[indexer] United all tries-by-locale into one trie.

indexer/categories_holder.cpp

@@ -203,6 +203,8 @@ void CategoriesHolder::AddCategory(Category & cat, vector<uint32_t> & types)
       auto const locale = synonym.m_locale;
       ASSERT_NOT_EQUAL(locale, kUnsupportedLocaleCode, ());
 
+      auto const localePrefix = String(1, static_cast<strings::UniChar>(locale));
+
       auto const uniName = search::NormalizeAndSimplifyString(synonym.m_name);
 
       vector<String> tokens;
@@ -213,10 +215,7 @@ void CategoriesHolder::AddCategory(Category & cat, vector<uint32_t> & types)
         if (!ValidKeyToken(token))
           continue;
         for (uint32_t const t : types)
-        {
-          auto it = m_name2type.emplace(locale, make_unique<Trie>()).first;
-          it->second->Add(token, t);
-        }
+          m_name2type->Add(localePrefix + token, t);
       }
     }
   }
@@ -243,7 +242,7 @@ bool CategoriesHolder::ValidKeyToken(String const & s)
 void CategoriesHolder::LoadFromStream(istream & s)
 {
   m_type2cat.clear();
-  m_name2type.clear();
+  m_name2type = make_unique<Trie>();
   m_groupTranslations.clear();
 
   State state = EParseTypes;

indexer/categories_holder.hpp

@@ -56,7 +56,8 @@ private:
   Type2CategoryCont m_type2cat;
 
   // Maps locale and category token to the list of corresponding types.
-  map<int8_t, unique_ptr<Trie>> m_name2type;
+  // Locale is treated as a special symbol prepended to the token.
+  unique_ptr<Trie> m_name2type;
 
   GroupTranslations m_groupTranslations;
 
@@ -109,10 +110,9 @@ public:
   template <class ToDo>
   void ForEachTypeByName(int8_t locale, String const & name, ToDo && toDo) const
   {
-    auto const it = m_name2type.find(locale);
-    if (it == m_name2type.end())
-      return;
-    it->second->ForEachInNode(name, my::MakeIgnoreFirstArgument(forward<ToDo>(toDo)));
+    auto const localePrefix = String(1, static_cast<strings::UniChar>(locale));
+    m_name2type->ForEachInNode(localePrefix + name,
+                               my::MakeIgnoreFirstArgument(forward<ToDo>(toDo)));
   }
 
   inline GroupTranslations const & GetGroupTranslations() const { return m_groupTranslations; }
@@ -126,14 +126,7 @@ public:
   string GetReadableFeatureType(uint32_t type, int8_t locale) const;
 
   // Exposes the tries that map category tokens to types.
-  Trie const * GetNameToTypesTrie(int8_t locale) const
-  {
-    auto const it = m_name2type.find(locale);
-    if (it == m_name2type.end())
-      return nullptr;
-    return it->second.get();
-  }
-
+  Trie const & GetNameToTypesTrie() const { return *m_name2type; }
   bool IsTypeExist(uint32_t type) const;
 
   inline void Swap(CategoriesHolder & r)

search/utils.hpp

@@ -15,13 +15,14 @@
 #include <cctype>
 #include <functional>
 #include <queue>
+#include <set>
 
 namespace search
 {
-// my::MemTrie<strings::UniString, uint32_t>
 // todo(@m, @y). Unite with the similar function in search/feature_offset_match.hpp.
 template <typename Trie, typename DFA, typename ToDo>
-bool MatchInTrie(Trie const & trie, DFA const & dfa, ToDo && toDo)
+bool MatchInTrie(Trie const & /* trie */, typename Trie::Iterator const & trieStartIt,
+                 DFA const & dfa, ToDo && toDo)
 {
   using Char = typename Trie::Char;
   using TrieIt = typename Trie::Iterator;
@@ -34,7 +35,7 @@ bool MatchInTrie(Trie const & trie, DFA const & dfa, ToDo && toDo)
     auto it = dfa.Begin();
     if (it.Rejects())
       return false;
-    q.emplace(trie.GetRootIterator(), it);
+    q.emplace(trieStartIt, it);
   }
 
   bool found = false;
@@ -97,17 +98,25 @@ template <typename ToDo>
 void ForEachCategoryTypeFuzzy(StringSliceBase const & slice, TLocales const & locales,
                               CategoriesHolder const & categories, ToDo && todo)
 {
+  using Trie = my::MemTrie<strings::UniString, uint32_t>;
+
+  auto const & trie = categories.GetNameToTypesTrie();
+  auto const & trieRootIt = trie.GetRootIterator();
+  std::set<int8_t> localeSet(locales.begin(), locales.end());
+
   for (size_t i = 0; i < slice.Size(); ++i)
   {
     auto const & token = slice.Get(i);
     auto const & dfa =
         strings::LevenshteinDFA(token, 1 /* prefixCharsToKeep */, GetMaxErrorsForToken(token));
-    for (int8_t const locale : locales)
-    {
-      auto const * trie = categories.GetNameToTypesTrie(locale);
-      if (trie != nullptr)
-        MatchInTrie(*trie, dfa, std::bind<void>(todo, i, std::placeholders::_1));
-    }
+
+    trieRootIt.ForEachMove([&](Trie::Char const & c, Trie::Iterator const & moveIt) {
+      if (localeSet.count(static_cast<int8_t>(c)) != 0)
+      {
+        MatchInTrie(trie /* passed to infer the iterator's type */, moveIt, dfa,
+                    std::bind<void>(todo, i, std::placeholders::_1));
+      }
+    });
   }
 }
 }  // namespace search