerService.js';
 import type * as Bull from 'bullmq';
+import type { CleanRemoteFilesJobData } from '../types.js';
+import { IdService } from '@/core/IdService.js';
 
 @Injectable()
 export class CleanRemoteFilesProcessorService {
@@ -23,35 +26,54 @@ export class CleanRemoteFilesProcessorService {
 
 		private driveService: DriveService,
 		private queueLoggerService: QueueLoggerService,
+		private idService: IdService,
 	) {
 		this.logger = this.queueLoggerService.logger.createSubLogger('clean-remote-files');
 	}
 
 	@bindThis
-	public async process(job: Bull.Job<Record<string, unknown>>): Promise<void> {
+	public async process(job: Bull.Job<CleanRemoteFilesJobData>): Promise<void> {
 		this.logger.info('Deleting cached remote files...');
 
+		const olderThanTimestamp = Date.now() - job.data.olderThanSeconds * 1000;
+		const olderThanDate = new Date(olderThanTimestamp);
+		const keepFilesInUse = job.data.keepFilesInUse;
 		let deletedCount = 0;
 		let cursor: MiDriveFile['id'] | null = null;
 		let errorCount = 0;
 
-		const total = await this.driveFilesRepository.countBy({
-			userHost: Not(IsNull()),
-			isLink: false,
-		});
+		const filesQuery = this.driveFilesRepository.createQueryBuilder('file')
+			.where('file.userHost IS NOT NULL') // remote files
+			.andWhere('file.isLink = FALSE') // cached
+			.andWhere('file.id <= :id', { id: this.idService.gen(olderThanTimestamp) }) // and old
+			.orderBy('file.id', 'ASC');
+
+		if (keepFilesInUse) {
+			filesQuery
+			// are they used as avatar&&c?
+				.leftJoinAndSelect(
+					MiUser, 'fileuser',
+					'fileuser."avatarId"="file"."id" OR fileuser."bannerId"="file"."id" OR fileuser."backgroundId"="file"."id"'
+				)
+				.andWhere(
+					new Brackets((qb) => {
+						qb.where('fileuser.id IS NULL') // not used
+							.orWhere( // or attached to a user
+								new Brackets((qb) => {
+									qb.where('fileuser.lastFetchedAt IS NOT NULL') // weird? maybe this only applies to local users
+										.andWhere('fileuser.lastFetchedAt < :old', { old: olderThanDate }); // old user
+								})
+							);
+					})
+				);
+		}
+
+		const total = await filesQuery.clone().getCount();
 
 		while (true) {
-			const files = await this.driveFilesRepository.find({
-				where: {
-					userHost: Not(IsNull()),
-					isLink: false,
-					...(cursor ? { id: MoreThan(cursor) } : {}),
-				},
-				take: 256,
-				order: {
-					id: 1,
-				},
-			});
+			const thisBatchQuery = filesQuery.clone();
+			if (cursor) thisBatchQuery.andWhere('file.id > :cursor', { cursor });
+			const files = await thisBatchQuery.take(256).getMany();
 
 			if (files.length === 0) {
 				job.updateProgress(100);

packages/backend/src/queue/types.ts

@@ -40,6 +40,11 @@ export type RelationshipJobData = {
 	withReplies?: boolean;
 };
 
+export type CleanRemoteFilesJobData = {
+	keepFilesInUse: boolean;
+	olderThanSeconds: number;
+};
+
 export type DbJobData<T extends keyof DbJobMap> = DbJobMap[T];
 
 export type DbJobMap = {