Add latest changes from gitlab-org/gitlab@13-10-stable-eev13.10.0-rc40

7 files changed, 244 insertions, 0 deletions

diff --git a/lib/gitlab/word_diff/chunk_collection.rb b/lib/gitlab/word_diff/chunk_collection.rb
new file mode 100644
index 00000000000..dd388f75302
--- /dev/null
+++ b/lib/gitlab/word_diff/chunk_collection.rb
@@ -0,0 +1,23 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module WordDiff
+    class ChunkCollection
+      def initialize
+        @chunks = []
+      end
+
+      def add(chunk)
+        @chunks << chunk
+      end
+
+      def content
+        @chunks.join('')
+      end
+
+      def reset
+        @chunks = []
+      end
+    end
+  end
+end
diff --git a/lib/gitlab/word_diff/line_processor.rb b/lib/gitlab/word_diff/line_processor.rb
new file mode 100644
index 00000000000..49263962dd6
--- /dev/null
+++ b/lib/gitlab/word_diff/line_processor.rb
@@ -0,0 +1,45 @@
+# frozen_string_literal: true
+
+# Converts a line from `git diff --word-diff=porcelain` output into a segment
+#
+# Possible options:
+# 1. Diff hunk
+# 2. Chunk
+# 3. Newline
+module Gitlab
+  module WordDiff
+    class LineProcessor
+      def initialize(line)
+        @line = line
+      end
+
+      def extract
+        return if empty_line?
+        return Segments::DiffHunk.new(full_line) if diff_hunk?
+        return Segments::Newline.new if newline_delimiter?
+
+        Segments::Chunk.new(full_line)
+      end
+
+      private
+
+      attr_reader :line
+
+      def diff_hunk?
+        line =~ /^@@ -/
+      end
+
+      def empty_line?
+        full_line == ' '
+      end
+
+      def newline_delimiter?
+        full_line == '~'
+      end
+
+      def full_line
+        @full_line ||= line.delete("\n")
+      end
+    end
+  end
+end
diff --git a/lib/gitlab/word_diff/parser.rb b/lib/gitlab/word_diff/parser.rb
new file mode 100644
index 00000000000..3b6d4d4d384
--- /dev/null
+++ b/lib/gitlab/word_diff/parser.rb
@@ -0,0 +1,57 @@
+# frozen_string_literal: true
+
+# Converts git diff --word-diff=porcelain output to Gitlab::Diff::Line objects
+# see: https://git-scm.com/docs/git-diff#Documentation/git-diff.txt-porcelain
+module Gitlab
+  module WordDiff
+    class Parser
+      include Enumerable
+
+      def parse(lines, diff_file: nil)
+        return [] if lines.blank?
+
+        # By returning an Enumerator we make it possible to search for a single line (with #find)
+        # without having to instantiate all the others that come after it.
+        Enumerator.new do |yielder|
+          @chunks = ChunkCollection.new
+          @counter = PositionsCounter.new
+
+          lines.each do |line|
+            segment = LineProcessor.new(line).extract
+
+            case segment
+            when Segments::DiffHunk
+              next if segment.first_line?
+
+              counter.set_pos_num(old: segment.pos_old, new: segment.pos_new)
+
+              yielder << build_line(segment.to_s, 'match', parent_file: diff_file)
+
+            when Segments::Chunk
+              @chunks.add(segment)
+
+            when Segments::Newline
+              yielder << build_line(@chunks.content, nil, parent_file: diff_file)
+
+              @chunks.reset
+              counter.increase_pos_num
+            end
+          end
+        end
+      end
+
+      private
+
+      attr_reader :counter
+
+      def build_line(content, type, options = {})
+        Gitlab::Diff::Line.new(
+          content, type,
+          counter.line_obj_index, counter.pos_old, counter.pos_new,
+          **options).tap do
+          counter.increase_obj_index
+        end
+      end
+    end
+  end
+end
diff --git a/lib/gitlab/word_diff/positions_counter.rb b/lib/gitlab/word_diff/positions_counter.rb
new file mode 100644
index 00000000000..ca66b43755f
--- /dev/null
+++ b/lib/gitlab/word_diff/positions_counter.rb
@@ -0,0 +1,30 @@
+# frozen_string_literal: true
+
+# Responsible for keeping track of line numbers and created Gitlab::Diff::Line objects
+module Gitlab
+  module WordDiff
+    class PositionsCounter
+      def initialize
+        @pos_old = 1
+        @pos_new = 1
+        @line_obj_index = 0
+      end
+
+      attr_reader :pos_old, :pos_new, :line_obj_index
+
+      def increase_pos_num
+        @pos_old += 1
+        @pos_new += 1
+      end
+
+      def increase_obj_index
+        @line_obj_index += 1
+      end
+
+      def set_pos_num(old:, new:)
+        @pos_old = old
+        @pos_new = new
+      end
+    end
+  end
+end
diff --git a/lib/gitlab/word_diff/segments/chunk.rb b/lib/gitlab/word_diff/segments/chunk.rb
new file mode 100644
index 00000000000..7c5850666f9
--- /dev/null
+++ b/lib/gitlab/word_diff/segments/chunk.rb
@@ -0,0 +1,36 @@
+# frozen_string_literal: true
+
+# Chunk is a part of the line that starts with ` `, `-`, `+`
+# Consecutive chunks build a line. Line that starts with `~` is an identifier of
+# end of the line.
+module Gitlab
+  module WordDiff
+    module Segments
+      class Chunk
+        def initialize(line)
+          @line = line
+        end
+
+        def removed?
+          line[0] == '-'
+        end
+
+        def added?
+          line[0] == '+'
+        end
+
+        def to_s
+          line[1..] || ''
+        end
+
+        def length
+          to_s.length
+        end
+
+        private
+
+        attr_reader :line
+      end
+    end
+  end
+end
diff --git a/lib/gitlab/word_diff/segments/diff_hunk.rb b/lib/gitlab/word_diff/segments/diff_hunk.rb
new file mode 100644
index 00000000000..88b6817676f
--- /dev/null
+++ b/lib/gitlab/word_diff/segments/diff_hunk.rb
@@ -0,0 +1,40 @@
+# frozen_string_literal: true
+
+# Diff hunk is line that starts with @@
+# It contains information about start line numbers
+#
+# Example:
+# @@ -1,4 +1,5 @@
+#
+# See more: https://www.gnu.org/software/diffutils/manual/html_node/Detailed-Unified.html
+module Gitlab
+  module WordDiff
+    module Segments
+      class DiffHunk
+        def initialize(line)
+          @line = line
+        end
+
+        def pos_old
+          line.match(/\-[0-9]*/)[0].to_i.abs rescue 0
+        end
+
+        def pos_new
+          line.match(/\+[0-9]*/)[0].to_i.abs rescue 0
+        end
+
+        def first_line?
+          pos_old <= 1 && pos_new <= 1
+        end
+
+        def to_s
+          line
+        end
+
+        private
+
+        attr_reader :line
+      end
+    end
+  end
+end
diff --git a/lib/gitlab/word_diff/segments/newline.rb b/lib/gitlab/word_diff/segments/newline.rb
new file mode 100644
index 00000000000..de8bbf252ff
--- /dev/null
+++ b/lib/gitlab/word_diff/segments/newline.rb
@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module Gitlab
+  module WordDiff
+    module Segments
+      class Newline
+        def to_s
+          ''
+        end
+      end
+    end
+  end
+end