Merge pull request #5629 from ygorshenin/extract-matching-partbeta-689beta-688

[search] Matching is extracted to search_quality lib.

5 files changed, 143 insertions, 92 deletions

diff --git a/search/search_quality/CMakeLists.txt b/search/search_quality/CMakeLists.txt
index e01dc02edf..4dae74599f 100644
--- a/search/search_quality/CMakeLists.txt
+++ b/search/search_quality/CMakeLists.txt
@@ -3,10 +3,12 @@ project(search_quality)
 include_directories(${OMIM_ROOT}/3party/jansson/src)
 
 set(SRC
-  helpers.hpp
   helpers.cpp
-  sample.hpp
+  helpers.hpp
+  matcher.cpp
+  matcher.hpp
   sample.cpp
+  sample.hpp
 )
 
 add_library(${PROJECT_NAME} ${SRC})
diff --git a/search/search_quality/features_collector_tool/features_collector_tool.cpp b/search/search_quality/features_collector_tool/features_collector_tool.cpp
index 270b26e860..f3fc477ce6 100644
--- a/search/search_quality/features_collector_tool/features_collector_tool.cpp
+++ b/search/search_quality/features_collector_tool/features_collector_tool.cpp
@@ -1,12 +1,12 @@
 #include "search/ranking_info.hpp"
 #include "search/result.hpp"
 #include "search/search_quality/helpers.hpp"
+#include "search/search_quality/matcher.hpp"
 #include "search/search_quality/sample.hpp"
 #include "search/search_tests_support/test_search_engine.hpp"
 #include "search/search_tests_support/test_search_request.hpp"
 
 #include "indexer/classificator_loader.hpp"
-#include "indexer/feature_algo.hpp"
 
 #include "storage/country_info_getter.hpp"
 #include "storage/index.hpp"
@@ -43,30 +43,12 @@ DEFINE_string(mwm_path, "", "Path to mwm files (writable dir)");
 DEFINE_string(stats_path, "", "Path to store stats about queries results (default: stderr)");
 DEFINE_string(json_in, "", "Path to the json file with samples (default: stdin)");
 
-size_t constexpr kInvalidId = numeric_limits<size_t>::max();
-
 struct Stats
 {
   // Indexes of not-found VITAL or RELEVANT results.
   vector<size_t> m_notFound;
 };
 
-struct Context
-{
-  Context(Index & index) : m_index(index) {}
-
-  WARN_UNUSED_RESULT bool GetFeature(FeatureID const & id, FeatureType & ft)
-  {
-    auto const & mwmId = id.m_mwmId;
-    if (!m_guard || m_guard->GetId() != mwmId)
-      m_guard = make_unique<Index::FeaturesLoaderGuard>(m_index, mwmId);
-    return m_guard->GetFeatureByIndex(id.m_index, ft);
-  }
-
-  Index & m_index;
-  unique_ptr<Index::FeaturesLoaderGuard> m_guard;
-};
-
 void GetContents(istream & is, string & contents)
 {
   string line;
@@ -77,73 +59,6 @@ void GetContents(istream & is, string & contents)
   }
 }
 
-bool Matches(Context & context, Sample::Result const & golden, search::Result const & actual)
-{
-  static double constexpr kToleranceMeters = 50;
-  if (actual.GetResultType() != Result::RESULT_FEATURE)
-    return false;
-
-  FeatureType ft;
-  if (!context.GetFeature(actual.GetFeatureID(), ft))
-    return false;
-
-  auto const houseNumber = ft.GetHouseNumber();
-  auto const center = feature::GetCenter(ft);
-
-  bool nameMatches = false;
-  if (golden.m_name.empty())
-  {
-    nameMatches = true;
-  }
-  else
-  {
-    ft.ForEachName([&golden, &nameMatches](int8_t /* lang */, string const & name) {
-      if (golden.m_name == strings::MakeUniString(name))
-      {
-        nameMatches = true;
-        return false;  // breaks the loop
-      }
-      return true;  // continues the loop
-    });
-  }
-
-  return nameMatches && golden.m_houseNumber == houseNumber &&
-         MercatorBounds::DistanceOnEarth(golden.m_pos, center) < kToleranceMeters;
-}
-
-void MatchResults(Context & context, vector<Sample::Result> const & golden,
-                  vector<search::Result> const & actual, vector<size_t> & goldenMatching,
-                  vector<size_t> & actualMatching)
-{
-  auto const n = golden.size();
-  auto const m = actual.size();
-
-  goldenMatching.assign(n, kInvalidId);
-  actualMatching.assign(m, kInvalidId);
-
-  // TODO (@y, @m): use Kuhn algorithm here for maximum matching.
-  for (size_t i = 0; i < n; ++i)
-  {
-    if (goldenMatching[i] != kInvalidId)
-      continue;
-    auto const & g = golden[i];
-
-    for (size_t j = 0; j < m; ++j)
-    {
-      if (actualMatching[j] != kInvalidId)
-        continue;
-
-      auto const & a = actual[j];
-      if (Matches(context, g, a))
-      {
-        goldenMatching[i] = j;
-        actualMatching[j] = i;
-        break;
-      }
-    }
-  }
-}
-
 void DidDownload(TCountryId const & /* countryId */,
                  shared_ptr<platform::LocalCountryFile> const & /* localFile */)
 {
@@ -247,7 +162,7 @@ int main(int argc, char * argv[])
   }
 
   vector<Stats> stats(samples.size());
-  Context context(engine);
+  Matcher matcher(engine);
 
   cout << "SampleId,";
   RankingInfo::PrintCSVHeader(cout);
@@ -274,7 +189,7 @@ int main(int argc, char * argv[])
 
     vector<size_t> goldenMatching;
     vector<size_t> actualMatching;
-    MatchResults(context, sample.m_results, results, goldenMatching, actualMatching);
+    matcher.Match(sample.m_results, results, goldenMatching, actualMatching);
 
     for (size_t j = 0; j < results.size(); ++j)
     {
@@ -285,7 +200,7 @@ int main(int argc, char * argv[])
       info.ToCSV(cout);
 
       auto relevance = Sample::Result::RELEVANCE_IRRELEVANT;
-      if (actualMatching[j] != kInvalidId)
+      if (actualMatching[j] != Matcher::kInvalidId)
         relevance = sample.m_results[actualMatching[j]].m_relevance;
       cout << "," << DebugPrint(relevance) << endl;
     }
@@ -293,7 +208,7 @@ int main(int argc, char * argv[])
     auto & s = stats[i];
     for (size_t j = 0; j < goldenMatching.size(); ++j)
     {
-      if (goldenMatching[j] == kInvalidId &&
+      if (goldenMatching[j] == Matcher::kInvalidId &&
           sample.m_results[j].m_relevance != Sample::Result::RELEVANCE_IRRELEVANT)
       {
         s.m_notFound.push_back(j);
diff --git a/search/search_quality/matcher.cpp b/search/search_quality/matcher.cpp
new file mode 100644
index 0000000000..5b73fbab14
--- /dev/null
+++ b/search/search_quality/matcher.cpp
@@ -0,0 +1,94 @@
+#include "search/search_quality/matcher.hpp"
+
+#include "indexer/feature.hpp"
+#include "indexer/feature_algo.hpp"
+#include "indexer/feature_decl.hpp"
+
+#include "base/string_utils.hpp"
+
+#include "geometry/mercator.hpp"
+
+#include "base/stl_add.hpp"
+
+namespace search
+{
+// static
+size_t constexpr Matcher::kInvalidId;
+
+Matcher::Matcher(Index & index) : m_index(index) {}
+
+void Matcher::Match(std::vector<Sample::Result> const & golden, std::vector<Result> const & actual,
+                    std::vector<size_t> & goldenMatching, std::vector<size_t> & actualMatching)
+{
+  auto const n = golden.size();
+  auto const m = actual.size();
+
+  goldenMatching.assign(n, kInvalidId);
+  actualMatching.assign(m, kInvalidId);
+
+  // TODO (@y, @m): use Kuhn algorithm here for maximum matching.
+  for (size_t i = 0; i < n; ++i)
+  {
+    if (goldenMatching[i] != kInvalidId)
+      continue;
+    auto const & g = golden[i];
+
+    for (size_t j = 0; j < m; ++j)
+    {
+      if (actualMatching[j] != kInvalidId)
+        continue;
+
+      auto const & a = actual[j];
+      if (Matches(g, a))
+      {
+        goldenMatching[i] = j;
+        actualMatching[j] = i;
+        break;
+      }
+    }
+  }
+}
+
+bool Matcher::GetFeature(FeatureID const & id, FeatureType & ft)
+{
+  auto const & mwmId = id.m_mwmId;
+  if (!m_guard || m_guard->GetId() != mwmId)
+    m_guard = my::make_unique<Index::FeaturesLoaderGuard>(m_index, mwmId);
+  return m_guard->GetFeatureByIndex(id.m_index, ft);
+}
+
+bool Matcher::Matches(Sample::Result const & golden, search::Result const & actual)
+{
+  static double constexpr kToleranceMeters = 50;
+
+  if (actual.GetResultType() != Result::RESULT_FEATURE)
+    return false;
+
+  FeatureType ft;
+  if (!GetFeature(actual.GetFeatureID(), ft))
+    return false;
+
+  auto const houseNumber = ft.GetHouseNumber();
+  auto const center = feature::GetCenter(ft);
+
+  bool nameMatches = false;
+  if (golden.m_name.empty())
+  {
+    nameMatches = true;
+  }
+  else
+  {
+    ft.ForEachName([&golden, &nameMatches](int8_t /* lang */, string const & name) {
+      if (golden.m_name == strings::MakeUniString(name))
+      {
+        nameMatches = true;
+        return false;  // breaks the loop
+      }
+      return true;  // continues the loop
+    });
+  }
+
+  return nameMatches && golden.m_houseNumber == houseNumber &&
+         MercatorBounds::DistanceOnEarth(golden.m_pos, center) < kToleranceMeters;
+}
+}  // namespace search
diff --git a/search/search_quality/matcher.hpp b/search/search_quality/matcher.hpp
new file mode 100644
index 0000000000..c99c3c079f
--- /dev/null
+++ b/search/search_quality/matcher.hpp
@@ -0,0 +1,38 @@
+#pragma once
+
+#include "search/result.hpp"
+#include "search/search_quality/sample.hpp"
+
+#include "indexer/index.hpp"
+
+#include "base/macros.hpp"
+
+#include <cstdint>
+#include <limits>
+#include <memory>
+#include <vector>
+
+class FeatureType;
+struct FeatureID;
+
+namespace search
+{
+class Matcher
+{
+public:
+  static size_t constexpr kInvalidId = std::numeric_limits<size_t>::max();
+
+  Matcher(Index & index);
+
+  void Match(std::vector<Sample::Result> const & golden, std::vector<Result> const & actual,
+             std::vector<size_t> & goldenMatching, std::vector<size_t> & actualMatching);
+
+private:
+  WARN_UNUSED_RESULT bool GetFeature(FeatureID const & id, FeatureType & ft);
+
+  bool Matches(Sample::Result const & golden, Result const & actual);
+
+  Index & m_index;
+  std::unique_ptr<Index::FeaturesLoaderGuard> m_guard;
+};
+}  // namespace search
diff --git a/search/search_quality/search_quality.pro b/search/search_quality/search_quality.pro
index 4cd8fe3de6..49a3fd922a 100644
--- a/search/search_quality/search_quality.pro
+++ b/search/search_quality/search_quality.pro
@@ -12,8 +12,10 @@ INCLUDEPATH += $$ROOT_DIR/3party/jansson/src
 
 HEADERS += \
     helpers.hpp \
+    matcher.hpp \
     sample.hpp \
 
 SOURCES += \
     helpers.cpp \
+    matcher.cpp \
     sample.cpp \