[search] Factor out approximate string matching and stop using it. There will be no spelling correction, but search works much faster.

indexer/search_index_builder.cpp

@@ -1,7 +1,7 @@
 #include "search_index_builder.hpp"
 #include "features_vector.hpp"
 #include "../search/search_trie.hpp"
-#include "../search/string_match.hpp"
+#include "../search/string_search_utils.hpp"
 #include "../coding/trie_builder.hpp"
 #include "../coding/writer.hpp"
 #include "../base/string_utils.hpp"

search/approximate_string_match.cpp

@@ -1,4 +1,4 @@
-#include "string_match.hpp"
+#include "approximate_string_match.hpp"
 
 // TODO: Сделать модель ошибок.
 // Учитывать соседние кнопки на клавиатуре.

search/approximate_string_match.hpp

@@ -1,28 +1,12 @@
 #pragma once
+#include "string_search_utils.hpp"
 #include "../base/base.hpp"
 #include "../base/buffer_vector.hpp"
-#include "../base/string_utils.hpp"
 #include "../std/queue.hpp"
 
 namespace search
 {
 
-using strings::UniChar;
-
-inline strings::UniString NormalizeAndSimplifyString(string const & s)
-{
-  strings::UniString uniS = strings::MakeLowerCase(strings::MakeUniString(s));
-  strings::Normalize(uniS);
-  return uniS;
-}
-
-template <class DelimsT, typename F>
-void SplitUniString(strings::UniString const & uniS, F f, DelimsT const & delims)
-{
-  for (strings::TokenizeIterator<DelimsT> iter(uniS, delims); iter; ++iter)
-    f(iter.GetUniString());
-}
-
 namespace impl
 {
 

search/keyword_matcher.cpp

@@ -1,6 +1,6 @@
 #include "keyword_matcher.hpp"
 #include "delimiters.hpp"
-#include "string_match.hpp"
+#include "string_search_utils.hpp"
 #include "../base/logging.hpp"
 #include "../base/string_utils.hpp"
 #include "../std/bind.hpp"

search/query.cpp

@@ -2,7 +2,7 @@
 #include "categories_holder.hpp"
 #include "delimiters.hpp"
 #include "latlon_match.hpp"
-#include "string_match.hpp"
+#include "string_search_utils.hpp"
 #include "search_trie_matching.hpp"
 #include "../indexer/feature_visibility.hpp"
 #include "../base/exception.hpp"
@@ -21,30 +21,26 @@ uint32_t KeywordMatch(strings::UniChar const * sA, uint32_t sizeA,
                       strings::UniChar const * sB, uint32_t sizeB,
                       uint32_t maxCost)
 {
-  /*
   if (sizeA != sizeB)
     return maxCost + 1;
-  for (uint32_t i = 0; i< sizeA; ++i)
-    if (sA[i] != sB[i])
+  strings::UniChar const * const endA = sA + sizeA;
+  while (sA != endA)
+    if (*sA++ != *sB++)
       return maxCost + 1;
   return 0;
-  */
-  return StringMatchCost(sA, sizeA, sB, sizeB, DefaultMatchCost(), maxCost, false);
 }
 
 uint32_t PrefixMatch(strings::UniChar const * sA, uint32_t sizeA,
                      strings::UniChar const * sB, uint32_t sizeB,
                      uint32_t maxCost)
 {
-  /*
   if (sizeA > sizeB)
     return maxCost + 1;
-  for (uint32_t i = 0; i< sizeA; ++i)
-    if (sA[i] != sB[i])
+  strings::UniChar const * const endA = sA + sizeA;
+  while (sA != endA)
+    if (*sA++ != *sB++)
       return maxCost + 1;
   return 0;
-  */
-  return StringMatchCost(sA, sizeA, sB, sizeB, DefaultMatchCost(), maxCost, true);
 }
 
 inline uint32_t GetMaxKeywordMatchScore() { return 512; }
@@ -243,7 +239,7 @@ void Query::Search(function<void (Result const &)> const & f)
     {
       FeatureProcessor featureProcessor(*this);
       /// @todo Tune depth scale search (1 is no enough)
-      m_pIndex->ForEachInRect(featureProcessor, m_viewport, min(scales::GetUpperScale(), scale + 1));
+      m_pIndex->ForEachInRect(featureProcessor, m_viewport, min(scales::GetUpperScale(), scale + 7));
     }
     catch (FeatureProcessor::StopException &)
     {

search/search.pro

@@ -16,11 +16,12 @@ HEADERS += \
     keyword_matcher.hpp \
     query.hpp \
     result.hpp \
-    string_match.hpp \
     latlon_match.hpp \
     categories_holder.hpp \
     search_trie.hpp \
     search_trie_matching.hpp \
+    string_search_utils.hpp \
+    approximate_string_match.hpp \
 
 SOURCES += \
     delimiters.cpp \
@@ -29,7 +30,7 @@ SOURCES += \
     keyword_matcher.cpp \
     query.cpp \
     result.cpp \
-    string_match.cpp \
     latlon_match.cpp \
     categories_holder.cpp \
     search_trie_matching.cpp \
+    approximate_string_match.cpp \

search/search_tests/keyword_matcher_test.cpp

@@ -1,7 +1,8 @@
 #include "../../testing/testing.hpp"
 #include "../keyword_matcher.hpp"
 #include "match_cost_mock.hpp"
-#include "../string_match.hpp"
+#include "../approximate_string_match.hpp"
+#include "../string_search_utils.hpp"
 #include "../../testing/testing_utils.hpp"
 #include "../../base/string_utils.hpp"
 #include "../../std/scoped_ptr.hpp"

search/search_trie_matching.cpp

@@ -1,6 +1,6 @@
 #include "search_trie_matching.hpp"
 #include "query.hpp"
-#include "string_match.hpp"
+#include "string_search_utils.hpp"
 
 #include "../indexer/feature_visibility.hpp"
 

search/string_search_utils.hpp

@@ -0,0 +1,24 @@
+#pragma once
+#include "../base/string_utils.hpp"
+#include "../base/base.hpp"
+
+namespace search
+{
+
+using strings::UniChar;
+
+inline strings::UniString NormalizeAndSimplifyString(string const & s)
+{
+  strings::UniString uniS = strings::MakeLowerCase(strings::MakeUniString(s));
+  strings::Normalize(uniS);
+  return uniS;
+}
+
+template <class DelimsT, typename F>
+void SplitUniString(strings::UniString const & uniS, F f, DelimsT const & delims)
+{
+  for (strings::TokenizeIterator<DelimsT> iter(uniS, delims); iter; ++iter)
+    f(iter.GetUniString());
+}
+
+}  // namespace search