From 338861eb8c4657f37a96a51503ddac62b79a8998 Mon Sep 17 00:00:00 2001
From: Markus Kahl <machisuji@gmail.com>
Date: Fri, 17 Dec 2021 15:43:19 +0000
Subject: [PATCH] delete cached attachments immediately after backup

---
 app/workers/backup_job.rb | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)

diff --git a/app/workers/backup_job.rb b/app/workers/backup_job.rb
index 5819218d34..3c3f13f67e 100644
--- a/app/workers/backup_job.rb
+++ b/app/workers/backup_job.rb
@@ -142,6 +142,9 @@ class BackupJob < ::ApplicationJob
   end
 
   def create_backup_archive!(file_name:, db_dump_file_name:, attachments: attachments_to_include)
+    paths_to_clean = []
+    clean_up = OpenProject::Configuration.remote_storage?
+
     Zip::File.open(file_name, Zip::File::CREATE) do |zipfile|
       attachments.each do |attachment|
         # If an attachment is destroyed on disk, skip i
@@ -151,16 +154,33 @@ class BackupJob < ::ApplicationJob
         path = diskfile.path
 
         zipfile.add "attachment/file/#{attachment.id}/#{attachment[:file]}", path
+
+        paths_to_clean << get_cache_folder_path(attachment) if clean_up && a.file.cached?
       end
 
       zipfile.get_output_stream("openproject.sql") { |f| f.write File.read(db_dump_file_name) }
     end
 
+    # delete locally cached files that were downloaded just for the backup
+    paths_to_clean.each do |path|
+      FileUtils.rm_rf path
+    end
+
     @archived = true
 
     file_name
   end
 
+  def get_cache_folder_path(attachment)
+    # expecting paths like /tmp/op_uploaded_files/1639754082-3468-0002-0911/file.ext
+    # just making extra sure so we don't delete anything wrong later on
+    unless attachment.diskfile.path =~ /#{attachment.file.cache_dir}\/[^\/]+\/[^\/]+/
+      raise "Unexpected cache path for attachment ##{attachment.id}: #{attachment.diskfile}"
+    end
+
+    Pathname(attachment.disfile.path).parent.to_s
+  end
+
   def attachments_to_include
     return Attachment.none if skip_attachments?