[search] Add only categorized features to search index. Do not add some categories of features without names to search index.

5 files changed, 69 insertions, 48 deletions

diff --git a/indexer/categories_holder.hpp b/indexer/categories_holder.hpp
index 552b60a3f4..71f96452dc 100644
--- a/indexer/categories_holder.hpp
+++ b/indexer/categories_holder.hpp
@@ -73,8 +73,18 @@ public:
     }
   }
 
+  /// Search name for type with preffered language.
+  /// If no name for this language, return first (en) name.
+  /// @return false if no categories for type.
   bool GetNameByType(uint32_t type, int8_t lang, string & name) const;
 
+  inline bool IsTypeExist(uint32_t type) const
+  {
+    // pass any language
+    string dummy;
+    return GetNameByType(type, 0, dummy);
+  }
+
   inline void Swap(CategoriesHolder & r)
   {
     m_type2cat.swap(r.m_type2cat);
diff --git a/indexer/feature_loader_base.hpp b/indexer/feature_loader_base.hpp
index 55fccbd396..962e8d71be 100644
--- a/indexer/feature_loader_base.hpp
+++ b/indexer/feature_loader_base.hpp
@@ -41,7 +41,6 @@ namespace feature
     inline int GetScalesCount() const { return static_cast<int>(m_header.GetScalesCount()); }
     inline int GetScale(int i) const { return m_header.GetScale(i); }
     inline int GetLastScale() const { return m_header.GetLastScale(); }
-    inline pair<int, int> GetScaleRange() const { return m_header.GetScaleRange(); }
   };
 
   class LoaderBase
diff --git a/indexer/features_vector.hpp b/indexer/features_vector.hpp
index 92766a1e8a..c66bbd1bee 100644
--- a/indexer/features_vector.hpp
+++ b/indexer/features_vector.hpp
@@ -28,12 +28,6 @@ public:
     m_RecordReader.ForEachRecord(DoGetFeatures<ToDo>(m_LoadInfo, toDo));
   }
 
-  inline serial::CodingParams const & GetCodingParams() const
-  {
-    return m_LoadInfo.GetDefCodingParams();
-  }
-  inline pair<int, int> GetScaleRange() const { return m_LoadInfo.GetScaleRange(); }
-
 private:
   template <class ToDo> class DoGetFeatures
   {
diff --git a/indexer/search_index_builder.cpp b/indexer/search_index_builder.cpp
index f1bfccc649..43dad95ec6 100644
--- a/indexer/search_index_builder.cpp
+++ b/indexer/search_index_builder.cpp
@@ -8,6 +8,7 @@
 #include "string_file.hpp"
 #include "classificator.hpp"
 #include "feature_visibility.hpp"
+#include "categories_holder.hpp"
 
 #include "../defines.hpp"
 
@@ -64,6 +65,8 @@ class FeatureInserter
 {
   StringsFile & m_names;
 
+  CategoriesHolder const & m_categories;
+
   typedef StringsFile::ValueT ValueT;
   typedef search::trie::ValueReader SaverT;
   SaverT m_valueSaver;
@@ -185,42 +188,52 @@ class FeatureInserter
 
   class AvoidEmptyName
   {
-    vector<uint32_t> m_vec;
+    vector<uint32_t> m_match[2];
 
-  public:
-    AvoidEmptyName()
+    template <size_t count, size_t ind>
+    void FillMatch(char const * (& arr)[count][ind])
     {
-      char const * arr[][3] = {
-        { "place", "city", ""},
-        { "place", "city", "capital"},
-        { "place", "town", ""},
-        { "place", "county", ""},
-        { "place", "state", ""},
-        { "place", "region", ""}
-      };
+      STATIC_ASSERT ( count > 0 );
+      STATIC_ASSERT ( ind > 0 );
 
       Classificator const & c = classif();
 
-      size_t const count = ARRAY_SIZE(arr);
-      m_vec.reserve(count);
-
+      m_match[ind-1].reserve(count);
       for (size_t i = 0; i < count; ++i)
       {
-        vector<string> v;
-        v.push_back(arr[i][0]);
-        v.push_back(arr[i][1]);
-        if (strlen(arr[i][2]) > 0)
-          v.push_back(arr[i][2]);
-
-        m_vec.push_back(c.GetTypeByPath(v));
+        vector<string> v(arr[i], arr[i] + ind);
+        m_match[ind-1].push_back(c.GetTypeByPath(v));
       }
     }
 
+  public:
+    AvoidEmptyName()
+    {
+      char const * arr1[][1] = { { "highway" }, { "natural" }, { "waterway"} };
+
+      char const * arr2[][2] = {
+        { "place", "city" },
+        { "place", "town" },
+        { "place", "county" },
+        { "place", "state" },
+        { "place", "region" }
+      };
+
+      FillMatch(arr1);
+      FillMatch(arr2);
+    }
+
     bool IsExist(feature::TypesHolder const & types) const
     {
-      for (size_t i = 0; i < m_vec.size(); ++i)
-        if (types.Has(m_vec[i]))
-          return true;
+      for (size_t i = 0; i < types.Size(); ++i)
+        for (size_t j = 0; j < ARRAY_SIZE(m_match); ++j)
+        {
+          uint32_t type = types[i];
+          ftype::TruncValue(type, j+1);
+
+          if (find(m_match[j].begin(), m_match[j].end(), type) != m_match[j].end())
+            return true;
+        }
 
       return false;
     }
@@ -228,9 +241,10 @@ class FeatureInserter
 
 public:
   FeatureInserter(StringsFile & names,
+                  CategoriesHolder const & catHolder,
                   serial::CodingParams const & cp,
                   pair<int, int> const & scales)
-    : m_names(names), m_valueSaver(cp), m_scales(scales)
+    : m_names(names), m_categories(catHolder), m_valueSaver(cp), m_scales(scales)
   {
   }
 
@@ -262,6 +276,10 @@ public:
       // highway-primary-oneway or amenity-parking-fee.
       ftype::TruncValue(type, 2);
 
+      // Push to index only categorized types.
+      if (!m_categories.IsTypeExist(type))
+        continue;
+
       // Do index only for visible types in mwm.
       pair<int, int> const r = feature::DrawableScaleRangeForType(type);
       if (my::between_s(m_scales.first, m_scales.second, r.first) ||
@@ -273,16 +291,19 @@ public:
   }
 };
 
-}  // unnamed namespace
-
-void indexer::BuildSearchIndex(FeaturesVector const & featuresV, Writer & writer,
-                               string const & tmpFilePath)
+void BuildSearchIndex(FilesContainerR const & cont, CategoriesHolder const & catHolder,
+                      Writer & writer, string const & tmpFilePath)
 {
   {
+    feature::DataHeader header;
+    header.Load(cont.GetReader(HEADER_FILE_TAG));
+    FeaturesVector featuresV(cont, header);
+
+    serial::CodingParams cp(search::GetCPForTrie(header.GetDefCodingParams()));
+
     StringsFile names(tmpFilePath);
-    serial::CodingParams cp(search::GetCPForTrie(featuresV.GetCodingParams()));
 
-    featuresV.ForEachOffset(FeatureInserter(names, cp, featuresV.GetScaleRange()));
+    featuresV.ForEachOffset(FeatureInserter(names, catHolder, cp, header.GetScaleRange()));
 
     names.EndAdding();
     names.OpenForRead();
@@ -295,6 +316,8 @@ void indexer::BuildSearchIndex(FeaturesVector const & featuresV, Writer & writer
   FileWriter::DeleteFileX(tmpFilePath);
 }
 
+}
+
 bool indexer::BuildSearchIndexFromDatFile(string const & fName, bool forceRebuild)
 {
   LOG(LINFO, ("Start building search index. Bits = ", search::POINT_CODING_BITS));
@@ -311,13 +334,12 @@ bool indexer::BuildSearchIndexFromDatFile(string const & fName, bool forceRebuil
       if (!forceRebuild && readCont.IsReaderExist(SEARCH_INDEX_FILE_TAG))
         return true;
 
-      feature::DataHeader header;
-      header.Load(readCont.GetReader(HEADER_FILE_TAG));
+      FileWriter writer(tmpFile);
 
-      FeaturesVector featuresVector(readCont, header);
+      CategoriesHolder catHolder(pl.GetReader(SEARCH_CATEGORIES_FILE_NAME));
 
-      FileWriter writer(tmpFile);
-      BuildSearchIndex(featuresVector, writer, pl.WritablePathForFile(fName + ".search_index_1.tmp"));
+      BuildSearchIndex(readCont, catHolder, writer,
+                       pl.WritablePathForFile(fName + ".search_index_1.tmp"));
 
       LOG(LINFO, ("Search index size = ", writer.Size()));
     }
diff --git a/indexer/search_index_builder.hpp b/indexer/search_index_builder.hpp
index baaed3d6c8..41ea88b8cb 100644
--- a/indexer/search_index_builder.hpp
+++ b/indexer/search_index_builder.hpp
@@ -1,12 +1,8 @@
 #pragma once
 #include "../std/string.hpp"
 
-class FeaturesVector;
-class Writer;
 
 namespace indexer
 {
-  void BuildSearchIndex(FeaturesVector const & featuresVector, Writer & writer,
-                        string const & tmpFilePath);
   bool BuildSearchIndexFromDatFile(string const & fName, bool forceRebuild = false);
-}  // namespace indexer
+}