Merge pull request #955 from ThibG/glitch-soc/merge-upstream

Merge upstream changes

Gemfile

@@ -15,7 +15,7 @@ gem 'makara', '~> 0.4'
 gem 'pghero', '~> 2.2'
 gem 'dotenv-rails', '~> 2.7'
 
-gem 'aws-sdk-s3', '~> 1.30', require: false
+gem 'aws-sdk-s3', '~> 1.31', require: false
 gem 'fog-core', '<= 2.1.0'
 gem 'fog-openstack', '~> 0.3', require: false
 gem 'paperclip', '~> 6.0'

Gemfile.lock

@@ -43,7 +43,7 @@ GEM
       activemodel (>= 4.1, < 6)
       case_transform (>= 0.2)
       jsonapi-renderer (>= 0.1.1.beta1, < 0.3)
-    active_record_query_trace (1.6)
+    active_record_query_trace (1.6.1)
     activejob (5.2.2)
       activesupport (= 5.2.2)
       globalid (>= 0.3.6)
@@ -76,8 +76,8 @@ GEM
     av (0.9.0)
       cocaine (~> 0.5.3)
     aws-eventstream (1.0.1)
-    aws-partitions (1.131.0)
+    aws-partitions (1.143.0)
-    aws-sdk-core (3.45.0)
+    aws-sdk-core (3.46.2)
       aws-eventstream (~> 1.0)
       aws-partitions (~> 1.0)
       aws-sigv4 (~> 1.0)
@@ -85,7 +85,7 @@ GEM
     aws-sdk-kms (1.13.0)
       aws-sdk-core (~> 3, >= 3.39.0)
       aws-sigv4 (~> 1.0)
-    aws-sdk-s3 (1.30.1)
+    aws-sdk-s3 (1.31.0)
       aws-sdk-core (~> 3, >= 3.39.0)
       aws-sdk-kms (~> 1)
       aws-sigv4 (~> 1.0)
@@ -402,7 +402,7 @@ GEM
     pg (1.1.4)
     pghero (2.2.0)
       activerecord
-    pkg-config (1.3.5)
+    pkg-config (1.3.6)
     powerpack (0.1.2)
     premailer (1.11.1)
       addressable
@@ -661,7 +661,7 @@ DEPENDENCIES
   active_record_query_trace (~> 1.6)
   addressable (~> 2.6)
   annotate (~> 2.7)
-  aws-sdk-s3 (~> 1.30)
+  aws-sdk-s3 (~> 1.31)
   better_errors (~> 2.5)
   binding_of_caller (~> 0.7)
   bootsnap (~> 1.4)

app/lib/request.rb

@@ -171,7 +171,7 @@ class Request
         outer_e = nil
 
         Resolv::DNS.open do |dns|
-          dns.timeouts = 1
+          dns.timeouts = 5
 
           addresses = dns.getaddresses(host).take(2)
           time_slot = 10.0 / addresses.size

lib/cli.rb

@@ -6,6 +6,7 @@ require_relative 'mastodon/emoji_cli'
 require_relative 'mastodon/accounts_cli'
 require_relative 'mastodon/feeds_cli'
 require_relative 'mastodon/settings_cli'
+require_relative 'mastodon/statuses_cli'
 require_relative 'mastodon/domains_cli'
 require_relative 'mastodon/version'
 
@@ -30,6 +31,9 @@ module Mastodon
     desc 'settings SUBCOMMAND ...ARGS', 'Manage dynamic settings'
     subcommand 'settings', Mastodon::SettingsCLI
 
+    desc 'statuses SUBCOMMAND ...ARGS', 'Manage statuses'
+    subcommand 'statuses', Mastodon::StatusesCLI
+
     desc 'domains SUBCOMMAND ...ARGS', 'Manage account domains'
     subcommand 'domains', Mastodon::DomainsCLI
 

lib/mastodon/statuses_cli.rb

@@ -0,0 +1,54 @@
+# frozen_string_literal: true
+
+require_relative '../../config/boot'
+require_relative '../../config/environment'
+require_relative 'cli_helper'
+
+module Mastodon
+  class StatusesCLI < Thor
+    include ActionView::Helpers::NumberHelper
+
+    def self.exit_on_failure?
+      true
+    end
+
+    option :days, type: :numeric, default: 90
+    desc 'remove', 'Remove statuses'
+    def remove
+      say('Creating temporary database indices...')
+
+      ActiveRecord::Base.connection.add_index(:accounts, :id, name: :index_accounts_local, where: 'domain is null', algorithm: :concurrently) unless ActiveRecord::Base.connection.index_name_exists?(:accounts, :index_accounts_local)
+      ActiveRecord::Base.connection.add_index(:status_pins, :status_id, name: :index_status_pins_status_id, algorithm: :concurrently) unless ActiveRecord::Base.connection.index_name_exists?(:status_pins, :index_status_pins_status_id)
+      ActiveRecord::Base.connection.add_index(:media_attachments, :remote_url, name: :index_media_attachments_remote_url, where: 'remote_url is not null', algorithm: :concurrently) unless ActiveRecord::Base.connection.index_name_exists?(:media_attachments, :index_media_attachments_remote_url)
+
+      max_id   = Mastodon::Snowflake.id_at(options[:days].days.ago)
+      start_at = Time.now.to_f
+
+      say('Beginning removal... This might take a while...')
+
+      Status.remote
+            .where('id < ?', max_id)
+            .where(reblog_of_id: nil)                                                                                                                                                                                              # Skip reblogs
+            .where(in_reply_to_id: nil)                                                                                                                                                                                            # Skip replies
+            .where('id NOT IN (SELECT status_pins.status_id FROM status_pins WHERE statuses.id = status_id)')                                                                                                                      # Skip statuses that are pinned on profiles
+            .where('id NOT IN (SELECT mentions.status_id FROM mentions WHERE statuses.id = mentions.status_id AND mentions.account_id IN (SELECT accounts.id FROM accounts WHERE domain IS NULL))')                                # Skip statuses that mention local accounts
+            .where('id NOT IN (SELECT statuses1.in_reply_to_id FROM statuses AS statuses1 WHERE statuses.id = statuses1.in_reply_to_id)')                                                                                          # Skip statuses favourited by local accounts
+            .where('id NOT IN (SELECT statuses1.reblog_of_id FROM statuses AS statuses1 WHERE statuses.id = statuses1.reblog_of_id AND statuses1.account_id IN (SELECT accounts.id FROM accounts WHERE accounts.domain IS NULL))') # Skip statuses reblogged by local accounts
+            .where('account_id NOT IN (SELECT follows.target_account_id FROM follows WHERE statuses.account_id = follows.target_account_id)')                                                                                      # Skip accounts followed by local accounts
+            .in_batches
+            .delete_all
+
+      say('Beginning removal of now-orphaned media attachments to free up disk space...')
+
+      Scheduler::MediaCleanupScheduler.new.perform
+
+      say("Done after #{Time.now.to_f - start_at}s", :green)
+    ensure
+      say('Removing temporary database indices to restore write performance...')
+
+      ActiveRecord::Base.connection.remove_index(:accounts, name: :index_accounts_local) if ActiveRecord::Base.connection.index_name_exists?(:accounts, :index_accounts_local)
+      ActiveRecord::Base.connection.remove_index(:status_pins, name: :index_status_pins_status_id) if ActiveRecord::Base.connection.index_name_exists?(:status_pins, :index_status_pins_status_id)
+      ActiveRecord::Base.connection.remove_index(:media_attachments, name: :index_media_attachments_remote_url) if ActiveRecord::Base.connection.index_name_exists?(:media_attachments, :index_media_attachments_remote_url)
+    end
+  end
+end