Merge branch 'master' into bootstrap4

32 files changed, 394 insertions, 158 deletions

diff --git a/lib/api/merge_requests.rb b/lib/api/merge_requests.rb
index 3264a26f7d2..d4cc18f622b 100644
--- a/lib/api/merge_requests.rb
+++ b/lib/api/merge_requests.rb
@@ -189,7 +189,7 @@ module API
       post ":id/merge_requests" do
         Gitlab::QueryLimiting.whitelist('https://gitlab.com/gitlab-org/gitlab-ce/issues/42316')
 
-        authorize! :create_merge_request, user_project
+        authorize! :create_merge_request_from, user_project
 
         mr_params = declared_params(include_missing: false)
         mr_params[:force_remove_source_branch] = mr_params.delete(:remove_source_branch)
diff --git a/lib/api/v3/merge_requests.rb b/lib/api/v3/merge_requests.rb
index ce216497996..9b0f70e2bfe 100644
--- a/lib/api/v3/merge_requests.rb
+++ b/lib/api/v3/merge_requests.rb
@@ -93,7 +93,7 @@ module API
         post ":id/merge_requests" do
           Gitlab::QueryLimiting.whitelist('https://gitlab.com/gitlab-org/gitlab-ce/issues/42126')
 
-          authorize! :create_merge_request, user_project
+          authorize! :create_merge_request_from, user_project
 
           mr_params = declared_params(include_missing: false)
           mr_params[:force_remove_source_branch] = mr_params.delete(:remove_source_branch) if mr_params[:remove_source_branch].present?
diff --git a/lib/banzai/commit_renderer.rb b/lib/banzai/commit_renderer.rb
index f5ff95e3eb3..c351a155ae5 100644
--- a/lib/banzai/commit_renderer.rb
+++ b/lib/banzai/commit_renderer.rb
@@ -3,7 +3,7 @@ module Banzai
     ATTRIBUTES = [:description, :title].freeze
 
     def self.render(commits, project, user = nil)
-      obj_renderer = ObjectRenderer.new(project, user)
+      obj_renderer = ObjectRenderer.new(user: user, default_project: project)
 
       ATTRIBUTES.each { |attr| obj_renderer.render(commits, attr) }
     end
diff --git a/lib/banzai/filter/issuable_state_filter.rb b/lib/banzai/filter/issuable_state_filter.rb
index 8f541dcfdb2..1a415232545 100644
--- a/lib/banzai/filter/issuable_state_filter.rb
+++ b/lib/banzai/filter/issuable_state_filter.rb
@@ -11,7 +11,8 @@ module Banzai
       def call
         return doc unless context[:issuable_state_filter_enabled]
 
-        extractor = Banzai::IssuableExtractor.new(project, current_user)
+        context = RenderContext.new(project, current_user)
+        extractor = Banzai::IssuableExtractor.new(context)
         issuables = extractor.extract([doc])
 
         issuables.each do |node, issuable|
diff --git a/lib/banzai/filter/redactor_filter.rb b/lib/banzai/filter/redactor_filter.rb
index 9f9882b3b40..caf11fe94c4 100644
--- a/lib/banzai/filter/redactor_filter.rb
+++ b/lib/banzai/filter/redactor_filter.rb
@@ -7,7 +7,11 @@ module Banzai
     #
     class RedactorFilter < HTML::Pipeline::Filter
       def call
-        Redactor.new(project, current_user).redact([doc]) unless context[:skip_redaction]
+        unless context[:skip_redaction]
+          context = RenderContext.new(project, current_user)
+
+          Redactor.new(context).redact([doc])
+        end
 
         doc
       end
diff --git a/lib/banzai/issuable_extractor.rb b/lib/banzai/issuable_extractor.rb
index 49603d0b363..ae7dc71e7eb 100644
--- a/lib/banzai/issuable_extractor.rb
+++ b/lib/banzai/issuable_extractor.rb
@@ -12,11 +12,11 @@ module Banzai
       [@data-reference-type="issue" or @data-reference-type="merge_request"]
     ).freeze
 
-    attr_reader :project, :user
+    attr_reader :context
 
-    def initialize(project, user)
-      @project = project
-      @user = user
+    # context - An instance of Banzai::RenderContext.
+    def initialize(context)
+      @context = context
     end
 
     # Returns Hash in the form { node => issuable_instance }
@@ -25,8 +25,10 @@ module Banzai
         document.xpath(QUERY)
       end
 
-      issue_parser = Banzai::ReferenceParser::IssueParser.new(project, user)
-      merge_request_parser = Banzai::ReferenceParser::MergeRequestParser.new(project, user)
+      issue_parser = Banzai::ReferenceParser::IssueParser.new(context)
+
+      merge_request_parser =
+        Banzai::ReferenceParser::MergeRequestParser.new(context)
 
       issuables_for_nodes = issue_parser.records_for_nodes(nodes).merge(
         merge_request_parser.records_for_nodes(nodes)
diff --git a/lib/banzai/object_renderer.rb b/lib/banzai/object_renderer.rb
index 2691be81623..a176f1e261b 100644
--- a/lib/banzai/object_renderer.rb
+++ b/lib/banzai/object_renderer.rb
@@ -13,14 +13,13 @@ module Banzai
   # As an example, rendering the attribute `note` would place the unredacted
   # HTML into `note_html` and the redacted HTML into `redacted_note_html`.
   class ObjectRenderer
-    attr_reader :project, :user
+    attr_reader :context
 
-    # project - A Project to use for redacting Markdown.
+    # default_project - A default Project to use for redacting Markdown.
     # user - The user viewing the Markdown/HTML documents, if any.
     # redaction_context - A Hash containing extra attributes to use during redaction
-    def initialize(project, user = nil, redaction_context = {})
-      @project = project
-      @user = user
+    def initialize(default_project: nil, user: nil, redaction_context: {})
+      @context = RenderContext.new(default_project, user)
       @redaction_context = base_context.merge(redaction_context)
     end
 
@@ -48,17 +47,21 @@ module Banzai
       pipeline = HTML::Pipeline.new([])
 
       objects.map do |object|
-        pipeline.to_document(Banzai.render_field(object, attribute))
+        document = pipeline.to_document(Banzai.render_field(object, attribute))
+
+        context.associate_document(document, object)
+
+        document
       end
     end
 
     def post_process_documents(documents, objects, attribute)
       # Called here to populate cache, refer to IssuableExtractor docs
-      IssuableExtractor.new(project, user).extract(documents)
+      IssuableExtractor.new(context).extract(documents)
 
       documents.zip(objects).map do |document, object|
-        context = context_for(object, attribute)
-        Banzai::Pipeline[:post_process].to_document(document, context)
+        pipeline_context = context_for(document, object, attribute)
+        Banzai::Pipeline[:post_process].to_document(document, pipeline_context)
       end
     end
 
@@ -66,20 +69,21 @@ module Banzai
     #
     # Returns an Array containing the redacted documents.
     def redact_documents(documents)
-      redactor = Redactor.new(project, user)
+      redactor = Redactor.new(context)
 
       redactor.redact(documents)
     end
 
     # Returns a Banzai context for the given object and attribute.
-    def context_for(object, attribute)
-      @redaction_context.merge(object.banzai_render_context(attribute))
+    def context_for(document, object, attribute)
+      @redaction_context.merge(object.banzai_render_context(attribute)).merge(
+        project: context.project_for_node(document)
+      )
     end
 
     def base_context
       {
-        current_user: user,
-        project: project,
+        current_user: context.current_user,
         skip_redaction: true
       }
     end
diff --git a/lib/banzai/redactor.rb b/lib/banzai/redactor.rb
index fd457bebf03..28928d6f376 100644
--- a/lib/banzai/redactor.rb
+++ b/lib/banzai/redactor.rb
@@ -2,13 +2,15 @@ module Banzai
   # Class for removing Markdown references a certain user is not allowed to
   # view.
   class Redactor
-    attr_reader :user, :project
+    attr_reader :context
 
-    # project - A Project to use for redacting links.
-    # user - The currently logged in user (if any).
-    def initialize(project, user = nil)
-      @project = project
-      @user = user
+    # context - An instance of `Banzai::RenderContext`.
+    def initialize(context)
+      @context = context
+    end
+
+    def user
+      context.current_user
     end
 
     # Redacts the references in the given Array of documents.
@@ -70,11 +72,11 @@ module Banzai
     end
 
     def redact_cross_project_references(documents)
-      extractor = Banzai::IssuableExtractor.new(project, user)
+      extractor = Banzai::IssuableExtractor.new(context)
       issuables = extractor.extract(documents)
 
       issuables.each do |node, issuable|
-        next if issuable.project == project
+        next if issuable.project == context.project_for_node(node)
 
         node['class'] = node['class'].gsub('has-tooltip', '')
         node['title'] = nil
@@ -95,7 +97,7 @@ module Banzai
       end
 
       per_type.each do |type, nodes|
-        parser = Banzai::ReferenceParser[type].new(project, user)
+        parser = Banzai::ReferenceParser[type].new(context)
 
         visible.merge(parser.nodes_visible_to_user(user, nodes))
       end
diff --git a/lib/banzai/reference_extractor.rb b/lib/banzai/reference_extractor.rb
index 7e6357f8a00..78588299c18 100644
--- a/lib/banzai/reference_extractor.rb
+++ b/lib/banzai/reference_extractor.rb
@@ -10,8 +10,8 @@ module Banzai
     end
 
     def references(type, project, current_user = nil)
-      processor = Banzai::ReferenceParser[type]
-        .new(project, current_user)
+      context = RenderContext.new(project, current_user)
+      processor = Banzai::ReferenceParser[type].new(context)
 
       processor.process(html_documents)
     end
diff --git a/lib/banzai/reference_parser/base_parser.rb b/lib/banzai/reference_parser/base_parser.rb
index 279fca8d043..68752f5bb5a 100644
--- a/lib/banzai/reference_parser/base_parser.rb
+++ b/lib/banzai/reference_parser/base_parser.rb
@@ -45,9 +45,13 @@ module Banzai
         @data_attribute ||= "data-#{reference_type.to_s.dasherize}"
       end
 
-      def initialize(project = nil, current_user = nil)
-        @project = project
-        @current_user = current_user
+      # context - An instance of `Banzai::RenderContext`.
+      def initialize(context)
+        @context = context
+      end
+
+      def project_for_node(node)
+        context.project_for_node(node)
       end
 
       # Returns all the nodes containing references that the user can refer to.
@@ -224,7 +228,11 @@ module Banzai
 
       private
 
-      attr_reader :current_user, :project
+      attr_reader :context
+
+      def current_user
+        context.current_user
+      end
 
       # When a feature is disabled or visible only for
       # team members we should not allow team members
diff --git a/lib/banzai/reference_parser/issue_parser.rb b/lib/banzai/reference_parser/issue_parser.rb
index 230827129b6..6bee5ea15b9 100644
--- a/lib/banzai/reference_parser/issue_parser.rb
+++ b/lib/banzai/reference_parser/issue_parser.rb
@@ -5,15 +5,10 @@ module Banzai
 
       def nodes_visible_to_user(user, nodes)
         issues = records_for_nodes(nodes)
-        issues_to_check = issues.values
+        issues_to_check, cross_project_issues = partition_issues(issues, user)
 
-        unless can?(user, :read_cross_project)
-          issues_to_check, cross_project_issues = issues_to_check.partition do |issue|
-            issue.project == project
-          end
-        end
-
-        readable_issues = Ability.issues_readable_by_user(issues_to_check, user).to_set
+        readable_issues =
+          Ability.issues_readable_by_user(issues_to_check, user).to_set
 
         nodes.select do |node|
           issue_in_node = issues[node]
@@ -25,7 +20,7 @@ module Banzai
           # but not the issue.
           if readable_issues.include?(issue_in_node)
             true
-          elsif cross_project_issues&.include?(issue_in_node)
+          elsif cross_project_issues.include?(issue_in_node)
             can_read_reference?(user, issue_in_node)
           else
             false
@@ -33,6 +28,32 @@ module Banzai
         end
       end
 
+      # issues - A Hash mapping HTML nodes to their corresponding Issue
+      #          instances.
+      # user - The current User.
+      def partition_issues(issues, user)
+        return [issues.values, []] if can?(user, :read_cross_project)
+
+        issues_to_check = []
+        cross_project_issues = []
+
+        # We manually partition the data since our input is a Hash and our
+        # output has to be an Array of issues; not an Array of (node, issue)
+        # pairs.
+        issues.each do |node, issue|
+          target =
+            if issue.project == project_for_node(node)
+              issues_to_check
+            else
+              cross_project_issues
+            end
+
+          target << issue
+        end
+
+        [issues_to_check, cross_project_issues]
+      end
+
       def records_for_nodes(nodes)
         @issues_for_nodes ||= grouped_objects_for_nodes(
           nodes,
diff --git a/lib/banzai/reference_parser/user_parser.rb b/lib/banzai/reference_parser/user_parser.rb
index 8932d4f2905..ceb7f1d165c 100644
--- a/lib/banzai/reference_parser/user_parser.rb
+++ b/lib/banzai/reference_parser/user_parser.rb
@@ -58,7 +58,7 @@ module Banzai
       def can_read_project_reference?(node)
         node_id = node.attr('data-project').to_i
 
-        project && project.id == node_id
+        project_for_node(node)&.id == node_id
       end
 
       def nodes_user_can_reference(current_user, nodes)
@@ -71,6 +71,7 @@ module Banzai
         nodes.select do |node|
           project_id = node.attr(project_attr)
           user_id = node.attr(author_attr)
+          project = project_for_node(node)
 
           if project && project_id && project.id == project_id.to_i
             true
diff --git a/lib/banzai/render_context.rb b/lib/banzai/render_context.rb
new file mode 100644
index 00000000000..e30fc9f469b
--- /dev/null
+++ b/lib/banzai/render_context.rb
@@ -0,0 +1,32 @@
+# frozen_string_literal: true
+
+module Banzai
+  # Object storing the current user, project, and other details used when
+  # parsing Markdown references.
+  class RenderContext
+    attr_reader :current_user
+
+    # default_project - The default project to use for all documents, if any.
+    # current_user - The user viewing the document, if any.
+    def initialize(default_project = nil, current_user = nil)
+      @current_user = current_user
+      @projects = Hash.new(default_project)
+    end
+
+    # Associates an HTML document with a Project.
+    #
+    # document - The HTML document to map to a Project.
+    # object - The object that produced the HTML document.
+    def associate_document(document, object)
+      # XML nodes respond to "document" but will return a Document instance,
+      # even when they belong to a DocumentFragment.
+      document = document.document if document.fragment?
+
+      @projects[document] = object.project if object.respond_to?(:project)
+    end
+
+    def project_for_node(node)
+      @projects[node.document]
+    end
+  end
+end
diff --git a/lib/gitlab/ci/status/build/common.rb b/lib/gitlab/ci/status/build/common.rb
index c0c7c7f5b5d..c1fc70ac266 100644
--- a/lib/gitlab/ci/status/build/common.rb
+++ b/lib/gitlab/ci/status/build/common.rb
@@ -3,6 +3,14 @@ module Gitlab
     module Status
       module Build
         module Common
+          def illustration
+            {
+              image: 'illustrations/skipped-job_empty.svg',
+              size: 'svg-430',
+              title: _('This job does not have a trace.')
+            }
+          end
+
           def has_details?
             can?(user, :read_build, subject)
           end
diff --git a/lib/gitlab/ci/status/build/erased.rb b/lib/gitlab/ci/status/build/erased.rb
index 3a5113b16b6..495227c2ffb 100644
--- a/lib/gitlab/ci/status/build/erased.rb
+++ b/lib/gitlab/ci/status/build/erased.rb
@@ -5,7 +5,7 @@ module Gitlab
         class Erased < Status::Extended
           def illustration
             {
-              image: 'illustrations/skipped-job_empty.svg',
+              image: 'illustrations/erased-log_empty.svg',
               size: 'svg-430',
               title: _('Job has been erased')
             }
diff --git a/lib/gitlab/ci/variables/collection/item.rb b/lib/gitlab/ci/variables/collection/item.rb
index 23ed71db8b0..d00e5b07f95 100644
--- a/lib/gitlab/ci/variables/collection/item.rb
+++ b/lib/gitlab/ci/variables/collection/item.rb
@@ -3,12 +3,9 @@ module Gitlab
     module Variables
       class Collection
         class Item
-          def initialize(**options)
+          def initialize(key:, value:, public: true, file: false)
             @variable = {
-              key: options.fetch(:key),
-              value: options.fetch(:value),
-              public: options.fetch(:public, true),
-              file: options.fetch(:files, false)
+              key: key, value: value, public: public, file: file
             }
           end
 
diff --git a/lib/gitlab/email/handler/create_merge_request_handler.rb b/lib/gitlab/email/handler/create_merge_request_handler.rb
index 3436306e122..2f864f2082b 100644
--- a/lib/gitlab/email/handler/create_merge_request_handler.rb
+++ b/lib/gitlab/email/handler/create_merge_request_handler.rb
@@ -23,7 +23,8 @@ module Gitlab
         def execute
           raise ProjectNotFound unless project
 
-          validate_permission!(:create_merge_request)
+          validate_permission!(:create_merge_request_in)
+          validate_permission!(:create_merge_request_from)
 
           verify_record!(
             record: create_merge_request,
diff --git a/lib/gitlab/git.rb b/lib/gitlab/git.rb
index d4e893b881c..c9abea90d21 100644
--- a/lib/gitlab/git.rb
+++ b/lib/gitlab/git.rb
@@ -1,5 +1,9 @@
 module Gitlab
   module Git
+    # The ID of empty tree.
+    # See http://stackoverflow.com/a/40884093/1856239 and
+    # https://github.com/git/git/blob/3ad8b5bf26362ac67c9020bf8c30eee54a84f56d/cache.h#L1011-L1012
+    EMPTY_TREE_ID = '4b825dc642cb6eb9a060e54bf8d69288fbee4904'.freeze
     BLANK_SHA = ('0' * 40).freeze
     TAG_REF_PREFIX = "refs/tags/".freeze
     BRANCH_REF_PREFIX = "refs/heads/".freeze
diff --git a/lib/gitlab/git/attributes_parser.rb b/lib/gitlab/git/attributes_parser.rb
index d8aeabb6cba..08f4d7d4f5c 100644
--- a/lib/gitlab/git/attributes_parser.rb
+++ b/lib/gitlab/git/attributes_parser.rb
@@ -3,12 +3,8 @@ module Gitlab
     # Class for parsing Git attribute files and extracting the attributes for
     # file patterns.
     class AttributesParser
-      def initialize(attributes_data)
+      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.
@@ -28,7 +24,7 @@ module Gitlab
 
       # Returns a Hash containing the file patterns and their attributes.
       def patterns
-        @patterns ||= parse_file
+        @patterns ||= parse_data
       end
 
       # Parses an attribute string.
@@ -91,8 +87,8 @@ module Gitlab
 
       private
 
-      # Parses the Git attributes file.
-      def parse_file
+      # Parses the Git attributes file contents.
+      def parse_data
         pairs = []
         comment = '#'
 
diff --git a/lib/gitlab/git/info_attributes.rb b/lib/gitlab/git/info_attributes.rb
deleted file mode 100644
index e79a440950b..00000000000
--- a/lib/gitlab/git/info_attributes.rb
+++ /dev/null
@@ -1,49 +0,0 @@
-# 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/raw_diff_change.rb b/lib/gitlab/git/raw_diff_change.rb
new file mode 100644
index 00000000000..eb3d8819239
--- /dev/null
+++ b/lib/gitlab/git/raw_diff_change.rb
@@ -0,0 +1,60 @@
+module Gitlab
+  module Git
+    # This class behaves like a struct with fields :blob_id, :blob_size, :operation, :old_path, :new_path
+    # All those fields are (binary) strings or integers
+    class RawDiffChange
+      attr_reader :blob_id, :blob_size, :old_path, :new_path, :operation
+
+      def initialize(raw_change)
+        parse(raw_change)
+      end
+
+      private
+
+      # Input data has the following format:
+      #
+      # When a file has been modified:
+      # 7e3e39ebb9b2bf433b4ad17313770fbe4051649c 669 M\tfiles/ruby/popen.rb
+      #
+      # When a file has been renamed:
+      # 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)
+        @operation = extract_operation
+        @old_path, @new_path = extract_paths(raw_paths)
+      end
+
+      def extract_paths(file_path)
+        case operation
+        when :renamed
+          file_path.split(/\t/)
+        when :deleted
+          [file_path, nil]
+        when :added
+          [nil, file_path]
+        else
+          [file_path, file_path]
+        end
+      end
+
+      def extract_operation
+        case @raw_operation&.first(1)
+        when 'A'
+          :added
+        when 'C'
+          :copied
+        when 'D'
+          :deleted
+        when 'M'
+          :modified
+        when 'R'
+          :renamed
+        when 'T'
+          :type_changed
+        else
+          :unknown
+        end
+      end
+    end
+  end
+end
diff --git a/lib/gitlab/git/repository.rb b/lib/gitlab/git/repository.rb
index f1b575bd872..36992cbcca0 100644
--- a/lib/gitlab/git/repository.rb
+++ b/lib/gitlab/git/repository.rb
@@ -105,7 +105,6 @@ module Gitlab
         )
         @path = File.join(storage_path, @relative_path)
         @name = @relative_path.split("/").last
-        @attributes = Gitlab::Git::InfoAttributes.new(path)
       end
 
       def ==(other)
@@ -559,6 +558,24 @@ module Gitlab
         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)
+        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
+
+        result
+      end
+
       # Returns the SHA of the most recent common ancestor of +from+ and +to+
       def merge_base(from, to)
         gitaly_migrate(:merge_base) do |is_enabled|
@@ -993,11 +1010,32 @@ module Gitlab
         raise InvalidRef
       end
 
+      def info_attributes
+        return @info_attributes if @info_attributes
+
+        content =
+          gitaly_migrate(:get_info_attributes) do |is_enabled|
+            if is_enabled
+              gitaly_repository_client.info_attributes
+            else
+              attributes_path = File.join(File.expand_path(@path), 'info', 'attributes')
+
+              if File.exist?(attributes_path)
+                File.read(attributes_path)
+              else
+                ""
+              end
+            end
+          end
+
+        @info_attributes = AttributesParser.new(content)
+      end
+
       # Returns the Git attributes for the given file path.
       #
       # See `Gitlab::Git::Attributes` for more information.
       def attributes(path)
-        @attributes.attributes(path)
+        info_attributes.attributes(path)
       end
 
       def gitattribute(path, name)
@@ -1385,7 +1423,8 @@ module Gitlab
       end
 
       def branch_names_contains_sha(sha)
-        gitaly_migrate(:branch_names_contains_sha) do |is_enabled|
+        gitaly_migrate(:branch_names_contains_sha,
+                      status: Gitlab::GitalyClient::MigrationStatus::OPT_OUT) do |is_enabled|
           if is_enabled
             gitaly_ref_client.branch_names_contains_sha(sha)
           else
@@ -1395,7 +1434,8 @@ module Gitlab
       end
 
       def tag_names_contains_sha(sha)
-        gitaly_migrate(:tag_names_contains_sha) do |is_enabled|
+        gitaly_migrate(:tag_names_contains_sha,
+                       status: Gitlab::GitalyClient::MigrationStatus::OPT_OUT) do |is_enabled|
           if is_enabled
             gitaly_ref_client.tag_names_contains_sha(sha)
           else
@@ -2461,6 +2501,35 @@ module Gitlab
 
         result.to_s(16)
       end
+
+      def build_git_cmd(*args)
+        object_directories = alternate_object_directories.join(File::PATH_SEPARATOR)
+
+        env = { 'PWD' => self.path }
+        env['GIT_ALTERNATE_OBJECT_DIRECTORIES'] = object_directories if object_directories.present?
+
+        [
+          env,
+          ::Gitlab.config.git.bin_path,
+          *args,
+          { chdir: self.path }
+        ]
+      end
+
+      def git_diff_cmd(old_rev, new_rev)
+        old_rev = old_rev == ::Gitlab::Git::BLANK_SHA ? ::Gitlab::Git::EMPTY_TREE_ID : old_rev
+
+        build_git_cmd('diff', old_rev, new_rev, '--raw')
+      end
+
+      def git_cat_file_cmd
+        format = '%(objectname) %(objectsize) %(rest)'
+        build_git_cmd('cat-file', "--batch-check=#{format}")
+      end
+
+      def format_git_cat_file_script
+        File.expand_path('../support/format-git-cat-file-input', __FILE__)
+      end
     end
   end
 end
diff --git a/lib/gitlab/git/support/format-git-cat-file-input b/lib/gitlab/git/support/format-git-cat-file-input
new file mode 100755
index 00000000000..2e93c646d0f
--- /dev/null
+++ b/lib/gitlab/git/support/format-git-cat-file-input
@@ -0,0 +1,21 @@
+#!/usr/bin/env ruby
+
+# This script formats the output of the `git diff <old_rev> <new_rev> --raw`
+# command so it can be processed by `git cat-file`
+
+# We need to convert this:
+# ":100644 100644 5f53439... 85bc2f9... R060\tfiles/js/commit.js.coffee\tfiles/js/commit.coffee"
+# To:
+# "85bc2f9 R\tfiles/js/commit.js.coffee\tfiles/js/commit.coffee"
+
+ARGF.each do |line|
+  _, _, old_blob_id, new_blob_id, rest = line.split(/\s/, 5)
+
+  old_blob_id.gsub!(/[^\h]/, '')
+  new_blob_id.gsub!(/[^\h]/, '')
+
+  # We can't pass '0000000...' to `git cat-file` given it will not return info about the deleted file
+  blob_id = new_blob_id =~ /\A0+\z/ ? old_blob_id : new_blob_id
+
+  $stdout.puts "#{blob_id} #{rest}"
+end
diff --git a/lib/gitlab/gitaly_client/commit_service.rb b/lib/gitlab/gitaly_client/commit_service.rb
index 456a8a1a2d6..a36e6c822f9 100644
--- a/lib/gitlab/gitaly_client/commit_service.rb
+++ b/lib/gitlab/gitaly_client/commit_service.rb
@@ -3,10 +3,6 @@ module Gitlab
     class CommitService
       include Gitlab::EncodingHelper
 
-      # The ID of empty tree.
-      # See http://stackoverflow.com/a/40884093/1856239 and https://github.com/git/git/blob/3ad8b5bf26362ac67c9020bf8c30eee54a84f56d/cache.h#L1011-L1012
-      EMPTY_TREE_ID = '4b825dc642cb6eb9a060e54bf8d69288fbee4904'.freeze
-
       def initialize(repository)
         @gitaly_repo = repository.gitaly_repository
         @repository = repository
@@ -37,7 +33,7 @@ module Gitlab
       def diff(from, to, options = {})
         from_id = case from
                   when NilClass
-                    EMPTY_TREE_ID
+                    Gitlab::Git::EMPTY_TREE_ID
                   else
                     if from.respond_to?(:oid)
                       # This is meant to match a Rugged::Commit. This should be impossible in
@@ -50,7 +46,7 @@ module Gitlab
 
         to_id = case to
                 when NilClass
-                  EMPTY_TREE_ID
+                  Gitlab::Git::EMPTY_TREE_ID
                 else
                   if to.respond_to?(:oid)
                     # This is meant to match a Rugged::Commit. This should be impossible in
@@ -352,7 +348,7 @@ module Gitlab
       end
 
       def diff_from_parent_request_params(commit, options = {})
-        parent_id = commit.parent_ids.first || EMPTY_TREE_ID
+        parent_id = commit.parent_ids.first || Gitlab::Git::EMPTY_TREE_ID
 
         diff_between_commits_request_params(parent_id, commit.id, options)
       end
diff --git a/lib/gitlab/gitaly_client/repository_service.rb b/lib/gitlab/gitaly_client/repository_service.rb
index 6441065f5fe..39057beefba 100644
--- a/lib/gitlab/gitaly_client/repository_service.rb
+++ b/lib/gitlab/gitaly_client/repository_service.rb
@@ -50,6 +50,15 @@ module Gitlab
         GitalyClient.call(@storage, :repository_service, :apply_gitattributes, request)
       end
 
+      def info_attributes
+        request = Gitaly::GetInfoAttributesRequest.new(repository: @gitaly_repo)
+
+        response = GitalyClient.call(@storage, :repository_service, :get_info_attributes, request)
+        response.each_with_object("") do |message, attributes|
+          attributes << message.attributes
+        end
+      end
+
       def fetch_remote(remote, ssh_auth:, forced:, no_tags:, timeout:, prune: true)
         request = Gitaly::FetchRemoteRequest.new(
           repository: @gitaly_repo, remote: remote, force: forced,
diff --git a/lib/gitlab/gitaly_client/wiki_service.rb b/lib/gitlab/gitaly_client/wiki_service.rb
index 0d8dd5cb8f4..7a698e4b3f3 100644
--- a/lib/gitlab/gitaly_client/wiki_service.rb
+++ b/lib/gitlab/gitaly_client/wiki_service.rb
@@ -136,7 +136,7 @@ module Gitlab
         wiki_file = nil
 
         response.each do |message|
-          next unless message.name.present?
+          next unless message.name.present? || wiki_file
 
           if wiki_file
             wiki_file.raw_data << message.raw_data
diff --git a/lib/gitlab/user_access.rb b/lib/gitlab/user_access.rb
index 24393f96d96..69952cbb47c 100644
--- a/lib/gitlab/user_access.rb
+++ b/lib/gitlab/user_access.rb
@@ -51,7 +51,7 @@ module Gitlab
       return false unless can_access_git?
 
       if protected?(ProtectedBranch, project, ref)
-        user.can?(:delete_protected_branch, project)
+        user.can?(:push_to_delete_protected_branch, project)
       else
         user.can?(:push_code, project)
       end
diff --git a/lib/gitlab/utils.rb b/lib/gitlab/utils.rb
index b0a492eaa58..aeda66763e8 100644
--- a/lib/gitlab/utils.rb
+++ b/lib/gitlab/utils.rb
@@ -73,6 +73,10 @@ module Gitlab
       nil
     end
 
+    def bytes_to_megabytes(bytes)
+      bytes.to_f / Numeric::MEGABYTE
+    end
+
     # Used in EE
     # Accepts either an Array or a String and returns an array
     def ensure_array_from_string(string_or_array)
diff --git a/lib/rspec_flaky/config.rb b/lib/rspec_flaky/config.rb
index a17ae55910e..06e96f969f1 100644
--- a/lib/rspec_flaky/config.rb
+++ b/lib/rspec_flaky/config.rb
@@ -1,9 +1,7 @@
-require 'json'
-
 module RspecFlaky
   class Config
     def self.generate_report?
-      ENV['FLAKY_RSPEC_GENERATE_REPORT'] == 'true'
+      !!(ENV['FLAKY_RSPEC_GENERATE_REPORT'] =~ /1|true/)
     end
 
     def self.suite_flaky_examples_report_path
diff --git a/lib/rspec_flaky/flaky_examples_collection.rb b/lib/rspec_flaky/flaky_examples_collection.rb
index 973c95b0212..dea23c325be 100644
--- a/lib/rspec_flaky/flaky_examples_collection.rb
+++ b/lib/rspec_flaky/flaky_examples_collection.rb
@@ -1,11 +1,9 @@
-require 'json'
+require 'active_support/hash_with_indifferent_access'
+
+require_relative 'flaky_example'
 
 module RspecFlaky
   class FlakyExamplesCollection < SimpleDelegator
-    def self.from_json(json)
-      new(JSON.parse(json))
-    end
-
     def initialize(collection = {})
       unless collection.is_a?(Hash)
         raise ArgumentError, "`collection` must be a Hash, #{collection.class} given!"
@@ -22,7 +20,7 @@ module RspecFlaky
       super(Hash[collection_of_flaky_examples])
     end
 
-    def to_report
+    def to_h
       Hash[map { |uid, example| [uid, example.to_h] }].deep_symbolize_keys
     end
 
diff --git a/lib/rspec_flaky/listener.rb b/lib/rspec_flaky/listener.rb
index 4a5bfec9967..5b5e4f7c7de 100644
--- a/lib/rspec_flaky/listener.rb
+++ b/lib/rspec_flaky/listener.rb
@@ -1,5 +1,11 @@
 require 'json'
 
+require_relative 'config'
+require_relative 'example'
+require_relative 'flaky_example'
+require_relative 'flaky_examples_collection'
+require_relative 'report'
+
 module RspecFlaky
   class Listener
     # - suite_flaky_examples: contains all the currently tracked flacky example
@@ -9,7 +15,7 @@ module RspecFlaky
     attr_reader :suite_flaky_examples, :flaky_examples
 
     def initialize(suite_flaky_examples_json = nil)
-      @flaky_examples = FlakyExamplesCollection.new
+      @flaky_examples = RspecFlaky::FlakyExamplesCollection.new
       @suite_flaky_examples = init_suite_flaky_examples(suite_flaky_examples_json)
     end
 
@@ -18,47 +24,36 @@ module RspecFlaky
 
       return unless current_example.attempts > 1
 
-      flaky_example = suite_flaky_examples.fetch(current_example.uid) { FlakyExample.new(current_example) }
+      flaky_example = suite_flaky_examples.fetch(current_example.uid) { RspecFlaky::FlakyExample.new(current_example) }
       flaky_example.update_flakiness!(last_attempts_count: current_example.attempts)
 
       flaky_examples[current_example.uid] = flaky_example
     end
 
     def dump_summary(_)
-      write_report_file(flaky_examples, RspecFlaky::Config.flaky_examples_report_path)
+      RspecFlaky::Report.new(flaky_examples).write(RspecFlaky::Config.flaky_examples_report_path)
+      # write_report_file(flaky_examples, RspecFlaky::Config.flaky_examples_report_path)
 
       new_flaky_examples = flaky_examples - suite_flaky_examples
       if new_flaky_examples.any?
         Rails.logger.warn "\nNew flaky examples detected:\n"
-        Rails.logger.warn JSON.pretty_generate(new_flaky_examples.to_report)
+        Rails.logger.warn JSON.pretty_generate(new_flaky_examples.to_h)
 
-        write_report_file(new_flaky_examples, RspecFlaky::Config.new_flaky_examples_report_path)
+        RspecFlaky::Report.new(new_flaky_examples).write(RspecFlaky::Config.new_flaky_examples_report_path)
+        # write_report_file(new_flaky_examples, RspecFlaky::Config.new_flaky_examples_report_path)
       end
     end
 
-    def to_report(examples)
-      Hash[examples.map { |k, ex| [k, ex.to_h] }]
-    end
-
     private
 
     def init_suite_flaky_examples(suite_flaky_examples_json = nil)
-      unless suite_flaky_examples_json
+      if suite_flaky_examples_json
+        RspecFlaky::Report.load_json(suite_flaky_examples_json).flaky_examples
+      else
         return {} unless File.exist?(RspecFlaky::Config.suite_flaky_examples_report_path)
 
-        suite_flaky_examples_json = File.read(RspecFlaky::Config.suite_flaky_examples_report_path)
+        RspecFlaky::Report.load(RspecFlaky::Config.suite_flaky_examples_report_path).flaky_examples
       end
-
-      FlakyExamplesCollection.from_json(suite_flaky_examples_json)
-    end
-
-    def write_report_file(examples_collection, file_path)
-      return unless RspecFlaky::Config.generate_report?
-
-      report_path_dir = File.dirname(file_path)
-      FileUtils.mkdir_p(report_path_dir) unless Dir.exist?(report_path_dir)
-
-      File.write(file_path, JSON.pretty_generate(examples_collection.to_report))
     end
   end
 end
diff --git a/lib/rspec_flaky/report.rb b/lib/rspec_flaky/report.rb
new file mode 100644
index 00000000000..a8730d3b7c7
--- /dev/null
+++ b/lib/rspec_flaky/report.rb
@@ -0,0 +1,54 @@
+require 'json'
+require 'time'
+
+require_relative 'config'
+require_relative 'flaky_examples_collection'
+
+module RspecFlaky
+  # This class is responsible for loading/saving JSON reports, and pruning
+  # outdated examples.
+  class Report < SimpleDelegator
+    OUTDATED_DAYS_THRESHOLD = 90
+
+    attr_reader :flaky_examples
+
+    def self.load(file_path)
+      load_json(File.read(file_path))
+    end
+
+    def self.load_json(json)
+      new(RspecFlaky::FlakyExamplesCollection.new(JSON.parse(json)))
+    end
+
+    def initialize(flaky_examples)
+      unless flaky_examples.is_a?(RspecFlaky::FlakyExamplesCollection)
+        raise ArgumentError, "`flaky_examples` must be a RspecFlaky::FlakyExamplesCollection, #{flaky_examples.class} given!"
+      end
+
+      @flaky_examples = flaky_examples
+      super(flaky_examples)
+    end
+
+    def write(file_path)
+      unless RspecFlaky::Config.generate_report?
+        puts "! Generating reports is disabled. To enable it, please set the `FLAKY_RSPEC_GENERATE_REPORT=1` !" # rubocop:disable Rails/Output
+        return
+      end
+
+      report_path_dir = File.dirname(file_path)
+      FileUtils.mkdir_p(report_path_dir) unless Dir.exist?(report_path_dir)
+
+      File.write(file_path, JSON.pretty_generate(flaky_examples.to_h))
+    end
+
+    def prune_outdated(days: OUTDATED_DAYS_THRESHOLD)
+      outdated_date_threshold = Time.now - (3600 * 24 * days)
+      updated_hash = flaky_examples.dup
+        .delete_if do |uid, hash|
+          hash[:last_flaky_at] && Time.parse(hash[:last_flaky_at]).to_i < outdated_date_threshold.to_i
+        end
+
+      self.class.new(RspecFlaky::FlakyExamplesCollection.new(updated_hash))
+    end
+  end
+end