Correctly parse non BMP char refs in the sax parser

Update the auto test accordingly, and at the same time remove
all uses of QTextStream (as they aren't required).

Change-Id: I71b7cf6a6b54ea59507f27d5d2d04cc5ae5885fc
Reviewed-by: Sona Kurazyan <sona.kurazyan@qt.io>

src/xml/sax/qxml.cpp

@@ -7475,7 +7475,12 @@ bool QXmlSimpleReaderPrivate::parseReference()
             case DoneD:
                 tmp = ref().toUInt(&ok, 10);
                 if (ok) {
-                    stringAddC(QChar(tmp));
+                    if (tmp > 0xffff) {
+                        stringAddC(QChar::highSurrogate(tmp));
+                        stringAddC(QChar::lowSurrogate(tmp));
+                    } else {
+                        stringAddC(QChar(tmp));
+                    }
                 } else {
                     reportParseError(QLatin1String(XMLERR_ERRORPARSINGREFERENCE));
                     return false;
@@ -7486,7 +7491,12 @@ bool QXmlSimpleReaderPrivate::parseReference()
             case DoneH:
                 tmp = ref().toUInt(&ok, 16);
                 if (ok) {
-                    stringAddC(QChar(tmp));
+                    if (tmp > 0xffff) {
+                        stringAddC(QChar::highSurrogate(tmp));
+                        stringAddC(QChar::lowSurrogate(tmp));
+                    } else {
+                        stringAddC(QChar(tmp));
+                    }
                 } else {
                     reportParseError(QLatin1String(XMLERR_ERRORPARSINGREFERENCE));
                     return false;

tests/auto/xml/sax/qxmlsimplereader/parser/main.cpp

@@ -37,7 +37,6 @@
 
 #include "parser.h"
 
-static QTextStream qout(stdout, QIODevice::WriteOnly);
 static QTextStream qerr(stderr, QIODevice::WriteOnly);
 
 static void usage()
@@ -79,19 +78,15 @@ int main(int argc, const char *argv[])
     if (out_file_name.isEmpty())
         out_file_name = file_name + ".ref";
 
-    QFile _out_file;
+    QFile out_file;
-    QTextStream _out_stream;
-    QTextStream *out_stream;
     if (out_file_name == "-") {
-        out_stream = &qout;
+        out_file.open(stdout, QFile::WriteOnly);
     } else {
-        _out_file.setFileName(out_file_name);
+        out_file.setFileName(out_file_name);
-        if (!_out_file.open(QIODevice::WriteOnly | QIODevice::Truncate)) {
+        if (!out_file.open(QIODevice::WriteOnly | QIODevice::Truncate)) {
             qerr << "Could not open " << out_file_name << ": " << strerror(errno) << Qt::endl;
             return 1;
         }
-        _out_stream.setDevice(&_out_file);
-        out_stream = &_out_stream;
     }
 
     Parser parser;
@@ -102,9 +97,7 @@ int main(int argc, const char *argv[])
 
     parser.parseFile(&in_file);
 
-    out_stream->setCodec("utf8");
+    out_file.write(parser.result().toUtf8());
-
-    *out_stream << parser.result();
 
     return 0;
 }

tests/auto/xml/sax/qxmlsimplereader/parser/parser.cpp

@@ -142,11 +142,12 @@ bool ContentHandler::startElement(const QString &namespaceURI,
 QString ContentHandler::escapeStr(const QString &s)
 {
     QString result = s;
-    result.replace(QRegularExpression("\""), "\\\"");
+    result.replace(QChar(0), "\\0");
-    result.replace(QRegularExpression("\\"), "\\\\");
+    result.replace("\\", "\\\\");
-    result.replace(QRegularExpression("\n"), "\\n");
+    result.replace("\"", "\\\"");
-    result.replace(QRegularExpression("\r"), "\\r");
+    result.replace("\n", "\\n");
-    result.replace(QRegularExpression("\t"), "\\t");
+    result.replace("\r", "\\r");
+    result.replace("\t", "\\t");
     return result;
 }
 

tests/auto/xml/sax/qxmlsimplereader/tst_qxmlsimplereader.cpp

@@ -311,14 +311,12 @@ void tst_QXmlSimpleReader::testGoodXmlFile()
     QVERIFY(file.open(QIODevice::ReadOnly));
     Parser parser;
 
-    QEXPECT_FAIL(QFINDTESTDATA("xmldocs/valid/sa/089.xml").toLocal8Bit().constData(), "a form feed character is not accepted in XML", Continue);
     QVERIFY(parser.parseFile(&file));
 
     QFile ref_file(file_name + ".ref");
     QVERIFY(ref_file.open(QIODevice::ReadOnly | QIODevice::Text));
-    QTextStream ref_stream(&ref_file);
+    QByteArray data = ref_file.readAll();
-    ref_stream.setCodec("UTF-8");
+    QString ref_file_contents = QString::fromUtf8(data.constData(), data.size());
-    QString ref_file_contents = ref_stream.readAll();
 
     QCOMPARE(parser.result(), ref_file_contents);
 }
@@ -393,9 +391,7 @@ void tst_QXmlSimpleReader::testBadXmlFile()
 
     QFile ref_file(file_name + ".ref");
     QVERIFY(ref_file.open(QIODevice::ReadOnly | QIODevice::Text));
-    QTextStream ref_stream(&ref_file);
+    QString ref_file_contents = QString::fromUtf8(ref_file.readAll());
-    ref_stream.setCodec("UTF-8");
-    QString ref_file_contents = ref_stream.readAll();
 
     QEXPECT_FAIL(QFINDTESTDATA("xmldocs/not-wf/sa/145.xml").toLocal8Bit().constData(), "Surrogate code point 0xD800 should be rejected", Continue);
 
@@ -469,9 +465,7 @@ void tst_QXmlSimpleReader::testIncrementalParsing()
 
     QFile ref_file(file_name + ".ref");
     QVERIFY(ref_file.open(QIODevice::ReadOnly | QIODevice::Text));
-    QTextStream ref_stream(&ref_file);
+    QString ref_file_contents = QString::fromUtf8(ref_file.readAll());
-    ref_stream.setCodec("UTF-8");
-    QString ref_file_contents = ref_stream.readAll();
 
     QCOMPARE(parser.result(), ref_file_contents);
 }