[search] Fixed streets matching.

3 files changed, 75 insertions, 35 deletions

diff --git a/search/geocoder.cpp b/search/geocoder.cpp
index 1f0ac2b8fa..1cafb50e73 100644
--- a/search/geocoder.cpp
+++ b/search/geocoder.cpp
@@ -466,6 +466,8 @@ void Geocoder::SetParams(Params const & params)
     }
   }
 
+  LOG(LDEBUG, ("Tokens = ", m_params.m_tokens));
+  LOG(LDEBUG, ("Prefix tokens = ", m_params.m_prefixTokens));
   LOG(LDEBUG, ("Languages =", m_params.m_langs));
 }
 
@@ -1075,61 +1077,82 @@ void Geocoder::GreedilyMatchStreets()
       continue;
 
     // Here we try to match as many tokens as possible while
-    // intersection is a non-empty bit vector of streets.  All tokens
-    // that are synonyms to streets are ignored.  Moreover, each time
-    // a token that looks like a beginning of a house number is met,
-    // we try to use current intersection of tokens as a street layer
-    // and try to match buildings or pois.
-    unique_ptr<coding::CompressedBitVector> allFeatures;
+    // intersection is a non-empty bit vector of streets. Single
+    // tokens that are synonyms to streets are ignored.  Moreover,
+    // each time a token that looks like a beginning of a house number
+    // is met, we try to use current intersection of tokens as a
+    // street layer and try to match BUILDINGs or POIs.
+    CBVPtr allFeatures(m_streets, false /* isOwner */);
 
     size_t curToken = startToken;
 
     // This variable is used for prevention of duplicate calls to
     // CreateStreetsLayerAndMatchLowerLayers() with the same
     // arguments.
-    size_t lastStopToken = curToken;
+    size_t lastToken = startToken;
 
-    for (; curToken < m_numTokens && !m_usedTokens[curToken]; ++curToken)
-    {
-      auto const & token = m_params.GetTokens(curToken).front();
-      if (IsStreetSynonymPrefix(token))
-        continue;
+    bool emptyIntersection = true;
+    bool incomplete = false;
 
-      bool const isPrefix = curToken >= m_params.m_tokens.size();
-      if (house_numbers::LooksLikeHouseNumber(token, isPrefix))
+    auto createStreetsLayerAndMatchLowerLayers = [&]()
+    {
+      if (!allFeatures.IsEmpty() && !emptyIntersection && !incomplete && lastToken != curToken)
       {
-        CreateStreetsLayerAndMatchLowerLayers(startToken, curToken, allFeatures);
-        lastStopToken = curToken;
+        CreateStreetsLayerAndMatchLowerLayers(startToken, curToken, *allFeatures);
+        lastToken = curToken;
       }
+    };
 
-      unique_ptr<coding::CompressedBitVector> buffer;
-      if (startToken == curToken || coding::CompressedBitVector::IsEmpty(allFeatures))
-        buffer = coding::CompressedBitVector::Intersect(*m_streets, *m_addressFeatures[curToken]);
-      else
-        buffer = coding::CompressedBitVector::Intersect(*allFeatures, *m_addressFeatures[curToken]);
+    StreetTokensFilter filter([&](strings::UniString const & /* token */, size_t tag)
+                              {
+                                auto buffer = coding::CompressedBitVector::Intersect(
+                                    *allFeatures, *m_addressFeatures[tag]);
+                                if (tag < curToken)
+                                {
+                                  allFeatures.Set(move(buffer));
+                                  emptyIntersection = false;
+                                  incomplete = true;
+                                  return;
+                                }
+                                ASSERT_EQUAL(tag, curToken, ());
+
+                                // |allFeatures| will become empty
+                                // after the intersection. Therefore
+                                // we need to create streets layer
+                                // right now.
+                                if (coding::CompressedBitVector::IsEmpty(buffer))
+                                  createStreetsLayerAndMatchLowerLayers();
+
+                                allFeatures.Set(move(buffer));
+                                emptyIntersection = false;
+                                incomplete = false;
+                              });
+
+    for (; curToken < m_numTokens && !m_usedTokens[curToken] && !allFeatures.IsEmpty(); ++curToken)
+    {
+      auto const & token = m_params.GetTokens(curToken).front();
+      bool const isPrefix = curToken >= m_params.m_tokens.size();
 
-      if (coding::CompressedBitVector::IsEmpty(buffer))
-        break;
+      if (house_numbers::LooksLikeHouseNumber(token, isPrefix))
+        createStreetsLayerAndMatchLowerLayers();
 
-      allFeatures.swap(buffer);
+      filter.Put(token, isPrefix, curToken);
     }
-
-    if (curToken != lastStopToken)
-      CreateStreetsLayerAndMatchLowerLayers(startToken, curToken, allFeatures);
+    createStreetsLayerAndMatchLowerLayers();
   }
 }
 
 void Geocoder::CreateStreetsLayerAndMatchLowerLayers(
-    size_t startToken, size_t endToken, unique_ptr<coding::CompressedBitVector> const & features)
+    size_t startToken, size_t endToken, coding::CompressedBitVector const & features)
 {
   ASSERT(m_layers.empty(), ());
 
-  if (coding::CompressedBitVector::IsEmpty(features))
+  if (coding::CompressedBitVector::IsEmpty(&features))
     return;
 
-  CBVPtr filtered(features.get(), false /* isOwner */);
-  if (m_filter->NeedToFilter(*features))
-    filtered.Set(m_filter->Filter(*features).release(), true /* isOwner */);
+  CBVPtr filtered(&features, false /* isOwner */);
+  if (m_filter->NeedToFilter(features))
+    filtered.Set(m_filter->Filter(features).release(), true /* isOwner */);
 
   m_layers.emplace_back();
   MY_SCOPE_GUARD(cleanupGuard, bind(&vector<FeaturesLayer>::pop_back, &m_layers));
@@ -1138,7 +1161,7 @@ void Geocoder::CreateStreetsLayerAndMatchLowerLayers(
   InitLayer(SearchModel::SEARCH_TYPE_STREET, startToken, endToken, layer);
 
   vector<uint32_t> sortedFeatures;
-  sortedFeatures.reserve(features->PopCount());
+  sortedFeatures.reserve(features.PopCount());
   filtered.ForEach(MakeBackInsertFunctor(sortedFeatures));
   layer.m_sortedFeatures = &sortedFeatures;
 
diff --git a/search/geocoder.hpp b/search/geocoder.hpp
index 439784a742..db8e905b98 100644
--- a/search/geocoder.hpp
+++ b/search/geocoder.hpp
@@ -245,8 +245,8 @@ private:
   // then performs geocoding in street vicinities.
   void GreedilyMatchStreets();
 
-  void CreateStreetsLayerAndMatchLowerLayers(
-      size_t startToken, size_t endToken, unique_ptr<coding::CompressedBitVector> const & features);
+  void CreateStreetsLayerAndMatchLowerLayers(size_t startToken, size_t endToken,
+                                             coding::CompressedBitVector const & features);
 
   // Tries to find all paths in a search tree, where each edge is
   // marked with some substring of the query tokens. These paths are
diff --git a/search/search_integration_tests/processor_test.cpp b/search/search_integration_tests/processor_test.cpp
index 89c9032634..419ed0025b 100644
--- a/search/search_integration_tests/processor_test.cpp
+++ b/search/search_integration_tests/processor_test.cpp
@@ -59,6 +59,8 @@ UNIT_CLASS_TEST(ProcessorTest, Smoke)
 
   TestCity losAlamosCity(m2::PointD(10, 10), "Los Alamos", "en", 100 /* rank */);
   TestCity mskCity(m2::PointD(0, 0), "Moscow", "en", 100 /* rank */);
+  TestCity torontoCity(m2::PointD(-10, -10), "Toronto", "en", 100 /* rank */);
+
   TestVillage longPondVillage(m2::PointD(15, 15), "Long Pond Village", "en", 10 /* rank */);
   TestStreet feynmanStreet(
       vector<m2::PointD>{m2::PointD(9.999, 9.999), m2::PointD(10, 10), m2::PointD(10.001, 10.001)},
@@ -94,16 +96,23 @@ UNIT_CLASS_TEST(ProcessorTest, Smoke)
   TestPOI lantern1(m2::PointD(10.0005, 10.0005), "lantern 1", "en");
   TestPOI lantern2(m2::PointD(10.0006, 10.0005), "lantern 2", "en");
 
+  TestStreet stradaDrive(vector<m2::PointD>{m2::PointD(-10.001, -10.001), m2::PointD(-10, -10),
+                                            m2::PointD(-9.999, -9.999)},
+                         "Strada drive", "en");
+  TestBuilding terranceHouse(m2::PointD(-10, -10), "", "155", stradaDrive, "en");
+
   BuildWorld([&](TestMwmBuilder & builder)
              {
                builder.Add(wonderlandCountry);
                builder.Add(losAlamosCity);
                builder.Add(mskCity);
+               builder.Add(torontoCity);
              });
   auto wonderlandId = BuildCountry(countryName, [&](TestMwmBuilder & builder)
                                    {
                                      builder.Add(losAlamosCity);
                                      builder.Add(mskCity);
+                                     builder.Add(torontoCity);
                                      builder.Add(longPondVillage);
 
                                      builder.Add(feynmanStreet);
@@ -125,6 +134,9 @@ UNIT_CLASS_TEST(ProcessorTest, Smoke)
                                      builder.Add(quantumCafe);
                                      builder.Add(lantern1);
                                      builder.Add(lantern2);
+
+                                     builder.Add(stradaDrive);
+                                     builder.Add(terranceHouse);
                                    });
 
   SetViewport(m2::RectD(m2::PointD(-1.0, -1.0), m2::PointD(1.0, 1.0)));
@@ -190,6 +202,11 @@ UNIT_CLASS_TEST(ProcessorTest, Smoke)
     TRules rules = {ExactMatch(wonderlandId, bornHouse)};
     TEST(ResultsMatch("long pond 1st april street 8", rules), ());
   }
+
+  {
+    TRules rules = {ExactMatch(wonderlandId, terranceHouse)};
+    TEST(ResultsMatch("Toronto strada drive 155", rules), ());
+  }
 }
 
 UNIT_CLASS_TEST(ProcessorTest, SearchInWorld)