Add utf_selector and is_utf_same.

ww898 · Oct 21, 2018 · 2ee0b7c · 2ee0b7c
1 parent 3e423a4
commit 2ee0b7c
Show file tree

Hide file tree

Showing 3 changed files with 190 additions and 92 deletions.
diff --git a/README.md b/README.md
@@ -29,16 +29,20 @@ Tested on following compilers:
 
 ```cpp
 // यूनिकोड
-static char const u8_orig[] = "\xE0\xA4\xAF\xE0\xA5\x82\xE0\xA4\xA8\xE0\xA4\xBF\xE0\xA4\x95\xE0\xA5\x8B\xE0\xA4\xA1";
-using namespace ww898;
+static char const u8s[] = "\xE0\xA4\xAF\xE0\xA5\x82\xE0\xA4\xA8\xE0\xA4\xBF\xE0\xA4\x95\xE0\xA5\x8B\xE0\xA4\xA1";
+using namespace ww898::utf;
 std::u16string u16;
-utf::convz<utf::utf8, utf::utf16>(u8_orig, std::back_inserter(u16));
+convz<utf_selector_t<decltype(*u8s)>, utf16>(u8s, std::back_inserter(u16));
 std::u32string u32;
-utf::conv<utf::utf16, utf::utf32>(u16.begin(), u16.end(), std::back_inserter(u32));
+conv<utf16, utf_selector_t<decltype(u32)::value_type>>(u16.begin(), u16.end(), std::back_inserter(u32));
 std::vector<char> u8;
-utf::convz<utf::utf32, utf::utf8>(u32.begin(), std::back_inserter(u8));
-std::wstring wstr;
-utf::convz<utf::utf8, utf::utfw>(u8.begin(), std::back_inserter(wstr));
+convz<utf32, utf8>(u32.data(), std::back_inserter(u8));
+std::wstring uw;
+conv<utf8, utfw>(u8s, u8s + sizeof(u8s), std::back_inserter(uw));
+static_assert(is_utf_same<decltype(*u8s), decltype(u8)::value_type>::value, "Fail");
+static_assert(1 ==
+    (is_utf_same<decltype(u16)::value_type, decltype(uw)::value_type>::value ? 1 : 0) +
+    (is_utf_same<decltype(u32)::value_type, decltype(uw)::value_type>::value ? 1 : 0), "Fail");
 ```
 
 ## Performance
@@ -237,33 +241,33 @@ UTF8  ==> UTFW : 0.340384930s (+196.02%)
 
 #### MacOS High Sierra v10.13.6 (Clang v6.0.0)
 ```cpp
-Running 489 test cases...
-sizeof wchar_t: 4
-UTFW: UTF32
-Resolution: 2793647583
-UTF8  ==> UTF8 : 0.111039205s
-UTF8  ==> UTF16: 0.143631552s
-UTF8  ==> UTF32: 0.105463425s
-UTF8  ==> UTFW : 0.105106640s
-UTF16 ==> UTF8 : 0.158074631s
-UTF16 ==> UTF16: 0.055528284s
-UTF16 ==> UTF32: 0.063203264s
-UTF16 ==> UTFW : 0.063167823s
-UTF32 ==> UTF8 : 0.123977591s
-UTF32 ==> UTF16: 0.061630976s
-UTF32 ==> UTF32: 0.027633560s
-UTF32 ==> UTFW : 0.029324893s
-UTFW  ==> UTF8 : 0.123948012s
-UTFW  ==> UTF16: 0.064873256s
-UTFW  ==> UTF32: 0.030606730s
-UTFW  ==> UTFW : 0.027596372s
-codecvt_utf8_utf16<char16_t>:
-UTF16 ==> UTF8 : 0.151798551s (-3.97%)
-UTF8  ==> UTF16: 0.256203078s (+78.38%)
-codecvt_utf8<wchar_t>:
-UTFW  ==> UTF8 : 0.137034385s (+10.56%)
-UTF8  ==> UTFW : 0.360953804s (+243.42%)
-
+Running 489 test cases...
+sizeof wchar_t: 4
+UTFW: UTF32
+Resolution: 2793647583
+UTF8  ==> UTF8 : 0.111039205s
+UTF8  ==> UTF16: 0.143631552s
+UTF8  ==> UTF32: 0.105463425s
+UTF8  ==> UTFW : 0.105106640s
+UTF16 ==> UTF8 : 0.158074631s
+UTF16 ==> UTF16: 0.055528284s
+UTF16 ==> UTF32: 0.063203264s
+UTF16 ==> UTFW : 0.063167823s
+UTF32 ==> UTF8 : 0.123977591s
+UTF32 ==> UTF16: 0.061630976s
+UTF32 ==> UTF32: 0.027633560s
+UTF32 ==> UTFW : 0.029324893s
+UTFW  ==> UTF8 : 0.123948012s
+UTFW  ==> UTF16: 0.064873256s
+UTFW  ==> UTF32: 0.030606730s
+UTFW  ==> UTFW : 0.027596372s
+codecvt_utf8_utf16<char16_t>:
+UTF16 ==> UTF8 : 0.151798551s (-3.97%)
+UTF8  ==> UTF16: 0.256203078s (+78.38%)
+codecvt_utf8<wchar_t>:
+UTFW  ==> UTF8 : 0.137034385s (+10.56%)
+UTF8  ==> UTFW : 0.360953804s (+243.42%)
+
 *** No errors detected
 ```
 

diff --git a/include/ww898/utf_converters.hpp b/include/ww898/utf_converters.hpp
@@ -456,8 +456,9 @@ template<
     typename Utf,
     typename Outf,
     typename It,
+    typename Eit,
     typename Oit>
-Oit conv(It && it, It && eit, Oit && oit)
+Oit conv(It && it, Eit && eit, Oit && oit)
 {
     return detail::conv_strategy<Utf, Outf,
             typename std::decay<It>::type,
@@ -468,7 +469,7 @@ Oit conv(It && it, It && eit, Oit && oit)
                     ? detail::conv_impl::random_interator
                     : detail::conv_impl::normal>()(
         std::forward<It>(it),
-        std::forward<It>(eit),
+        std::forward<Eit>(eit),
         std::forward<Oit>(oit));
 }
 
@@ -478,11 +479,39 @@ template<
     size_t wchar_size>
 struct wchar_selector {};
 
-template<> struct wchar_selector<2> { typedef utf16 utfw_type; };
-template<> struct wchar_selector<4> { typedef utf32 utfw_type; };
+template<> struct wchar_selector<2> { typedef utf16 type; };
+template<> struct wchar_selector<4> { typedef utf32 type; };
 
 }
 
-typedef detail::wchar_selector<sizeof(wchar_t)>::utfw_type utfw;
+typedef detail::wchar_selector<sizeof(wchar_t)>::type utfw;
+
+namespace detail {
+
+template<
+    typename Ch>
+struct utf_selector {};
+
+template<> struct utf_selector<         char> { typedef utf8  type; };
+template<> struct utf_selector<unsigned char> { typedef utf8  type; };
+template<> struct utf_selector<signed   char> { typedef utf8  type; };
+template<> struct utf_selector<char16_t     > { typedef utf16 type; };
+template<> struct utf_selector<char32_t     > { typedef utf32 type; };
+template<> struct utf_selector<wchar_t      > { typedef utfw  type; };
+
+}
+
+template<
+    typename Ch>
+using utf_selector = detail::utf_selector<typename std::decay<Ch>::type>;
+
+template<
+    typename Ch>
+using utf_selector_t = typename utf_selector<Ch>::type;
+
+template<
+    typename Ch1,
+    typename Ch2>
+using is_utf_same = std::is_same<utf_selector_t<Ch1>, utf_selector_t<Ch2>>;
 
 }}