Cache commits on the repository model

Now, when requesting a commit from the Repository model, the results are
not cached. This means we're fetching the same commit by oid multiple times
during the same request. To prevent us from doing this, we now cache
results. Caching is done only based on object id (aka SHA).

Given we cache on the Repository model, results are scoped to the
associated project, eventhough the change of two repositories having the
same oids for different commits is small.

3 files changed, 31 insertions, 17 deletions

diff --git a/app/models/ci/pipeline.rb b/app/models/ci/pipeline.rb
index cf3ce3c9e54..ca65e81f27a 100644
--- a/app/models/ci/pipeline.rb
+++ b/app/models/ci/pipeline.rb
@@ -249,9 +249,7 @@ module Ci
     end
 
     def commit
-      @commit ||= project.commit(sha)
-    rescue
-      nil
+      @commit ||= project.commit_by(oid: sha)
     end
 
     def branch?
diff --git a/app/models/project.rb b/app/models/project.rb
index 4689b588906..7185b4d44fc 100644
--- a/app/models/project.rb
+++ b/app/models/project.rb
@@ -540,6 +540,10 @@ class Project < ActiveRecord::Base
     repository.commit(ref)
   end
 
+  def commit_by(oid:)
+    repository.commit_by(oid: oid)
+  end
+
   # ref can't be HEAD, can only be branch/tag name or SHA
   def latest_successful_builds_for(ref = default_branch)
     latest_pipeline = pipelines.latest_successful_for(ref)
@@ -553,7 +557,7 @@ class Project < ActiveRecord::Base
 
   def merge_base_commit(first_commit_id, second_commit_id)
     sha = repository.merge_base(first_commit_id, second_commit_id)
-    repository.commit(sha) if sha
+    commit_by(oid: sha) if sha
   end
 
   def saved?
diff --git a/app/models/repository.rb b/app/models/repository.rb
index 4324ea46aac..1d1987ec322 100644
--- a/app/models/repository.rb
+++ b/app/models/repository.rb
@@ -76,6 +76,7 @@ class Repository
     @full_path = full_path
     @disk_path = disk_path || full_path
     @project = project
+    @commit_cache = {}
   end
 
   def ==(other)
@@ -103,18 +104,17 @@ class Repository
 
   def commit(ref = 'HEAD')
     return nil unless exists?
+    return ref if ref.is_a?(::Commit)
 
-    commit =
-      if ref.is_a?(Gitlab::Git::Commit)
-        ref
-      else
-        Gitlab::Git::Commit.find(raw_repository, ref)
-      end
+    find_commit(ref)
+  end
 
-    commit = ::Commit.new(commit, @project) if commit
-    commit
-  rescue Rugged::OdbError, Rugged::TreeError
-    nil
+  # Finding a commit by the passed SHA
+  # Also takes care of caching, based on the SHA
+  def commit_by(oid:)
+    return @commit_cache[oid] if @commit_cache.key?(oid)
+
+    @commit_cache[oid] = find_commit(oid)
   end
 
   def commits(ref, path: nil, limit: nil, offset: nil, skip_merges: false, after: nil, before: nil)
@@ -231,7 +231,7 @@ class Repository
   # branches or tags, but we want to keep some of these commits around, for
   # example if they have comments or CI builds.
   def keep_around(sha)
-    return unless sha && commit(sha)
+    return unless sha && commit_by(oid: sha)
 
     return if kept_around?(sha)
 
@@ -1069,6 +1069,18 @@ class Repository
 
   private
 
+  # TODO Generice finder, later split this on finders by Ref or Oid
+  # gitlab-org/gitlab-ce#39239
+  def find_commit(oid_or_ref)
+    commit = if oid_or_ref.is_a?(Gitlab::Git::Commit)
+               oid_or_ref
+             else
+               Gitlab::Git::Commit.find(raw_repository, oid_or_ref)
+             end
+
+    ::Commit.new(commit, @project) if commit
+  end
+
   def blob_data_at(sha, path)
     blob = blob_at(sha, path)
     return unless blob
@@ -1107,12 +1119,12 @@ class Repository
 
   def last_commit_for_path_by_gitaly(sha, path)
     c = raw_repository.gitaly_commit_client.last_commit_for_path(sha, path)
-    commit(c)
+    commit_by(oid: c)
   end
 
   def last_commit_for_path_by_rugged(sha, path)
     sha = last_commit_id_for_path_by_shelling_out(sha, path)
-    commit(sha)
+    commit_by(oid: sha)
   end
 
   def last_commit_id_for_path_by_shelling_out(sha, path)