core: fix deletion of referenced files

packages/hydrooj/src/model/storage.ts

@@ -4,6 +4,7 @@ import moment from 'moment-timezone';
 import { nanoid } from 'nanoid';
 import type { Readable } from 'stream';
 import { Context } from '../context';
+import { FileNode } from '../interface';
 import mime from '../lib/mime';
 import db from '../service/db';
 import storage from '../service/storage';
@@ -49,25 +50,40 @@ export class StorageModel {
         );
     }
 
+    private static async _swapId(fileA: FileNode, fileB: FileNode) {
+        // try to swap the two file with same content to make sure the first file is no longer being referenced
+        // Example
+        // _id: A, path: C, meta: R
+        // _id: B, path: D, link: A, meta: P
+        // and we want to delete path=B, the result should be
+        // _id: A, path: D, meta: P
+        // _id: B, path: C, link: A, meta: R
+        await Promise.all([
+            StorageModel.coll.updateOne({ _id: fileA._id }, { $set: omit(fileB, ['_id', 'link']) }),
+            StorageModel.coll.updateOne({ _id: fileB._id }, { $set: omit(fileA, ['_id', 'link']) }),
+        ]);
+    }
+
     static async del(path: string[], operator = 1) {
         if (!path.length) return;
-        const affected = await StorageModel.coll.find({ path: { $in: path } }).toArray();
-        if (!affected.length) return;
-        const linked = await StorageModel.coll.find({ link: { $in: affected.map((i) => i._id) }, path: { $nin: path } }).toArray();
-        const processedIds = [];
-        for (const i of linked || []) {
-            if (processedIds.includes(i.link)) continue;
-            const current = affected.find((t) => t._id === i.link); // to be deleted
-            // eslint-disable-next-line no-await-in-loop
-            await Promise.all([
-                StorageModel.coll.updateOne({ _id: current._id }, { $set: omit(i, ['_id']) }),
-                StorageModel.coll.updateOne({ _id: i._id }, { $set: omit(current, ['_id']), $unset: { link: '' } }),
-            ]);
-            processedIds.push(i.link);
+        // files pending to be deleted
+        const pendingDelete = await StorageModel.coll.find({ path: { $in: path }, autoDelete: null }).toArray();
+        if (!pendingDelete.length) return;
+        // all files not going to be deleted but affected (referencing to the files to be deleted)
+        const linked = await StorageModel.coll.find({ link: { $in: pendingDelete.map((i) => i._id) }, path: { $nin: path } }).toArray();
+        const fileIds = [];
+        for (const i of pendingDelete) {
+            const affected = linked.filter((t) => t.link === i._id);
+            if (!affected.length) {
+                fileIds.push(i._id);
+                continue;
+            }
+            await StorageModel._swapId(i, affected[0]); // eslint-disable-line no-await-in-loop
+            fileIds.push(affected[0]._id); // we already swapped the two files, so we need to delete the other one
         }
         const autoDelete = moment().add(7, 'day').toDate();
         await StorageModel.coll.updateMany(
-            { path: { $in: path }, autoDelete: null },
+            { _id: { $in: fileIds } },
             { $set: { autoDelete }, $push: { operator } },
         );
     }