Emit moveToTrash event only for the deleting user

[juliusknorr]

This is the same issue as in #16624 which causes a huge loading time caused by iterating over all mounts for a file id for all users if getById is called on the RootFolder:

server/lib/private/Files/Node/Folder.php

Line 292 in e43b341

$mountsContainingFile = $mountCache->getMountsForFileId((int)$id, $user);

Since the moveToTrash event is only used in e2ee it should be fine to just check for the file in the users mount who is trying to delete the file.

[blizzz]

don't we need the user exists check anymore when requesting the user folder instance?

[juliusknorr]

Hm I currently fail to understand why that unit test is failing with that change :/

[rullzer]

ok so the issue with the failing test is that the users storage gets initisialized thus for the final change of the lock the user mount is there. Causing the lock to be different.

So assume 'testUser'

• Then the first read lock is on '/' with path 'testUser/files/test.txt'
• Second change is on '/' with 'testUser/files/test.txt'
• Then the userfolder gets initialized
• Thirs lock change is on '/testUser' with path 'files/test.txt'

The last one fails because it wants to release the exclusive lock (which isn't there)

[juliusknorr]

/backport to stable16

[juliusknorr]

/backport to stable15

[backportbot-nextcloud]

backport to stable16 in #17099

[backportbot-nextcloud]

backport to stable15 in #17100