Add language metainfo generation routine. Use unicode supplementalData.xml as source.

storage/country.cpp

@@ -142,11 +142,11 @@ namespace
     }
   };
 
-  class DoStoreNames
+  class DoStoreFile2Name
   {
-    map<string, string> & m_id2name;
+    map<string, string> & m_file2name;
   public:
-    DoStoreNames(map<string, string> & id2name) : m_id2name(id2name) {}
+    DoStoreFile2Name(map<string, string> & file2name) : m_file2name(file2name) {}
 
     void operator() (string name, string const & file, string const &,
                      uint32_t size, int64_t, int)
@@ -159,10 +159,23 @@ namespace
           name = file.substr(0, i) + '_' + name;
 
         if (name != file)
-          m_id2name[file] = name;
+          m_file2name[file] = name;
       }
     }
   };
+
+  class DoStoreCode2File
+  {
+    multimap<string, string> & m_code2file;
+  public:
+    DoStoreCode2File(multimap<string, string> & code2file) : m_code2file(code2file) {}
+
+    void operator() (string const &, string const & file, string const & flag,
+                     uint32_t, int64_t, int)
+    {
+      m_code2file.insert(make_pair(flag, file));
+    }
+  };
 }
 
 int64_t LoadCountries(string const & jsonBuffer, CountriesContainerT & countries)
@@ -172,10 +185,17 @@ int64_t LoadCountries(string const & jsonBuffer, CountriesContainerT & countries
   return LoadCountriesImpl(jsonBuffer, doStore);
 }
 
-void LoadCountryNames(string const & jsonBuffer, map<string, string> & id2name)
+void LoadCountryFile2Name(string const & jsonBuffer, map<string, string> & id2name)
 {
   ASSERT ( id2name.empty(), () );
-  DoStoreNames doStore(id2name);
+  DoStoreFile2Name doStore(id2name);
+  LoadCountriesImpl(jsonBuffer, doStore);
+}
+
+void LoadCountryCode2File(string const & jsonBuffer, multimap<string, string> & code2file)
+{
+  ASSERT ( code2file.empty(), () );
+  DoStoreCode2File doStore(code2file);
   LoadCountriesImpl(jsonBuffer, doStore);
 }
 

storage/country.hpp

@@ -70,7 +70,8 @@ namespace storage
 
   /// @return version of country file or -1 if error was encountered
   int64_t LoadCountries(string const & jsonBuffer, CountriesContainerT & countries);
-  void LoadCountryNames(string const & jsonBuffer, map<string, string> & id2name);
+  void LoadCountryFile2Name(string const & jsonBuffer, map<string, string> & id2name);
+  void LoadCountryCode2File(string const & jsonBuffer, multimap<string, string> & code2file);
 
   bool SaveCountries(int64_t version, CountriesContainerT const & countries, string & jsonBuffer);
 }

storage/country_info.cpp

@@ -20,7 +20,7 @@ namespace storage
 
     string buffer;
     countryR.ReadAsString(buffer);
-    LoadCountryNames(buffer, m_id2name);
+    LoadCountryFile2Name(buffer, m_id2name);
   }
 
   template <class ToDo>

storage/storage_tests/country_info_test.cpp

@@ -27,7 +27,7 @@ UNIT_TEST(CountryInfo_ValidName_Smoke)
   ReaderPtr<Reader>(GetPlatform().GetReader(COUNTRIES_FILE)).ReadAsString(buffer);
 
   map<string, string> id2name;
-  storage::LoadCountryNames(buffer, id2name);
+  storage::LoadCountryFile2Name(buffer, id2name);
 
   TEST(id2name.count("Germany_Baden-Wurttemberg") == 1, ());
   TEST(id2name.count("France_Paris & Ile-de-France") == 1, ());

storage/storage_tests/generate_langs.cpp

@@ -0,0 +1,139 @@
+#include "../../testing/testing.hpp"
+
+#include "../country.hpp"
+
+#include "../../platform/platform.hpp"
+
+#include "../../coding/file_reader.hpp"
+#include "../../coding/parse_xml.hpp"
+
+#include "../../base/string_utils.hpp"
+
+#include "../../std/fstream.hpp"
+
+
+/*
+namespace
+{
+  class LangXMLGetter
+  {
+    string m_path;
+    multimap<string, string> & m_code2file;
+
+    int m_state;
+    string m_country, m_lang;
+    double m_percent;
+    bool m_official;
+
+    string m_res;
+
+  public:
+    LangXMLGetter(string const & path, multimap<string, string> & code2file)
+      : m_path(path), m_code2file(code2file), m_state(0)
+    {
+    }
+
+    bool Push(string const & name)
+    {
+      if (m_state == 0 && name == "territoryInfo")
+      {
+        m_state = 1;
+      }
+      else if (m_state == 1 && name == "territory")
+      {
+        m_country.clear();
+        m_res.clear();
+
+        m_state = 2;
+      }
+      else if (m_state == 2 && name == "languagePopulation")
+      {
+        m_lang.clear();
+        m_percent = 0.0;
+        m_official = false;
+
+        m_state = 3;
+      }
+
+      return true;
+    }
+
+    void AddAttr(string const & name, string const & value)
+    {
+      switch (m_state)
+      {
+      case 2:
+        if (name == "type")
+          m_country = value;
+        break;
+      case 3:
+        if (name == "type")
+          m_lang = value;
+        else if (name == "populationPercent")
+          strings::to_double(value, m_percent);
+        else
+        {
+          if (name == "officialStatus" &&
+              (value == "official" || value == "de_facto_official"))
+          {
+            m_official = true;
+          }
+        }
+      }
+    }
+
+    void Pop(string const &)
+    {
+      switch (m_state)
+      {
+      case 3:
+        // emit language
+        if (!m_lang.empty() && (m_percent >= 10.0 || m_official))
+        {
+          if (m_res.empty()) m_res = m_lang;
+          else m_res = m_res + "|" + m_lang;
+        }
+        break;
+
+      case 2:
+        // save result languages
+        if (!m_country.empty() && !m_res.empty())
+        {
+          typedef multimap<string, string>::const_iterator iter_t;
+
+          strings::MakeLowerCase(m_country);
+          pair<iter_t, iter_t> r = m_code2file.equal_range(m_country);
+          while (r.first != r.second)
+          {
+            ofstream file((m_path + r.first->second + ".meta").c_str());
+            file << m_res;
+
+            ++r.first;
+          }
+        }
+        break;
+      }
+
+      if (m_state > 0) --m_state;
+    }
+
+    void CharData(string const &) {}
+  };
+}
+
+UNIT_TEST(GenerateLanguages)
+{
+  string buffer;
+  ReaderPtr<Reader>(GetPlatform().GetReader(COUNTRIES_FILE)).ReadAsString(buffer);
+
+  multimap<string, string> code2file;
+  storage::LoadCountryCode2File(buffer, code2file);
+
+  string const path = "/Users/alena/omim/omim/data/metainfo_test/";
+  FileReader reader(path + "supplementalData.xml");
+
+  ReaderSource<FileReader> src(reader);
+  LangXMLGetter parser(path, code2file);
+  ParseXML(src, parser);
+}
+*/

storage/storage_tests/storage_tests.pro

@@ -6,7 +6,7 @@ CONFIG -= app_bundle
 TEMPLATE = app
 
 ROOT_DIR = ../..
-DEPENDENCIES = storage indexer platform coding base jansson tomcrypt
+DEPENDENCIES = storage indexer platform coding base jansson tomcrypt expat
 
 include($$ROOT_DIR/common.pri)
 
@@ -22,3 +22,4 @@ SOURCES += \
   country_test.cpp \
   simple_tree_test.cpp \
   country_info_test.cpp \
+  generate_langs.cpp \