Add isValidUtf8() methods to QUtf8StringView and QByteArray{,View}

The new methods return true if the string contains valid UTF-8
encoded data, or false otherwise.

[ChangeLog][QtCore][QByteArray] Added isValidUtf8() method.

[ChangeLog][QtCore][QByteArrayView] Added isValidUtf8() method.

[ChangeLog][QtCore][QUtf8StringView] Added isValidUtf8() method.

Task-number: QTBUG-92021
Change-Id: I5d0cb613265d98b1f189c5f5cc09c1f7db302272
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>

src/corelib/text/qbytearray.cpp

@@ -51,6 +51,7 @@
 #include "qstringalgorithms_p.h"
 #include "qscopedpointer.h"
 #include "qbytearray_p.h"
+#include "qstringconverter_p.h"
 #include <qdatastream.h>
 #include <qmath.h>
 
@@ -426,6 +427,14 @@ int QtPrivate::compareMemory(QByteArrayView lhs, QByteArrayView rhs)
     return lhs.size() == rhs.size() ? 0 : lhs.size() > rhs.size() ? 1 : -1;
 }
 
+/*!
+    \internal
+*/
+bool QtPrivate::isValidUtf8(QByteArrayView s) noexcept
+{
+    return QUtf8::isValidUtf8(s).isValidUtf8;
+}
+
 // the CRC table below is created by the following piece of code
 #if 0
 static void createCRC16Table()                        // build CRC16 lookup table
@@ -2783,6 +2792,15 @@ bool QByteArray::isLower() const
     return true;
 }
 
+/*!
+    \fn QByteArray::isValidUtf8() const
+
+    Returns \c true if this byte array contains valid UTF-8 encoded data,
+    or \c false otherwise.
+
+    \since 6.3
+*/
+
 /*!
     Returns a byte array that contains the first \a len bytes of this byte
     array.

src/corelib/text/qbytearray.h

@@ -200,6 +200,11 @@ public:
     bool isUpper() const;
     bool isLower() const;
 
+    [[nodiscard]] bool isValidUtf8() const noexcept
+    {
+        return QtPrivate::isValidUtf8(qToByteArrayViewIgnoringNull(*this));
+    }
+
     void truncate(qsizetype pos);
     void chop(qsizetype n);
 

src/corelib/text/qbytearrayalgorithms.h

@@ -74,6 +74,8 @@ qsizetype count(QByteArrayView haystack, QByteArrayView needle) noexcept;
 
 [[nodiscard]] Q_CORE_EXPORT Q_DECL_PURE_FUNCTION QByteArrayView trimmed(QByteArrayView s) noexcept;
 
+[[nodiscard]] Q_CORE_EXPORT Q_DECL_PURE_FUNCTION bool isValidUtf8(QByteArrayView s) noexcept;
+
 } // namespace QtPrivate
 
 /*****************************************************************************

src/corelib/text/qbytearrayview.h

@@ -288,6 +288,8 @@ public:
 
     inline int compare(QByteArrayView a, Qt::CaseSensitivity cs = Qt::CaseSensitive) const noexcept;
 
+    [[nodiscard]] inline bool isValidUtf8() const noexcept { return QtPrivate::isValidUtf8(*this); }
+
     //
     // STL compatibility API:
     //

src/corelib/text/qbytearrayview.qdoc

@@ -376,6 +376,15 @@
     \sa operator==()
 */
 
+/*!
+    \fn QByteArrayView::isValidUtf8() const
+
+    Returns \c true if this byte array view contains valid UTF-8 encoded data,
+    or \c false otherwise.
+
+    \since 6.3
+*/
+
 /*!
     \fn QByteArrayView::const_iterator QByteArrayView::begin() const
 

src/corelib/text/qutf8stringview.h

@@ -285,6 +285,11 @@ public:
     constexpr void chop(qsizetype n)
     { verify(n); m_size -= n; }
 
+    [[nodiscard]] inline bool isValidUtf8() const noexcept
+    {
+        return QByteArrayView(reinterpret_cast<const char *>(data()), size()).isValidUtf8();
+    }
+
     //
     // STL compatibility API:
     //

src/corelib/text/qutf8stringview.qdoc

@@ -678,6 +678,15 @@
     \sa sliced(), first(), last(), chopped(), truncate()
 */
 
+/*!
+    \fn QUtf8StringView::isValidUtf8() const
+
+    Returns \c true if this string contains valid UTF-8 encoded data,
+    or \c false otherwise.
+
+    \since 6.3
+*/
+
 /*!
     \fn template <typename QStringLike> qToUtf8StringViewIgnoringNull(const QStringLike &s);
     \relates QUtf8StringView

tests/auto/corelib/text/qstringapisymmetry/tst_qstringapisymmetry.cpp

@@ -898,6 +898,19 @@ private Q_SLOTS:
     void indexOf_regexp_QString() { indexOf_contains_lastIndexOf_count_regexp_impl<QString>(); }
     void indexOf_regexp_QStringView_data() { indexOf_contains_lastIndexOf_count_regexp_data(); }
     void indexOf_regexp_QStringView() { indexOf_contains_lastIndexOf_count_regexp_impl<QStringView>(); }
+
+private:
+    void isValidUtf8_data();
+    template<typename String>
+    void isValidUtf8_impl() const;
+
+private Q_SLOTS:
+    void isValidUtf8_QByteArray_data() { isValidUtf8_data(); }
+    void isValidUtf8_QByteArray() { isValidUtf8_impl<QByteArray>(); }
+    void isValidUtf8_QByteArrayView_data() { isValidUtf8_data(); }
+    void isValidUtf8_QByteArrayView() { isValidUtf8_impl<QByteArrayView>(); }
+    void isValidUtf8_QUtf8StringView_data() { isValidUtf8_data(); }
+    void isValidUtf8_QUtf8StringView() { isValidUtf8_impl<QUtf8StringView>(); }
 };
 
 namespace {
@@ -2855,6 +2868,76 @@ void tst_QStringApiSymmetry::indexOf_contains_lastIndexOf_count_regexp_impl() co
     }
 }
 
+void tst_QStringApiSymmetry::isValidUtf8_data()
+{
+    QTest::addColumn<QByteArray>("ba");
+    QTest::addColumn<bool>("valid");
+
+    int row = 0;
+    QTest::addRow("valid-%02d", row++) << QByteArray() << true;
+    QTest::addRow("valid-%02d", row++) << QByteArray("ascii") << true;
+    QTest::addRow("valid-%02d", row++)
+            << QByteArray("\xc2\xa2\xe0\xa4\xb9\xf0\x90\x8d\x88") << true; // U+00A2 U+0939 U+10348
+    QTest::addRow("valid-%02d", row++) << QByteArray("\xf4\x8f\xbf\xbf") << true; // U+10FFFF
+
+    row = 0;
+    QTest::addRow("overlong-%02d", row++) << QByteArray("\xc0\x00") << false;
+    QTest::addRow("overlong-%02d", row++) << QByteArray("\xc1\xff") << false;
+    QTest::addRow("overlong-%02d", row++) << QByteArray("\xe0\x00\x00") << false;
+    QTest::addRow("overlong-%02d", row++) << QByteArray("\xe0\xa0\x7f") << false;
+    QTest::addRow("overlong-%02d", row++) << QByteArray("\xf0\x00\x00\x00") << false;
+    QTest::addRow("overlong-%02d", row++) << QByteArray("\xf0\x90\x80\x7f") << false;
+
+    row = 0;
+    QTest::addRow("short-%02d", row++) << QByteArray("\xc2") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("x\xc2") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("x\xc2y") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("\xc2y") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("\xe0\xa4") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("x\xe0\xa4") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("x\xe0\xa4y") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("\xe0\xa4y") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("\xe0") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("x\xe0") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("x\xe0y") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("\xe0y") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("\xf4\x8f\xbf") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("x\xf4\x8f\xbf") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("x\xf4\x8f\xbfy") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("\xf4\x8f\xbfy") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("\xf4\x8f") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("x\xf4\x8f") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("x\xf4\x8fy") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("\xf4\x8fy") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("\xf4") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("x\xf4") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("x\xf4y") << false;
+    QTest::addRow("short-%02d", row++) << QByteArray("\xf4y") << false;
+
+    row = 0;
+    QTest::addRow("surrogates-%02d", row++) << QByteArray("\xed\x9f\xc0\xee\x80\x7f") << false;
+    QTest::addRow("surrogates-%02d", row++) << QByteArray("\xed\x9f\xc0") << false;
+    QTest::addRow("surrogates-%02d", row++) << QByteArray("\xee\x80\x7f") << false;
+    QTest::addRow("surrogates-%02d", row++) << QByteArray("\xee\x80\x7f\xed\x9f\xc0") << false;
+
+    row = 0;
+    QTest::addRow("other-%02d", row++) << QByteArray("\xf4\x8f\xbf\xc0") << false;
+    QTest::addRow("other-%02d", row++) << QByteArray("\xf7\x80\x80\x80") << false;
+    QTest::addRow("other-%02d", row++) << QByteArray("\xfd\xbf\xbf\xbf\xbf") << false;
+    QTest::addRow("other-%02d", row++) << QByteArray("\xfe\xbf\xbf\xbf\xbf\xbf") << false;
+    QTest::addRow("other-%02d", row++) << QByteArray("\xff\xbf\xbf\xbf\xbf\xbf\xbf") << false;
+    QTest::addRow("other-%02d", row++) << QByteArray("\x80") << false;
+    QTest::addRow("other-%02d", row++) << QByteArray("\xbf") << false;
+}
+
+template<typename String>
+void tst_QStringApiSymmetry::isValidUtf8_impl() const
+{
+    QFETCH(QByteArray, ba);
+    const String string(ba);
+    QTEST(string.isValidUtf8(), "valid");
+}
+
 QTEST_APPLESS_MAIN(tst_QStringApiSymmetry)
 
 #include "tst_qstringapisymmetry.moc"