Add separate cache directory for non-local uploads (#12821)

5 files changed, 182 insertions, 7 deletions

diff --git a/lib/cli.rb b/lib/cli.rb
index 19cc5d6b5..313a36a3d 100644
--- a/lib/cli.rb
+++ b/lib/cli.rb
@@ -11,6 +11,7 @@ require_relative 'mastodon/statuses_cli'
 require_relative 'mastodon/domains_cli'
 require_relative 'mastodon/preview_cards_cli'
 require_relative 'mastodon/cache_cli'
+require_relative 'mastodon/upgrade_cli'
 require_relative 'mastodon/version'
 
 module Mastodon
@@ -49,6 +50,9 @@ module Mastodon
     desc 'cache SUBCOMMAND ...ARGS', 'Manage cache'
     subcommand 'cache', Mastodon::CacheCLI
 
+    desc 'upgrade SUBCOMMAND ...ARGS', 'Various version upgrade utilities'
+    subcommand 'upgrade', Mastodon::UpgradeCLI
+
     option :dry_run, type: :boolean
     desc 'self-destruct', 'Erase the server from the federation'
     long_desc <<~LONG_DESC
diff --git a/lib/mastodon/cli_helper.rb b/lib/mastodon/cli_helper.rb
index ec4d9a81e..4a20fa8d6 100644
--- a/lib/mastodon/cli_helper.rb
+++ b/lib/mastodon/cli_helper.rb
@@ -10,6 +10,10 @@ Paperclip.options[:log]      = false
 
 module Mastodon
   module CLIHelper
+    def dry_run?
+      options[:dry_run]
+    end
+
     def create_progress_bar(total = nil)
       ProgressBar.create(total: total, format: '%c/%u |%b%i| %e')
     end
diff --git a/lib/mastodon/media_cli.rb b/lib/mastodon/media_cli.rb
index 0f211f272..424d65a5f 100644
--- a/lib/mastodon/media_cli.rb
+++ b/lib/mastodon/media_cli.rb
@@ -85,7 +85,9 @@ module Mastodon
           record_map = preload_records_from_mixed_objects(objects)
 
           objects.each do |object|
-            path_segments   = object.key.split('/')
+            path_segments = object.key.split('/')
+            path_segments.delete('cache')
+
             model_name      = path_segments.first.classify
             attachment_name = path_segments[1].singularize
             record_id       = path_segments[2..-2].join.to_i
@@ -120,8 +122,11 @@ module Mastodon
         Find.find(File.join(*[root_path, prefix].compact)) do |path|
           next if File.directory?(path)
 
-          key             = path.gsub("#{root_path}#{File::SEPARATOR}", '')
-          path_segments   = key.split(File::SEPARATOR)
+          key = path.gsub("#{root_path}#{File::SEPARATOR}", '')
+
+          path_segments = key.split(File::SEPARATOR)
+          path_segments.delete('cache')
+
           model_name      = path_segments.first.classify
           record_id       = path_segments[2..-2].join.to_i
           attachment_name = path_segments[1].singularize
@@ -229,10 +234,13 @@ module Mastodon
 
     desc 'lookup URL', 'Lookup where media is displayed by passing a media URL'
     def lookup(url)
-      path          = Addressable::URI.parse(url).path
+      path = Addressable::URI.parse(url).path
+
       path_segments = path.split('/')[2..-1]
-      model_name    = path_segments.first.classify
-      record_id     = path_segments[2..-2].join.to_i
+      path_segments.delete('cache')
+
+      model_name = path_segments.first.classify
+      record_id  = path_segments[2..-2].join.to_i
 
       unless PRELOAD_MODEL_WHITELIST.include?(model_name)
         say("Cannot find corresponding model: #{model_name}", :red)
@@ -276,7 +284,9 @@ module Mastodon
       preload_map = Hash.new { |hash, key| hash[key] = [] }
 
       objects.map do |object|
-        segments   = object.key.split('/')
+        segments = object.key.split('/')
+        segments.delete('cache')
+
         model_name = segments.first.classify
         record_id  = segments[2..-2].join.to_i
 
diff --git a/lib/mastodon/upgrade_cli.rb b/lib/mastodon/upgrade_cli.rb
new file mode 100644
index 000000000..74d13f62d
--- /dev/null
+++ b/lib/mastodon/upgrade_cli.rb
@@ -0,0 +1,148 @@
+# frozen_string_literal: true
+
+require_relative '../../config/boot'
+require_relative '../../config/environment'
+require_relative 'cli_helper'
+
+module Mastodon
+  class UpgradeCLI < Thor
+    include CLIHelper
+
+    def self.exit_on_failure?
+      true
+    end
+
+    CURRENT_STORAGE_SCHEMA_VERSION = 1
+
+    option :dry_run, type: :boolean, default: false
+    option :verbose, type: :boolean, default: false, aliases: [:v]
+    desc 'storage-schema', 'Upgrade storage schema of various file attachments to the latest version'
+    long_desc <<~LONG_DESC
+      Iterates over every file attachment of every record and, if its storage schema is outdated, performs the
+      necessary upgrade to the latest one. In practice this means e.g. moving files to different directories.
+
+      Will most likely take a long time.
+    LONG_DESC
+    def storage_schema
+      progress = create_progress_bar(nil)
+      dry_run  = dry_run? ? ' (DRY RUN)' : ''
+      records  = 0
+
+      klasses = [
+        Account,
+        CustomEmoji,
+        MediaAttachment,
+        PreviewCard,
+      ]
+
+      klasses.each do |klass|
+        attachment_names = klass.attachment_definitions.keys
+
+        klass.find_each do |record|
+          attachment_names.each do |attachment_name|
+            attachment = record.public_send(attachment_name)
+
+            next if attachment.blank? || attachment.storage_schema_version >= CURRENT_STORAGE_SCHEMA_VERSION
+
+            attachment.styles.each_key do |style|
+              case Paperclip::Attachment.default_options[:storage]
+              when :s3
+                upgrade_storage_s3(progress, attachment, style)
+              when :fog
+                upgrade_storage_fog(progress, attachment, style)
+              when :filesystem
+                upgrade_storage_filesystem(progress, attachment, style)
+              end
+
+              progress.increment
+            end
+
+            attachment.instance_write(:storage_schema_version, CURRENT_STORAGE_SCHEMA_VERSION)
+          end
+
+          if record.changed?
+            record.save unless dry_run?
+            records += 1
+          end
+        end
+      end
+
+      progress.total = progress.progress
+      progress.finish
+
+      say("Upgraded storage schema of #{records} records#{dry_run}", :green, true)
+    end
+
+    private
+
+    def upgrade_storage_s3(progress, attachment, style)
+      previous_storage_schema_version = attachment.storage_schema_version
+      object                          = attachment.s3_object(style)
+
+      attachment.instance_write(:storage_schema_version, CURRENT_STORAGE_SCHEMA_VERSION)
+
+      upgraded_path = attachment.path(style)
+
+      if upgraded_path != object.key && object.exists?
+        progress.log("Moving #{object.key} to #{upgraded_path}") if options[:verbose]
+
+        begin
+          object.move_to(upgraded_path) unless dry_run?
+        rescue => e
+          progress.log(pastel.red("Error processing #{object.key}: #{e}"))
+        end
+      end
+
+      # Because we move files style-by-style, it's important to restore
+      # previous version at the end. The upgrade will be recorded after
+      # all styles are updated
+      attachment.instance_write(:storage_schema_version, previous_storage_schema_version)
+    end
+
+    def upgrade_storage_fog(_progress, _attachment, _style)
+      say('The fog storage driver is not supported for this operation at this time', :red)
+      exit(1)
+    end
+
+    def upgrade_storage_filesystem(progress, attachment, style)
+      previous_storage_schema_version = attachment.storage_schema_version
+      previous_path                   = attachment.path(style)
+
+      attachment.instance_write(:storage_schema_version, CURRENT_STORAGE_SCHEMA_VERSION)
+
+      upgraded_path = attachment.path(style)
+
+      if upgraded_path != previous_path && File.exist?(previous_path)
+        progress.log("Moving #{previous_path} to #{upgraded_path}") if options[:verbose]
+
+        begin
+          unless dry_run?
+            FileUtils.mkdir_p(File.dirname(upgraded_path))
+            FileUtils.mv(previous_path, upgraded_path)
+
+            begin
+              FileUtils.rmdir(previous_path, parents: true)
+            rescue Errno::ENOTEMPTY
+              # OK
+            end
+          end
+        rescue => e
+          progress.log(pastel.red("Error processing #{previous_path}: #{e}"))
+
+          unless dry_run?
+            begin
+              FileUtils.rmdir(upgraded_path, parents: true)
+            rescue Errno::ENOTEMPTY
+              # OK
+            end
+          end
+        end
+      end
+
+      # Because we move files style-by-style, it's important to restore
+      # previous version at the end. The upgrade will be recorded after
+      # all styles are updated
+      attachment.instance_write(:storage_schema_version, previous_storage_schema_version)
+    end
+  end
+end
diff --git a/lib/paperclip/attachment_extensions.rb b/lib/paperclip/attachment_extensions.rb
index ce5780557..f3e51dbd3 100644
--- a/lib/paperclip/attachment_extensions.rb
+++ b/lib/paperclip/attachment_extensions.rb
@@ -14,6 +14,15 @@ module Paperclip
       end
     end
 
+    def storage_schema_version
+      instance_read(:storage_schema_version) || 0
+    end
+
+    def assign_attributes
+      super
+      instance_write(:storage_schema_version, 1)
+    end
+
     def variant?(other_filename)
       return true  if original_filename == other_filename
       return false if original_filename.nil?