[search] Avoid fancy categories matching (like cafe <-> care).

Signed-off-by: Viktor Govako <viktor.govako@gmail.com>

indexer/search_string_utils.cpp

@@ -55,12 +55,26 @@ size_t GetMaxErrorsForToken(strings::UniString const & token)
 
 strings::LevenshteinDFA BuildLevenshteinDFA(strings::UniString const & s)
 {
+  ASSERT(!s.empty(), ());
   // In search we use LevenshteinDFAs for fuzzy matching. But due to
   // performance reasons, we limit prefix misprints to fixed set of substitutions defined in
   // kAllowedMisprints and skipped letters.
   return strings::LevenshteinDFA(s, 1 /* prefixSize */, kAllowedMisprints, GetMaxErrorsForToken(s));
 }
 
+strings::LevenshteinDFA BuildLevenshteinDFA_Category(strings::UniString const & s)
+{
+  // https://github.com/organicmaps/organicmaps/issues/3655
+  // Separate DFA for categories (token's length <= 4 means no errors allowed) to avoid fancy matchings like:
+  // cafe <-> care
+  // ecco -> eco
+  // shop <-> shoe
+  /// @todo "hote" doesn't match "hotel" now. Should allow _adding_ symbols when size == 4.
+
+  ASSERT(!s.empty(), ());
+  return strings::LevenshteinDFA(s, 1 /* prefixSize */, kAllowedMisprints, GetMaxErrorsForTokenLength(s.size() - 1));
+}
+
 UniString NormalizeAndSimplifyString(string_view s)
 {
   UniString uniString = MakeUniString(s);

indexer/search_string_utils.hpp

@@ -22,6 +22,7 @@ inline constexpr size_t GetMaxErrorsForTokenLength(size_t length)
 size_t GetMaxErrorsForToken(strings::UniString const & token);
 
 strings::LevenshteinDFA BuildLevenshteinDFA(strings::UniString const & s);
+strings::LevenshteinDFA BuildLevenshteinDFA_Category(strings::UniString const & s);
 
 // This function should be used for all search strings normalization.
 // It does some magic text transformation which greatly helps us to improve our search.

search/search_integration_tests/processor_test.cpp

@@ -917,7 +917,8 @@ UNIT_CLASS_TEST(ProcessorTest, TestCategorialSearch)
   }
 
   {
-    Rules const rules = {ExactMatch(wonderlandId, hotel1), ExactMatch(wonderlandId, hotel2),
+    /// @todo We updated fuzzy match for categories: hote -> hotel is not matched now (4 letters input token).
+    Rules const rules = {/*ExactMatch(wonderlandId, hotel1),*/ ExactMatch(wonderlandId, hotel2),
                          ExactMatch(wonderlandId, hotelCafe), ExactMatch(testWorldId, homel),
                          ExactMatch(wonderlandId, hotelDeVille)};
     // A prefix token.
@@ -3207,4 +3208,45 @@ UNIT_CLASS_TEST(ProcessorTest, Place_Region)
   TEST(ResultsMatch("carth", rules, "en"), ());
 }
 
+UNIT_CLASS_TEST(ProcessorTest, FuzzyCategories)
+{
+  TestPOI cafe({0, 0.01}, "xxx", "en");
+  cafe.SetTypes({{"amenity", "cafe"}});
+
+  TestPOI cosmetics({0, 0.02}, "yyy", "en");
+  cosmetics.SetTypes({{"shop", "cosmetics"}});
+
+  TestPOI shoes({0, 0.03}, "ecco", "en");
+  shoes.SetTypes({{"shop", "shoes"}});
+
+  TestPOI organic({0, 0.04}, "zzz", "en");
+  organic.SetTypes({{"shop", "grocery"}, {"organic", "yes"}});
+
+  auto wonderlandId = BuildCountry("Wonderland", [&](TestMwmBuilder & builder)
+  {
+    builder.Add(cafe);
+    builder.Add(cosmetics);
+    builder.Add(shoes);
+    builder.Add(organic);
+  });
+
+  SetViewport(m2::RectD(-0.5, -0.5, 0.5, 0.5));
+
+  {
+    Rules const rules = {ExactMatch(wonderlandId, cafe)};
+    TEST(ResultsMatch("cafe", rules), ());
+  }
+
+  {
+    Rules const rules = {ExactMatch(wonderlandId, shoes)};
+    TEST(ResultsMatch("shoe", rules), ());
+    TEST(ResultsMatch("shoes", rules), ());
+  }
+
+  {
+    Rules const rules = {ExactMatch(wonderlandId, shoes)};
+    TEST(ResultsMatch("ecco", rules), ());
+  }
+}
+
 } // namespace processor_test

search/utils.hpp

@@ -63,7 +63,7 @@ void ForEachCategoryTypeFuzzy(StringSliceBase const & slice, Locales const & loc
     // A possible optimization is to build each dfa once and save it. Note that
     // dfas for the prefix tokens differ, i.e. we ignore slice.IsPrefix(i) here.
     SearchTrieRequest<strings::LevenshteinDFA> request;
-    request.m_names.push_back(BuildLevenshteinDFA(slice.Get(i)));
+    request.m_names.push_back(BuildLevenshteinDFA_Category(slice.Get(i)));
     request.SetLangs(locales);
 
     MatchFeaturesInTrie(request, iterator,