Merge branch 'jej/gitattributes-check-at-ref' into 'master'

Can parse root .gitattributes file for a ref

See merge request gitlab-org/gitlab-ce!16418

4 files changed, 92 insertions, 38 deletions

diff --git a/lib/gitlab/git/attributes_at_ref_parser.rb b/lib/gitlab/git/attributes_at_ref_parser.rb
new file mode 100644
index 00000000000..26b5bd520d5
--- /dev/null
+++ b/lib/gitlab/git/attributes_at_ref_parser.rb
@@ -0,0 +1,14 @@
+module Gitlab
+  module Git
+    # Parses root .gitattributes file at a given ref
+    class AttributesAtRefParser
+      delegate :attributes, to: :@parser
+
+      def initialize(repository, ref)
+        blob = repository.blob_at(ref, '.gitattributes')
+
+        @parser = AttributesParser.new(blob&.data)
+      end
+    end
+  end
+end
diff --git a/lib/gitlab/git/attributes.rb b/lib/gitlab/git/attributes_parser.rb
index 2d20cd473a7..d8aeabb6cba 100644
--- a/lib/gitlab/git/attributes.rb
+++ b/lib/gitlab/git/attributes_parser.rb
@@ -1,42 +1,26 @@
-# Gitaly note: JV: not sure what to make of this class. Why does it use
-# the full disk path of the repository to look up attributes This is
-# problematic in Gitaly, because Gitaly hides the full disk path to the
-# repository from gitlab-ce.
-
 module Gitlab
   module Git
     # Class for parsing Git attribute files and extracting the attributes for
     # file patterns.
-    #
-    # Unlike Rugged this parser only needs a single IO call (a call to `open`),
-    # vastly reducing the time spent in extracting attributes.
-    #
-    # This class _only_ supports parsing the attributes file located at
-    # `$GIT_DIR/info/attributes` as GitLab doesn't use any other files
-    # (`.gitattributes` is copied to this particular path).
-    #
-    # Basic usage:
-    #
-    #     attributes = Gitlab::Git::Attributes.new(some_repo.path)
-    #
-    #     attributes.attributes('README.md') # => { "eol" => "lf }
-    class Attributes
-      # path - The path to the Git repository.
-      def initialize(path)
-        @path = File.expand_path(path)
-        @patterns = nil
+    class AttributesParser
+      def initialize(attributes_data)
+        @data = attributes_data || ""
+
+        if @data.is_a?(File)
+          @patterns = parse_file
+        end
       end
 
       # Returns all the Git attributes for the given path.
       #
-      # path - A path to a file for which to get the attributes.
+      # file_path - A path to a file for which to get the attributes.
       #
       # Returns a Hash.
-      def attributes(path)
-        full_path = File.join(@path, path)
+      def attributes(file_path)
+        absolute_path = File.join('/', file_path)
 
         patterns.each do |pattern, attrs|
-          return attrs if File.fnmatch?(pattern, full_path)
+          return attrs if File.fnmatch?(pattern, absolute_path)
         end
 
         {}
@@ -98,16 +82,10 @@ module Gitlab
 
       # Iterates over every line in the attributes file.
       def each_line
-        full_path = File.join(@path, 'info/attributes')
+        @data.each_line do |line|
+          break unless line.valid_encoding?
 
-        return unless File.exist?(full_path)
-
-        File.open(full_path, 'r') do |handle|
-          handle.each_line do |line|
-            break unless line.valid_encoding?
-
-            yield line.strip
-          end
+          yield line.strip
         end
       end
 
@@ -125,7 +103,8 @@ module Gitlab
 
           parsed = attrs ? parse_attributes(attrs) : {}
 
-          pairs << [File.join(@path, pattern), parsed]
+          absolute_pattern = File.join('/', pattern)
+          pairs << [absolute_pattern, parsed]
         end
 
         # Newer entries take precedence over older entries.
diff --git a/lib/gitlab/git/info_attributes.rb b/lib/gitlab/git/info_attributes.rb
new file mode 100644
index 00000000000..e79a440950b
--- /dev/null
+++ b/lib/gitlab/git/info_attributes.rb
@@ -0,0 +1,49 @@
+# Gitaly note: JV: not sure what to make of this class. Why does it use
+# the full disk path of the repository to look up attributes This is
+# problematic in Gitaly, because Gitaly hides the full disk path to the
+# repository from gitlab-ce.
+
+module Gitlab
+  module Git
+    # Parses gitattributes at `$GIT_DIR/info/attributes`
+    #
+    # Unlike Rugged this parser only needs a single IO call (a call to `open`),
+    # vastly reducing the time spent in extracting attributes.
+    #
+    # This class _only_ supports parsing the attributes file located at
+    # `$GIT_DIR/info/attributes` as GitLab doesn't use any other files
+    # (`.gitattributes` is copied to this particular path).
+    #
+    # Basic usage:
+    #
+    #     attributes = Gitlab::Git::InfoAttributes.new(some_repo.path)
+    #
+    #     attributes.attributes('README.md') # => { "eol" => "lf }
+    class InfoAttributes
+      delegate :attributes, :patterns, to: :parser
+
+      # path - The path to the Git repository.
+      def initialize(path)
+        @repo_path = File.expand_path(path)
+      end
+
+      def parser
+        @parser ||= begin
+          if File.exist?(attributes_path)
+            File.open(attributes_path, 'r') do |file_handle|
+              AttributesParser.new(file_handle)
+            end
+          else
+            AttributesParser.new("")
+          end
+        end
+      end
+
+      private
+
+      def attributes_path
+        @attributes_path ||= File.join(@repo_path, 'info/attributes')
+      end
+    end
+  end
+end
diff --git a/lib/gitlab/git/repository.rb b/lib/gitlab/git/repository.rb
index d7c712e75c5..540fe010127 100644
--- a/lib/gitlab/git/repository.rb
+++ b/lib/gitlab/git/repository.rb
@@ -102,7 +102,7 @@ module Gitlab
         )
         @path = File.join(storage_path, @relative_path)
         @name = @relative_path.split("/").last
-        @attributes = Gitlab::Git::Attributes.new(path)
+        @attributes = Gitlab::Git::InfoAttributes.new(path)
       end
 
       def ==(other)
@@ -991,6 +991,18 @@ module Gitlab
         attributes(path)[name]
       end
 
+      # Check .gitattributes for a given ref
+      #
+      # This only checks the root .gitattributes file,
+      # it does not traverse subfolders to find additional .gitattributes files
+      #
+      # This method is around 30 times slower than `attributes`,
+      # which uses `$GIT_DIR/info/attributes`
+      def attributes_at(ref, file_path)
+        parser = AttributesAtRefParser.new(self, ref)
+        parser.attributes(file_path)
+      end
+
       def languages(ref = nil)
         Gitlab::GitalyClient.migrate(:commit_languages) do |is_enabled|
           if is_enabled