QLocal8Bit::convertToUnicode[win]: Support stateless flag

By just setting state to nullptr.

Pick-to: 6.5
Task-number: QTBUG-105105
Change-Id: I6b4f8fe39f1ba51dcfaf98ce7e42c2acd4c4cf98
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>
(cherry picked from commit 10f5e4f8099136bce59f5f859df89c5b2e695fc9)
Reviewed-by: Qt Cherry-pick Bot <cherrypick_bot@qt-project.org>

2 files changed, 41 insertions, 1 deletions

diff --git a/src/corelib/text/qstringconverter.cpp b/src/corelib/text/qstringconverter.cpp
index 7d7dd35dce..0d0a885d70 100644
--- a/src/corelib/text/qstringconverter.cpp
+++ b/src/corelib/text/qstringconverter.cpp
@@ -1270,6 +1270,9 @@ QString QLocal8Bit::convertToUnicode_sys(QByteArrayView in, quint32 codePage,
     const char *mb = in.data();
     int mblen = length;
 
+    if (state && state->flags & QStringConverter::Flag::Stateless)
+        state = nullptr;
+
     if (!mb || !mblen)
         return QString();
 
@@ -1327,7 +1330,7 @@ QString QLocal8Bit::convertToUnicode_sys(QByteArrayView in, quint32 codePage,
     }
 
     Q_ASSERT(mblen > 0);
-    Q_ASSERT(state->remainingChars == 0);
+    Q_ASSERT(!state || state->remainingChars == 0);
 
     while (!(len = MultiByteToWideChar(codePage, MB_ERR_INVALID_CHARS, mb, mblen, out,
                                           int(outlen)))) {
diff --git a/tests/auto/corelib/text/qstringconverter/tst_qstringconverter.cpp b/tests/auto/corelib/text/qstringconverter/tst_qstringconverter.cpp
index 9b33b46421..3202f553e8 100644
--- a/tests/auto/corelib/text/qstringconverter/tst_qstringconverter.cpp
+++ b/tests/auto/corelib/text/qstringconverter/tst_qstringconverter.cpp
@@ -2523,6 +2523,14 @@ void tst_QStringConverter::fromLocal8Bit()
         result += QLocal8Bit::convertToUnicode_sys({&c, 1}, codePage, &state);
     QCOMPARE(result, utf16);
     QCOMPARE(state.remainingChars, 0);
+
+    result.clear();
+    state.clear();
+    // Decode the full string again, this time without state
+    state.flags |= QStringConverter::Flag::Stateless;
+    result = QLocal8Bit::convertToUnicode_sys(eightBit, codePage, &state);
+    QCOMPARE(result, utf16);
+    QCOMPARE(state.remainingChars, 0);
 }
 
 void tst_QStringConverter::fromLocal8Bit_special_cases()
@@ -2539,6 +2547,17 @@ void tst_QStringConverter::fromLocal8Bit_special_cases()
     QCOMPARE(result, u"こ");
     QCOMPARE(state.remainingChars, 0);
 
+    // And without state:
+    result.clear();
+    QStringConverter::State statelessState;
+    statelessState.flags |= QStringConverter::Flag::Stateless;
+    result = QLocal8Bit::convertToUnicode_sys("\x82", SHIFT_JIS, &statelessState);
+    result += QLocal8Bit::convertToUnicode_sys("\xb1", SHIFT_JIS, &statelessState);
+    // 0xb1 is a valid single-octet character in Shift-JIS, so the output
+    // isn't really what you would expect:
+    QCOMPARE(result, QString(QChar::ReplacementCharacter) + u'ｱ');
+    QCOMPARE(statelessState.remainingChars, 0);
+
     // Now try a 3-octet UTF-8 sequence:
     result.clear();
     state.clear();
@@ -2600,6 +2619,14 @@ void tst_QStringConverter::toLocal8Bit()
         result += QLocal8Bit::convertFromUnicode_sys(QStringView(&c, 1), codePage, &state);
     QCOMPARE(result, eightBit);
     QCOMPARE(state.remainingChars, 0);
+
+    result.clear();
+    state.clear();
+    // Decode the full string again, this time without state
+    state.flags |= QStringConverter::Flag::Stateless;
+    result = QLocal8Bit::convertFromUnicode_sys(utf16, codePage, &state);
+    QCOMPARE(result, eightBit);
+    QCOMPARE(state.remainingChars, 0);
 }
 
 void tst_QStringConverter::toLocal8Bit_special_cases()
@@ -2622,6 +2649,16 @@ void tst_QStringConverter::toLocal8Bit_special_cases()
     // Retain compat with the behavior for toLocal8Bit:
     QCOMPARE(codeUnits.first(1).toLocal8Bit(), "?");
 
+    // QString::toLocal8Bit is already stateless, but test stateless handling
+    // explicitly anyway:
+    result.clear();
+    QStringConverter::State statelessState;
+    statelessState.flags |= QStringConverter::Flag::Stateless;
+    result = QLocal8Bit::convertFromUnicode_sys(codeUnits.first(1), UTF8, &statelessState);
+    result += QLocal8Bit::convertFromUnicode_sys(codeUnits.sliced(1), UTF8, &statelessState);
+    // Windows uses the replacement character for invalid characters:
+    QCOMPARE(result, "\ufffd\ufffd");
+
     // Now do the same, but the second time we feed in a character, we also
     // provide many more so the internal stack buffer is not large enough.
     result.clear();