normalize fa cdn hosts to avoid redownloading files

app/jobs/domain/fa/job/scan_post_job.rb

@@ -117,11 +117,25 @@ class Domain::Fa::Job::ScanPostJob < Domain::Fa::Job::Base
     uri = Addressable::URI.parse(submission.full_res_img)
     uri.scheme = "https" if uri.scheme.blank?
 
-    if (file = post.file) && (file.url_str != uri.to_s)
-      file = post.files.build(url_str: uri.to_s)
-    else
-      file = post.file || post.build_file(url_str: uri.to_s)
+    # resolve the existing file and check if the URL has changed.
+    # sometimes, the domain will change from `d.facdn.net` to
+    # `d.furaffinity.net`, and we want to ignore the change in that case
+    file = post.file
+    if file && (old_url_str = file.url_str) && (old_url_str != uri.to_s)
+      if self.class.uri_same_with_normalized_facdn_host?(old_url_str, uri.to_s)
+        logger.info(
+          format_tags(
+            make_tag("old_url_str", old_url_str),
+            make_tag("new_url_str", uri.to_s),
+            "file url has changed, but is the same domain",
+          ),
+        )
+      else
+        file = post.files.build(url_str: uri.to_s)
+      end
     end
+    file ||= post.build_file(url_str: uri.to_s)
+
     if file.url_str_changed?
       file.enqueue_job_after_save(
         Domain::Fa::Job::ScanFileJob,
@@ -142,4 +156,22 @@ class Domain::Fa::Job::ScanPostJob < Domain::Fa::Job::Base
     post.posted_at = submission.posted_date&.in_time_zone("UTC")
     post.scanned_at = Time.now
   end
+
+  FA_CDN_HOSTS = %w[d.facdn.net d.furaffinity.net].freeze
+  sig { params(url_str: String, new_url_str: String).returns(T::Boolean) }
+  def self.uri_same_with_normalized_facdn_host?(url_str, new_url_str)
+    uri = Addressable::URI.parse(url_str)
+    new_uri = Addressable::URI.parse(new_url_str)
+    uri.scheme = nil
+    new_uri.scheme = nil
+
+    if [uri, new_uri].all? { |uri| FA_CDN_HOSTS.include?(uri.host) }
+      # both URIs have an facdn host, so compare them but ignore the host
+      uri.host = nil
+      new_uri.host = nil
+      uri == new_uri
+    else
+      url_str == new_url_str
+    end
+  end
 end

spec/jobs/domain/fa/job/scan_post_job_spec.rb

@@ -219,4 +219,81 @@ describe Domain::Fa::Job::ScanPostJob do
       expect(post.state).to eq("removed")
     end
   end
+
+  describe "#uri_same_with_normalized_facdn_host?" do
+    let(:client_mock_config) { [] }
+
+    shared_examples "has result" do |result|
+      it "is #{result.to_s}, both have schema" do
+        url1 = "https://#{host1}#{path1}"
+        url2 = "https://#{host2}#{path2}"
+        expect(
+          described_class.uri_same_with_normalized_facdn_host?(url1, url2),
+        ).to eq(result)
+      end
+
+      it "is #{result.to_s}, both missing schema" do
+        url1 = "//#{host1}#{path1}"
+        url2 = "//#{host2}#{path2}"
+        expect(
+          described_class.uri_same_with_normalized_facdn_host?(url1, url2),
+        ).to eq(result)
+      end
+
+      it "is #{result.to_s}, one has schema" do
+        url1 = "https://#{host1}#{path1}"
+        url2 = "//#{host2}#{path2}"
+        expect(
+          described_class.uri_same_with_normalized_facdn_host?(url1, url2),
+        ).to eq(result)
+      end
+    end
+
+    shared_context "host: different cdn hosts" do
+      let(:host1) { "d.facdn.net" }
+      let(:host2) { "d.furaffinity.net" }
+    end
+
+    shared_context "host: both hosts are d.facdn.net" do
+      let(:host1) { "d.facdn.net" }
+      let(:host2) { "d.facdn.net" }
+    end
+
+    shared_context "host: both hosts are d.furaffinity.net" do
+      let(:host1) { "d.furaffinity.net" }
+      let(:host2) { "d.furaffinity.net" }
+    end
+
+    shared_context "host: one domain is not a cdn" do
+      let(:host1) { "d.facdn.net" }
+      let(:host2) { "example.com" }
+    end
+
+    shared_context "paths: are the same" do
+      let(:path1) { "/art/user/1234567890/image.jpg" }
+      let(:path2) { "/art/user/1234567890/image.jpg" }
+    end
+
+    shared_context "paths: are different" do
+      let(:path1) { "/art/user/1234567890/image.jpg" }
+      let(:path2) { "/art/user/1234567890/some_other_image.jpg" }
+    end
+
+    [
+      ["host: different cdn hosts", "paths: are the same", true],
+      ["host: both hosts are d.facdn.net", "paths: are the same", true],
+      ["host: both hosts are d.furaffinity.net", "paths: are the same", true],
+      ["host: one domain is not a cdn", "paths: are the same", false],
+      ["host: different cdn hosts", "paths: are different", false],
+      ["host: both hosts are d.facdn.net", "paths: are different", false],
+      ["host: both hosts are d.furaffinity.net", "paths: are different", false],
+      ["host: one domain is not a cdn", "paths: are different", false],
+    ].each do |host_context, path_context, result|
+      context "#{host_context} and #{path_context}" do
+        include_context host_context
+        include_context path_context
+        include_examples "has result", result
+      end
+    end
+  end
 end