diff --git a/generator/search_index_builder.cpp b/generator/search_index_builder.cpp
index 9ba6112884..e4e32ed20b 100644
--- a/generator/search_index_builder.cpp
+++ b/generator/search_index_builder.cpp
@@ -261,9 +261,11 @@ public:
 
   void operator() (FeatureType const & f, uint32_t index) const
   {
+    using namespace search;
+
     feature::TypesHolder types(f);
 
-    static search::TypesSkipper skipIndex;
+    static TypesSkipper skipIndex;
 
     skipIndex.SkipTypes(types);
     if (types.Empty())
@@ -284,10 +286,10 @@ public:
       // See OSM TagInfo or Wiki about modern postcodes format. The average number of tokens is less
       // than two.
       buffer_vector<strings::UniString, 2> tokens;
-      SplitUniString(search::NormalizeAndSimplifyString(postcode), MakeBackInsertFunctor(tokens),
-                     search::Delimiters());
+      SplitUniString(NormalizeAndSimplifyString(postcode), MakeBackInsertFunctor(tokens),
+                     Delimiters());
       for (auto const & token : tokens)
-        inserter.AddToken(search::kCategoriesLang, search::PostcodeToString(token));
+        inserter.AddToken(kPostcodesLang, PostcodeToString(token));
     }
 
     // Skip types for features without names.
@@ -303,7 +305,7 @@ public:
 
     // add names of categories of the feature
     for (uint32_t t : categoryTypes)
-      inserter.AddToken(search::kCategoriesLang, search::FeatureTypeToString(c.GetIndexForType(t)));
+      inserter.AddToken(kCategoriesLang, FeatureTypeToString(c.GetIndexForType(t)));
   }
 };
 
diff --git a/search/feature_offset_match.hpp b/search/feature_offset_match.hpp
index 232bec7b7e..c6f52f89e7 100644
--- a/search/feature_offset_match.hpp
+++ b/search/feature_offset_match.hpp
@@ -3,6 +3,7 @@
 #include "search/search_index_values.hpp"
 #include "search/search_query.hpp"
 #include "search/search_query_params.hpp"
+#include "search/v2/token_slice.hpp"
 
 #include "indexer/trie.hpp"
 
@@ -103,6 +104,25 @@ bool CheckMatchString(strings::UniChar const * rootPrefix, size_t rootPrefixSize
 
   return false;
 }
+
+template <typename TValue>
+bool FindLangIndex(trie::Iterator<ValueList<TValue>> const & trieRoot, uint8_t lang, uint32_t & langIx)
+{
+  ASSERT_LESS(trieRoot.m_edge.size(), numeric_limits<uint32_t>::max(), ());
+
+  uint32_t const numLangs = static_cast<uint32_t>(trieRoot.m_edge.size());
+  for (uint32_t i = 0; i < numLangs; ++i)
+  {
+    auto const & edge = trieRoot.m_edge[i].m_label;
+    ASSERT_GREATER_OR_EQUAL(edge.size(), 1, ());
+    if (edge[0] == lang)
+    {
+      langIx = i;
+      return true;
+    }
+  }
+  return false;
+}
 }  // namespace
 
 template <typename TValue, typename TF>
@@ -222,7 +242,7 @@ public:
       toDo(value);
   }
 };
-}  // namespace search::impl
+}  // impl
 
 template <typename TValue>
 struct TrieRootPrefix
@@ -345,27 +365,23 @@ template <typename TValue, typename THolder>
 bool MatchCategoriesInTrie(SearchQueryParams const & params,
                            trie::Iterator<ValueList<TValue>> const & trieRoot, THolder && holder)
 {
-  ASSERT_LESS(trieRoot.m_edge.size(), numeric_limits<uint32_t>::max(), ());
-  uint32_t const numLangs = static_cast<uint32_t>(trieRoot.m_edge.size());
-  for (uint32_t langIx = 0; langIx < numLangs; ++langIx)
-  {
-    auto const & edge = trieRoot.m_edge[langIx].m_label;
-    ASSERT_GREATER_OR_EQUAL(edge.size(), 1, ());
-    if (edge[0] == search::kCategoriesLang)
-    {
-      auto const catRoot = trieRoot.GoToEdge(langIx);
-      MatchTokensInTrie(params.m_tokens, TrieRootPrefix<TValue>(*catRoot, edge), holder);
+  uint32_t langIx = 0;
+  if (!impl::FindLangIndex(trieRoot, search::kCategoriesLang, langIx))
+    return false;
 
-      // Last token's prefix is used as a complete token here, to
-      // limit the number of features in the last bucket of a
-      // holder. Probably, this is a false optimization.
-      holder.Resize(params.m_tokens.size() + 1);
-      holder.SwitchTo(params.m_tokens.size());
-      MatchTokenInTrie(params.m_prefixTokens, TrieRootPrefix<TValue>(*catRoot, edge), holder);
-      return true;
-    }
-  }
-  return false;
+  auto const & edge = trieRoot.m_edge[langIx].m_label;
+  ASSERT_GREATER_OR_EQUAL(edge.size(), 1, ());
+
+  auto const catRoot = trieRoot.GoToEdge(langIx);
+  MatchTokensInTrie(params.m_tokens, TrieRootPrefix<TValue>(*catRoot, edge), holder);
+
+  // Last token's prefix is used as a complete token here, to limit
+  // the number of features in the last bucket of a holder. Probably,
+  // this is a false optimization.
+  holder.Resize(params.m_tokens.size() + 1);
+  holder.SwitchTo(params.m_tokens.size());
+  MatchTokenInTrie(params.m_prefixTokens, TrieRootPrefix<TValue>(*catRoot, edge), holder);
+  return true;
 }
 
 // Calls toDo with trie root prefix and language code on each language
@@ -427,4 +443,29 @@ void MatchFeaturesInTrie(SearchQueryParams const & params,
 
   intersecter.ForEachResult(forward<ToDo>(toDo));
 }
+
+template <typename TValue, typename TFilter, typename ToDo>
+void MatchPostcodesInTrie(v2::TokenSlice const & slice,
+                          trie::Iterator<ValueList<TValue>> const & trieRoot,
+                          TFilter const & filter, ToDo && toDo)
+{
+  uint32_t langIx = 0;
+  if (!impl::FindLangIndex(trieRoot, search::kPostcodesLang, langIx))
+    return;
+
+  auto const & edge = trieRoot.m_edge[langIx].m_label;
+  auto const postcodesRoot = trieRoot.GoToEdge(langIx);
+
+  impl::OffsetIntersecter<TFilter, TValue> intersecter(filter);
+  for (size_t i = 0; i < slice.Size(); ++i)
+  {
+    if (slice.IsPrefix(i))
+      MatchTokenPrefixInTrie(slice.Get(i), TrieRootPrefix<TValue>(*postcodesRoot, edge), intersecter);
+    else
+      MatchTokenInTrie(slice.Get(i), TrieRootPrefix<TValue>(*postcodesRoot, edge), intersecter);
+    intersecter.NextStep();
+  }
+
+  intersecter.ForEachResult(forward<ToDo>(toDo));
+}
 }  // namespace search
diff --git a/search/retrieval.cpp b/search/retrieval.cpp
index 3d4ac39a62..c4376b5996 100644
--- a/search/retrieval.cpp
+++ b/search/retrieval.cpp
@@ -7,6 +7,7 @@
 #include "search_trie.hpp"
 
 #include "v2/mwm_context.hpp"
+#include "v2/token_slice.hpp"
 
 #include "indexer/feature.hpp"
 #include "indexer/feature_algo.hpp"
@@ -29,8 +30,65 @@ using osm::Editor;
 
 namespace search
 {
+namespace v2
+{
 namespace
 {
+class FeaturesCollector
+{
+public:
+  FeaturesCollector(my::Cancellable const & cancellable, vector<uint64_t> & features)
+    : m_cancellable(cancellable), m_features(features), m_counter(0)
+  {
+  }
+
+  template <typename TValue>
+  void operator()(TValue const & value)
+  {
+    if ((++m_counter & 0xFF) == 0)
+      BailIfCancelled(m_cancellable);
+    m_features.push_back(value.m_featureId);
+  }
+
+  inline void operator()(uint32_t feature) { m_features.push_back(feature); }
+
+  inline void operator()(uint64_t feature) { m_features.push_back(feature); }
+
+private:
+  my::Cancellable const & m_cancellable;
+  vector<uint64_t> & m_features;
+  uint32_t m_counter;
+};
+
+class EditedFeaturesHolder
+{
+public:
+  EditedFeaturesHolder(MwmSet::MwmId const & id)
+  {
+    Editor & editor = Editor::Instance();
+    m_deleted = editor.GetFeaturesByStatus(id, Editor::FeatureStatus::Deleted);
+    m_modified = editor.GetFeaturesByStatus(id, Editor::FeatureStatus::Modified);
+    m_created = editor.GetFeaturesByStatus(id, Editor::FeatureStatus::Created);
+  }
+
+  bool ModifiedOrDeleted(uint32_t featureIndex) const
+  {
+    return binary_search(m_deleted.begin(), m_deleted.end(), featureIndex) ||
+           binary_search(m_modified.begin(), m_modified.end(), featureIndex);
+  }
+
+  template <typename TFn>
+  void ForEachModifiedOrCreated(TFn & fn)
+  {
+    for_each(m_modified.begin(), m_modified.end(), fn);
+    for_each(m_created.begin(), m_created.end(), fn);
+  }
+
+private:
+  vector<uint32_t> m_deleted;
+  vector<uint32_t> m_modified;
+  vector<uint32_t> m_created;
+};
 
 unique_ptr<coding::CompressedBitVector> SortFeaturesAndBuildCBV(vector<uint64_t> && features)
 {
@@ -89,6 +147,13 @@ bool MatchFeatureByName(FeatureType const & ft, SearchQueryParams const & params
   return matched;
 }
 
+bool MatchFeatureByPostcode(FeatureType const & ft, v2::TokenSlice const & slice)
+{
+  string const postcode = ft.GetMetadata().Get(feature::Metadata::FMD_POSTCODE);
+  // TODO(@y): implement this.
+  return false;
+}
+
 // Retrieves from the search index corresponding to |value| all
 // features matching to |params|.
 template <typename TValue>
@@ -96,16 +161,7 @@ unique_ptr<coding::CompressedBitVector> RetrieveAddressFeaturesImpl(
     MwmSet::MwmId const & id, MwmValue & value, my::Cancellable const & cancellable,
     SearchQueryParams const & params)
 {
-  // Exclude from search all deleted/modified features and match all edited/created features separately.
-  Editor & editor = Editor::Instance();
-
-  auto const deleted = editor.GetFeaturesByStatus(id, Editor::FeatureStatus::Deleted);
-  auto const modified = editor.GetFeaturesByStatus(id, Editor::FeatureStatus::Modified);
-  auto const filter = [&](uint32_t featureIndex) -> bool
-  {
-    return (!binary_search(deleted.begin(), deleted.end(), featureIndex) &&
-            !binary_search(modified.begin(), modified.end(), featureIndex));
-  };
+  EditedFeaturesHolder holder(id);
 
   serial::CodingParams codingParams(trie::GetCodingParams(value.GetHeader().GetDefCodingParams()));
   ModelReaderPtr searchReader = value.m_cont.GetReader(SEARCH_INDEX_FILE_TAG);
@@ -116,17 +172,16 @@ unique_ptr<coding::CompressedBitVector> RetrieveAddressFeaturesImpl(
   // TODO (@y, @m): This code may be optimized in the case where
   // bit vectors are sorted in the search index.
   vector<uint64_t> features;
-  uint32_t counter = 0;
-  auto const collector = [&](TValue const & value)
-  {
-    if ((++counter & 0xFF) == 0)
-      BailIfCancelled(cancellable);
-    features.push_back(value.m_featureId);
-  };
+  FeaturesCollector collector(cancellable, features);
 
-  MatchFeaturesInTrie(params, *trieRoot, filter, collector);
+  MatchFeaturesInTrie(params, *trieRoot, [&holder](uint32_t featureIndex)
+                      {
+                        return !holder.ModifiedOrDeleted(featureIndex);
+                      },
+                      collector);
 
   // Match all edited/created features separately.
+  Editor & editor = Editor::Instance();
   auto const matcher = [&](uint32_t featureIndex)
   {
     FeatureType ft;
@@ -136,9 +191,47 @@ unique_ptr<coding::CompressedBitVector> RetrieveAddressFeaturesImpl(
       features.push_back(featureIndex);
   };
 
-  for_each(modified.begin(), modified.end(), matcher);
-  auto const created = editor.GetFeaturesByStatus(id, Editor::FeatureStatus::Created);
-  for_each(created.begin(), created.end(), matcher);
+  holder.ForEachModifiedOrCreated(matcher);
+
+  return SortFeaturesAndBuildCBV(move(features));
+}
+
+template <typename TValue>
+unique_ptr<coding::CompressedBitVector> RetrievePostcodeFeaturesImpl(
+    MwmSet::MwmId const & id, MwmValue & value, my::Cancellable const & cancellable,
+    TokenSlice const & slice)
+{
+  EditedFeaturesHolder holder(id);
+
+  serial::CodingParams codingParams(trie::GetCodingParams(value.GetHeader().GetDefCodingParams()));
+  ModelReaderPtr searchReader = value.m_cont.GetReader(SEARCH_INDEX_FILE_TAG);
+
+  auto const trieRoot = trie::ReadTrie<SubReaderWrapper<Reader>, ValueList<TValue>>(
+      SubReaderWrapper<Reader>(searchReader.GetPtr()), SingleValueSerializer<TValue>(codingParams));
+
+  // TODO (@y, @m): This code may be optimized in the case where
+  // bit vectors are sorted in the search index.
+  vector<uint64_t> features;
+  FeaturesCollector collector(cancellable, features);
+
+  MatchPostcodesInTrie(slice, *trieRoot, [&holder](uint32_t featureIndex)
+                       {
+                         return !holder.ModifiedOrDeleted(featureIndex);
+                       },
+                       collector);
+
+  // Match all edited/created features separately.
+  Editor & editor = Editor::Instance();
+  auto const matcher = [&](uint32_t featureIndex)
+  {
+    FeatureType ft;
+    VERIFY(editor.GetEditedFeature(id, featureIndex, ft), ());
+    // TODO(AlexZ): Should we match by some feature's metafields too?
+    if (MatchFeatureByPostcode(ft, slice))
+      features.push_back(featureIndex);
+  };
+
+  holder.ForEachModifiedOrCreated(matcher);
 
   return SortFeaturesAndBuildCBV(move(features));
 }
@@ -149,16 +242,11 @@ unique_ptr<coding::CompressedBitVector> RetrieveGeometryFeaturesImpl(
     v2::MwmContext const & context, my::Cancellable const & cancellable,
     covering::IntervalsT const & coverage, int scale)
 {
-  uint32_t counter = 0;
   vector<uint64_t> features;
 
-  context.ForEachIndex(coverage, scale, [&](uint64_t featureId)
-  {
-    if ((++counter & 0xFF) == 0)
-      BailIfCancelled(cancellable);
-    features.push_back(featureId);
-  });
+  FeaturesCollector collector(cancellable, features);
 
+  context.ForEachIndex(coverage, scale, collector);
   return SortFeaturesAndBuildCBV(move(features));
 }
 
@@ -172,6 +260,16 @@ struct RetrieveAddressFeaturesAdaptor
   }
 };
 
+template <typename T>
+struct RetrievePostcodeFeaturesAdaptor
+{
+  template <typename... TArgs>
+  unique_ptr<coding::CompressedBitVector> operator()(TArgs &&... args)
+  {
+    return RetrievePostcodeFeaturesImpl<T>(forward<TArgs>(args)...);
+  }
+};
+
 template <template <typename> class T>
 struct Selector
 {
@@ -198,8 +296,6 @@ struct Selector
 };
 }  // namespace
 
-namespace v2
-{
 unique_ptr<coding::CompressedBitVector> RetrieveAddressFeatures(
     MwmSet::MwmId const & id, MwmValue & value, my::Cancellable const & cancellable,
     SearchQueryParams const & params)
@@ -208,6 +304,14 @@ unique_ptr<coding::CompressedBitVector> RetrieveAddressFeatures(
   return selector(id, value, cancellable, params);
 }
 
+unique_ptr<coding::CompressedBitVector> RetrievePostcodeFeatures(
+    MwmSet::MwmId const & id, MwmValue & value, my::Cancellable const & cancellable,
+    TokenSlice const & slice)
+{
+  Selector<RetrievePostcodeFeaturesAdaptor> selector;
+  return selector(id, value, cancellable, slice);
+}
+
 unique_ptr<coding::CompressedBitVector> RetrieveGeometryFeatures(
     MwmContext const & context, my::Cancellable const & cancellable,
     m2::RectD const & rect, int scale)
diff --git a/search/retrieval.hpp b/search/retrieval.hpp
index e0c55d6cf5..7540211db3 100644
--- a/search/retrieval.hpp
+++ b/search/retrieval.hpp
@@ -21,6 +21,7 @@ namespace search
 namespace v2
 {
 class MwmContext;
+class TokenSlice;
 
 // Retrieves from the search index corresponding to |value| all
 // features matching to |params|.
@@ -29,6 +30,12 @@ unique_ptr<coding::CompressedBitVector> RetrieveAddressFeatures(MwmSet::MwmId co
                                                                 my::Cancellable const & cancellable,
                                                                 SearchQueryParams const & params);
 
+// Retrieves from the search index corresponding to |value| all
+// postcodes matching to |slice|.
+unique_ptr<coding::CompressedBitVector> RetrievePostcodeFeatures(
+    MwmSet::MwmId const & id, MwmValue & value, my::Cancellable const & cancellable,
+    TokenSlice const & slice);
+
 // Retrieves from the geometry index corresponding to |value| all features belonging to |rect|.
 unique_ptr<coding::CompressedBitVector> RetrieveGeometryFeatures(
     MwmContext const & context, my::Cancellable const & cancellable, m2::RectD const & rect,
diff --git a/search/search.pro b/search/search.pro
index 0b4a76bfa0..6ba422fe32 100644
--- a/search/search.pro
+++ b/search/search.pro
@@ -63,7 +63,7 @@ HEADERS += \
     v2/search_query_v2.hpp \
     v2/stats_cache.hpp \
     v2/street_vicinity_loader.hpp \
-    v2/tokens_slice.hpp \
+    v2/token_slice.hpp \
 
 SOURCES += \
     approximate_string_match.cpp \
@@ -109,4 +109,4 @@ SOURCES += \
     v2/search_model.cpp \
     v2/search_query_v2.cpp \
     v2/street_vicinity_loader.cpp \
-    v2/tokens_slice.cpp \
+    v2/token_slice.cpp \
diff --git a/search/search_integration_tests/search_query_v2_test.cpp b/search/search_integration_tests/search_query_v2_test.cpp
index 7ed19b2c58..f8a728cb7b 100644
--- a/search/search_integration_tests/search_query_v2_test.cpp
+++ b/search/search_integration_tests/search_query_v2_test.cpp
@@ -1,10 +1,15 @@
 #include "testing/testing.hpp"
 
+#include "search/retrieval.hpp"
 #include "search/search_integration_tests/helpers.hpp"
 #include "search/search_tests_support/test_feature.hpp"
 #include "search/search_tests_support/test_mwm_builder.hpp"
 #include "search/search_tests_support/test_results_matching.hpp"
 #include "search/search_tests_support/test_search_request.hpp"
+#include "search/v2/token_slice.hpp"
+
+#include "indexer/feature.hpp"
+#include "indexer/index.hpp"
 
 #include "geometry/point2d.hpp"
 #include "geometry/rect2d.hpp"
@@ -377,6 +382,32 @@ UNIT_CLASS_TEST(SearchQueryV2Test, TestPostcodes)
                                   builder.Add(street);
                                   builder.Add(building);
                                 });
+
+  // Tests that postcode is added to the search index.
+  {
+    auto handle = m_engine.GetMwmHandleById(countryId);
+    TEST(handle.IsAlive(), ());
+    my::Cancellable cancellable;
+
+    SearchQueryParams params;
+    params.m_tokens.emplace_back();
+    params.m_tokens.back().push_back(PostcodeToString(strings::MakeUniString("141701")));
+    auto * value = handle.GetValue<MwmValue>();
+    auto features = v2::RetrievePostcodeFeatures(countryId, *value, cancellable,
+                                                 TokenSlice(params, 0, params.m_tokens.size()));
+    TEST_EQUAL(1, features->PopCount(), ());
+
+    uint64_t index = 0;
+    while (!features->GetBit(index))
+      ++index;
+
+    Index::FeaturesLoaderGuard loader(m_engine, countryId);
+    FeatureType ft;
+    loader.GetFeatureByIndex(index, ft);
+
+    auto rule = ExactMatch(countryId, building);
+    TEST(rule->Matches(ft), ());
+  }
   {
     TRules rules{ExactMatch(countryId, building)};
     TEST(ResultsMatch("Долгопрудный первомайская 28а", "ru" /* locale */, rules), ());
diff --git a/search/search_query.cpp b/search/search_query.cpp
index d60c3f7120..9c90c5e62f 100644
--- a/search/search_query.cpp
+++ b/search/search_query.cpp
@@ -12,7 +12,7 @@
 #include "search/v2/pre_ranking_info.hpp"
 #include "search/v2/ranking_info.hpp"
 #include "search/v2/ranking_utils.hpp"
-#include "search/v2/tokens_slice.hpp"
+#include "search/v2/token_slice.hpp"
 
 #include "storage/country_info_getter.hpp"
 #include "storage/index.hpp"
@@ -622,7 +622,7 @@ class PreResult2Maker
 
     info.m_nameScore = v2::NAME_SCORE_ZERO;
 
-    v2::TokensSliceNoCategories slice(m_params, preInfo.m_startToken, preInfo.m_endToken);
+    v2::TokenSliceNoCategories slice(m_params, preInfo.m_startToken, preInfo.m_endToken);
 
     for (auto const & lang : m_params.m_langs)
     {
diff --git a/search/search_tests/postcodes_matcher_tests.cpp b/search/search_tests/postcodes_matcher_tests.cpp
index 79013bc289..0438b5cdbb 100644
--- a/search/search_tests/postcodes_matcher_tests.cpp
+++ b/search/search_tests/postcodes_matcher_tests.cpp
@@ -2,7 +2,7 @@
 
 #include "search/search_query_params.hpp"
 #include "search/v2/postcodes_matcher.hpp"
-#include "search/v2/tokens_slice.hpp"
+#include "search/v2/token_slice.hpp"
 
 #include "indexer/search_delimiters.hpp"
 #include "indexer/search_string_utils.hpp"
@@ -42,7 +42,7 @@ bool LooksLikePostcode(string const & s, bool checkPrefix)
     params.m_tokens.back().push_back(token);
   }
 
-  return LooksLikePostcode(TokensSlice(params, 0, numTokens));
+  return LooksLikePostcode(TokenSlice(params, 0, numTokens));
 }
 
 UNIT_TEST(PostcodesMatcher_Smoke)
diff --git a/search/search_tests/ranking_tests.cpp b/search/search_tests/ranking_tests.cpp
index 37a2b15da0..cb82359f11 100644
--- a/search/search_tests/ranking_tests.cpp
+++ b/search/search_tests/ranking_tests.cpp
@@ -2,7 +2,7 @@
 
 #include "search/search_query_params.hpp"
 #include "search/v2/ranking_utils.hpp"
-#include "search/v2/tokens_slice.hpp"
+#include "search/v2/token_slice.hpp"
 
 #include "indexer/search_delimiters.hpp"
 #include "indexer/search_string_utils.hpp"
@@ -33,7 +33,7 @@ NameScore GetScore(string const & name, string const & query, size_t startToken,
     params.m_prefixTokens.swap(params.m_tokens.back());
     params.m_tokens.pop_back();
   }
-  return GetNameScore(name, TokensSlice(params, startToken, endToken));
+  return GetNameScore(name, TokenSlice(params, startToken, endToken));
 }
 
 UNIT_TEST(NameTest_Smoke)
diff --git a/search/search_tests_support/test_feature.hpp b/search/search_tests_support/test_feature.hpp
index 933d128609..b038998a1e 100644
--- a/search/search_tests_support/test_feature.hpp
+++ b/search/search_tests_support/test_feature.hpp
@@ -19,6 +19,7 @@ public:
 
   bool Matches(FeatureType const & feature) const;
   inline void SetPostcode(string const & postcode) { m_postcode = postcode; }
+  inline uint64_t GetId() const { return m_id; }
   inline string const & GetName() const { return m_name; }
 
   virtual void Serialize(FeatureBuilder1 & fb) const;
diff --git a/search/search_trie.hpp b/search/search_trie.hpp
index 513f50007d..2248d1694d 100644
--- a/search/search_trie.hpp
+++ b/search/search_trie.hpp
@@ -5,8 +5,8 @@
 
 namespace search
 {
-static const uint8_t kPostcodeLang = 127;
 static const uint8_t kCategoriesLang = 128;
+static const uint8_t kPostcodesLang = 129;
 static const uint8_t kPointCodingBits = 20;
 }  // namespace search
 
diff --git a/search/v2/locality_scorer.cpp b/search/v2/locality_scorer.cpp
index 053d57a595..5c790ab6aa 100644
--- a/search/v2/locality_scorer.cpp
+++ b/search/v2/locality_scorer.cpp
@@ -1,6 +1,6 @@
 #include "search/v2/locality_scorer.hpp"
 
-#include "search/v2/tokens_slice.hpp"
+#include "search/v2/token_slice.hpp"
 
 #include "std/algorithm.hpp"
 
@@ -101,8 +101,8 @@ void LocalityScorer::SortByName(vector<ExLocality> & ls) const
     auto score = NAME_SCORE_ZERO;
     for (auto const & name : names)
     {
-      score = max(score, GetNameScore(name, v2::TokensSlice(m_params, l.m_locality.m_startToken,
-                                                            l.m_locality.m_endToken)));
+      score = max(score, GetNameScore(name, v2::TokenSlice(m_params, l.m_locality.m_startToken,
+                                                           l.m_locality.m_endToken)));
     }
     l.m_nameScore = score;
   }
diff --git a/search/v2/postcodes_matcher.cpp b/search/v2/postcodes_matcher.cpp
index 84bfa1fcae..8c28d83eb2 100644
--- a/search/v2/postcodes_matcher.cpp
+++ b/search/v2/postcodes_matcher.cpp
@@ -1,6 +1,6 @@
 #include "search/v2/postcodes_matcher.hpp"
 
-#include "search/v2/tokens_slice.hpp"
+#include "search/v2/token_slice.hpp"
 
 #include "indexer/search_delimiters.hpp"
 #include "indexer/search_string_utils.hpp"
@@ -24,7 +24,8 @@ namespace v2
 namespace
 {
 // Top patterns for postcodes. See
-// search/search_quality/clusterize_postcodes.lisp for details.
+// search/search_quality/clusterize_postcodes.lisp for details how
+// these patterns were constructed.
 char const * const g_patterns[] = {
     "aa nnnn",   "aa nnnnn",   "aaa nnnn",    "aan",      "aan naa",  "aana naa", "aann",
     "aann naa",  "aannaa",     "aannnaa",     "aannnn",   "an naa",   "ana naa",  "ana nan",
@@ -90,6 +91,8 @@ struct Node
   DISALLOW_COPY(Node);
 };
 
+// This class puts all strings from g_patterns to a trie with a low
+// branching factor and matches queries against these patterns.
 class PostcodesMatcher
 {
 public:
@@ -100,7 +103,11 @@ public:
       AddString(MakeUniString(pattern), delimiters);
   }
 
-  bool HasString(TokensSlice const & slice) const
+  // Checks that given tokens match to at least one of postcodes
+  // patterns.
+  //
+  // Complexity: O(total length of tokens in |slice|).
+  bool HasString(TokenSlice const & slice) const
   {
     Node const * cur = &m_root;
     for (size_t i = 0; i < slice.Size() && cur; ++i)
@@ -154,7 +161,7 @@ PostcodesMatcher const & GetPostcodesMatcher()
 }
 }  // namespace
 
-bool LooksLikePostcode(TokensSlice const & slice) { return GetPostcodesMatcher().HasString(slice); }
+bool LooksLikePostcode(TokenSlice const & slice) { return GetPostcodesMatcher().HasString(slice); }
 
 size_t GetMaxNumTokensInPostcode() { return GetPostcodesMatcher().GetMaxNumTokensInPostcode(); }
 }  // namespace v2
diff --git a/search/v2/postcodes_matcher.hpp b/search/v2/postcodes_matcher.hpp
index 266e63b334..b0e2398e8e 100644
--- a/search/v2/postcodes_matcher.hpp
+++ b/search/v2/postcodes_matcher.hpp
@@ -6,9 +6,9 @@ namespace search
 {
 namespace v2
 {
-class TokensSlice;
+class TokenSlice;
 
-bool LooksLikePostcode(TokensSlice const & slice);
+bool LooksLikePostcode(TokenSlice const & slice);
 
 size_t GetMaxNumTokensInPostcode();
 }  // namespace v2
diff --git a/search/v2/tokens_slice.cpp b/search/v2/token_slice.cpp
similarity index 60%
rename from search/v2/tokens_slice.cpp
rename to search/v2/token_slice.cpp
index 9a523dcc02..38556c40df 100644
--- a/search/v2/tokens_slice.cpp
+++ b/search/v2/token_slice.cpp
@@ -1,17 +1,17 @@
-#include "search/v2/tokens_slice.hpp"
+#include "search/v2/token_slice.hpp"
 
 namespace search
 {
 namespace v2
 {
-TokensSlice::TokensSlice(SearchQueryParams const & params, size_t startToken, size_t endToken)
+TokenSlice::TokenSlice(SearchQueryParams const & params, size_t startToken, size_t endToken)
   : m_params(params), m_offset(startToken), m_size(endToken - startToken)
 {
   ASSERT_LESS_OR_EQUAL(startToken, endToken, ());
 }
 
-TokensSliceNoCategories::TokensSliceNoCategories(SearchQueryParams const & params,
-                                                 size_t startToken, size_t endToken)
+TokenSliceNoCategories::TokenSliceNoCategories(SearchQueryParams const & params, size_t startToken,
+                                               size_t endToken)
   : m_params(params)
 {
   ASSERT_LESS_OR_EQUAL(startToken, endToken, ());
diff --git a/search/v2/tokens_slice.hpp b/search/v2/token_slice.hpp
similarity index 84%
rename from search/v2/tokens_slice.hpp
rename to search/v2/token_slice.hpp
index 70173fffdd..7b9553e2f4 100644
--- a/search/v2/tokens_slice.hpp
+++ b/search/v2/token_slice.hpp
@@ -11,10 +11,10 @@ namespace search
 {
 namespace v2
 {
-class TokensSlice
+class TokenSlice
 {
 public:
-  TokensSlice(SearchQueryParams const & params, size_t startToken, size_t endToken);
+  TokenSlice(SearchQueryParams const & params, size_t startToken, size_t endToken);
 
   inline SearchQueryParams::TSynonymsVector const & Get(size_t i) const
   {
@@ -38,10 +38,10 @@ private:
   size_t const m_size;
 };
 
-class TokensSliceNoCategories
+class TokenSliceNoCategories
 {
 public:
-  TokensSliceNoCategories(SearchQueryParams const & params, size_t startToken, size_t endToken);
+  TokenSliceNoCategories(SearchQueryParams const & params, size_t startToken, size_t endToken);
 
   inline SearchQueryParams::TSynonymsVector const & Get(size_t i) const
   {