Add qt_string_normalize to do in-place Unicode normalization.

This way, we can improve QUrl parsing performance by avoiding
unnecessary copies.

2 files changed, 37 insertions, 37 deletions

diff --git a/src/corelib/tools/qchar.cpp b/src/corelib/tools/qchar.cpp
index 88053d6..1558f7d 100644
--- a/src/corelib/tools/qchar.cpp
+++ b/src/corelib/tools/qchar.cpp
@@ -1421,16 +1421,15 @@ QDataStream &operator>>(QDataStream &in, QChar &chr)
 // ---------------------------------------------------------------------------
 
 
-static QString decomposeHelper
-    (const QString &str, bool canonical, QChar::UnicodeVersion version)
+static void decomposeHelper(QString *str, bool canonical, QChar::UnicodeVersion version, int from)
 {
     unsigned short buffer[3];
 
-    QString s = str;
+    QString &s = *str;
 
     const unsigned short *utf16 = s.utf16();
     const unsigned short *uc = utf16 + s.length();
-    while (uc != utf16) {
+    while (uc != utf16 + from) {
         uint ucs4 = *(--uc);
         if (QChar(ucs4).isLowSurrogate() && uc != utf16) {
             ushort high = *(uc - 1);
@@ -1453,8 +1452,6 @@ static QString decomposeHelper
         utf16 = s.utf16();
         uc = utf16 + pos + length;
     }
-
-    return s;
 }
 
 
@@ -1489,17 +1486,17 @@ static ushort ligatureHelper(ushort u1, ushort u2)
     return 0;
 }
 
-static QString composeHelper(const QString &str)
+static void composeHelper(QString *str, int from)
 {
-    QString s = str;
+    QString &s = *str;
 
-    if (s.length() < 2)
-        return s;
+    if (s.length() - from < 2)
+        return;
 
     // the loop can partly ignore high Unicode as all ligatures are in the BMP
     int starter = 0;
     int lastCombining = 0;
-    int pos = 0;
+    int pos = from;
     while (pos < s.length()) {
         uint uc = s.utf16()[pos];
         if (QChar(uc).isHighSurrogate() && pos < s.length()-1) {
@@ -1524,16 +1521,14 @@ static QString composeHelper(const QString &str)
         lastCombining = combining;
         ++pos;
     }
-    return s;
 }
 
 
-static QString canonicalOrderHelper
-    (const QString &str, QChar::UnicodeVersion version)
+static void canonicalOrderHelper(QString *str, QChar::UnicodeVersion version, int from)
 {
-    QString s = str;
+    QString &s = *str;
     const int l = s.length()-1;
-    int pos = 0;
+    int pos = from;
     while (pos < l) {
         int p2 = pos+1;
         uint u1 = s.at(pos).unicode();
@@ -1593,7 +1588,6 @@ static QString canonicalOrderHelper
                 ++pos;
         }
     }
-    return s;
 }
 
 int QT_FASTCALL QUnicodeTables::script(unsigned int uc)
diff --git a/src/corelib/tools/qstring.cpp b/src/corelib/tools/qstring.cpp
index b97ba45..99fbaa9 100644
--- a/src/corelib/tools/qstring.cpp
+++ b/src/corelib/tools/qstring.cpp
@@ -6028,6 +6028,7 @@ QString QString::repeated(int times) const
     return result;
 }
 
+void qt_string_normalize(QString *data, QString::NormalizationForm mode, QChar::UnicodeVersion version, int from);
 /*!
     \overload
     \fn QString QString::normalized(NormalizationForm mode, QChar::UnicodeVersion version) const
@@ -6037,42 +6038,48 @@ QString QString::repeated(int times) const
 */
 QString QString::normalized(QString::NormalizationForm mode, QChar::UnicodeVersion version) const
 {
+    QString copy = *this;
+    qt_string_normalize(&copy, mode, version, 0);
+    return copy;
+}
+
+void qt_string_normalize(QString *data, QString::NormalizationForm mode, QChar::UnicodeVersion version, int from)
+{
     bool simple = true;
-    for (int i = 0; i < d->size; ++i) {
-        if (d->data[i] >= 0x80) {
+    const QChar *p = data->constData();
+    int len = data->length();
+    for (int i = from; i < len; ++i) {
+        if (p[i].unicode() >= 0x80) {
             simple = false;
             break;
         }
     }
     if (simple)
-        return *this;
+        return;
 
-    QString s = *this;
+    QString &s = *data;
     if (version != CURRENT_VERSION) {
         for (int i = 0; i < NumNormalizationCorrections; ++i) {
             const NormalizationCorrection &n = uc_normalization_corrections[i];
             if (n.version > version) {
+                int pos = from;
                 if (n.ucs4 > 0xffff) {
                     ushort ucs4High = QChar::highSurrogate(n.ucs4);
                     ushort ucs4Low = QChar::lowSurrogate(n.ucs4);
                     ushort oldHigh = QChar::highSurrogate(n.old_mapping);
                     ushort oldLow = QChar::lowSurrogate(n.old_mapping);
-                    int pos = 0;
-                    while (pos < s.d->size - 1) {
-                        if (s.d->data[pos] == ucs4High && s.d->data[pos + 1] == ucs4Low) {
-                            s.detach();
-                            s.d->data[pos] = oldHigh;
-                            s.d->data[pos + 1] = oldLow;
+                    while (pos < s.length() - 1) {
+                        if (s.at(pos).unicode() == ucs4High && s.at(pos + 1).unicode() == ucs4Low) {
+                            s[pos] = oldHigh;
+                            s[pos + 1] = oldLow;
                             ++pos;
                         }
                         ++pos;
                     }
                 } else {
-                    int pos = 0;
-                    while (pos < s.d->size) {
-                        if (s.d->data[pos] == n.ucs4) {
-                            s.detach();
-                            s.d->data[pos] = n.old_mapping;
+                    while (pos < s.length()) {
+                        if (s.at(pos).unicode() == n.ucs4) {
+                            s[pos] = n.old_mapping;
                         }
                         ++pos;
                     }
@@ -6080,15 +6087,14 @@ QString QString::normalized(QString::NormalizationForm mode, QChar::UnicodeVersi
             }
         }
     }
-    s = decomposeHelper(s, mode < QString::NormalizationForm_KD, version);
+    decomposeHelper(data, mode < QString::NormalizationForm_KD, version, from);
 
-    s = canonicalOrderHelper(s, version);
+    canonicalOrderHelper(data, version, from);
 
     if (mode == QString::NormalizationForm_D || mode == QString::NormalizationForm_KD)
-        return s;
-
-    return composeHelper(s);
+        return;
 
+    composeHelper(data, from);
 }