String-view-ify qt_splitLocaleName() and simplify its callers

Two of its callers didn't even care about the fragments it was
returning, one only cared about two of them. The parsing could be more
straightforward and less verbose. Parsing into QStringView saves the
need to allocate copies of the substrings parsed.  Cleaned up around
the code that called it.

Change-Id: Ie65d2b1b3eb891c20bc82275d7a0da00c30d7b8d
Reviewed-by: Mårten Nordheim <marten.nordheim@qt.io>

5 files changed, 73 insertions, 81 deletions

diff --git a/src/corelib/text/qlocale.cpp b/src/corelib/text/qlocale.cpp
index b751ed4579..a630f3a5db 100644
--- a/src/corelib/text/qlocale.cpp
+++ b/src/corelib/text/qlocale.cpp
@@ -491,105 +491,102 @@ int QLocaleData::findLocaleIndex(QLocaleId lid)
     return locale_index[fallback];
 }
 
-static bool parse_locale_tag(const QString &input, int &i, QString *result,
-                             const QString &separators)
-{
-    *result = QString(8, Qt::Uninitialized); // worst case according to BCP47
-    QChar *pch = result->data();
-    const QChar *uc = input.data() + i;
-    const int l = input.length();
-    int size = 0;
-    for (; i < l && size < 8; ++i, ++size) {
-        if (separators.contains(*uc))
-            break;
-        if (! ((uc->unicode() >= 'a' && uc->unicode() <= 'z') ||
-               (uc->unicode() >= 'A' && uc->unicode() <= 'Z') ||
-               (uc->unicode() >= '0' && uc->unicode() <= '9')) ) // latin only
+static QStringView findTag(QStringView name)
+{
+    const QString separators = QStringLiteral("_-.@");
+    int i = 0;
+    while (i < name.size() && !separators.contains(name[i]))
+        i++;
+    return name.first(i);
+}
+
+static bool validTag(QStringView tag)
+{
+    // Returns false if any character in tag is not an ASCII letter or digit
+    for (const QChar uc : tag) {
+        const char16_t ch = uc.unicode();
+        if (!((ch >= 'a' && ch <= 'z') || (ch >= 'A' && ch <= 'Z') || (ch >= '0' && ch <= '9')))
             return false;
-        *pch++ = *uc++;
     }
-    result->truncate(size);
-    return true;
+    return tag.size() > 0;
 }
 
-bool qt_splitLocaleName(const QString &name, QString &lang, QString &script, QString &cntry)
+static bool isScript(QStringView tag)
 {
-    const int length = name.length();
-
-    lang = script = cntry = QString();
+    // Every script name is 4 characters, a capital followed by three lower-case;
+    // so a search for tag in allScripts *can* only match if it's aligned.
+    static const QString allScripts =
+        QString::fromLatin1(reinterpret_cast<const char *>(script_code_list),
+                            sizeof(script_code_list) - 1);
+    return tag.length() == 4 && allScripts.indexOf(tag) % 4 == 0;
+}
 
-    const QString separators = QStringLiteral("_-.@");
+bool qt_splitLocaleName(QStringView name, QStringView *lang, QStringView *script, QStringView *land)
+{
+    // Assume each of lang, script and land is nullptr or points to an empty QStringView.
     enum ParserState { NoState, LangState, ScriptState, CountryState };
     ParserState state = LangState;
-    for (int i = 0; i < length && state != NoState; ) {
-        QString value;
-        if (!parse_locale_tag(name, i, &value, separators) ||value.isEmpty())
+    while (name.size() && state != NoState) {
+        const QStringView tag = findTag(name);
+        if (!validTag(tag))
             break;
-        QChar sep = i < length ? name.at(i) : QChar();
+        name = name.sliced(tag.size());
+        const bool sep = name.size() > 0;
+        if (sep) // tag wasn't all that remained; there was a separator
+            name = name.sliced(1);
+
         switch (state) {
         case LangState:
-            if (!sep.isNull() && !separators.contains(sep)) {
-                state = NoState;
-                break;
-            }
-            lang = value;
-            if (i == length) {
-                // just language was specified
-                state = NoState;
-                break;
-            }
-            state = ScriptState;
+            if (tag.size() != 2 && tag.size() != 3)
+                return false;
+            if (lang)
+                *lang = tag;
+            state = sep ? ScriptState : NoState;
             break;
-        case ScriptState: {
-            QString scripts = QString::fromLatin1((const char *)script_code_list,
-                                                  sizeof(script_code_list) - 1);
-            if (value.length() == 4 && scripts.indexOf(value) % 4 == 0) {
-                // script name is always 4 characters
-                script = value;
-                state = CountryState;
-            } else {
-                // it wasn't a script, maybe it is a country then?
-                cntry = value;
-                state = NoState;
+        case ScriptState:
+            if (isScript(tag)) {
+                if (script)
+                    *script = tag;
+                state = sep ? CountryState : NoState;
+                break;
             }
-            break;
-        }
+            // It wasn't a script, assume it's a country.
+            Q_FALLTHROUGH();
         case CountryState:
-            cntry = value;
+            if (land)
+                *land = tag;
             state = NoState;
             break;
-        case NoState:
-            // shouldn't happen
-            qWarning("QLocale: This should never happen");
+        case NoState: // Precluded by loop condition !
+            Q_ASSERT(!"QLocale: This should never happen");
             break;
         }
-        ++i;
     }
-    return lang.length() == 2 || lang.length() == 3;
+    return state != LangState;
 }
 
 // TODO: kill this !  Still in use by qttools, patch submitted (2020 October).
 void QLocalePrivate::getLangAndCountry(const QString &name, QLocale::Language &lang,
-                                       QLocale::Script &script, QLocale::Country &cntry)
+                                       QLocale::Script &script, QLocale::Country &land)
 {
     const auto id = QLocaleId::fromName(name);
     lang = QLocale::Language(id.language_id);
     script = QLocale::Script(id.script_id);
-    cntry = QLocale::Country(id.country_id);
+    land = QLocale::Country(id.country_id);
 }
 
 QLocaleId QLocaleId::fromName(const QString &name)
 {
-    QString lang;
-    QString script;
-    QString cntry;
-    if (!qt_splitLocaleName(name, lang, script, cntry))
+    QStringView lang;
+    QStringView script;
+    QStringView land;
+    if (!qt_splitLocaleName(name, &lang, &script, &land))
         return { QLocale::C, 0, 0 };
 
     QLocale::Language langId = QLocalePrivate::codeToLanguage(lang);
     if (langId == QLocale::C)
         return QLocaleId { langId, 0, 0 };
-    return { langId, QLocalePrivate::codeToScript(script), QLocalePrivate::codeToCountry(cntry) };
+    return { langId, QLocalePrivate::codeToScript(script), QLocalePrivate::codeToCountry(land) };
 }
 
 QString qt_readEscapedFormatString(QStringView format, int *idx)
diff --git a/src/corelib/text/qlocale_mac.mm b/src/corelib/text/qlocale_mac.mm
index 7ed7832564..281fedcfab 100644
--- a/src/corelib/text/qlocale_mac.mm
+++ b/src/corelib/text/qlocale_mac.mm
@@ -70,10 +70,8 @@ static QByteArray envVarLocale()
 static QString getMacLocaleName()
 {
     QString result = QString::fromLocal8Bit(envVarLocale());
-
-    QString lang, script, cntry;
     if (result.isEmpty()
-        || (result != QLatin1String("C") && !qt_splitLocaleName(result, lang, script, cntry))) {
+        || (result != QLatin1String("C") && !qt_splitLocaleName(result))) {
         QCFType<CFLocaleRef> l = CFLocaleCopyCurrent();
         CFStringRef locale = CFLocaleGetIdentifier(l);
         result = QString::fromCFString(locale);
diff --git a/src/corelib/text/qlocale_p.h b/src/corelib/text/qlocale_p.h
index eda8bd0ca4..aad344eccc 100644
--- a/src/corelib/text/qlocale_p.h
+++ b/src/corelib/text/qlocale_p.h
@@ -489,7 +489,8 @@ inline char QLocaleData::numericToCLocale(QStringView in) const
 }
 
 QString qt_readEscapedFormatString(QStringView format, int *idx);
-bool qt_splitLocaleName(const QString &name, QString &lang, QString &script, QString &cntry);
+bool qt_splitLocaleName(QStringView name, QStringView *lang = nullptr,
+                        QStringView *script = nullptr, QStringView *cntry = nullptr);
 int qt_repeatCount(QStringView s);
 
 enum { AsciiSpaceMask = (1u << (' ' - 1)) |
diff --git a/src/corelib/text/qlocale_unix.cpp b/src/corelib/text/qlocale_unix.cpp
index b1edd0dd3c..adc73ad19a 100644
--- a/src/corelib/text/qlocale_unix.cpp
+++ b/src/corelib/text/qlocale_unix.cpp
@@ -269,12 +269,10 @@ QVariant QSystemLocale::query(QueryType type, QVariant in) const
         // than one script in the same country, e.g. Sindhi in India.
         // However, can clients of the UILanguage query cope if we include script ?
         for (int i = 0; i < lst.size(); ++i) {
-            QString lang, script, cntry;
-            if (qt_splitLocaleName(lst.at(i), lang, script, cntry)) {
-                if (!cntry.length())
-                    d->uiLanguages.append(lang);
-                else
-                    d->uiLanguages.append(lang % QLatin1Char('-') % cntry);
+            QStringView lang, cntry;
+            if (qt_splitLocaleName(lst.at(i), &lang, nullptr, &cntry)) {
+                d->uiLanguages.append(
+                    cntry.size() ? lang % QLatin1Char('-') % cntry : lang.toString());
             }
         }
         return d->uiLanguages.isEmpty() ? QVariant() : QVariant(d->uiLanguages);
diff --git a/src/corelib/text/qlocale_win.cpp b/src/corelib/text/qlocale_win.cpp
index cb0ef52c6f..f244826c33 100644
--- a/src/corelib/text/qlocale_win.cpp
+++ b/src/corelib/text/qlocale_win.cpp
@@ -1022,15 +1022,13 @@ static QByteArray getWinLocaleName(LCID id)
 {
     QByteArray result;
     if (id == LOCALE_USER_DEFAULT) {
-        static QByteArray langEnvVar = qgetenv("LANG");
+        static const QByteArray langEnvVar = qgetenv("LANG");
         result = langEnvVar;
-        QString lang, script, cntry;
-        if ( result == "C" || (!result.isEmpty()
-                && qt_splitLocaleName(QString::fromLocal8Bit(result), lang, script, cntry)) ) {
-            long id = 0;
-            bool ok = false;
-            id = qstrtoll(result.data(), 0, 0, &ok);
-            if ( !ok || id == 0 || id < INT_MIN || id > INT_MAX )
+        if (result == "C"
+            || (!result.isEmpty() && qt_splitLocaleName(QString::fromLocal8Bit(result)))) {
+            bool ok = false; // See if we have a Windows locale code instead of a locale name:
+            long id = qstrtoll(result.data(), 0, 0, &ok);
+            if (!ok || id == 0 || id < INT_MIN || id > INT_MAX) // Assume real locale name
                 return result;
             return winLangCodeToIsoName(int(id));
         }