Improve search filter by taking space in file path into account

changelogs/unreleased/281677_space_in_filename.yml

+---
+title: Improve search filter by taking space in file path into account
+merge_request: 52392
+author:
+type: fixed

lib/gitlab/search/query.rb

@@ -5,6 +5,9 @@ module Gitlab
     class Query < SimpleDelegator
       include EncodingHelper
 
+      QUOTES_REGEXP = %r{\A"|"\Z}.freeze
+      TOKEN_WITH_QUOTES_REGEXP = %r{\s(?=(?:[^"]|"[^"]*")*$)}.freeze
+
       def initialize(query, filter_opts = {}, &block)
         @raw_query = query.dup
         @filters = []
@@ -35,22 +38,24 @@ module Gitlab
       def extract_filters
         fragments = []
 
+        query_tokens = parse_raw_query
         filters = @filters.each_with_object([]) do |filter, parsed_filters|
-          match = @raw_query.split.find { |part| part =~ /\A-?#{filter[:name]}:/ }
+          match = query_tokens.find { |part| part =~ /\A-?#{filter[:name]}:/ }
+
           next unless match
 
           input = match.split(':')[1..-1].join
           next if input.empty?
 
           filter[:negated] = match.start_with?("-")
-          filter[:value] = parse_filter(filter, input)
+          filter[:value] = parse_filter(filter, input.gsub(QUOTES_REGEXP, ''))
           filter[:regex_value] = Regexp.escape(filter[:value]).gsub('\*', '.*?')
           fragments << match
 
           parsed_filters << filter
         end
 
-        query = (@raw_query.split - fragments).join(' ')
+        query = (query_tokens - fragments).join(' ')
         query = '*' if query.empty?
 
         [query, filters]
@@ -61,6 +66,13 @@ module Gitlab
 
         @filter_options[:encode_binary] ? encode_binary(result) : result
       end
+
+      def parse_raw_query
+        # Positive lookahead for any non-quote char or even number of quotes
+        # for example '"search term" path:"foo bar.txt"' would break into
+        # ["search term", "path:\"foo bar.txt\""]
+        @raw_query.split(TOKEN_WITH_QUOTES_REGEXP).reject(&:empty?)
+      end
     end
   end
 end

spec/lib/gitlab/file_finder_spec.rb

@@ -53,6 +53,14 @@ RSpec.describe Gitlab::FileFinder do
       end
     end
 
+    context 'with white space in the path' do
+      it 'filters by path correctly' do
+        results = subject.find('directory path:"with space/README.md"')
+
+        expect(results.count).to eq(1)
+      end
+    end
+
     it 'does not cause N+1 query' do
       expect(Gitlab::GitalyClient).to receive(:call).at_most(10).times.and_call_original
 

spec/lib/gitlab/search/query_spec.rb

@@ -46,4 +46,22 @@ RSpec.describe Gitlab::Search::Query do
       expect(subject.filters).to all(include(negated: true))
     end
   end
+
+  context 'with filter value in quotes' do
+    let(:query) { '"foo bar" name:"my test script.txt"' }
+
+    it 'does not break the filter value in quotes' do
+      expect(subject.term).to eq('"foo bar"')
+      expect(subject.filters[0]).to include(name: :name, negated: false, value: "MY TEST SCRIPT.TXT")
+    end
+  end
+
+  context 'with extra white spaces between the query words' do
+    let(:query) { ' foo = bar  name:"my test.txt"' }
+
+    it 'removes the extra whitespace between tokens' do
+      expect(subject.term).to eq('foo = bar')
+      expect(subject.filters[0]).to include(name: :name, negated: false, value: "MY TEST.TXT")
+    end
+  end
 end