Refactor ActivityPub::FetchRepliesService and ActivityPub::FetchAllRepliesService (#34149)

app/services/activitypub/fetch_all_replies_service.rb

@@ -6,25 +6,15 @@ class ActivityPub::FetchAllRepliesService < ActivityPub::FetchRepliesService
   # Limit of replies to fetch per status
   MAX_REPLIES = (ENV['FETCH_REPLIES_MAX_SINGLE'] || 500).to_i
 
-  def call(collection_or_uri, status_uri, max_pages = nil, request_id: nil)
-    @allow_synchronous_requests = true
-    @collection_or_uri = collection_or_uri
+  def call(status_uri, collection_or_uri, max_pages: 1, request_id: nil)
     @status_uri = status_uri
 
-    @items, n_pages = collection_items(collection_or_uri, max_pages)
-    @items = filtered_replies
-    return if @items.nil?
-
-    FetchReplyWorker.push_bulk(@items) { |reply_uri| [reply_uri, { 'request_id' => request_id }] }
-
-    [@items, n_pages]
+    super
   end
 
   private
 
-  def filtered_replies
-    return if @items.nil?
-
+  def filter_replies(items)
     # Find all statuses that we *shouldn't* update the replies for, and use that as a filter.
     # We don't assume that we have the statuses before they're created,
     # hence the negative filter -
@@ -34,7 +24,7 @@ class ActivityPub::FetchAllRepliesService < ActivityPub::FetchRepliesService
     #
     # Typically we assume the number of replies we *shouldn't* fetch is smaller than the
     # replies we *should* fetch, so we also minimize the number of uris we should load here.
-    uris = @items.map { |item| value_or_id(item) }
+    uris = items.map { |item| value_or_id(item) }
 
     # Expand collection to get replies in the DB that were
     # - not included in the collection,
@@ -61,8 +51,4 @@ class ActivityPub::FetchAllRepliesService < ActivityPub::FetchRepliesService
     Rails.logger.debug { "FetchAllRepliesService - #{@collection_or_uri}: Fetching filtered statuses: #{uris}" }
     uris
   end
-
-  def filter_by_host?
-    false
-  end
 end

app/services/activitypub/fetch_replies_service.rb

@@ -6,53 +6,56 @@ class ActivityPub::FetchRepliesService < BaseService
   # Limit of fetched replies
   MAX_REPLIES = 5
 
-  def call(parent_status, collection_or_uri, allow_synchronous_requests: true, request_id: nil)
-    @account = parent_status.account
+  def call(reference_uri, collection_or_uri, max_pages: 1, allow_synchronous_requests: true, request_id: nil)
+    @reference_uri = reference_uri
     @allow_synchronous_requests = allow_synchronous_requests
 
-    @items, = collection_items(collection_or_uri)
+    @items, n_pages = collection_items(collection_or_uri, max_pages: max_pages)
     return if @items.nil?
 
-    FetchReplyWorker.push_bulk(filtered_replies) { |reply_uri| [reply_uri, { 'request_id' => request_id }] }
+    @items = filter_replies(@items)
+    FetchReplyWorker.push_bulk(@items) { |reply_uri| [reply_uri, { 'request_id' => request_id }] }
 
-    @items
+    [@items, n_pages]
   end
 
   private
 
-  def collection_items(collection_or_uri, max_pages = nil)
+  def collection_items(collection_or_uri, max_pages: 1)
     collection = fetch_collection(collection_or_uri)
     return unless collection.is_a?(Hash)
 
     collection = fetch_collection(collection['first']) if collection['first'].present?
     return unless collection.is_a?(Hash)
 
-    all_items = []
+    items = []
     n_pages = 1
     while collection.is_a?(Hash)
-      items = case collection['type']
-              when 'Collection', 'CollectionPage'
-                collection['items']
-              when 'OrderedCollection', 'OrderedCollectionPage'
-                collection['orderedItems']
-              end
+      items.concat(as_array(collection_page_items(collection)))
 
-      all_items.concat(as_array(items))
-
-      break if all_items.size >= MAX_REPLIES
-      break if !max_pages.nil? && n_pages >= max_pages
+      break if items.size >= MAX_REPLIES
+      break if n_pages >= max_pages
 
       collection = collection['next'].present? ? fetch_collection(collection['next']) : nil
       n_pages += 1
     end
 
-    [all_items, n_pages]
+    [items, n_pages]
+  end
+
+  def collection_page_items(collection)
+    case collection['type']
+    when 'Collection', 'CollectionPage'
+      collection['items']
+    when 'OrderedCollection', 'OrderedCollectionPage'
+      collection['orderedItems']
+    end
   end
 
   def fetch_collection(collection_or_uri)
     return collection_or_uri if collection_or_uri.is_a?(Hash)
     return unless @allow_synchronous_requests
-    return if filter_by_host? && non_matching_uri_hosts?(@account.uri, collection_or_uri)
+    return if non_matching_uri_hosts?(@reference_uri, collection_or_uri)
 
     # NOTE: For backward compatibility reasons, Mastodon signs outgoing
     # queries incorrectly by default.
@@ -70,20 +73,11 @@ class ActivityPub::FetchRepliesService < BaseService
     end
   end
 
-  def filtered_replies
-    if filter_by_host?
-      # Only fetch replies to the same server as the original status to avoid
-      # amplification attacks.
+  def filter_replies(items)
+    # Only fetch replies to the same server as the original status to avoid
+    # amplification attacks.
 
-      # Also limit to 5 fetched replies to limit potential for DoS.
-      @items.map { |item| value_or_id(item) }.reject { |uri| non_matching_uri_hosts?(@account.uri, uri) }.take(MAX_REPLIES)
-    else
-      @items.map { |item| value_or_id(item) }.take(MAX_REPLIES)
-    end
-  end
-
-  # Whether replies with a different domain than the replied_to post should be rejected
-  def filter_by_host?
-    true
+    # Also limit to 5 fetched replies to limit potential for DoS.
+    items.map { |item| value_or_id(item) }.reject { |uri| non_matching_uri_hosts?(@reference_uri, uri) }.take(MAX_REPLIES)
   end
 end