Merge branch 'backport-rd-5566-improve-performance-of-repository-size-limit-check' into 'master'

Backport some changes from gitlab-ee!5476

See merge request gitlab-org/gitlab-ce!18760

5 files changed, 55 insertions, 17 deletions

diff --git a/lib/gitlab/git/blob.rb b/lib/gitlab/git/blob.rb
index eabcf46cf58..d78d29b7ac6 100644
--- a/lib/gitlab/git/blob.rb
+++ b/lib/gitlab/git/blob.rb
@@ -62,6 +62,12 @@ module Gitlab
           end
         end
 
+        # Returns an array of Blob instances just with the metadata, that means
+        # the data attribute has no content.
+        def batch_metadata(repository, blob_references)
+          batch(repository, blob_references, blob_size_limit: 0)
+        end
+
         # Find LFS blobs given an array of sha ids
         # Returns array of Gitlab::Git::Blob
         # Does not guarantee blob data will be set
diff --git a/lib/gitlab/git/raw_diff_change.rb b/lib/gitlab/git/raw_diff_change.rb
index 6042e993113..98de9328071 100644
--- a/lib/gitlab/git/raw_diff_change.rb
+++ b/lib/gitlab/git/raw_diff_change.rb
@@ -28,13 +28,14 @@ module Gitlab
       # 85bc2f9753afd5f4fc5d7c75f74f8d526f26b4f3 107 R060\tfiles/js/commit.js.coffee\tfiles/js/commit.coffee
       def parse(raw_change)
         @blob_id, @blob_size, @raw_operation, raw_paths = raw_change.split(' ', 4)
+        @blob_size = @blob_size.to_i
         @operation = extract_operation
         @old_path, @new_path = extract_paths(raw_paths)
       end
 
       def extract_paths(file_path)
         case operation
-        when :renamed
+        when :copied, :renamed
           file_path.split(/\t/)
         when :deleted
           [file_path, nil]
diff --git a/lib/gitlab/git/repository.rb b/lib/gitlab/git/repository.rb
index b145001a024..5d47f8b2075 100644
--- a/lib/gitlab/git/repository.rb
+++ b/lib/gitlab/git/repository.rb
@@ -579,29 +579,40 @@ module Gitlab
         count_commits(from: from, to: to, **options)
       end
 
+      # Counts the amount of commits between `from` and `to`.
+      def count_commits_between(from, to, options = {})
+        count_commits(from: from, to: to, **options)
+      end
+
       # old_rev and new_rev are commit ID's
       # the result of this method is an array of Gitlab::Git::RawDiffChange
       def raw_changes_between(old_rev, new_rev)
-        gitaly_migrate(:raw_changes_between) do |is_enabled|
-          if is_enabled
-            gitaly_repository_client.raw_changes_between(old_rev, new_rev)
-              .each_with_object([]) do |msg, arr|
-              msg.raw_changes.each { |change| arr << ::Gitlab::Git::RawDiffChange.new(change) }
-            end
-          else
-            result = []
+        @raw_changes_between ||= {}
 
-            circuit_breaker.perform do
-              Open3.pipeline_r(git_diff_cmd(old_rev, new_rev), format_git_cat_file_script, git_cat_file_cmd) do |last_stdout, wait_threads|
-                last_stdout.each_line { |line| result << ::Gitlab::Git::RawDiffChange.new(line.chomp!) }
+        @raw_changes_between[[old_rev, new_rev]] ||= begin
+          return [] if new_rev.blank? || new_rev == Gitlab::Git::BLANK_SHA
 
-                if wait_threads.any? { |waiter| !waiter.value&.success? }
-                  raise ::Gitlab::Git::Repository::GitError, "Unabled to obtain changes between #{old_rev} and #{new_rev}"
+          gitaly_migrate(:raw_changes_between) do |is_enabled|
+            if is_enabled
+              gitaly_repository_client.raw_changes_between(old_rev, new_rev)
+                .each_with_object([]) do |msg, arr|
+                msg.raw_changes.each { |change| arr << ::Gitlab::Git::RawDiffChange.new(change) }
+              end
+            else
+              result = []
+
+              circuit_breaker.perform do
+                Open3.pipeline_r(git_diff_cmd(old_rev, new_rev), format_git_cat_file_script, git_cat_file_cmd) do |last_stdout, wait_threads|
+                  last_stdout.each_line { |line| result << ::Gitlab::Git::RawDiffChange.new(line.chomp!) }
+
+                  if wait_threads.any? { |waiter| !waiter.value&.success? }
+                    raise ::Gitlab::Git::Repository::GitError, "Unabled to obtain changes between #{old_rev} and #{new_rev}"
+                  end
                 end
               end
-            end
 
-            result
+              result
+            end
           end
         end
       rescue ArgumentError => e
diff --git a/spec/lib/gitlab/git/blob_spec.rb b/spec/lib/gitlab/git/blob_spec.rb
index 67d898e787e..e2547ed0311 100644
--- a/spec/lib/gitlab/git/blob_spec.rb
+++ b/spec/lib/gitlab/git/blob_spec.rb
@@ -251,6 +251,26 @@ describe Gitlab::Git::Blob, seed_helper: true do
     end
   end
 
+  describe '.batch_metadata' do
+    let(:blob_references) do
+      [
+        [SeedRepo::Commit::ID, "files/ruby/popen.rb"],
+        [SeedRepo::Commit::ID, 'six']
+      ]
+    end
+
+    subject { described_class.batch_metadata(repository, blob_references) }
+
+    it 'returns an empty data attribute' do
+      first_blob, last_blob = subject
+
+      expect(first_blob.data).to be_blank
+      expect(first_blob.path).to eq("files/ruby/popen.rb")
+      expect(last_blob.data).to be_blank
+      expect(last_blob.path).to eq("six")
+    end
+  end
+
   describe '.batch_lfs_pointers' do
     let(:tree_object) { repository.rugged.rev_parse('master^{tree}') }
 
diff --git a/spec/lib/gitlab/git/raw_diff_change_spec.rb b/spec/lib/gitlab/git/raw_diff_change_spec.rb
index eedde34534f..a0bb37fd84a 100644
--- a/spec/lib/gitlab/git/raw_diff_change_spec.rb
+++ b/spec/lib/gitlab/git/raw_diff_change_spec.rb
@@ -12,7 +12,7 @@ describe Gitlab::Git::RawDiffChange do
       expect(change.operation).to eq(:unknown)
       expect(change.old_path).to be_blank
       expect(change.new_path).to be_blank
-      expect(change.blob_size).to be_blank
+      expect(change.blob_size).to eq(0)
     end
   end