Merge branch '3011-fix-advanced-search-syntax-commits' into 'master'

Fix advanced search syntax for commits

Closes #3011

See merge request !2709

changelogs/unreleased-ee/3011-fix-advanced-search-syntax-commits.yml

+---
+title: Fix global code search when using negation queries
+merge_request: 2709
+author:
+type: fixed

lib/elasticsearch/git/repository.rb

@@ -407,13 +407,14 @@ module Elasticsearch
           query_hash = {
             query: {
               bool: {
-                must: [{
+                must: {
                   simple_query_string: {
                     fields: fields,
                     query: query,
-                    default_operator: :or
+                    default_operator: :and
                   }
-                }]
+                },
+                filter: [{ term: { 'commit.type' => 'commit' } }]
               }
             },
             size: per,
@@ -426,7 +427,7 @@ module Elasticsearch
           end
 
           if options[:repository_id]
-            query_hash[:query][:bool][:filter] = {
+            query_hash[:query][:bool][:filter] << {
               terms: {
                 'commit.rid' => [options[:repository_id]].flatten
               }
@@ -434,7 +435,6 @@ module Elasticsearch
           end
 
           if options[:additional_filter]
-            query_hash[:query][:bool][:filter] ||= []
             query_hash[:query][:bool][:filter] << options[:additional_filter]
           end
 
@@ -474,15 +474,14 @@ module Elasticsearch
                     default_operator: :and,
                     fields: %w[blob.content blob.file_name]
                   }
-                }
+                },
+                filter: [{ term: { 'blob.type' => 'blob' } }]
               }
             },
             size: per,
             from: per * (page - 1)
           }
 
-          query_hash[:query][:bool][:filter] = []
-
           if options[:repository_id]
             query_hash[:query][:bool][:filter] << {
               terms: {
@@ -492,7 +491,6 @@ module Elasticsearch
           end
 
           if options[:additional_filter]
-            query_hash[:query][:bool][:filter] ||= []
             query_hash[:query][:bool][:filter] << options[:additional_filter]
           end
 
@@ -527,26 +525,6 @@ module Elasticsearch
             total_count: res.size
           }
         end
-
-        def search_file_names(query, page: 1, per: 20, options: {})
-          query_hash = {
-            fields: ['blob.path'],
-            query: {
-                fuzzy: {
-                    'repository.blob.path' => { value: query }
-                }
-            },
-            filter: {
-                term: {
-                    'repository.blob.rid' => [options[:repository_id]].flatten
-                }
-            },
-            size: per,
-            from: per * (page - 1)
-          }
-
-          self.__elasticsearch__.search(query_hash)
-        end
       end
     end
   end

lib/gitlab/elastic/search_results.rb

@@ -75,8 +75,11 @@ module Gitlab
         content = result["_source"]["blob"]["content"]
         total_lines = content.lines.size
 
-        highlighted_content = result["highlight"]["blob.content"]
-        term = highlighted_content && highlighted_content[0].match(/gitlabelasticsearch→(.*?)←gitlabelasticsearch/)[1]
+        term =
+          if result['highlight']
+            highlighted = result['highlight']['blob.content']
+            highlighted && highlighted[0].match(/gitlabelasticsearch→(.*?)←gitlabelasticsearch/)[1]
+          end
 
         found_line_number = 0
 

spec/lib/gitlab/elastic/search_results_spec.rb

@@ -16,6 +16,49 @@ describe Gitlab::Elastic::SearchResults do
   let(:project_2) { create(:project, :repository) }
   let(:limit_project_ids) { [project_1.id] }
 
+  describe 'parse_search_result' do
+    let(:blob) do
+      {
+        'blob' => {
+          'commit_sha' => 'sha',
+          'content' => "foo\nbar\nbaz\n",
+          'path' => 'path/file.ext'
+        }
+      }
+    end
+
+    it 'returns an unhighlighted blob when no highlight data is present' do
+      parsed = described_class.parse_search_result('_source' => blob)
+
+      expect(parsed).to be_kind_of(::Gitlab::SearchResults::FoundBlob)
+      expect(parsed).to have_attributes(
+        startline: 1,
+        data: "foo\n"
+      )
+    end
+
+    it 'parses the blob with highlighting' do
+      result = {
+        '_source' => blob,
+        'highlight' => {
+          'blob.content' => ["foo\ngitlabelasticsearch→bar←gitlabelasticsearch\nbaz\n"]
+        }
+      }
+
+      parsed = described_class.parse_search_result(result)
+
+      expect(parsed).to be_kind_of(::Gitlab::SearchResults::FoundBlob)
+      expect(parsed).to have_attributes(
+        id: nil,
+        filename: 'path/file.ext',
+        basename: 'path/file',
+        ref: 'sha',
+        startline: 2,
+        data: "bar\n"
+      )
+    end
+  end
+
   describe 'issues' do
     before do
       @issue_1 = create(

spec/models/concerns/elastic/repository_spec.rb

@@ -11,21 +11,53 @@ describe Repository, elastic: true do
     stub_application_setting(elasticsearch_search: false, elasticsearch_indexing: false)
   end
 
-  it "searches blobs and commits" do
-    project = create :project, :repository
-
+  def index!(project)
     Sidekiq::Testing.inline! do
       project.repository.index_blobs
       project.repository.index_commits
 
       Gitlab::Elastic::Helper.refresh_index
     end
+  end
+
+  it "searches blobs and commits" do
+    project = create :project, :repository
+    index!(project)
 
     expect(project.repository.search('def popen')[:blobs][:total_count]).to eq(1)
     expect(project.repository.search('def | popen')[:blobs][:total_count] > 1).to be_truthy
     expect(project.repository.search('initial')[:commits][:total_count]).to eq(1)
   end
 
+  def search_and_check!(on, query, type:, per: 1000)
+    results = on.search(query, type: type, per: per)["#{type}s".to_sym][:results]
+
+    blobs, commits = results.partition { |result| result['_source']['blob'].present? }
+
+    case type
+    when :blob
+      expect(blobs).not_to be_empty
+      expect(commits).to be_empty
+    when :commit
+      expect(blobs).to be_empty
+      expect(commits).not_to be_empty
+    else
+      raise ArgumentError
+    end
+  end
+
+  # A negation query can match both commits and blobs as they both have _type
+  # 'repository'. Ensure this doesn't happen, in both global and project search
+  it 'filters commits from blobs, and vice-versa' do
+    project = create :project, :repository
+    index!(project)
+
+    search_and_check!(Repository, '-foo', type: :blob)
+    search_and_check!(Repository, '-foo', type: :commit)
+    search_and_check!(project.repository, '-foo', type: :blob)
+    search_and_check!(project.repository, '-foo', type: :commit)
+  end
+
   describe "class method find_commits_by_message_with_elastic" do
     it "returns commits" do
       project = create :project, :repository