[search] Using SmallSet for Locales.

indexer/categories_holder.cpp

@@ -141,8 +141,9 @@ void ProcessCategory(string const & line, vector<string> & groups, vector<uint32
 }  // namespace
 
 // static
-int8_t const CategoriesHolder::kEnglishCode = 1;
-int8_t const CategoriesHolder::kUnsupportedLocaleCode = -1;
+int8_t constexpr CategoriesHolder::kEnglishCode;
+int8_t constexpr CategoriesHolder::kUnsupportedLocaleCode;
+uint8_t constexpr CategoriesHolder::kMaxSupportedLocaleIndex;
 
 // *NOTE* These constants should be updated when adding new
 // translation to categories.txt. When editing, keep in mind to check
@@ -186,6 +187,11 @@ CategoriesHolder::CategoriesHolder(unique_ptr<Reader> && reader)
   ReaderStreamBuf buffer(move(reader));
   istream s(&buffer);
   LoadFromStream(s);
+
+#if defined(DEBUG)
+  for (auto const & entry : kLocaleMapping)
+    ASSERT_LESS_OR_EQUAL(entry.m_code, kMaxSupportedLocaleIndex, ());
+#endif
 }
 
 void CategoriesHolder::AddCategory(Category & cat, vector<uint32_t> & types)

indexer/categories_holder.hpp

@@ -63,8 +63,9 @@ private:
   GroupTranslations m_groupTranslations;
 
 public:
-  static int8_t const kEnglishCode;
-  static int8_t const kUnsupportedLocaleCode;
+  static int8_t constexpr kEnglishCode = 1;
+  static int8_t constexpr kUnsupportedLocaleCode = -1;
+  static uint8_t constexpr kMaxSupportedLocaleIndex = 30;
   static vector<Mapping> const kLocaleMapping;
 
   // List of languages that are currently disabled in the application

search/processor.cpp

@@ -397,12 +397,11 @@ Locales Processor::GetCategoryLocales() const
   Locales result;
 
   // Prepare array of processing locales. English locale is always present for category matching.
+  result.Insert(static_cast<uint64_t>(enLocaleCode));
   if (m_currentLocaleCode != -1)
-    result.push_back(m_currentLocaleCode);
-  if (m_inputLocaleCode != -1 && m_inputLocaleCode != m_currentLocaleCode)
-    result.push_back(m_inputLocaleCode);
-  if (enLocaleCode != m_currentLocaleCode && enLocaleCode != m_inputLocaleCode)
-    result.push_back(enLocaleCode);
+    result.Insert(static_cast<uint64_t>(m_currentLocaleCode));
+  if (m_inputLocaleCode != -1)
+    result.Insert(static_cast<uint64_t>(m_inputLocaleCode));
 
   return result;
 }

search/processor.hpp

@@ -12,6 +12,7 @@
 #include "search/search_trie.hpp"
 #include "search/suggest.hpp"
 #include "search/token_slice.hpp"
+#include "search/utils.hpp"
 
 #include "indexer/ftypes_matcher.hpp"
 #include "indexer/index.hpp"
@@ -132,9 +133,8 @@ protected:
   using TMWMVector = vector<shared_ptr<MwmInfo>>;
   using TOffsetsVector = map<MwmSet::MwmId, vector<uint32_t>>;
   using TFHeader = feature::DataHeader;
-  using TLocales = buffer_vector<int8_t, 3>;
 
-  TLocales GetCategoryLocales() const;
+  Locales GetCategoryLocales() const;
 
   template <typename ToDo>
   void ForEachCategoryType(StringSliceBase const & slice, ToDo && toDo) const;

search/ranker.hpp

@@ -10,6 +10,7 @@
 #include "search/reverse_geocoder.hpp"
 #include "search/search_params.hpp"
 #include "search/suggest.hpp"
+#include "search/utils.hpp"
 
 #include "indexer/categories_holder.hpp"
 #include "indexer/feature_decl.hpp"
@@ -43,8 +44,6 @@ class Ranker
 public:
   struct Params
   {
-    using TLocales = buffer_vector<int8_t, 3>;
-
     int8_t m_currentLocaleCode = CategoriesHolder::kEnglishCode;
     m2::RectD m_viewport;
     m2::PointD m_position;
@@ -65,7 +64,7 @@ public:
     // filtering of indentical search results.
     double m_minDistanceOnMapBetweenResults = 0.0;
 
-    TLocales m_categoryLocales;
+    Locales m_categoryLocales;
 
     size_t m_limit = 0;
   };

search/utils.hpp

@@ -8,6 +8,7 @@
 
 #include "base/buffer_vector.hpp"
 #include "base/levenshtein_dfa.hpp"
+#include "base/small_set.hpp"
 #include "base/stl_helpers.hpp"
 #include "base/string_utils.hpp"
 
@@ -20,6 +21,8 @@
 
 namespace search
 {
+using Locales = base::SafeSmallSet<CategoriesHolder::kMaxSupportedLocaleIndex + 1>;
+
 // todo(@m, @y). Unite with the similar function in search/feature_offset_match.hpp.
 template <typename TrieIt, typename DFA, typename ToDo>
 bool MatchInTrie(TrieIt const & trieStartIt, DFA const & dfa, ToDo && toDo)
@@ -64,8 +67,6 @@ bool MatchInTrie(TrieIt const & trieStartIt, DFA const & dfa, ToDo && toDo)
   return found;
 }
 
-using Locales = buffer_vector<int8_t, 3>;
-
 size_t GetMaxErrorsForToken(strings::UniString const & token);
 
 strings::LevenshteinDFA BuildLevenshteinDFA(strings::UniString const & s);
@@ -103,8 +104,6 @@ void ForEachCategoryTypeFuzzy(StringSliceBase const & slice, Locales const & loc
 
   auto const & trie = categories.GetNameToTypesTrie();
   auto const & trieRootIt = trie.GetRootIterator();
-  vector<int8_t> sortedLocales(locales.begin(), locales.end());
-  my::SortUnique(sortedLocales);
 
   for (size_t i = 0; i < slice.Size(); ++i)
   {
@@ -115,7 +114,7 @@ void ForEachCategoryTypeFuzzy(StringSliceBase const & slice, Locales const & loc
     strings::LevenshteinDFA const dfa(BuildLevenshteinDFA(token));
 
     trieRootIt.ForEachMove([&](Trie::Char const & c, Trie::Iterator const & trieStartIt) {
-      if (std::binary_search(sortedLocales.begin(), sortedLocales.end(), static_cast<int8_t>(c)))
+      if (locales.Contains(static_cast<uint64_t>(c)))
         MatchInTrie(trieStartIt, dfa, std::bind<void>(todo, i, std::placeholders::_1));
     });
   }
@@ -136,7 +135,7 @@ bool IsCategorialRequest(QuerySliceOnRawStrings<T> const & slice, Locales const
   bool found = false;
   auto token = slice.Get(0);
   catHolder.ForEachName([&](CategoriesHolder::Category::Name const & categorySynonym) {
-    if (std::find(locales.begin(), locales.end(), categorySynonym.m_locale) == locales.end())
+    if (!locales.Contains(static_cast<uint64_t>(categorySynonym.m_locale)))
       return;
 
     if (token != strings::MakeUniString(categorySynonym.m_name))