diff options
| -rw-r--r-- | src/corelib/io/qurl.cpp | 287 | ||||
| -rw-r--r-- | tests/auto/qurl/tst_qurl.cpp | 64 | ||||
| -rw-r--r-- | tests/auto/utf8/tst_utf8.cpp | 4 |
3 files changed, 222 insertions, 133 deletions
diff --git a/src/corelib/io/qurl.cpp b/src/corelib/io/qurl.cpp index aac1f10..20ec995 100644 --- a/src/corelib/io/qurl.cpp +++ b/src/corelib/io/qurl.cpp @@ -966,14 +966,14 @@ static void QT_FASTCALL _fragment(const char **ptr, QUrlParseData *parseData) } struct NameprepCaseFoldingEntry { - int uc; + uint uc; ushort mapping[4]; }; -inline bool operator<(int one, const NameprepCaseFoldingEntry &other) +inline bool operator<(uint one, const NameprepCaseFoldingEntry &other) { return one < other.uc; } -inline bool operator<(const NameprepCaseFoldingEntry &one, int other) +inline bool operator<(const NameprepCaseFoldingEntry &one, uint other) { return one.uc < other; } static const NameprepCaseFoldingEntry NameprepCaseFolding[] = { @@ -1862,45 +1862,44 @@ static const NameprepCaseFoldingEntry NameprepCaseFolding[] = { { 0xFF38, { 0xFF58, 0x0000, 0x0000, 0x0000 } }, { 0xFF39, { 0xFF59, 0x0000, 0x0000, 0x0000 } }, { 0xFF3A, { 0xFF5A, 0x0000, 0x0000, 0x0000 } }, - // ##### -/* { 0x10400, { 0x10428, 0x0000, 0x0000, 0x0000 } }, - { 0x10401, { 0x10429, 0x0000, 0x0000, 0x0000 } }, - { 0x10402, { 0x1042A, 0x0000, 0x0000, 0x0000 } }, - { 0x10403, { 0x1042B, 0x0000, 0x0000, 0x0000 } }, - { 0x10404, { 0x1042C, 0x0000, 0x0000, 0x0000 } }, - { 0x10405, { 0x1042D, 0x0000, 0x0000, 0x0000 } }, - { 0x10406, { 0x1042E, 0x0000, 0x0000, 0x0000 } }, - { 0x10407, { 0x1042F, 0x0000, 0x0000, 0x0000 } }, - { 0x10408, { 0x10430, 0x0000, 0x0000, 0x0000 } }, - { 0x10409, { 0x10431, 0x0000, 0x0000, 0x0000 } }, - { 0x1040A, { 0x10432, 0x0000, 0x0000, 0x0000 } }, - { 0x1040B, { 0x10433, 0x0000, 0x0000, 0x0000 } }, - { 0x1040C, { 0x10434, 0x0000, 0x0000, 0x0000 } }, - { 0x1040D, { 0x10435, 0x0000, 0x0000, 0x0000 } }, - { 0x1040E, { 0x10436, 0x0000, 0x0000, 0x0000 } }, - { 0x1040F, { 0x10437, 0x0000, 0x0000, 0x0000 } }, - { 0x10410, { 0x10438, 0x0000, 0x0000, 0x0000 } }, - { 0x10411, { 0x10439, 0x0000, 0x0000, 0x0000 } }, - { 0x10412, { 0x1043A, 0x0000, 0x0000, 0x0000 } }, - { 0x10413, { 0x1043B, 0x0000, 0x0000, 0x0000 } }, - { 0x10414, { 0x1043C, 0x0000, 0x0000, 0x0000 } }, - { 0x10415, { 0x1043D, 0x0000, 0x0000, 0x0000 } }, - { 0x10416, { 0x1043E, 0x0000, 0x0000, 0x0000 } }, - { 0x10417, { 0x1043F, 0x0000, 0x0000, 0x0000 } }, - { 0x10418, { 0x10440, 0x0000, 0x0000, 0x0000 } }, - { 0x10419, { 0x10441, 0x0000, 0x0000, 0x0000 } }, - { 0x1041A, { 0x10442, 0x0000, 0x0000, 0x0000 } }, - { 0x1041B, { 0x10443, 0x0000, 0x0000, 0x0000 } }, - { 0x1041C, { 0x10444, 0x0000, 0x0000, 0x0000 } }, - { 0x1041D, { 0x10445, 0x0000, 0x0000, 0x0000 } }, - { 0x1041E, { 0x10446, 0x0000, 0x0000, 0x0000 } }, - { 0x1041F, { 0x10447, 0x0000, 0x0000, 0x0000 } }, - { 0x10420, { 0x10448, 0x0000, 0x0000, 0x0000 } }, - { 0x10421, { 0x10449, 0x0000, 0x0000, 0x0000 } }, - { 0x10422, { 0x1044A, 0x0000, 0x0000, 0x0000 } }, - { 0x10423, { 0x1044B, 0x0000, 0x0000, 0x0000 } }, - { 0x10424, { 0x1044C, 0x0000, 0x0000, 0x0000 } }, - { 0x10425, { 0x1044D, 0x0000, 0x0000, 0x0000 } },*/ + { 0x10400, { 0xd801, 0xdc28, 0x0000, 0x0000 } }, + { 0x10401, { 0xd801, 0xdc29, 0x0000, 0x0000 } }, + { 0x10402, { 0xd801, 0xdc2A, 0x0000, 0x0000 } }, + { 0x10403, { 0xd801, 0xdc2B, 0x0000, 0x0000 } }, + { 0x10404, { 0xd801, 0xdc2C, 0x0000, 0x0000 } }, + { 0x10405, { 0xd801, 0xdc2D, 0x0000, 0x0000 } }, + { 0x10406, { 0xd801, 0xdc2E, 0x0000, 0x0000 } }, + { 0x10407, { 0xd801, 0xdc2F, 0x0000, 0x0000 } }, + { 0x10408, { 0xd801, 0xdc30, 0x0000, 0x0000 } }, + { 0x10409, { 0xd801, 0xdc31, 0x0000, 0x0000 } }, + { 0x1040A, { 0xd801, 0xdc32, 0x0000, 0x0000 } }, + { 0x1040B, { 0xd801, 0xdc33, 0x0000, 0x0000 } }, + { 0x1040C, { 0xd801, 0xdc34, 0x0000, 0x0000 } }, + { 0x1040D, { 0xd801, 0xdc35, 0x0000, 0x0000 } }, + { 0x1040E, { 0xd801, 0xdc36, 0x0000, 0x0000 } }, + { 0x1040F, { 0xd801, 0xdc37, 0x0000, 0x0000 } }, + { 0x10410, { 0xd801, 0xdc38, 0x0000, 0x0000 } }, + { 0x10411, { 0xd801, 0xdc39, 0x0000, 0x0000 } }, + { 0x10412, { 0xd801, 0xdc3A, 0x0000, 0x0000 } }, + { 0x10413, { 0xd801, 0xdc3B, 0x0000, 0x0000 } }, + { 0x10414, { 0xd801, 0xdc3C, 0x0000, 0x0000 } }, + { 0x10415, { 0xd801, 0xdc3D, 0x0000, 0x0000 } }, + { 0x10416, { 0xd801, 0xdc3E, 0x0000, 0x0000 } }, + { 0x10417, { 0xd801, 0xdc3F, 0x0000, 0x0000 } }, + { 0x10418, { 0xd801, 0xdc40, 0x0000, 0x0000 } }, + { 0x10419, { 0xd801, 0xdc41, 0x0000, 0x0000 } }, + { 0x1041A, { 0xd801, 0xdc42, 0x0000, 0x0000 } }, + { 0x1041B, { 0xd801, 0xdc43, 0x0000, 0x0000 } }, + { 0x1041C, { 0xd801, 0xdc44, 0x0000, 0x0000 } }, + { 0x1041D, { 0xd801, 0xdc45, 0x0000, 0x0000 } }, + { 0x1041E, { 0xd801, 0xdc46, 0x0000, 0x0000 } }, + { 0x1041F, { 0xd801, 0xdc47, 0x0000, 0x0000 } }, + { 0x10420, { 0xd801, 0xdc48, 0x0000, 0x0000 } }, + { 0x10421, { 0xd801, 0xdc49, 0x0000, 0x0000 } }, + { 0x10422, { 0xd801, 0xdc4A, 0x0000, 0x0000 } }, + { 0x10423, { 0xd801, 0xdc4B, 0x0000, 0x0000 } }, + { 0x10424, { 0xd801, 0xdc4C, 0x0000, 0x0000 } }, + { 0x10425, { 0xd801, 0xdc4D, 0x0000, 0x0000 } }, { 0x1D400, { 0x0061, 0x0000, 0x0000, 0x0000 } }, { 0x1D401, { 0x0062, 0x0000, 0x0000, 0x0000 } }, { 0x1D402, { 0x0063, 0x0000, 0x0000, 0x0000 } }, @@ -2355,17 +2354,23 @@ static void mapToLowerCase(QString *str, int from) { int N = sizeof(NameprepCaseFolding) / sizeof(NameprepCaseFolding[0]); - QChar *d = 0; + ushort *d = 0; for (int i = from; i < str->size(); ++i) { - int uc = str->at(i).unicode(); + uint uc = str->at(i).unicode(); if (uc < 0x80) { if (uc <= 'Z' && uc >= 'A') { - uc |= 0x20; if (!d) - d = str->data(); - d[i] = QChar(uc); + d = reinterpret_cast<ushort *>(str->data()); + d[i] = (uc | 0x20); } } else { + if (QChar(uc).isHighSurrogate() && i < str->size() - 1) { + ushort low = str->at(i + 1).unicode(); + if (QChar(low).isLowSurrogate()) { + uc = QChar::surrogateToUcs4(uc, low); + ++i; + } + } const NameprepCaseFoldingEntry *entry = qBinaryFind(NameprepCaseFolding, NameprepCaseFolding + N, uc); @@ -2374,23 +2379,26 @@ static void mapToLowerCase(QString *str, int from) while (l < 4 && entry->mapping[l]) ++l; if (l > 1) { - str->replace(i, 1, (const QChar *)&entry->mapping[0], l); + if (uc <= 0xffff) + str->replace(i, 1, reinterpret_cast<const QChar *>(&entry->mapping[0]), l); + else + str->replace(i-1, 2, reinterpret_cast<const QChar *>(&entry->mapping[0]), l); d = 0; } else { if (!d) - d = str->data(); - d[i] = QChar(entry->mapping[0]); + d = reinterpret_cast<ushort *>(str->data()); + d[i] = entry->mapping[0]; } } } } } -static bool isMappedToNothing(const QChar &ch) +static bool isMappedToNothing(uint uc) { - if (ch.unicode() < 0xad) + if (uc < 0xad) return false; - switch (ch.unicode()) { + switch (uc) { case 0x00AD: case 0x034F: case 0x1806: case 0x180B: case 0x180C: case 0x180D: case 0x200B: case 0x200C: case 0x200D: case 0x2060: case 0xFE00: case 0xFE01: case 0xFE02: case 0xFE03: case 0xFE04: case 0xFE05: case 0xFE06: case 0xFE07: @@ -2409,66 +2417,72 @@ static void stripProhibitedOutput(QString *str, int from) const ushort *in = out; const ushort *end = (ushort *)str->data() + str->size(); while (in < end) { - ushort uc = *in; - if (uc < 0x80 || - !(uc <= 0x009F - || uc == 0x00A0 - || uc == 0x0340 - || uc == 0x0341 - || uc == 0x06DD - || uc == 0x070F - || uc == 0x1680 - || uc == 0x180E - || (uc >= 0x2000 && uc <= 0x200B) - || uc == 0x200C - || uc == 0x200D - || uc == 0x200E - || uc == 0x200F - || (uc >= 0x2028 && uc <= 0x202F) - || uc == 0x205F - || (uc >= 0x2060 && uc <= 0x2063) - || uc == 0x206A - || (uc >= 0x206A && uc <= 0x206F) - || (uc >= 0x2FF0 && uc <= 0x2FFB) - || uc == 0x3000 - || (uc >= 0xD800 && uc <= 0xDFFF) - || (uc >= 0xE000 && uc <= 0xF8FF) - || (uc >= 0xFDD0 && uc <= 0xFDEF) - || uc == 0xFEFF - || (uc >= 0xFFF9 && uc <= 0xFFFC) - || (uc >= 0xFFFA && (uc <= 0xFFFE || uc == 0xFFFF)) - /* ### Add NAMEPREP support for surrogates - || uc == 0xE0001 - || (uc >= 0x2FFFE && uc <= 0x2FFFF) - || (uc >= 0x1D173 && uc <= 0x1D17A) - || (uc >= 0x1FFFE && uc <= 0x1FFFF) - || (uc >= 0x3FFFE && uc <= 0x3FFFF) - || (uc >= 0x4FFFE && uc <= 0x4FFFF) - || (uc >= 0x5FFFE && uc <= 0x5FFFF) - || (uc >= 0x6FFFE && uc <= 0x6FFFF) - || (uc >= 0x7FFFE && uc <= 0x7FFFF) - || (uc >= 0x8FFFE && uc <= 0x8FFFF) - || (uc >= 0x9FFFE && uc <= 0x9FFFF) - || (uc >= 0xAFFFE && uc <= 0xAFFFF) - || (uc >= 0xBFFFE && uc <= 0xBFFFF) - || (uc >= 0xCFFFE && uc <= 0xCFFFF) - || (uc >= 0xDFFFE && uc <= 0xDFFFF) - || (uc >= 0xE0020 && uc <= 0xE007F) - || (uc >= 0xEFFFE && uc <= 0xEFFFF) - || (uc >= 0xF0000 && uc <= 0xFFFFD) - || (uc >= 0xFFFFE && uc <= 0xFFFFF) - || (uc >= 0x100000 && uc <= 0x10FFFD) - || (uc >= 0x10FFFE && uc <= 0x10FFFF)*/)) - *out++ = *in; + uint uc = *in; + if (QChar(uc).isHighSurrogate() && in < end - 1) { + ushort low = *(in + 1); + if (QChar(low).isLowSurrogate()) { + ++in; + uc = QChar::surrogateToUcs4(uc, low); + } + } + if (uc <= 0xFFFF) { + if (uc < 0x80 || + !(uc <= 0x009F + || uc == 0x00A0 + || uc == 0x0340 + || uc == 0x0341 + || uc == 0x06DD + || uc == 0x070F + || uc == 0x1680 + || uc == 0x180E + || (uc >= 0x2000 && uc <= 0x200F) + || (uc >= 0x2028 && uc <= 0x202F) + || uc == 0x205F + || (uc >= 0x2060 && uc <= 0x2063) + || (uc >= 0x206A && uc <= 0x206F) + || (uc >= 0x2FF0 && uc <= 0x2FFB) + || uc == 0x3000 + || (uc >= 0xD800 && uc <= 0xDFFF) + || (uc >= 0xE000 && uc <= 0xF8FF) + || (uc >= 0xFDD0 && uc <= 0xFDEF) + || uc == 0xFEFF + || (uc >= 0xFFF9 && uc <= 0xFFFF))) { + *out++ = *in; + } + } else { + if (!((uc >= 0x1D173 && uc <= 0x1D17A) + || (uc >= 0x1FFFE && uc <= 0x1FFFF) + || (uc >= 0x2FFFE && uc <= 0x2FFFF) + || (uc >= 0x3FFFE && uc <= 0x3FFFF) + || (uc >= 0x4FFFE && uc <= 0x4FFFF) + || (uc >= 0x5FFFE && uc <= 0x5FFFF) + || (uc >= 0x6FFFE && uc <= 0x6FFFF) + || (uc >= 0x7FFFE && uc <= 0x7FFFF) + || (uc >= 0x8FFFE && uc <= 0x8FFFF) + || (uc >= 0x9FFFE && uc <= 0x9FFFF) + || (uc >= 0xAFFFE && uc <= 0xAFFFF) + || (uc >= 0xBFFFE && uc <= 0xBFFFF) + || (uc >= 0xCFFFE && uc <= 0xCFFFF) + || (uc >= 0xDFFFE && uc <= 0xDFFFF) + || uc == 0xE0001 + || (uc >= 0xE0020 && uc <= 0xE007F) + || (uc >= 0xEFFFE && uc <= 0xEFFFF) + || (uc >= 0xF0000 && uc <= 0xFFFFD) + || (uc >= 0xFFFFE && uc <= 0xFFFFF) + || (uc >= 0x100000 && uc <= 0x10FFFD) + || (uc >= 0x10FFFE && uc <= 0x10FFFF))) { + *out++ = QChar::highSurrogate(uc); + *out++ = QChar::lowSurrogate(uc); + } + } ++in; } if (in != out) str->truncate(out - str->utf16()); } -static bool isBidirectionalRorAL(const QChar &c) +static bool isBidirectionalRorAL(uint uc) { - ushort uc = c.unicode(); if (uc < 0x5b0) return false; return uc == 0x05BE @@ -2507,9 +2521,8 @@ static bool isBidirectionalRorAL(const QChar &c) || (uc >= 0xFE76 && uc <= 0xFEFC); } -static bool isBidirectionalL(const QChar &ch) +static bool isBidirectionalL(uint uc) { - ushort uc = ch.unicode(); if (uc < 0xaa) return (uc >= 0x0041 && uc <= 0x005A) || (uc >= 0x0061 && uc <= 0x007A); @@ -2874,8 +2887,7 @@ static bool isBidirectionalL(const QChar &ch) return true; } - /* ### Add NAMEPREP support for surrogates - || (uc >= 0x10300 && uc <= 0x1031E) + if ((uc >= 0x10300 && uc <= 0x1031E) || (uc >= 0x10320 && uc <= 0x10323) || (uc >= 0x10330 && uc <= 0x1034A) || (uc >= 0x10400 && uc <= 0x10425) @@ -2911,7 +2923,9 @@ static bool isBidirectionalL(const QChar &ch) || (uc >= 0x20000 && uc <= 0x2A6D6) || (uc >= 0x2F800 && uc <= 0x2FA1D) || (uc >= 0xF0000 && uc <= 0xFFFFD) - || (uc >= 0x100000 && uc <= 0x10FFFD)*/ + || (uc >= 0x100000 && uc <= 0x10FFFD)) { + return true; + } return false; } @@ -2944,13 +2958,37 @@ void qt_nameprep(QString *source, int from) return; // everything was mapped easily (lowercased, actually) int firstNonAscii = out - src; + // Characters unassigned in Unicode 3.2 are not allowed in "stored string" scheme + // but allowed in "query" scheme + // (Table A.1) + const bool isUnassignedAllowed = false; // ### // Characters commonly mapped to nothing are simply removed // (Table B.1) const QChar *in = out; - while (in < e) { - if (!isMappedToNothing(*in)) - *out++ = *in; - ++in; + for ( ; in < e; ++in) { + uint uc = in->unicode(); + if (QChar(uc).isHighSurrogate() && in < e - 1) { + ushort low = in[1].unicode(); + if (QChar(low).isLowSurrogate()) { + ++in; + uc = QChar::surrogateToUcs4(uc, low); + } + } + if (!isUnassignedAllowed) { + QChar::UnicodeVersion version = QChar::unicodeVersion(uc); + if (version == QChar::Unicode_Unassigned || version > QChar::Unicode_3_2) { + source->resize(from); // not allowed, clear the label + return; + } + } + if (!isMappedToNothing(uc)) { + if (uc <= 0xFFFF) { + *out++ = *in; + } else { + *out++ = QChar::highSurrogate(uc); + *out++ = QChar::lowSurrogate(uc); + } + } } if (out != in) source->truncate(out - src); @@ -2961,7 +2999,8 @@ void qt_nameprep(QString *source, int from) // Normalize to Unicode 3.2 form KC extern void qt_string_normalize(QString *data, QString::NormalizationForm mode, QChar::UnicodeVersion version, int from); - qt_string_normalize(source, QString::NormalizationForm_KC, QChar::Unicode_3_2, firstNonAscii); + qt_string_normalize(source, QString::NormalizationForm_KC, QChar::Unicode_3_2, + firstNonAscii > from ? firstNonAscii - 1 : from); // Strip prohibited output stripProhibitedOutput(source, firstNonAscii); @@ -2972,14 +3011,22 @@ void qt_nameprep(QString *source, int from) src = source->data(); e = src + source->size(); for (in = src + from; in < e && (!containsLCat || !containsRandALCat); ++in) { - if (isBidirectionalL(*in)) + uint uc = in->unicode(); + if (QChar(uc).isHighSurrogate() && in < e - 1) { + ushort low = in[1].unicode(); + if (QChar(low).isLowSurrogate()) { + ++in; + uc = QChar::surrogateToUcs4(uc, low); + } + } + if (isBidirectionalL(uc)) containsLCat = true; - else if (isBidirectionalRorAL(*in)) + else if (isBidirectionalRorAL(uc)) containsRandALCat = true; } if (containsRandALCat) { - if (containsLCat || (!isBidirectionalRorAL(src[from]) - || !isBidirectionalRorAL(e[-1]))) + if (containsLCat || (!isBidirectionalRorAL(src[from].unicode()) + || !isBidirectionalRorAL(e[-1].unicode()))) source->resize(from); // not allowed, clear the label } } diff --git a/tests/auto/qurl/tst_qurl.cpp b/tests/auto/qurl/tst_qurl.cpp index d055b6b..b2b28bd 100644 --- a/tests/auto/qurl/tst_qurl.cpp +++ b/tests/auto/qurl/tst_qurl.cpp @@ -161,6 +161,8 @@ private slots: void idna_testsuite(); void nameprep_testsuite_data(); void nameprep_testsuite(); + void nameprep_highcodes_data(); + void nameprep_highcodes(); void ace_testsuite_data(); void ace_testsuite(); void std3violations_data(); @@ -2912,7 +2914,6 @@ void tst_QUrl::nameprep_testsuite_data() << QString() << 0 << 0; QTest::newRow("Case folding 8bit U+00DF (german sharp s)") -// << QString::fromUtf8("\xC3\xDF") ### typo in the original testsuite << QString::fromUtf8("\xC3\x9F") << QString("ss") << QString() << 0 << 0; @@ -2943,7 +2944,8 @@ void tst_QUrl::nameprep_testsuite_data() << QString() << 0 << 0; QTest::newRow("Self-reverting case folding U+01F0 and normalization") - << QString::fromUtf8("\xC7\xF0") +// << QString::fromUtf8("\xC7\xF0") ### typo in the original testsuite + << QString::fromUtf8("\xC7\xB0") << QString::fromUtf8("\xC7\xB0") << QString() << 0 << 0; @@ -3118,13 +3120,13 @@ void tst_QUrl::nameprep_testsuite_data() << QString("Nameprep") << STRINGPREP_NO_UNASSIGNED << STRINGPREP_CONTAINS_UNASSIGNED; QTest::newRow("Larger test (shrinking)") - << QString::fromUtf8("X\xC2\xAD\xC3\xDF\xC4\xB0\xE2\x84\xA1\x6a\xcc\x8c\xc2\xa0\xc2" + << QString::fromUtf8("X\xC2\xAD\xC3\x9F\xC4\xB0\xE2\x84\xA1\x6a\xcc\x8c\xc2\xa0\xc2" "\xaa\xce\xb0\xe2\x80\x80") << QString::fromUtf8("xssi\xcc\x87""tel\xc7\xb0 a\xce\xb0 ") << QString("Nameprep") << 0 << 0; QTest::newRow("Larger test (expanding)") - << QString::fromUtf8("X\xC3\xDF\xe3\x8c\x96\xC4\xB0\xE2\x84\xA1\xE2\x92\x9F\xE3\x8c\x80") + << QString::fromUtf8("X\xC3\x9F\xe3\x8c\x96\xC4\xB0\xE2\x84\xA1\xE2\x92\x9F\xE3\x8c\x80") << QString::fromUtf8("xss\xe3\x82\xad\xe3\x83\xad\xe3\x83\xa1\xe3\x83\xbc\xe3\x83\x88" "\xe3\x83\xab""i\xcc\x87""tel\x28""d\x29\xe3\x82\xa2\xe3\x83\x91" "\xe3\x83\xbc\xe3\x83\x88") @@ -3145,20 +3147,58 @@ void tst_QUrl::nameprep_testsuite() QFETCH(QString, out); QFETCH(QString, profile); - QEXPECT_FAIL("Case folding U+2121 U+33C6 U+1D7BB", - ">0xffff unicode points are not supported", Continue); - QEXPECT_FAIL("Self-reverting case folding U+01F0 and normalization", - "Investigate further", Continue); QEXPECT_FAIL("Left-to-right mark U+200E", "Investigate further", Continue); QEXPECT_FAIL("Deprecated U+202A", "Investigate further", Continue); QEXPECT_FAIL("Language tagging character U+E0001", "Investigate further", Continue); - QEXPECT_FAIL("Larger test (shrinking)", - "Investigate further", Continue); - QEXPECT_FAIL("Larger test (expanding)", - "Investigate further", Continue); + qt_nameprep(&in, 0); + QCOMPARE(in, out); +#endif +} + +void tst_QUrl::nameprep_highcodes_data() +{ + QTest::addColumn<QString>("in"); + QTest::addColumn<QString>("out"); + QTest::addColumn<QString>("profile"); + QTest::addColumn<int>("flags"); + QTest::addColumn<int>("rc"); + + { + QChar st[] = { '-', 0xd801, 0xdc1d, 'a' }; + QChar se[] = { '-', 0xd801, 0xdc45, 'a' }; + QTest::newRow("highcodes (U+1041D)") + << QString(st, sizeof(st)/sizeof(st[0])) + << QString(se, sizeof(se)/sizeof(se[0])) + << QString() << 0 << 0; + } + { + QChar st[] = { 0x011C, 0xd835, 0xdf6e, 0x0110 }; + QChar se[] = { 0x011D, 0x03C9, 0x0111 }; + QTest::newRow("highcodes (U+1D76E)") + << QString(st, sizeof(st)/sizeof(st[0])) + << QString(se, sizeof(se)/sizeof(se[0])) + << QString() << 0 << 0; + } + { + QChar st[] = { 'D', 0xdb40, 0xdc20, 'o', 0xd834, 0xdd7a, '\'', 0x2060, 'h' }; + QChar se[] = { 'd', 'o', '\'', 'h' }; + QTest::newRow("highcodes (D, U+E0020, o, U+1D17A, ', U+2060, h)") + << QString(st, sizeof(st)/sizeof(st[0])) + << QString(se, sizeof(se)/sizeof(se[0])) + << QString() << 0 << 0; + } +} + +void tst_QUrl::nameprep_highcodes() +{ +#ifdef QT_BUILD_INTERNAL + QFETCH(QString, in); + QFETCH(QString, out); + QFETCH(QString, profile); + qt_nameprep(&in, 0); QCOMPARE(in, out); #endif diff --git a/tests/auto/utf8/tst_utf8.cpp b/tests/auto/utf8/tst_utf8.cpp index 7bbbfab..9b6b8c1 100644 --- a/tests/auto/utf8/tst_utf8.cpp +++ b/tests/auto/utf8/tst_utf8.cpp @@ -210,7 +210,9 @@ void tst_Utf8::invalidUtf8_data() QTest::addColumn<QByteArray>("utf8"); QTest::newRow("1char") << QByteArray("\x80"); - QTest::newRow("2chars") << QByteArray("\xC2\xC0"); + QTest::newRow("2chars-1") << QByteArray("\xC2\xC0"); + QTest::newRow("2chars-2") << QByteArray("\xC3\xDF"); + QTest::newRow("2chars-3") << QByteArray("\xC7\xF0"); QTest::newRow("3chars-1") << QByteArray("\xE0\xA0\xC0"); QTest::newRow("3chars-2") << QByteArray("\xE0\xC0\xA0"); QTest::newRow("4chars-1") << QByteArray("\xF0\x90\x80\xC0"); |