diff options
Diffstat (limited to 'src/corelib/tools/qbytearray.cpp')
| -rw-r--r-- | src/corelib/tools/qbytearray.cpp | 295 |
1 files changed, 279 insertions, 16 deletions
diff --git a/src/corelib/tools/qbytearray.cpp b/src/corelib/tools/qbytearray.cpp index 424420204a..10e9b22161 100644 --- a/src/corelib/tools/qbytearray.cpp +++ b/src/corelib/tools/qbytearray.cpp @@ -47,6 +47,7 @@ #include "qlocale_p.h" #include "qlocale_tools_p.h" #include "private/qnumeric_p.h" +#include "private/qsimd_p.h" #include "qstringalgorithms_p.h" #include "qscopedpointer.h" #include "qbytearray_p.h" @@ -356,7 +357,8 @@ char *qstrncpy(char *dst, const char *src, uint len) Special case 2: Returns an arbitrary non-zero value if \a str1 is nullptr or \a str2 is nullptr (but not both). - \sa qstrncmp(), qstricmp(), qstrnicmp(), {8-bit Character Comparisons} + \sa qstrncmp(), qstricmp(), qstrnicmp(), {8-bit Character Comparisons}, + QByteArray::compare() */ int qstrcmp(const char *str1, const char *str2) { @@ -381,7 +383,8 @@ int qstrcmp(const char *str1, const char *str2) Special case 2: Returns a random non-zero value if \a str1 is nullptr or \a str2 is nullptr (but not both). - \sa qstrcmp(), qstricmp(), qstrnicmp(), {8-bit Character Comparisons} + \sa qstrcmp(), qstricmp(), qstrnicmp(), {8-bit Character Comparisons}, + QByteArray::compare() */ /*! \relates QByteArray @@ -400,21 +403,80 @@ int qstrcmp(const char *str1, const char *str2) Special case 2: Returns a random non-zero value if \a str1 is nullptr or \a str2 is nullptr (but not both). - \sa qstrcmp(), qstrncmp(), qstrnicmp(), {8-bit Character Comparisons} + \sa qstrcmp(), qstrncmp(), qstrnicmp(), {8-bit Character Comparisons}, + QByteArray::compare() */ int qstricmp(const char *str1, const char *str2) { const uchar *s1 = reinterpret_cast<const uchar *>(str1); const uchar *s2 = reinterpret_cast<const uchar *>(str2); - int res; - uchar c; - if (!s1 || !s2) - return s1 ? 1 : (s2 ? -1 : 0); - for (; !(res = (c = latin1_lowercased[*s1]) - latin1_lowercased[*s2]); s1++, s2++) - if (!c) // strings are equal - break; - return res; + if (!s1) + return s2 ? -1 : 0; + if (!s2) + return 1; + + enum { Incomplete = 256 }; + qptrdiff offset = 0; + auto innerCompare = [=, &offset](qptrdiff max, bool unlimited) { + max += offset; + do { + uchar c = latin1_lowercased[s1[offset]]; + int res = c - latin1_lowercased[s2[offset]]; + if (Q_UNLIKELY(res)) + return res; + if (Q_UNLIKELY(!c)) + return 0; + ++offset; + } while (unlimited || offset < max); + return int(Incomplete); + }; + +#ifdef __SSE4_1__ + enum { PageSize = 4096, PageMask = PageSize - 1 }; + const __m128i zero = _mm_setzero_si128(); + forever { + // Calculate how many bytes we can load until we cross a page boundary + // for either source. This isn't an exact calculation, just something + // very quick. + quintptr u1 = quintptr(s1 + offset); + quintptr u2 = quintptr(s2 + offset); + uint n = PageSize - ((u1 | u2) & PageMask); + + qptrdiff maxoffset = offset + n; + for ( ; offset + 16 <= maxoffset; offset += sizeof(__m128i)) { + // load 16 bytes from either source + __m128i a = _mm_loadu_si128(reinterpret_cast<const __m128i *>(s1 + offset)); + __m128i b = _mm_loadu_si128(reinterpret_cast<const __m128i *>(s2 + offset)); + + // compare the two against each oher + __m128i cmp = _mm_cmpeq_epi8(a, b); + + // find NUL terminators too + cmp = _mm_min_epu8(cmp, a); + cmp = _mm_cmpeq_epi8(cmp, zero); + + // was there any difference or a NUL? + uint mask = _mm_movemask_epi8(cmp); + if (mask) { + // yes, find out where + uint start = qCountTrailingZeroBits(mask); + uint end = sizeof(mask) * 8 - qCountLeadingZeroBits(mask); + Q_ASSUME(end >= start); + offset += start; + n = end - start; + break; + } + } + + // using SIMD could cause a page fault, so iterate byte by byte + int res = innerCompare(n, false); + if (res != Incomplete) + return res; + } +#endif + + return innerCompare(-1, true); } /*! \relates QByteArray @@ -434,7 +496,8 @@ int qstricmp(const char *str1, const char *str2) Special case 2: Returns a random non-zero value if \a str1 is nullptr or \a str2 is nullptr (but not both). - \sa qstrcmp(), qstrncmp(), qstricmp(), {8-bit Character Comparisons} + \sa qstrcmp(), qstrncmp(), qstricmp(), {8-bit Character Comparisons}, + QByteArray::compare() */ int qstrnicmp(const char *str1, const char *str2, uint len) @@ -456,6 +519,55 @@ int qstrnicmp(const char *str1, const char *str2, uint len) /*! \internal + \since 5.12 + + A helper for QByteArray::compare. Compares \a len1 bytes from \a str1 to \a + len2 bytes from \a str2. If \a len2 is -1, then \a str2 is expected to be + null-terminated. + */ +int qstrnicmp(const char *str1, qsizetype len1, const char *str2, qsizetype len2) +{ + Q_ASSERT(str1); + Q_ASSERT(len1 >= 0); + Q_ASSERT(len2 >= -1); + const uchar *s1 = reinterpret_cast<const uchar *>(str1); + const uchar *s2 = reinterpret_cast<const uchar *>(str2); + if (!s2) + return len1 == 0 ? 0 : 1; + + int res; + uchar c; + if (len2 == -1) { + // null-terminated str2 + qsizetype i; + for (i = 0; i < len1; ++i) { + c = latin1_lowercased[s2[i]]; + if (!c) + return 1; + + res = latin1_lowercased[s1[i]] - c; + if (res) + return res; + } + c = latin1_lowercased[s2[i]]; + return c ? -1 : 0; + } else { + // not null-terminated + for (qsizetype i = 0; i < qMin(len1, len2); ++i) { + c = latin1_lowercased[s2[i]]; + res = latin1_lowercased[s1[i]] - c; + if (res) + return res; + } + if (len1 == len2) + return 0; + return len1 < len2 ? -1 : 1; + } +} + +/*! + \internal + ### Qt6: replace the QByteArray parameter with [pointer,len] pair */ int qstrcmp(const QByteArray &str1, const char *str2) { @@ -483,6 +595,7 @@ int qstrcmp(const QByteArray &str1, const char *str2) /*! \internal + ### Qt6: replace the QByteArray parameter with [pointer,len] pair */ int qstrcmp(const QByteArray &str1, const QByteArray &str2) { @@ -950,7 +1063,7 @@ QByteArray qUncompress(const uchar* data, int nbytes) $LC_CTYPE is set, most Unix systems do "the right thing".) Functions that this affects include contains(), indexOf(), lastIndexOf(), operator<(), operator<=(), operator>(), - operator>=(), toLower() and toUpper(). + operator>=(), isLower(), isUpper(), toLower() and toUpper(). This issue does not apply to \l{QString}s since they represent characters using Unicode. @@ -2864,6 +2977,31 @@ int QByteArray::count(char ch) const */ /*! + \fn int QByteArray::compare(const char *c, Qt::CaseSensitivity cs = Qt::CaseSensitive) const + \since 5.12 + + Returns an integer less than, equal to, or greater than zero depending on + whether this QByteArray sorts before, at the same position, or after the + string pointed to by \a c. The comparison is performed according to case + sensitivity \a cs. + + \sa operator== +*/ + +/*! + \fn int QByteArray::compare(const QByteArray &a, Qt::CaseSensitivity cs = Qt::CaseSensitive) const + \overload + \since 5.12 + + Returns an integer less than, equal to, or greater than zero depending on + whether this QByteArray sorts before, at the same position, or after the + QByteArray \a a. The comparison is performed according to case sensitivity + \a cs. + + \sa operator== +*/ + +/*! Returns \c true if this byte array starts with byte array \a ba; otherwise returns \c false. @@ -2941,6 +3079,78 @@ bool QByteArray::endsWith(const char *str) const return qstrncmp(d->data() + d->size - len, str, len) == 0; } +/*! + Returns true if \a c is an uppercase Latin1 letter. + \note The multiplication sign 0xD7 and the sz ligature 0xDF are not + treated as uppercase Latin1. + */ +static inline bool isUpperCaseLatin1(char c) +{ + if (c >= 'A' && c <= 'Z') + return true; + + return (uchar(c) >= 0xC0 && uchar(c) <= 0xDE && uchar(c) != 0xD7); +} + +/*! + Returns \c true if this byte array contains only uppercase letters, + otherwise returns \c false. The byte array is interpreted as a Latin-1 + encoded string. + \since 5.12 + + \sa isLower(), toUpper() +*/ +bool QByteArray::isUpper() const +{ + if (isEmpty()) + return false; + + const char *d = data(); + + for (int i = 0, max = size(); i < max; ++i) { + if (!isUpperCaseLatin1(d[i])) + return false; + } + + return true; +} + +/*! + Returns true if \a c is an lowercase Latin1 letter. + \note The division sign 0xF7 is not treated as lowercase Latin1, + but the small y dieresis 0xFF is. + */ +static inline bool isLowerCaseLatin1(char c) +{ + if (c >= 'a' && c <= 'z') + return true; + + return (uchar(c) >= 0xD0 && uchar(c) != 0xF7); +} + +/*! + Returns \c true if this byte array contains only lowercase letters, + otherwise returns \c false. The byte array is interpreted as a Latin-1 + encoded string. + \since 5.12 + + \sa isUpper(), toLower() + */ +bool QByteArray::isLower() const +{ + if (isEmpty()) + return false; + + const char *d = data(); + + for (int i = 0, max = size(); i < max; ++i) { + if (!isLowerCaseLatin1(d[i])) + return false; + } + + return true; +} + /*! \overload Returns \c true if this byte array ends with character \a ch; @@ -3052,7 +3262,7 @@ QByteArray QByteArray::mid(int pos, int len) const Example: \snippet code/src_corelib_tools_qbytearray.cpp 30 - \sa toUpper(), {8-bit Character Comparisons} + \sa isLower(), toUpper(), {8-bit Character Comparisons} */ // prevent the compiler from inlining the function in each of @@ -3106,7 +3316,7 @@ QByteArray QByteArray::toLower_helper(QByteArray &a) Example: \snippet code/src_corelib_tools_qbytearray.cpp 31 - \sa toLower(), {8-bit Character Comparisons} + \sa isUpper(), toLower(), {8-bit Character Comparisons} */ QByteArray QByteArray::toUpper_helper(const QByteArray &a) @@ -3295,6 +3505,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if byte array \a a1 is equal to byte array \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator==(const QByteArray &a1, const char *a2) @@ -3304,6 +3516,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if byte array \a a1 is equal to string \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator==(const char *a1, const QByteArray &a2) @@ -3313,6 +3527,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if string \a a1 is equal to byte array \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator!=(const QByteArray &a1, const QByteArray &a2) @@ -3322,6 +3538,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if byte array \a a1 is not equal to byte array \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator!=(const QByteArray &a1, const char *a2) @@ -3331,6 +3549,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if byte array \a a1 is not equal to string \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator!=(const char *a1, const QByteArray &a2) @@ -3340,6 +3560,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if string \a a1 is not equal to byte array \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator<(const QByteArray &a1, const QByteArray &a2) @@ -3349,6 +3571,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if byte array \a a1 is lexically less than byte array \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn inline bool operator<(const QByteArray &a1, const char *a2) @@ -3358,6 +3582,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if byte array \a a1 is lexically less than string \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator<(const char *a1, const QByteArray &a2) @@ -3367,6 +3593,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if string \a a1 is lexically less than byte array \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator<=(const QByteArray &a1, const QByteArray &a2) @@ -3376,6 +3604,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if byte array \a a1 is lexically less than or equal to byte array \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator<=(const QByteArray &a1, const char *a2) @@ -3385,6 +3615,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if byte array \a a1 is lexically less than or equal to string \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator<=(const char *a1, const QByteArray &a2) @@ -3394,6 +3626,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if string \a a1 is lexically less than or equal to byte array \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator>(const QByteArray &a1, const QByteArray &a2) @@ -3403,6 +3637,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if byte array \a a1 is lexically greater than byte array \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator>(const QByteArray &a1, const char *a2) @@ -3412,6 +3648,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if byte array \a a1 is lexically greater than string \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator>(const char *a1, const QByteArray &a2) @@ -3421,6 +3659,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if string \a a1 is lexically greater than byte array \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator>=(const QByteArray &a1, const QByteArray &a2) @@ -3430,6 +3670,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if byte array \a a1 is lexically greater than or equal to byte array \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator>=(const QByteArray &a1, const char *a2) @@ -3439,6 +3681,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if byte array \a a1 is lexically greater than or equal to string \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn bool operator>=(const char *a1, const QByteArray &a2) @@ -3448,6 +3692,8 @@ QDataStream &operator>>(QDataStream &in, QByteArray &ba) Returns \c true if string \a a1 is lexically greater than or equal to byte array \a a2; otherwise returns \c false. + + \sa QByteArray::compare() */ /*! \fn const QByteArray operator+(const QByteArray &a1, const QByteArray &a2) @@ -3878,9 +4124,16 @@ ushort QByteArray::toUShort(bool *ok, int base) const \snippet code/src_corelib_tools_qbytearray.cpp 38 + \warning The QByteArray content may only contain valid numerical characters + which includes the plus/minus sign, the character e used in scientific + notation, and the decimal point. Including the unit or additional characters + leads to a conversion error. + \note The conversion of the number is performed in the default C locale, irrespective of the user's locale. + This function ignores leading and trailing whitespace. + \sa number() */ @@ -3889,7 +4142,8 @@ double QByteArray::toDouble(bool *ok) const QByteArray nulled = nulTerminated(); bool nonNullOk = false; int processed = 0; - double d = asciiToDouble(nulled.constData(), nulled.length(), nonNullOk, processed); + double d = qt_asciiToDouble(nulled.constData(), nulled.length(), + nonNullOk, processed, WhitespacesAllowed); if (ok) *ok = nonNullOk; return d; @@ -3903,9 +4157,18 @@ double QByteArray::toDouble(bool *ok) const If \a ok is not \c nullptr, failure is reported by setting *\a{ok} to \c false, and success by setting *\a{ok} to \c true. + \snippet code/src_corelib_tools_qbytearray.cpp 38float + + \warning The QByteArray content may only contain valid numerical characters + which includes the plus/minus sign, the character e used in scientific + notation, and the decimal point. Including the unit or additional characters + leads to a conversion error. + \note The conversion of the number is performed in the default C locale, irrespective of the user's locale. + This function ignores leading and trailing whitespace. + \sa number() */ |