Add Fetch All Replies Part 1: Backend (#32615)

Signed-off-by: sneakers-the-rat <sneakers-the-rat@protonmail.com>
Co-authored-by: jonny <j@nny.fyi>
Co-authored-by: Claire <claire.github-309c@sitedethib.com>
Co-authored-by: Kouhai <66407198+kouhaidev@users.noreply.github.com>

spec/services/activitypub/fetch_all_replies_service_spec.rb

@@ -0,0 +1,90 @@
+# frozen_string_literal: true
+
+require 'rails_helper'
+
+RSpec.describe ActivityPub::FetchAllRepliesService do
+  subject { described_class.new }
+
+  let(:actor)          { Fabricate(:account, domain: 'example.com', uri: 'http://example.com/account') }
+  let(:status)         { Fabricate(:status, account: actor) }
+  let(:collection_uri) { 'http://example.com/replies/1' }
+
+  let(:items) do
+    [
+      'http://example.com/self-reply-1',
+      'http://example.com/self-reply-2',
+      'http://example.com/self-reply-3',
+      'http://other.com/other-reply-1',
+      'http://other.com/other-reply-2',
+      'http://other.com/other-reply-3',
+      'http://example.com/self-reply-4',
+      'http://example.com/self-reply-5',
+      'http://example.com/self-reply-6',
+    ]
+  end
+
+  let(:payload) do
+    {
+      '@context': 'https://www.w3.org/ns/activitystreams',
+      type: 'Collection',
+      id: collection_uri,
+      items: items,
+    }.with_indifferent_access
+  end
+
+  describe '#call' do
+    it 'fetches more than the default maximum and from multiple domains' do
+      allow(FetchReplyWorker).to receive(:push_bulk)
+
+      subject.call(payload, status.uri)
+
+      expect(FetchReplyWorker).to have_received(:push_bulk).with(%w(http://example.com/self-reply-1 http://example.com/self-reply-2 http://example.com/self-reply-3 http://other.com/other-reply-1 http://other.com/other-reply-2 http://other.com/other-reply-3 http://example.com/self-reply-4
+                                                                    http://example.com/self-reply-5 http://example.com/self-reply-6))
+    end
+
+    context 'with a recent status' do
+      before do
+        Fabricate(:status, uri: 'http://example.com/self-reply-2', fetched_replies_at: 1.second.ago, local: false)
+      end
+
+      it 'skips statuses that have been updated recently' do
+        allow(FetchReplyWorker).to receive(:push_bulk)
+
+        subject.call(payload, status.uri)
+
+        expect(FetchReplyWorker).to have_received(:push_bulk).with(%w(http://example.com/self-reply-1 http://example.com/self-reply-3 http://other.com/other-reply-1 http://other.com/other-reply-2 http://other.com/other-reply-3 http://example.com/self-reply-4 http://example.com/self-reply-5 http://example.com/self-reply-6))
+      end
+    end
+
+    context 'with an old status' do
+      before do
+        Fabricate(:status, uri: 'http://other.com/other-reply-1', fetched_replies_at: 1.year.ago, created_at: 1.year.ago, account: actor)
+      end
+
+      it 'updates the time that fetched statuses were last fetched' do
+        allow(FetchReplyWorker).to receive(:push_bulk)
+
+        subject.call(payload, status.uri)
+
+        expect(Status.find_by(uri: 'http://other.com/other-reply-1').fetched_replies_at).to be >= 1.minute.ago
+      end
+    end
+
+    context 'with unsubscribed replies' do
+      before do
+        remote_actor = Fabricate(:account, domain: 'other.com', uri: 'http://other.com/account')
+        # reply not in the collection from the remote instance, but we know about anyway without anyone following the account
+        Fabricate(:status, account: remote_actor, in_reply_to_id: status.id, uri: 'http://other.com/account/unsubscribed', fetched_replies_at: 1.year.ago, created_at: 1.year.ago)
+      end
+
+      it 'updates the unsubscribed replies' do
+        allow(FetchReplyWorker).to receive(:push_bulk)
+
+        subject.call(payload, status.uri)
+
+        expect(FetchReplyWorker).to have_received(:push_bulk).with(%w(http://example.com/self-reply-1 http://example.com/self-reply-2 http://example.com/self-reply-3 http://other.com/other-reply-1 http://other.com/other-reply-2 http://other.com/other-reply-3 http://example.com/self-reply-4
+                                                                      http://example.com/self-reply-5 http://example.com/self-reply-6 http://other.com/account/unsubscribed))
+      end
+    end
+  end
+end

spec/services/activitypub/fetch_remote_status_service_spec.rb

@@ -9,6 +9,9 @@ RSpec.describe ActivityPub::FetchRemoteStatusService do
 
   let!(:sender) { Fabricate(:account, domain: 'foo.bar', uri: 'https://foo.bar') }
 
+  let(:follower) { Fabricate(:account, username: 'alice') }
+  let(:follow) { nil }
+  let(:response) { { body: Oj.dump(object), headers: { 'content-type': 'application/activity+json' } } }
   let(:existing_status) { nil }
 
   let(:note) do
@@ -23,13 +26,14 @@ RSpec.describe ActivityPub::FetchRemoteStatusService do
 
   before do
     stub_request(:get, 'https://foo.bar/watch?v=12345').to_return(status: 404, body: '')
-    stub_request(:get, object[:id]).to_return(body: Oj.dump(object))
+    stub_request(:get, object[:id]).to_return(**response)
   end
 
   describe '#call' do
     before do
+      follow
       existing_status
-      subject.call(object[:id], prefetched_body: Oj.dump(object))
+      subject.call(object[:id])
     end
 
     context 'with Note object' do
@@ -254,6 +258,51 @@ RSpec.describe ActivityPub::FetchRemoteStatusService do
           expect(existing_status.text).to eq 'Lorem ipsum'
           expect(existing_status.edits).to_not be_empty
         end
+
+        context 'when the status appears to have been deleted at source' do
+          let(:response) { { status: 404, body: '' } }
+
+          shared_examples 'no delete' do
+            it 'does not delete the status' do
+              existing_status.reload
+              expect(existing_status.text).to eq 'Foo'
+              expect(existing_status.edits).to be_empty
+            end
+          end
+
+          context 'when the status is orphaned/unsubscribed' do
+            it 'deletes the orphaned status' do
+              expect { existing_status.reload }.to raise_error(ActiveRecord::RecordNotFound)
+            end
+          end
+
+          context 'when the status is from an account with only remote followers' do
+            let(:follower) { Fabricate(:account, username: 'alice', domain: 'foo.bar') }
+            let(:follow) { Fabricate(:follow, account: follower, target_account: sender, created_at: 2.days.ago) }
+
+            it 'deletes the orphaned status' do
+              expect { existing_status.reload }.to raise_error(ActiveRecord::RecordNotFound)
+            end
+
+            context 'when the status is private' do
+              let(:existing_status) { Fabricate(:status, account: sender, text: 'Foo', uri: note[:id], visibility: :private) }
+
+              it_behaves_like 'no delete'
+            end
+
+            context 'when the status is direct' do
+              let(:existing_status) { Fabricate(:status, account: sender, text: 'Foo', uri: note[:id], visibility: :direct) }
+
+              it_behaves_like 'no delete'
+            end
+          end
+
+          context 'when the status is from an account with local followers' do
+            let(:follow) { Fabricate(:follow, account: follower, target_account: sender, created_at: 2.days.ago) }
+
+            it_behaves_like 'no delete'
+          end
+        end
       end
 
       context 'with a Create activity' do