|
|
@ -5,6 +5,11 @@ |
|
|
|
|
|
|
|
using namespace pEp; |
|
|
|
|
|
|
|
namespace pEp |
|
|
|
{ |
|
|
|
std::string escape_utf16(u16string_view s); |
|
|
|
} |
|
|
|
|
|
|
|
namespace { |
|
|
|
|
|
|
|
struct TestEntry |
|
|
@ -20,7 +25,7 @@ typedef TestEntry TE; |
|
|
|
|
|
|
|
std::ostream& operator<<(std::ostream& o, const TestEntry& tt) |
|
|
|
{ |
|
|
|
return o << "input=«" << tt.input << "», isNfc=" << tt.is_nfc << ", quick=" << tt.quick << ". "; |
|
|
|
return o << "input=«" << pEp::escape_utf16(tt.input) << "», isNfc=" << tt.is_nfc << ", quick=" << tt.quick << ". "; |
|
|
|
} |
|
|
|
|
|
|
|
|
|
|
@ -28,28 +33,28 @@ const char16_t nullo[4] = {0,0,0,0}; |
|
|
|
|
|
|
|
const std::vector<TestEntry> testValues = |
|
|
|
{ |
|
|
|
{ "" , true, IsNFC::Yes, "" }, // always start with the simple case ;-)
|
|
|
|
{ "123" , true, IsNFC::Yes, "123" }, // some ASCII digits. Still easy.
|
|
|
|
{ "\n\\\b" , true, IsNFC::Yes, "\n\\\b" }, // backslash escapes for ASCII and control chars
|
|
|
|
{ "ä" , true, IsNFC::Yes, "ä" }, // <U+00E4> small a with diaeresis
|
|
|
|
{ "\xc4\x85" , true, IsNFC::Yes, "\xc4\x85" }, // <U+0105> small a with ogonek
|
|
|
|
{ u"" , true, IsNFC::Yes, u"" }, // always start with the simple case ;-)
|
|
|
|
{ u"123" , true, IsNFC::Yes, u"123" }, // some ASCII digits. Still easy.
|
|
|
|
{ u"\n\\\b" , true, IsNFC::Yes, u"\n\\\b" }, // backslash escapes for ASCII and control chars
|
|
|
|
{ u"ä" , true, IsNFC::Yes, u"ä" }, // <U+00E4> small a with diaeresis
|
|
|
|
{ u"\u0105" , true, IsNFC::Yes, u"\u0105" }, // <U+0105> small a with ogonek
|
|
|
|
|
|
|
|
{ "a\xcc\x88", false, IsNFC::Maybe, "ä" }, // a + <U+0308> combining diaresis
|
|
|
|
{ "a\xcc\xa8", false, IsNFC::Maybe, "\xc4\x85" }, // a + <U+0328> combining ogonek
|
|
|
|
{ "a\xcc\xa8\xcc\x88", false, IsNFC::Maybe, "\xc4\x85\xcc\x88" }, // a + <U+0328> + <U+0308> (ogonek + diaeresis)
|
|
|
|
{ "a\xcc\x88\xcc\xa8", false, IsNFC::Maybe, "\xc4\x85\xcc\x88" }, // a + <U+0308> + <U+0328> (diaeresis + ogonek)
|
|
|
|
{ u"a\u0308" , false, IsNFC::Maybe, u"ä" }, // a + <U+0308> combining diaresis
|
|
|
|
{ u"a\u0328" , false, IsNFC::Maybe, u"\u0105" }, // a + <U+0328> combining ogonek
|
|
|
|
{ u"a\u0328\u0308", false, IsNFC::Maybe, u"\u0105\u0308" }, // a + <U+0328> + <U+0308> (ogonek + diaeresis)
|
|
|
|
{ u"a\u0308\u0328", false, IsNFC::Maybe, u"\u0105\u0308" }, // a + <U+0308> + <U+0328> (diaeresis + ogonek)
|
|
|
|
|
|
|
|
{ "\xc4\x85\xcc\x88" , true, IsNFC::Maybe, "\xc4\x85\xcc\x88" }, // <U+0105> small a with ogonek + combining diaeresis
|
|
|
|
{ "ä\xcc\xa8" , false, IsNFC::Maybe, "\xc4\x85\xcc\x88" }, // a diaeresis + <U+0328> combining ogonek
|
|
|
|
{ u"\u0105\u0308" , true, IsNFC::Maybe, u"\u0105\u0308" }, // <U+0105> small a with ogonek + combining diaeresis
|
|
|
|
{ u"ä\u0328" , false, IsNFC::Maybe, u"\u0105\u0308" }, // a diaeresis + <U+0328> combining ogonek
|
|
|
|
|
|
|
|
// Already implemented, because <U+305> and <U+33C> have neither "No" nor "Maybe" NFC class:
|
|
|
|
{ "a\xcc\x85\xcc\xbc", false, IsNFC::No , "a\xcc\xbc\xcc\x85"}, // a + <U+0305> + <U+033C> (overline + seagull_below)
|
|
|
|
{ "a\xcc\xbc\xcc\x85", true, IsNFC::Yes , "a\xcc\xbc\xcc\x85"}, // a + <U+033C> + <U+0305> (seagull_below + overline)
|
|
|
|
{ u"a\u0305\u033c", false, IsNFC::No , u"a\u033c\u0305"}, // a + <U+0305> + <U+033C> (overline + seagull_below)
|
|
|
|
{ u"a\u033c\u0305", true, IsNFC::Yes , u"a\u033c\u0305"}, // a + <U+033C> + <U+0305> (seagull_below + overline)
|
|
|
|
|
|
|
|
{ string_view(nullo, 1), true, IsNFC::Yes, string_view(nullo, 1) }, // Yeah, 1 NUL byte
|
|
|
|
{ string_view(nullo, 4), true, IsNFC::Yes, string_view(nullo, 4) }, // Yeah, 4 NUL bytes
|
|
|
|
{ u16string_view(nullo, 1), true, IsNFC::Yes, u16string_view(nullo, 1) }, // Yeah, 1 NUL byte
|
|
|
|
{ u16string_view(nullo, 4), true, IsNFC::Yes, u16string_view(nullo, 4) }, // Yeah, 4 NUL bytes
|
|
|
|
|
|
|
|
{ "EOF", true, IsNFC::Yes, "EOF" } |
|
|
|
{ u"EOF", true, IsNFC::Yes, u"EOF" } |
|
|
|
}; |
|
|
|
|
|
|
|
} // end of anonymous namespace
|
|
|
|