Improve natural sort algorithm

1. Use proper case folding function instead of `toLower()`.
2. Use locale aware comparison instead of comparing unicode code points.
   Now `a` comes before `A` which is the same as the result from QCollator. A nice side effect
   is now it properly compares locale specific characters (for example `C`, `Č`).
3. Improve testing. Now the test is runnable and stable on all platforms.

PR  #20208.
This commit is contained in:
Chocobo1
2024-01-08 14:47:00 +08:00
committed by GitHub
parent e69f857828
commit 5b3b56c918
3 changed files with 42 additions and 34 deletions

View File

@@ -31,7 +31,6 @@
#include <QChar>
#include <QString>
#ifndef QBT_USE_QCOLLATOR
int Utils::Compare::naturalCompare(const QString &left, const QString &right, const Qt::CaseSensitivity caseSensitivity)
{
// Return value <0: `left` is smaller than `right`
@@ -45,8 +44,8 @@ int Utils::Compare::naturalCompare(const QString &left, const QString &right, co
if ((posL == left.size()) || (posR == right.size()))
return (left.size() - right.size()); // when a shorter string is another string's prefix, shorter string place before longer string
const QChar leftChar = (caseSensitivity == Qt::CaseSensitive) ? left[posL] : left[posL].toLower();
const QChar rightChar = (caseSensitivity == Qt::CaseSensitive) ? right[posR] : right[posR].toLower();
const QChar leftChar = (caseSensitivity == Qt::CaseSensitive) ? left[posL] : left[posL].toCaseFolded();
const QChar rightChar = (caseSensitivity == Qt::CaseSensitive) ? right[posR] : right[posR].toCaseFolded();
// Compare only non-digits.
// Numbers should be compared as a whole
// otherwise the string->int conversion can yield a wrong value
@@ -89,8 +88,7 @@ int Utils::Compare::naturalCompare(const QString &left, const QString &right, co
}
else
{
return (leftChar.unicode() - rightChar.unicode());
return QString::localeAwareCompare(leftChar, rightChar);
}
}
}
#endif

View File

@@ -34,24 +34,35 @@
// for QT_FEATURE_xxx, see: https://wiki.qt.io/Qt5_Build_System#How_to
#include <QtCore/private/qtcore-config_p.h>
#ifndef QBT_USE_QCOLLATOR
// macOS and Windows support 'case sensitivity' and 'numeric mode' natively
// https://github.com/qt/qtbase/blob/6.0/src/corelib/CMakeLists.txt#L777-L793
// https://github.com/qt/qtbase/blob/6.0/src/corelib/text/qcollator_macx.cpp#L74-L77
// https://github.com/qt/qtbase/blob/6.0/src/corelib/text/qcollator_win.cpp#L72-L78
#if ((QT_FEATURE_icu == 1) || defined(Q_OS_MACOS) || defined(Q_OS_WIN))
#define QBT_USE_QCOLLATOR
#define QBT_USE_QCOLLATOR 1
#include <QCollator>
#else
#define QBT_USE_QCOLLATOR 0
#endif
#endif
class QString;
namespace Utils::Compare
{
#ifdef QBT_USE_QCOLLATOR
int naturalCompare(const QString &left, const QString &right, Qt::CaseSensitivity caseSensitivity);
template <Qt::CaseSensitivity caseSensitivity>
class NaturalCompare
{
public:
#if (QBT_USE_QCOLLATOR == 0)
int operator()(const QString &left, const QString &right) const
{
return naturalCompare(left, right, caseSensitivity);
}
#else
NaturalCompare()
{
m_collator.setNumericMode(true);
@@ -65,20 +76,8 @@ namespace Utils::Compare
private:
QCollator m_collator;
};
#else
int naturalCompare(const QString &left, const QString &right, Qt::CaseSensitivity caseSensitivity);
template <Qt::CaseSensitivity caseSensitivity>
class NaturalCompare
{
public:
int operator()(const QString &left, const QString &right) const
{
return naturalCompare(left, right, caseSensitivity);
}
};
#endif
};
template <Qt::CaseSensitivity caseSensitivity>
class NaturalLessThan