Merge tag 'v4.2.9' into lets-bump-hometown-to-mastodon-4.2

CHANGELOG.md

@@ -2,6 +2,41 @@
 
 All notable changes to this project will be documented in this file.
 
+## [4.2.9] - 2024-05-30
+
+### Security
+
+- Update dependencies
+- Fix private mention filtering ([GHSA-5fq7-3p3j-9vrf](https://github.com/mastodon/mastodon/security/advisories/GHSA-5fq7-3p3j-9vrf))
+- Fix password change endpoint not being rate-limited ([GHSA-q3rg-xx5v-4mxh](https://github.com/mastodon/mastodon/security/advisories/GHSA-q3rg-xx5v-4mxh))
+- Add hardening around rate-limit bypass ([GHSA-c2r5-cfqr-c553](https://github.com/mastodon/mastodon/security/advisories/GHSA-c2r5-cfqr-c553))
+
+### Added
+
+- Add rate-limit on OAuth application registration ([ThisIsMissEm](https://github.com/mastodon/mastodon/pull/30316))
+- Add fallback redirection when getting a webfinger query `WEB_DOMAIN@WEB_DOMAIN` ([ClearlyClaire](https://github.com/mastodon/mastodon/pull/28592))
+- Add `digest` attribute to `Admin::DomainBlock` entity in REST API ([ThisIsMissEm](https://github.com/mastodon/mastodon/pull/29092))
+
+### Removed
+
+- Remove superfluous application-level caching in some controllers ([ClearlyClaire](https://github.com/mastodon/mastodon/pull/29862))
+- Remove aggressive OAuth application vacuuming ([ThisIsMissEm](https://github.com/mastodon/mastodon/pull/30316))
+
+### Fixed
+
+- Fix leaking Elasticsearch connections in Sidekiq processes ([ClearlyClaire](https://github.com/mastodon/mastodon/pull/30450))
+- Fix language of remote posts not being recognized when using unusual casing ([ClearlyClaire](https://github.com/mastodon/mastodon/pull/30403))
+- Fix off-by-one in `tootctl media` commands ([ClearlyClaire](https://github.com/mastodon/mastodon/pull/30306))
+- Fix removal of allowed domains (in `LIMITED_FEDERATION_MODE`) not being recorded in the audit log ([ThisIsMissEm](https://github.com/mastodon/mastodon/pull/30125))
+- Fix not being able to block a subdomain of an already-blocked domain through the API ([ClearlyClaire](https://github.com/mastodon/mastodon/pull/30119))
+- Fix `Idempotency-Key` being ignored when scheduling a post ([ClearlyClaire](https://github.com/mastodon/mastodon/pull/30084))
+- Fix crash when supplying the `FFMPEG_BINARY` environment variable ([timothyjrogers](https://github.com/mastodon/mastodon/pull/30022))
+- Fix improper email address validation ([ClearlyClaire](https://github.com/mastodon/mastodon/pull/29838))
+- Fix results/query in `api/v1/featured_tags/suggestions` ([mjankowski](https://github.com/mastodon/mastodon/pull/29597))
+- Fix unblocking internationalized domain names under certain conditions ([tribela](https://github.com/mastodon/mastodon/pull/29530))
+- Fix admin account created by `mastodon:setup` not being auto-approved ([ClearlyClaire](https://github.com/mastodon/mastodon/pull/29379))
+- Fix reference to non-existent var in CLI maintenance command ([mjankowski](https://github.com/mastodon/mastodon/pull/28363))
+
 ## [4.2.8] - 2024-02-23
 
 ### Added

Gemfile

@@ -204,3 +204,5 @@ gem 'net-http', '~> 0.3.2'
 gem 'rubyzip', '~> 2.3'
 
 gem 'hcaptcha', '~> 7.1'
+
+gem 'mail', '~> 2.8'

Gemfile.lock

@@ -288,7 +288,7 @@ GEM
     faraday_middleware (1.2.0)
       faraday (~> 1.0)
     fast_blank (1.0.1)
-    fastimage (2.3.0)
+    fastimage (2.3.1)
     ffi (1.15.5)
     ffi-compiler (1.0.1)
       ffi (>= 1.0.0)
@@ -368,7 +368,7 @@ GEM
     jmespath (1.6.2)
     json (2.6.3)
     json-canonicalization (1.0.0)
-    json-jwt (1.15.3)
+    json-jwt (1.15.3.1)
       activesupport (>= 4.2)
       aes_key_wrap
       bindata
@@ -469,7 +469,7 @@ GEM
       net-protocol
     net-ssh (7.1.0)
     nio4r (2.7.0)
-    nokogiri (1.16.2)
+    nokogiri (1.16.5)
       mini_portile2 (~> 2.8.2)
       racc (~> 1.4)
     nsa (0.3.0)
@@ -537,7 +537,7 @@ GEM
     rack (2.2.8.1)
     rack-attack (6.7.0)
       rack (>= 1.0, < 4)
-    rack-cors (2.0.1)
+    rack-cors (2.0.2)
       rack (>= 2.0.0)
     rack-oauth2 (1.21.3)
       activesupport
@@ -605,8 +605,9 @@ GEM
     responders (3.1.0)
       actionpack (>= 5.2)
       railties (>= 5.2)
-    rexml (3.2.6)
-    rotp (6.2.2)
+    rexml (3.2.8)
+      strscan (>= 3.0.9)
+    rotp (6.3.0)
     rouge (4.1.2)
     rpam2 (4.0.2)
     rqrcode (2.2.0)
@@ -731,6 +732,7 @@ GEM
       redlock (~> 1.0)
     strong_migrations (0.8.0)
       activerecord (>= 5.2)
+    strscan (3.0.9)
     swd (1.3.0)
       activesupport (>= 3)
       attr_required (>= 0.0.5)
@@ -871,6 +873,7 @@ DEPENDENCIES
   letter_opener_web (~> 2.0)
   link_header (~> 0.0)
   lograge (~> 0.12)
+  mail (~> 2.8)
   mario-redis-lock (~> 1.2)
   md-paperclip-azure (~> 2.2)
   memory_profiler

app/controllers/admin/domain_allows_controller.rb

@@ -25,6 +25,8 @@ class Admin::DomainAllowsController < Admin::BaseController
   def destroy
     authorize @domain_allow, :destroy?
     UnallowDomainService.new.call(@domain_allow)
+    log_action :destroy, @domain_allow
+
     redirect_to admin_instances_path, notice: I18n.t('admin.domain_allows.destroyed_msg')
   end
 

app/controllers/api/v1/admin/domain_blocks_controller.rb

@@ -29,10 +29,11 @@ class Api::V1::Admin::DomainBlocksController < Api::BaseController
   def create
     authorize :domain_block, :create?
 
+    @domain_block = DomainBlock.new(resource_params)
     existing_domain_block = resource_params[:domain].present? ? DomainBlock.rule_for(resource_params[:domain]) : nil
-    return render json: existing_domain_block, serializer: REST::Admin::ExistingDomainBlockErrorSerializer, status: 422 if existing_domain_block.present?
+    return render json: existing_domain_block, serializer: REST::Admin::ExistingDomainBlockErrorSerializer, status: 422 if conflicts_with_existing_block?(@domain_block, existing_domain_block)
 
-    @domain_block = DomainBlock.create!(resource_params)
+    @domain_block.save!
     DomainBlockWorker.perform_async(@domain_block.id)
     log_action :create, @domain_block
     render json: @domain_block, serializer: REST::Admin::DomainBlockSerializer
@@ -55,6 +56,10 @@ class Api::V1::Admin::DomainBlocksController < Api::BaseController
 
   private
 
+  def conflicts_with_existing_block?(domain_block, existing_domain_block)
+    existing_domain_block.present? && (existing_domain_block.domain == TagManager.instance.normalize_domain(domain_block.domain) || !domain_block.stricter_than?(existing_domain_block))
+  end
+
   def set_domain_blocks
     @domain_blocks = filtered_domain_blocks.order(id: :desc).to_a_paginated_by_id(limit_param(LIMIT), params_slice(:max_id, :since_id, :min_id))
   end

app/controllers/api/v1/featured_tags/suggestions_controller.rb

@@ -12,6 +12,10 @@ class Api::V1::FeaturedTags::SuggestionsController < Api::BaseController
   private
 
   def set_recently_used_tags
-    @recently_used_tags = Tag.recently_used(current_account).where.not(id: current_account.featured_tags).limit(10)
+    @recently_used_tags = Tag.recently_used(current_account).where.not(id: featured_tag_ids).limit(10)
+  end
+
+  def featured_tag_ids
+    current_account.featured_tags.pluck(:tag_id)
   end
 end

app/controllers/concerns/cache_concern.rb

@@ -198,34 +198,19 @@ module CacheConcern
     end
   end
 
+  # TODO: Rename this method, as it does not perform any caching anymore.
   def cache_collection(raw, klass)
-    return raw unless klass.respond_to?(:with_includes)
+    return raw unless klass.respond_to?(:preload_cacheable_associations)
 
-    raw = raw.cache_ids.to_a if raw.is_a?(ActiveRecord::Relation)
-    return [] if raw.empty?
+    records = raw.to_a
 
-    cached_keys_with_value = begin
-      Rails.cache.read_multi(*raw).transform_keys(&:id).transform_values { |r| ActiveRecordCoder.load(r) }
-    rescue ActiveRecordCoder::Error
-      {} # The serialization format may have changed, let's pretend it's a cache miss.
-    end
+    klass.preload_cacheable_associations(records)
 
-    uncached_ids = raw.map(&:id) - cached_keys_with_value.keys
-
-    klass.reload_stale_associations!(cached_keys_with_value.values) if klass.respond_to?(:reload_stale_associations!)
-
-    unless uncached_ids.empty?
-      uncached = klass.where(id: uncached_ids).with_includes.index_by(&:id)
-
-      uncached.each_value do |item|
-        Rails.cache.write(item, ActiveRecordCoder.dump(item))
-      end
-    end
-
-    raw.filter_map { |item| cached_keys_with_value[item.id] || uncached[item.id] }
+    records
   end
 
+  # TODO: Rename this method, as it does not perform any caching anymore.
   def cache_collection_paginated_by_id(raw, klass, limit, options)
-    cache_collection raw.cache_ids.to_a_paginated_by_id(limit, options), klass
+    cache_collection raw.to_a_paginated_by_id(limit, options), klass
   end
 end

app/controllers/well_known/webfinger_controller.rb

@@ -21,7 +21,7 @@ module WellKnown
       username = username_from_resource
 
       @account = begin
-        if username == Rails.configuration.x.local_domain
+        if username == Rails.configuration.x.local_domain || username == Rails.configuration.x.web_domain
           Account.representative
         else
           Account.find_local!(username)

app/lib/activitypub/parser/status_parser.rb

@@ -3,6 +3,8 @@
 class ActivityPub::Parser::StatusParser
   include JsonLdHelper
 
+  NORMALIZED_LOCALE_NAMES = LanguagesHelper::SUPPORTED_LOCALES.keys.index_by(&:downcase).freeze
+
   # @param [Hash] json
   # @param [Hash] magic_values
   # @option magic_values [String] :followers_collection
@@ -86,6 +88,13 @@ class ActivityPub::Parser::StatusParser
   end
 
   def language
+    lang = raw_language_code
+    lang.presence && NORMALIZED_LOCALE_NAMES.fetch(lang.downcase.to_sym, lang)
+  end
+
+  private
+
+  def raw_language_code
     if content_language_map?
       @object['contentMap'].keys.first
     elsif name_language_map?
@@ -95,8 +104,6 @@ class ActivityPub::Parser::StatusParser
     end
   end
 
-  private
-
   def audience_to
     as_array(@object['to'] || @json['to']).map { |x| value_or_id(x) }
   end

app/lib/vacuum/applications_vacuum.rb

@@ -1,10 +0,0 @@
-# frozen_string_literal: true
-
-class Vacuum::ApplicationsVacuum
-  def perform
-    Doorkeeper::Application.where(owner_id: nil)
-                           .where.missing(:created_users, :access_tokens, :access_grants)
-                           .where(created_at: ...1.day.ago)
-                           .in_batches.delete_all
-  end
-end

app/lib/video_metadata_extractor.rb

@@ -22,7 +22,7 @@ class VideoMetadataExtractor
   private
 
   def ffmpeg_command_output
-    command = Terrapin::CommandLine.new('ffprobe', '-i :path -print_format :format -show_format -show_streams -show_error -loglevel :loglevel')
+    command = Terrapin::CommandLine.new(Rails.configuration.x.ffprobe_binary, '-i :path -print_format :format -show_format -show_streams -show_error -loglevel :loglevel')
     command.run(path: @path, format: 'json', loglevel: 'fatal')
   end
 

app/models/concerns/account_interactions.rb

@@ -185,7 +185,7 @@ module AccountInteractions
   end
 
   def unblock_domain!(other_domain)
-    block = domain_blocks.find_by(domain: other_domain)
+    block = domain_blocks.find_by(domain: normalized_domain(other_domain))
     block&.destroy
   end
 
@@ -313,4 +313,8 @@ module AccountInteractions
   def remove_potential_friendship(other_account)
     PotentialFriendshipTracker.remove(id, other_account.id)
   end
+
+  def normalized_domain(domain)
+    TagManager.instance.normalize_domain(domain)
+  end
 end

app/models/concerns/cacheable.rb

@@ -14,6 +14,10 @@ module Cacheable
       includes(@cache_associated)
     end
 
+    def preload_cacheable_associations(records)
+      ActiveRecord::Associations::Preloader.new(records: records, associations: @cache_associated).call
+    end
+
     def cache_ids
       select(:id, :updated_at)
     end

app/models/feed.rb

@@ -28,7 +28,7 @@ class Feed
       unhydrated = redis.zrangebyscore(key, "(#{min_id}", "(#{max_id}", limit: [0, limit], with_scores: true).map(&:first).map(&:to_i)
     end
 
-    Status.where(id: unhydrated).cache_ids
+    Status.where(id: unhydrated)
   end
 
   def key

app/models/public_feed.rb

@@ -33,7 +33,7 @@ class PublicFeed
     scope.merge!(media_only_scope) if media_only?
     scope.merge!(language_scope) if account&.chosen_languages.present?
 
-    scope.cache_ids.to_a_paginated_by_id(limit, max_id: max_id, since_id: since_id, min_id: min_id)
+    scope.to_a_paginated_by_id(limit, max_id: max_id, since_id: since_id, min_id: min_id)
   end
 
   private

app/models/status.rb

@@ -347,38 +347,6 @@ class Status < ApplicationRecord
       StatusPin.select('status_id').where(status_id: status_ids).where(account_id: account_id).each_with_object({}) { |p, h| h[p.status_id] = true }
     end
 
-    def reload_stale_associations!(cached_items)
-      account_ids = []
-
-      cached_items.each do |item|
-        account_ids << item.account_id
-        account_ids << item.reblog.account_id if item.reblog?
-      end
-
-      account_ids.uniq!
-
-      status_ids = cached_items.map { |item| item.reblog? ? item.reblog_of_id : item.id }.uniq
-
-      return if account_ids.empty?
-
-      accounts = Account.where(id: account_ids).includes(:account_stat, :user).index_by(&:id)
-
-      status_stats = StatusStat.where(status_id: status_ids).index_by(&:status_id)
-
-      cached_items.each do |item|
-        item.account = accounts[item.account_id]
-        item.reblog.account = accounts[item.reblog.account_id] if item.reblog?
-
-        if item.reblog?
-          status_stat = status_stats[item.reblog.id]
-          item.reblog.status_stat = status_stat if status_stat.present?
-        else
-          status_stat = status_stats[item.id]
-          item.status_stat = status_stat if status_stat.present?
-        end
-      end
-    end
-
     def from_text(text)
       return [] if text.blank?
 

app/models/tag_feed.rb

@@ -34,7 +34,7 @@ class TagFeed < PublicFeed
     scope.merge!(media_only_scope) if media_only?
     scope.merge!(without_local_only_scope) if without_local_only?
 
-    scope.cache_ids.to_a_paginated_by_id(limit, max_id: max_id, since_id: since_id, min_id: min_id)
+    scope.to_a_paginated_by_id(limit, max_id: max_id, since_id: since_id, min_id: min_id)
   end
 
   private

app/models/user.rb

@@ -96,6 +96,8 @@ class User < ApplicationRecord
   accepts_nested_attributes_for :invite_request, reject_if: ->(attributes) { attributes['text'].blank? && !Setting.require_invite_text }
   validates :invite_request, presence: true, on: :create, if: :invite_text_required?
 
+  validates :email, presence: true, email_address: true
+
   validates_with BlacklistedEmailValidator, if: -> { ENV['EMAIL_DOMAIN_LISTS_APPLY_AFTER_CONFIRMATION'] == 'true' || !confirmed? }
   validates_with EmailMxValidator, if: :validate_email_dns?
   validates :agreement, acceptance: { allow_nil: false, accept: [true, 'true', '1'] }, on: :create

app/serializers/rest/admin/domain_block_serializer.rb

@@ -1,11 +1,15 @@
 # frozen_string_literal: true
 
 class REST::Admin::DomainBlockSerializer < ActiveModel::Serializer
-  attributes :id, :domain, :created_at, :severity,
+  attributes :id, :domain, :digest, :created_at, :severity,
              :reject_media, :reject_reports,
              :private_comment, :public_comment, :obfuscate
 
   def id
     object.id.to_s
   end
+
+  def digest
+    object.domain_digest
+  end
 end

app/services/notify_service.rb

@@ -71,16 +71,17 @@ class NotifyService < BaseService
           LEFT JOIN mentions m ON m.silent = FALSE AND m.account_id = :sender_id AND m.status_id = s.id
           WHERE s.id = :id
         UNION ALL
-          SELECT s.id, s.in_reply_to_id, m.id, st.path || s.id, st.depth + 1
-          FROM ancestors st
-          JOIN statuses s ON s.id = st.in_reply_to_id
-          LEFT JOIN mentions m ON m.silent = FALSE AND m.account_id = :sender_id AND m.status_id = s.id
-          WHERE st.mention_id IS NULL AND NOT s.id = ANY(path) AND st.depth < :depth_limit
+          SELECT s.id, s.in_reply_to_id, m.id, ancestors.path || s.id, ancestors.depth + 1
+          FROM ancestors
+          JOIN statuses s ON s.id = ancestors.in_reply_to_id
+          /* early exit if we already have a mention matching our requirements */
+          LEFT JOIN mentions m ON m.silent = FALSE AND m.account_id = :sender_id AND m.status_id = s.id AND s.account_id = :recipient_id
+          WHERE ancestors.mention_id IS NULL AND NOT s.id = ANY(path) AND ancestors.depth < :depth_limit
       )
       SELECT COUNT(*)
-      FROM ancestors st
-      JOIN statuses s ON s.id = st.id
-      WHERE st.mention_id IS NOT NULL AND s.visibility = 3
+      FROM ancestors
+      JOIN statuses s ON s.id = ancestors.id
+      WHERE ancestors.mention_id IS NOT NULL AND s.account_id = :recipient_id AND s.visibility = 3
     SQL
   end
 

app/services/post_status_service.rb

@@ -177,7 +177,7 @@ class PostStatusService < BaseService
 
   def idempotency_duplicate
     if scheduled?
-      @account.schedule_statuses.find(@idempotency_duplicate)
+      @account.scheduled_statuses.find(@idempotency_duplicate)
     else
       @account.statuses.find(@idempotency_duplicate)
     end
@@ -232,7 +232,7 @@ class PostStatusService < BaseService
   end
 
   def scheduled_options
-    @options.tap do |options_hash|
+    @options.dup.tap do |options_hash|
       options_hash[:in_reply_to_id]  = options_hash.delete(:thread)&.id
       options_hash[:application_id]  = options_hash.delete(:application)&.id
       options_hash[:scheduled_at]    = nil

app/validators/email_address_validator.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+# NOTE: I initially wrote this as `EmailValidator` but it ended up clashing
+# with an indirect dependency of ours, `validate_email`, which, turns out,
+# has the same approach as we do, but with an extra check disallowing
+# single-label domains. Decided to not switch to `validate_email` because
+# we do want to allow at least `localhost`.
+
+class EmailAddressValidator < ActiveModel::EachValidator
+  def validate_each(record, attribute, value)
+    value = value.strip
+
+    address = Mail::Address.new(value)
+    record.errors.add(attribute, :invalid) if address.address != value
+  rescue Mail::Field::FieldError
+    record.errors.add(attribute, :invalid)
+  end
+end

app/workers/scheduler/vacuum_scheduler.rb

@@ -22,7 +22,6 @@ class Scheduler::VacuumScheduler
       preview_cards_vacuum,
       backups_vacuum,
       access_tokens_vacuum,
-      applications_vacuum,
       feeds_vacuum,
       imports_vacuum,
     ]
@@ -56,10 +55,6 @@ class Scheduler::VacuumScheduler
     Vacuum::ImportsVacuum.new
   end
 
-  def applications_vacuum
-    Vacuum::ApplicationsVacuum.new
-  end
-
   def content_retention_policy
     ContentRetentionPolicy.current
   end

config/application.rb

@@ -48,6 +48,7 @@ require_relative '../lib/chewy/strategy/bypass_with_warning'
 require_relative '../lib/webpacker/manifest_extensions'
 require_relative '../lib/webpacker/helper_extensions'
 require_relative '../lib/rails/engine_extensions'
+require_relative '../lib/action_dispatch/remote_ip_extensions'
 require_relative '../lib/active_record/database_tasks_extensions'
 require_relative '../lib/active_record/batches'
 require_relative '../lib/simple_navigation/item_extensions'

config/initializers/ffmpeg.rb

@@ -1,5 +1,6 @@
 # frozen_string_literal: true
 
-if ENV['FFMPEG_BINARY'].present?
-  FFMPEG.ffmpeg_binary = ENV['FFMPEG_BINARY']
+Rails.application.configure do
+  config.x.ffmpeg_binary = ENV['FFMPEG_BINARY'] || 'ffmpeg'
+  config.x.ffprobe_binary = ENV['FFPROBE_BINARY'] || 'ffprobe'
 end

config/initializers/rack_attack.rb

@@ -37,6 +37,10 @@ class Rack::Attack
       authenticated_token&.id
     end
 
+    def warden_user_id
+      @env['warden']&.user&.id
+    end
+
     def unauthenticated?
       !authenticated_user_id
     end
@@ -58,10 +62,6 @@ class Rack::Attack
     end
   end
 
-  Rack::Attack.safelist('allow from localhost') do |req|
-    req.remote_ip == '127.0.0.1' || req.remote_ip == '::1'
-  end
-
   Rack::Attack.blocklist('deny from blocklist') do |req|
     IpBlock.blocked?(req.remote_ip)
   end
@@ -105,6 +105,10 @@ class Rack::Attack
     req.authenticated_user_id if (req.post? && req.path.match?(API_DELETE_REBLOG_REGEX)) || (req.delete? && req.path.match?(API_DELETE_STATUS_REGEX))
   end
 
+  throttle('throttle_oauth_application_registrations/ip', limit: 5, period: 10.minutes) do |req|
+    req.throttleable_remote_ip if req.post? && req.path == '/api/v1/apps'
+  end
+
   throttle('throttle_sign_up_attempts/ip', limit: 25, period: 5.minutes) do |req|
     req.throttleable_remote_ip if req.post? && req.path_matches?('/auth')
   end
@@ -137,6 +141,10 @@ class Rack::Attack
     req.session[:attempt_user_id] || req.params.dig('user', 'email').presence if req.post? && req.path_matches?('/auth/sign_in')
   end
 
+  throttle('throttle_password_change/account', limit: 10, period: 10.minutes) do |req|
+    req.warden_user_id if req.put? || (req.patch? && req.path_matches?('/auth'))
+  end
+
   self.throttled_responder = lambda do |request|
     now        = Time.now.utc
     match_data = request.env['rack.attack.match_data']

docker-compose.yml

@@ -56,7 +56,7 @@ services:
 
   web:
     build: .
-    image: ghcr.io/mastodon/mastodon:v4.2.8
+    image: ghcr.io/mastodon/mastodon:v4.2.9
     restart: always
     env_file: .env.production
     command: bash -c "rm -f /mastodon/tmp/pids/server.pid; bundle exec rails s -p 3000"
@@ -77,7 +77,7 @@ services:
 
   streaming:
     build: .
-    image: ghcr.io/mastodon/mastodon:v4.2.8
+    image: ghcr.io/mastodon/mastodon:v4.2.9
     restart: always
     env_file: .env.production
     command: node ./streaming
@@ -95,7 +95,7 @@ services:
 
   sidekiq:
     build: .
-    image: ghcr.io/mastodon/mastodon:v4.2.8
+    image: ghcr.io/mastodon/mastodon:v4.2.9
     restart: always
     env_file: .env.production
     command: bundle exec sidekiq

lib/action_dispatch/remote_ip_extensions.rb

@@ -0,0 +1,72 @@
+# frozen_string_literal: true
+
+# Mastodon is not made to be directly accessed without a reverse proxy.
+# This monkey-patch prevents remote IP address spoofing when being accessed
+# directly.
+#
+# See PR: https://github.com/rails/rails/pull/51610
+
+# In addition to the PR above, it also raises an error if a request with
+# `X-Forwarded-For` or `Client-Ip` comes directly from a client without
+# going through a trusted proxy.
+
+# rubocop:disable all -- This is a mostly vendored file
+
+module ActionDispatch
+  class RemoteIp
+    module GetIpExtensions
+      def calculate_ip
+        # Set by the Rack web server, this is a single value.
+        remote_addr = ips_from(@req.remote_addr).last
+
+        # Could be a CSV list and/or repeated headers that were concatenated.
+        client_ips    = ips_from(@req.client_ip).reverse!
+        forwarded_ips = ips_from(@req.x_forwarded_for).reverse!
+
+        # `Client-Ip` and `X-Forwarded-For` should not, generally, both be set. If they
+        # are both set, it means that either:
+        #
+        # 1) This request passed through two proxies with incompatible IP header
+        #     conventions.
+        #
+        # 2) The client passed one of `Client-Ip` or `X-Forwarded-For`
+        #     (whichever the proxy servers weren't using) themselves.
+        #
+        # Either way, there is no way for us to determine which header is the right one
+        # after the fact. Since we have no idea, if we are concerned about IP spoofing
+        # we need to give up and explode. (If you're not concerned about IP spoofing you
+        # can turn the `ip_spoofing_check` option off.)
+        should_check_ip = @check_ip && client_ips.last && forwarded_ips.last
+        if should_check_ip && !forwarded_ips.include?(client_ips.last)
+          # We don't know which came from the proxy, and which from the user
+          raise IpSpoofAttackError, "IP spoofing attack?! " \
+            "HTTP_CLIENT_IP=#{@req.client_ip.inspect} " \
+            "HTTP_X_FORWARDED_FOR=#{@req.x_forwarded_for.inspect}"
+        end
+
+        # NOTE: Mastodon addition to make sure we don't get requests from a non-trusted client
+        if @check_ip && (forwarded_ips.last || client_ips.last) && !@proxies.any? { |proxy| proxy === remote_addr }
+          raise IpSpoofAttackError, "IP spoofing attack?! client #{remote_addr} is not a trusted proxy " \
+            "HTTP_CLIENT_IP=#{@req.client_ip.inspect} " \
+            "HTTP_X_FORWARDED_FOR=#{@req.x_forwarded_for.inspect}"
+        end
+
+        # We assume these things about the IP headers:
+        #
+        #     - X-Forwarded-For will be a list of IPs, one per proxy, or blank
+        #     - Client-Ip is propagated from the outermost proxy, or is blank
+        #     - REMOTE_ADDR will be the IP that made the request to Rack
+        ips = forwarded_ips + client_ips
+        ips.compact!
+
+        # If every single IP option is in the trusted list, return the IP that's
+        # furthest away
+        filter_proxies([remote_addr] + ips).first || ips.last || remote_addr
+      end
+    end
+  end
+end
+
+ActionDispatch::RemoteIp::GetIp.prepend(ActionDispatch::RemoteIp::GetIpExtensions)
+
+# rubocop:enable all

lib/mastodon/cli/maintenance.rb

@@ -224,7 +224,7 @@ module Mastodon::CLI
         users = User.where(id: row['ids'].split(',')).sort_by(&:updated_at).reverse
         ref_user = users.shift
         say "Multiple users registered with e-mail address #{ref_user.email}.", :yellow
-        say "e-mail will be disabled for the following accounts: #{user.map(&:account).map(&:acct).join(', ')}", :yellow
+        say "e-mail will be disabled for the following accounts: #{users.map { |user| user.account.acct }.join(', ')}", :yellow
         say 'Please reach out to them and set another address with `tootctl account modify` or delete them.', :yellow
 
         users.each_with_index do |user, index|

lib/mastodon/cli/media.rb

@@ -134,7 +134,7 @@ module Mastodon::CLI
 
             model_name      = path_segments.first.classify
             attachment_name = path_segments[1].singularize
-            record_id       = path_segments[2..-2].join.to_i
+            record_id       = path_segments[2...-2].join.to_i
             file_name       = path_segments.last
             record          = record_map.dig(model_name, record_id)
             attachment      = record&.public_send(attachment_name)
@@ -180,7 +180,7 @@ module Mastodon::CLI
           end
 
           model_name      = path_segments.first.classify
-          record_id       = path_segments[2..-2].join.to_i
+          record_id       = path_segments[2...-2].join.to_i
           attachment_name = path_segments[1].singularize
           file_name       = path_segments.last
 
@@ -311,7 +311,7 @@ module Mastodon::CLI
       end
 
       model_name = path_segments.first.classify
-      record_id  = path_segments[2..-2].join.to_i
+      record_id  = path_segments[2...-2].join.to_i
 
       unless PRELOAD_MODEL_WHITELIST.include?(model_name)
         say("Cannot find corresponding model: #{model_name}", :red)
@@ -361,7 +361,7 @@ module Mastodon::CLI
         next unless VALID_PATH_SEGMENTS_SIZE.include?(segments.size)
 
         model_name = segments.first.classify
-        record_id  = segments[2..-2].join.to_i
+        record_id  = segments[2...-2].join.to_i
 
         next unless PRELOAD_MODEL_WHITELIST.include?(model_name)
 

lib/mastodon/sidekiq_middleware.rb

@@ -8,6 +8,7 @@ class Mastodon::SidekiqMiddleware
   rescue Mastodon::HostValidationError
     # Do not retry
   rescue => e
+    clean_up_elasticsearch_connections!
     limit_backtrace_and_raise(e)
   ensure
     clean_up_sockets!
@@ -25,6 +26,32 @@ class Mastodon::SidekiqMiddleware
     clean_up_statsd_socket!
   end
 
+  # This is a hack to immediately free up unused Elasticsearch connections.
+  #
+  # Indeed, Chewy creates one `Elasticsearch::Client` instance per thread,
+  # and each such client manages its long-lasting connection to
+  # Elasticsearch.
+  #
+  # As far as I know, neither `chewy`,  `elasticsearch-transport` or even
+  # `faraday` provide a reliable way to immediately close a connection, and
+  # rely on the underlying object to be garbage-collected instead.
+  #
+  # Furthermore, `sidekiq` creates a new thread each time a job throws an
+  # exception, meaning that each failure will create a new connection, and
+  # the old one will only be closed on full garbage collection.
+  def clean_up_elasticsearch_connections!
+    return unless Chewy.enabled? && Chewy.current[:chewy_client].present?
+
+    Chewy.client.transport.connections.each do |connection|
+      # NOTE: This bit of code is tailored for the HTTPClient Faraday adapter
+      connection.connection.app.instance_variable_get(:@client)&.reset_all
+    end
+
+    Chewy.current.delete(:chewy_client)
+  rescue
+    nil
+  end
+
   def clean_up_redis_socket!
     RedisConfiguration.pool.checkin if Thread.current[:redis]
     Thread.current[:redis] = nil

lib/mastodon/version.rb

@@ -13,7 +13,7 @@ module Mastodon
     end
 
     def patch
-      8
+      9
     end
 
     def default_prerelease

lib/paperclip/image_extractor.rb

@@ -35,7 +35,7 @@ module Paperclip
       dst.binmode
 
       begin
-        command = Terrapin::CommandLine.new('ffmpeg', '-i :source -loglevel :loglevel -y :destination', logger: Paperclip.logger)
+        command = Terrapin::CommandLine.new(Rails.configuration.x.ffmpeg_binary, '-i :source -loglevel :loglevel -y :destination', logger: Paperclip.logger)
         command.run(source: @file.path, destination: dst.path, loglevel: 'fatal')
       rescue Terrapin::ExitStatusError
         dst.close(true)

lib/paperclip/transcoder.rb

@@ -61,7 +61,7 @@ module Paperclip
       command_arguments, interpolations = prepare_command(destination)
 
       begin
-        command = Terrapin::CommandLine.new('ffmpeg', command_arguments.join(' '), logger: Paperclip.logger)
+        command = Terrapin::CommandLine.new(Rails.configuration.x.ffmpeg_binary, command_arguments.join(' '), logger: Paperclip.logger)
         command.run(interpolations)
       rescue Terrapin::ExitStatusError => e
         raise Paperclip::Error, "Error while transcoding #{@basename}: #{e}"

lib/tasks/mastodon.rake

@@ -515,6 +515,7 @@ namespace :mastodon do
           owner_role = UserRole.find_by(name: 'Owner')
           user = User.new(email: email, password: password, confirmed_at: Time.now.utc, account_attributes: { username: username }, bypass_invite_request_check: true, role: owner_role)
           user.save(validate: false)
+          user.approve!
 
           Setting.site_contact_username = username
 

spec/config/initializers/rack_attack_spec.rb

@@ -103,4 +103,46 @@ describe Rack::Attack, type: :request do
       it_behaves_like 'throttled endpoint'
     end
   end
+
+  describe 'throttle excessive oauth application registration requests by IP address' do
+    let(:throttle) { 'throttle_oauth_application_registrations/ip' }
+    let(:limit)  { 5 }
+    let(:period) { 10.minutes }
+    let(:path)   { '/api/v1/apps' }
+    let(:params) do
+      {
+        client_name: 'Throttle Test',
+        redirect_uris: 'urn:ietf:wg:oauth:2.0:oob',
+        scopes: 'read',
+      }
+    end
+
+    let(:request) { -> { post path, params: params, headers: { 'REMOTE_ADDR' => remote_ip } } }
+
+    it_behaves_like 'throttled endpoint'
+  end
+
+  describe 'throttle excessive password change requests by account' do
+    let(:user) { Fabricate(:user, email: 'user@host.example') }
+    let(:limit) { 10 }
+    let(:period) { 10.minutes }
+    let(:request) { -> { put path, headers: { 'REMOTE_ADDR' => remote_ip } } }
+    let(:path) { '/auth' }
+
+    before do
+      sign_in user, scope: :user
+
+      # Unfortunately, devise's `sign_in` helper causes the `session` to be
+      # loaded in the next request regardless of whether it's actually accessed
+      # by the client code.
+      #
+      # So, we make an extra query to clear issue a session cookie instead.
+      #
+      # A less resource-intensive way to deal with that would be to generate the
+      # session cookie manually, but this seems pretty involved.
+      get '/'
+    end
+
+    it_behaves_like 'throttled endpoint'
+  end
 end

spec/controllers/api/v1/featured_tags/suggestions_controller_spec.rb

@@ -7,17 +7,39 @@ describe Api::V1::FeaturedTags::SuggestionsController do
 
   let(:user)    { Fabricate(:user) }
   let(:token)   { Fabricate(:accessible_access_token, resource_owner_id: user.id, scopes: 'read:accounts') }
-  let(:account) { Fabricate(:account) }
+  let(:account) { Fabricate(:account, user: user) }
 
   before do
     allow(controller).to receive(:doorkeeper_token) { token }
   end
 
   describe 'GET #index' do
-    it 'returns http success' do
+    let!(:unused_featured_tag) { Fabricate(:tag, name: 'unused_featured_tag') }
+    let!(:used_tag) { Fabricate(:tag, name: 'used_tag') }
+    let!(:used_featured_tag) { Fabricate(:tag, name: 'used_featured_tag') }
+
+    before do
+      _unused_tag = Fabricate(:tag, name: 'unused_tag')
+
+      # Make relevant tags used by account
+      status = Fabricate(:status, account: account)
+      status.tags << used_tag
+      status.tags << used_featured_tag
+
+      # Feature the relevant tags
+      Fabricate :featured_tag, account: account, name: unused_featured_tag.name
+      Fabricate :featured_tag, account: account, name: used_featured_tag.name
+    end
+
+    it 'returns http success and recently used but not featured tags', :aggregate_failures do
       get :index, params: { account_id: account.id, limit: 2 }
 
-      expect(response).to have_http_status(200)
+      expect(response)
+        .to have_http_status(200)
+      expect(body_as_json)
+        .to contain_exactly(
+          include(name: used_tag.name)
+        )
     end
   end
 end

spec/controllers/application_controller_spec.rb

@@ -221,39 +221,4 @@ describe ApplicationController do
 
     include_examples 'respond_with_error', 422
   end
-
-  describe 'cache_collection' do
-    subject do
-      Class.new(ApplicationController) do
-        public :cache_collection
-      end
-    end
-
-    shared_examples 'receives :with_includes' do |fabricator, klass|
-      it 'uses raw if it is not an ActiveRecord::Relation' do
-        record = Fabricate(fabricator)
-        expect(subject.new.cache_collection([record], klass)).to eq [record]
-      end
-    end
-
-    shared_examples 'cacheable' do |fabricator, klass|
-      include_examples 'receives :with_includes', fabricator, klass
-
-      it 'calls cache_ids of raw if it is an ActiveRecord::Relation' do
-        record = Fabricate(fabricator)
-        relation = klass.none
-        allow(relation).to receive(:cache_ids).and_return([record])
-        expect(subject.new.cache_collection(relation, klass)).to eq [record]
-      end
-    end
-
-    it 'returns raw unless class responds to :with_includes' do
-      raw = Object.new
-      expect(subject.new.cache_collection(raw, Object)).to eq raw
-    end
-
-    context 'with a Status' do
-      include_examples 'cacheable', :status, Status
-    end
-  end
 end

spec/fabricators/featured_tag_fabricator.rb

@@ -2,6 +2,6 @@
 
 Fabricator(:featured_tag) do
   account { Fabricate.build(:account) }
-  tag { Fabricate.build(:tag) }
+  tag { nil }
   name { sequence(:name) { |i| "Tag#{i}" } }
 end

spec/lib/activitypub/parser/status_parser_spec.rb

@@ -0,0 +1,50 @@
+# frozen_string_literal: true
+
+require 'rails_helper'
+
+RSpec.describe ActivityPub::Parser::StatusParser do
+  subject { described_class.new(json) }
+
+  let(:sender) { Fabricate(:account, followers_url: 'http://example.com/followers', domain: 'example.com', uri: 'https://example.com/actor') }
+  let(:follower) { Fabricate(:account, username: 'bob') }
+
+  let(:json) do
+    {
+      '@context': 'https://www.w3.org/ns/activitystreams',
+      id: [ActivityPub::TagManager.instance.uri_for(sender), '#foo'].join,
+      type: 'Create',
+      actor: ActivityPub::TagManager.instance.uri_for(sender),
+      object: object_json,
+    }.with_indifferent_access
+  end
+
+  let(:object_json) do
+    {
+      id: [ActivityPub::TagManager.instance.uri_for(sender), 'post1'].join('/'),
+      type: 'Note',
+      to: [
+        'https://www.w3.org/ns/activitystreams#Public',
+        ActivityPub::TagManager.instance.uri_for(follower),
+      ],
+      content: '@bob lorem ipsum',
+      contentMap: {
+        EN: '@bob lorem ipsum',
+      },
+      published: 1.hour.ago.utc.iso8601,
+      updated: 1.hour.ago.utc.iso8601,
+      tag: {
+        type: 'Mention',
+        href: ActivityPub::TagManager.instance.uri_for(follower),
+      },
+    }
+  end
+
+  it 'correctly parses status' do
+    expect(subject).to have_attributes(
+      text: '@bob lorem ipsum',
+      uri: [ActivityPub::TagManager.instance.uri_for(sender), 'post1'].join('/'),
+      reply: false,
+      language: :en
+    )
+  end
+end

spec/lib/vacuum/applications_vacuum_spec.rb

@@ -1,48 +0,0 @@
-# frozen_string_literal: true
-
-require 'rails_helper'
-
-RSpec.describe Vacuum::ApplicationsVacuum do
-  subject { described_class.new }
-
-  describe '#perform' do
-    let!(:app_with_token)  { Fabricate(:application, created_at: 1.month.ago) }
-    let!(:app_with_grant)  { Fabricate(:application, created_at: 1.month.ago) }
-    let!(:app_with_signup) { Fabricate(:application, created_at: 1.month.ago) }
-    let!(:app_with_owner)  { Fabricate(:application, created_at: 1.month.ago, owner: Fabricate(:user)) }
-    let!(:unused_app)      { Fabricate(:application, created_at: 1.month.ago) }
-    let!(:recent_app)      { Fabricate(:application, created_at: 1.hour.ago) }
-
-    let!(:active_access_token) { Fabricate(:access_token, application: app_with_token) }
-    let!(:active_access_grant) { Fabricate(:access_grant, application: app_with_grant) }
-    let!(:user) { Fabricate(:user, created_by_application: app_with_signup) }
-
-    before do
-      subject.perform
-    end
-
-    it 'does not delete applications with valid access tokens' do
-      expect { app_with_token.reload }.to_not raise_error
-    end
-
-    it 'does not delete applications with valid access grants' do
-      expect { app_with_grant.reload }.to_not raise_error
-    end
-
-    it 'does not delete applications that were used to create users' do
-      expect { app_with_signup.reload }.to_not raise_error
-    end
-
-    it 'does not delete owned applications' do
-      expect { app_with_owner.reload }.to_not raise_error
-    end
-
-    it 'does not delete applications registered less than a day ago' do
-      expect { recent_app.reload }.to_not raise_error
-    end
-
-    it 'deletes unused applications' do
-      expect { unused_app.reload }.to raise_error ActiveRecord::RecordNotFound
-    end
-  end
-end

spec/models/concerns/account_interactions_spec.rb

@@ -250,6 +250,24 @@ describe AccountInteractions do
     end
   end
 
+  describe '#block_idna_domain!' do
+    subject do
+      [
+        account.block_domain!(idna_domain),
+        account.block_domain!(punycode_domain),
+      ]
+    end
+
+    let(:idna_domain) { '대한민국.한국' }
+    let(:punycode_domain) { 'xn--3e0bs9hfvinn1a.xn--3e0b707e' }
+
+    it 'creates single AccountDomainBlock' do
+      expect do
+        expect(subject).to all(be_a AccountDomainBlock)
+      end.to change { account.domain_blocks.count }.by 1
+    end
+  end
+
   describe '#unfollow!' do
     subject { account.unfollow!(target_account) }
 
@@ -345,6 +363,28 @@ describe AccountInteractions do
     end
   end
 
+  describe '#unblock_idna_domain!' do
+    subject { account.unblock_domain!(punycode_domain) }
+
+    let(:idna_domain) { '대한민국.한국' }
+    let(:punycode_domain) { 'xn--3e0bs9hfvinn1a.xn--3e0b707e' }
+
+    context 'when blocking the domain' do
+      it 'returns destroyed AccountDomainBlock' do
+        account_domain_block = Fabricate(:account_domain_block, domain: idna_domain)
+        account.domain_blocks << account_domain_block
+        expect(subject).to be_a AccountDomainBlock
+        expect(subject).to be_destroyed
+      end
+    end
+
+    context 'when unblocking idna domain' do
+      it 'returns nil' do
+        expect(subject).to be_nil
+      end
+    end
+  end
+
   describe '#following?' do
     subject { account.following?(target_account) }
 

spec/models/home_feed_spec.rb

@@ -27,7 +27,6 @@ RSpec.describe HomeFeed do
         results = subject.get(3)
 
         expect(results.map(&:id)).to eq [3, 2]
-        expect(results.first.attributes.keys).to eq %w(id updated_at)
       end
     end
 

spec/models/user_spec.rb

@@ -39,6 +39,12 @@ RSpec.describe User do
       expect(user.valid?).to be true
     end
 
+    it 'is valid with a localhost e-mail address' do
+      user = Fabricate.build(:user, email: 'admin@localhost')
+      user.valid?
+      expect(user.valid?).to be true
+    end
+
     it 'cleans out invalid locale' do
       user = Fabricate.build(:user, locale: 'toto')
       expect(user.valid?).to be true

spec/requests/api/v1/admin/domain_blocks_spec.rb

@@ -49,6 +49,7 @@ RSpec.describe 'Domain Blocks' do
           {
             id: domain_block.id.to_s,
             domain: domain_block.domain,
+            digest: domain_block.domain_digest,
             created_at: domain_block.created_at.strftime('%Y-%m-%dT%H:%M:%S.%LZ'),
             severity: domain_block.severity.to_s,
             reject_media: domain_block.reject_media,
@@ -102,6 +103,7 @@ RSpec.describe 'Domain Blocks' do
         {
           id: domain_block.id.to_s,
           domain: domain_block.domain,
+          digest: domain_block.domain_digest,
           created_at: domain_block.created_at.strftime('%Y-%m-%dT%H:%M:%S.%LZ'),
           severity: domain_block.severity.to_s,
           reject_media: domain_block.reject_media,
@@ -133,14 +135,10 @@ RSpec.describe 'Domain Blocks' do
     it_behaves_like 'forbidden for wrong role', ''
     it_behaves_like 'forbidden for wrong role', 'Moderator'
 
-    it 'returns http success' do
+    it 'creates a domain block with the expected domain name and severity', :aggregate_failures do
       subject
 
       expect(response).to have_http_status(200)
-    end
-
-    it 'returns expected domain name and severity' do
-      subject
 
       body = body_as_json
 
@@ -158,7 +156,44 @@ RSpec.describe 'Domain Blocks' do
       expect(DomainBlock.find_by(domain: 'foo.bar.com')).to be_present
     end
 
-    context 'when a stricter domain block already exists' do
+    context 'when a looser domain block already exists on a higher level domain' do
+      let(:params) { { domain: 'foo.bar.com', severity: :suspend } }
+
+      before do
+        Fabricate(:domain_block, domain: 'bar.com', severity: :silence)
+      end
+
+      it 'creates a domain block with the expected domain name and severity', :aggregate_failures do
+        subject
+
+        body = body_as_json
+
+        expect(response).to have_http_status(200)
+        expect(body).to match a_hash_including(
+          {
+            domain: 'foo.bar.com',
+            severity: 'suspend',
+          }
+        )
+
+        expect(DomainBlock.find_by(domain: 'foo.bar.com')).to be_present
+      end
+    end
+
+    context 'when a domain block already exists on the same domain' do
+      before do
+        Fabricate(:domain_block, domain: 'foo.bar.com', severity: :silence)
+      end
+
+      it 'returns existing domain block in error', :aggregate_failures do
+        subject
+
+        expect(response).to have_http_status(422)
+        expect(body_as_json[:existing_domain_block][:domain]).to eq('foo.bar.com')
+      end
+    end
+
+    context 'when a stricter domain block already exists on a higher level domain' do
       before do
         Fabricate(:domain_block, domain: 'bar.com', severity: :suspend)
       end
@@ -212,6 +247,7 @@ RSpec.describe 'Domain Blocks' do
         {
           id: domain_block.id.to_s,
           domain: domain_block.domain,
+          digest: domain_block.domain_digest,
           severity: 'suspend',
         }
       )

spec/services/notify_service_spec.rb

@@ -76,10 +76,10 @@ RSpec.describe NotifyService, type: :service do
       end
 
       context 'when the message chain is initiated by recipient, but without a mention to the sender, even if the sender sends multiple messages in a row' do
-        let(:reply_to) { Fabricate(:status, account: recipient) }
-        let!(:mention) { Fabricate(:mention, account: sender, status: reply_to) }
-        let(:dummy_reply) { Fabricate(:status, account: sender, visibility: :direct, thread: reply_to) }
-        let(:activity) { Fabricate(:mention, account: recipient, status: Fabricate(:status, account: sender, visibility: :direct, thread: dummy_reply)) }
+        let(:public_status) { Fabricate(:status, account: recipient) }
+        let(:intermediate_reply) { Fabricate(:status, account: sender, thread: public_status, visibility: :direct) }
+        let!(:intermediate_mention) { Fabricate(:mention, account: sender, status: intermediate_reply) }
+        let(:activity) { Fabricate(:mention, account: recipient, status: Fabricate(:status, account: sender, visibility: :direct, thread: intermediate_reply)) }
 
         it 'does not notify' do
           expect { subject }.to_not change(Notification, :count)

spec/services/post_status_service_spec.rb

@@ -54,6 +54,13 @@ RSpec.describe PostStatusService, type: :service do
     it 'does not change statuses count' do
       expect { subject.call(account, text: 'Hi future!', scheduled_at: future, thread: previous_status) }.to_not(change { [account.statuses_count, previous_status.replies_count] })
     end
+
+    it 'returns existing status when used twice with idempotency key' do
+      account = Fabricate(:account)
+      status1 = subject.call(account, text: 'test', idempotency: 'meepmeep', scheduled_at: future)
+      status2 = subject.call(account, text: 'test', idempotency: 'meepmeep', scheduled_at: future)
+      expect(status2.id).to eq status1.id
+    end
   end
 
   it 'creates response to the original status of boost' do