Use regex to skip unnecessary reference processing in ProcessCommitWorker

4 files changed, 45 insertions, 2 deletions

diff --git a/app/models/concerns/mentionable.rb b/app/models/concerns/mentionable.rb
index 7e56e371b27..5ac56ac6fa0 100644
--- a/app/models/concerns/mentionable.rb
+++ b/app/models/concerns/mentionable.rb
@@ -78,6 +78,8 @@ module Mentionable
 
   # Extract GFM references to other Mentionables from this Mentionable. Always excludes its #local_reference.
   def referenced_mentionables(current_user = self.author)
+    return [] unless matches_cross_reference_regex?
+
     refs = all_references(current_user)
     refs = (refs.issues + refs.merge_requests + refs.commits)
 
@@ -87,6 +89,20 @@ module Mentionable
     refs.reject { |ref| ref == local_reference }
   end
 
+  # Uses regex to quickly determine if mentionables might be referenced
+  # Allows heavy processing to be skipped
+  def matches_cross_reference_regex?
+    reference_pattern = if project.default_issues_tracker?
+                          ReferenceRegexes::DEFAULT_PATTERN
+                        else
+                          ReferenceRegexes::EXTERNAL_PATTERN
+                        end
+
+    self.class.mentionable_attrs.any? do |attr, _|
+      __send__(attr) =~ reference_pattern
+    end
+  end
+
   # Create a cross-reference Note for each GFM reference to another Mentionable found in the +mentionable_attrs+.
   def create_cross_references!(author = self.author, without = [])
     refs = referenced_mentionables(author)
diff --git a/app/models/concerns/mentionable/reference_regexes.rb b/app/models/concerns/mentionable/reference_regexes.rb
new file mode 100644
index 00000000000..1848230ec7e
--- /dev/null
+++ b/app/models/concerns/mentionable/reference_regexes.rb
@@ -0,0 +1,22 @@
+module Mentionable
+  module ReferenceRegexes
+    def self.reference_pattern(link_patterns, issue_pattern)
+      Regexp.union(link_patterns,
+                   issue_pattern,
+                   Commit.reference_pattern,
+                   MergeRequest.reference_pattern)
+    end
+
+    DEFAULT_PATTERN = begin
+      issue_pattern = Issue.reference_pattern
+      link_patterns = Regexp.union([Issue, Commit, MergeRequest].map(&:link_reference_pattern))
+      reference_pattern(link_patterns, issue_pattern)
+    end
+
+    EXTERNAL_PATTERN = begin
+      issue_pattern = ExternalIssue.reference_pattern
+      link_patterns = URI.regexp(%w(http https))
+      reference_pattern(link_patterns, issue_pattern)
+    end
+  end
+end
diff --git a/app/services/git_push_service.rb b/app/services/git_push_service.rb
index 45411c779cc..b8bd7720265 100644
--- a/app/services/git_push_service.rb
+++ b/app/services/git_push_service.rb
@@ -85,8 +85,10 @@ class GitPushService < BaseService
     default = is_default_branch?
 
     push_commits.last(PROCESS_COMMIT_LIMIT).each do |commit|
-      ProcessCommitWorker.
-        perform_async(project.id, current_user.id, commit.to_hash, default)
+      if commit.matches_cross_reference_regex?
+        ProcessCommitWorker.
+          perform_async(project.id, current_user.id, commit.to_hash, default)
+      end
     end
   end
 
diff --git a/app/workers/process_commit_worker.rb b/app/workers/process_commit_worker.rb
index 2f7967cf531..d6ed0e253ad 100644
--- a/app/workers/process_commit_worker.rb
+++ b/app/workers/process_commit_worker.rb
@@ -23,6 +23,9 @@ class ProcessCommitWorker
     return unless user
 
     commit = build_commit(project, commit_hash)
+
+    return unless commit.matches_cross_reference_regex?
+
     author = commit.author || user
 
     process_commit_message(project, commit, user, author, default)