libstdc++-v3/testsuite/ext/unicode/grapheme_view.cc

   1 // { dg-do compile { target c++20 } }
   2
   3 #include <format> // includes <bits/unicode.h>
   4 #include <string_view>
   5 #include <ranges>
   6 #include <testsuite_hooks.h>
   7
   8 namespace uc = std::__unicode;
   9 using namespace std::string_view_literals;
  10
  11 constexpr void
  12 test_breaks()
  13 {
  14   VERIFY(uc::__field_width(u8"\N{LATIN SMALL LETTER E WITH ACUTE}"sv) == 1 );
  15
  16   auto sv = u8"ee\N{COMBINING ACUTE ACCENT}e"sv;
  17   auto data = sv.data();
  18   VERIFY( uc::__field_width(sv) == 3 );
  19   VERIFY( uc::__truncate(sv, 3) == 3 );
  20   VERIFY( uc::__truncate(sv, 4) == 3 );
  21   VERIFY( sv == data );
  22
  23   VERIFY( uc::__truncate(sv, 2) == 2 );
  24   VERIFY( sv == u8"ee\N{COMBINING ACUTE ACCENT}"sv );
  25
  26   sv = data;
  27   sv.remove_prefix(1);
  28   VERIFY( uc::__field_width(sv) == 2 );
  29   VERIFY( uc::__truncate(sv, 3) == 2 );
  30   VERIFY( sv == data+1 );
  31
  32   sv = u8"\N{REGIONAL INDICATOR SYMBOL LETTER G}"
  33          "\N{REGIONAL INDICATOR SYMBOL LETTER B}"; // GB flag emoji
  34   data = sv.data();
  35   VERIFY( uc::__field_width(sv) == 1 );
  36   VERIFY( uc::__truncate(sv, 2) == 1 );
  37   VERIFY( sv == data );
  38   VERIFY( uc::__truncate(sv, 1) == 1 ); // Do not break inside a flag emoji.
  39   VERIFY( sv == data );
  40
  41   sv = u8"abcd"
  42     "\N{REGIONAL INDICATOR SYMBOL LETTER G}" // 4 bytes
  43     "\N{REGIONAL INDICATOR SYMBOL LETTER B}" // 4 bytes
  44     "\N{DEVANAGARI LETTER KA}"               // 3 bytes
  45     "\N{DEVANAGARI SIGN VIRAMA}"             // 3 bytes
  46     "\N{DEVANAGARI LETTER RA}"               // 3 bytes
  47     "\N{MAN}\N{ZERO WIDTH JOINER}"           // 4+3 bytes
  48     "\N{WOMAN}\N{ZERO WIDTH JOINER}"         // 4+3 bytes
  49     "\N{GIRL}\N{ZERO WIDTH JOINER}"          // 4+3 bytes
  50     "\N{BOY}\N{ZERO WIDTH JOINER}"           // 4+3 bytes
  51     "\N{HANGUL CHOSEONG KIYEOK}"             // 3 bytes
  52     "\N{HANGUL CHOSEONG KIYEOK}"             // 3 bytes
  53     "\N{HANGUL CHOSEONG KIYEOK}"             // 3 bytes
  54     "\N{HANGUL CHOSEONG KIYEOK}"             // 3 bytes
  55     "\N{HANGUL JUNGSEONG A}"                 // 3 bytes
  56     "\N{HANGUL JONGSEONG KIYEOK}"            // 3 bytes
  57     "\N{HANGUL JONGSEONG KIYEOK}"            // 3 bytes
  58     "\N{HANGUL JONGSEONG KIYEOK}";           // 3 bytes
  59
  60   uc::_Grapheme_cluster_view gv(sv);
  61   auto iter = gv.begin();
  62   VERIFY( iter.base() == sv.data() );
  63   VERIFY( *iter == U'a' );
  64   std::ranges::advance(iter, 3);
  65   VERIFY( *iter == U'd' );
  66   VERIFY( iter.base() == sv.data() + 3 );
  67   ++iter;
  68   VERIFY( *iter == U'\N{REGIONAL INDICATOR SYMBOL LETTER G}' );
  69   VERIFY( iter.base() == sv.data() + 4 );
  70   ++iter;
  71   VERIFY( *iter == U'\N{DEVANAGARI LETTER KA}' );
  72   VERIFY( iter.base() == sv.data() + 4 + 8 );
  73   ++iter;
  74   VERIFY( *iter == U'\N{MAN}' );
  75   VERIFY( iter.base() == sv.data() + 4 + 8 + 9 );
  76   ++iter;
  77   VERIFY( iter.base() == sv.data() + 4 + 8 + 9 + 28 );
  78   VERIFY( *iter == U'\N{HANGUL CHOSEONG KIYEOK}' );
  79   ++iter;
  80   VERIFY( iter.base() == sv.data() + 4 + 8 + 9 + 28 + 24 );
  81   VERIFY( iter == gv.end() );
  82   ++iter;
  83   VERIFY( iter == gv.end() );
  84 }
  85
  86 int main()
  87 {
  88   auto run_tests = []{
  89     test_breaks();
  90     return true;
  91   };
  92
  93   VERIFY( run_tests() );
  94   static_assert( run_tests() );
  95 }