Port QDomDocument to QXmlStreamReader

Reimplement QDomDocument using QXmlStreamReader and switch to the new
implementation starting from Qt 6.

The changes in the behavior are reflected in tests: some test cases
which were marked as "expected to fail" are now passing.

Task-number: QTBUG-76178
Change-Id: I5ace2f13c036a9a778de922b47a1ce35957ce5f6
Reviewed-by: Kai Koehne <kai.koehne@qt.io>

5 files changed, 396 insertions, 0 deletions

diff --git a/src/xml/dom/qdom.cpp b/src/xml/dom/qdom.cpp
index bbbdf1c6f3..81f33b0693 100644
--- a/src/xml/dom/qdom.cpp
+++ b/src/xml/dom/qdom.cpp
@@ -59,6 +59,7 @@
 #include <qvariant.h>
 #include <qshareddata.h>
 #include <qdebug.h>
+#include <qxmlstream.h>
 #include <stdio.h>
 
 QT_BEGIN_NAMESPACE
@@ -5734,6 +5735,34 @@ bool QDomDocumentPrivate::setContent(QXmlInputSource *source, QXmlReader *reader
     return true;
 }
 
+bool QDomDocumentPrivate::setContent(QXmlStreamReader *reader, bool namespaceProcessing,
+                                     QString *errorMsg, int *errorLine, int *errorColumn)
+{
+    clear();
+    impl = new QDomImplementationPrivate;
+    type = new QDomDocumentTypePrivate(this, this);
+    type->ref.deref();
+
+    if (!reader) {
+        qWarning("Failed to set content, XML reader is not initialized");
+        return false;
+    }
+
+    QDomParser domParser(this, reader, namespaceProcessing);
+
+    if (!domParser.parse()) {
+        if (errorMsg)
+            *errorMsg = std::get<0>(domParser.errorInfo());
+        if (errorLine)
+            *errorLine = std::get<1>(domParser.errorInfo());
+        if (errorColumn)
+            *errorColumn = std::get<2>(domParser.errorInfo());
+        return false;
+    }
+
+    return true;
+}
+
 QDomNodePrivate* QDomDocumentPrivate::cloneNode(bool deep)
 {
     QDomNodePrivate *p = new QDomDocumentPrivate(this, deep);
@@ -6153,9 +6182,16 @@ bool QDomDocument::setContent(const QString& text, bool namespaceProcessing, QSt
 {
     if (!impl)
         impl = new QDomDocumentPrivate();
+
+#if QT_VERSION < QT_VERSION_CHECK(6, 0, 0)
     QXmlInputSource source;
     source.setData(text);
     return IMPL->setContent(&source, namespaceProcessing, errorMsg, errorLine, errorColumn);
+#else
+    QXmlStreamReader streamReader(text);
+    streamReader.setNamespaceProcessing(namespaceProcessing);
+    return IMPL->setContent(&streamReader, namespaceProcessing, errorMsg, errorLine, errorColumn);
+#endif
 }
 
 /*!
@@ -6215,10 +6251,17 @@ bool QDomDocument::setContent(const QByteArray &data, bool namespaceProcessing,
 {
     if (!impl)
         impl = new QDomDocumentPrivate();
+
+#if QT_VERSION < QT_VERSION_CHECK(6, 0, 0)
     QBuffer buf;
     buf.setData(data);
     QXmlInputSource source(&buf);
     return IMPL->setContent(&source, namespaceProcessing, errorMsg, errorLine, errorColumn);
+#else
+    QXmlStreamReader streamReader(data);
+    streamReader.setNamespaceProcessing(namespaceProcessing);
+    return IMPL->setContent(&streamReader, namespaceProcessing, errorMsg, errorLine, errorColumn);
+#endif
 }
 
 /*!
@@ -6231,8 +6274,15 @@ bool QDomDocument::setContent(QIODevice* dev, bool namespaceProcessing, QString
 {
     if (!impl)
         impl = new QDomDocumentPrivate();
+
+#if QT_VERSION < QT_VERSION_CHECK(6, 0, 0)
     QXmlInputSource source(dev);
     return IMPL->setContent(&source, namespaceProcessing, errorMsg, errorLine, errorColumn);
+#else
+    QXmlStreamReader streamReader(dev);
+    streamReader.setNamespaceProcessing(namespaceProcessing);
+    return IMPL->setContent(&streamReader, namespaceProcessing, errorMsg, errorLine, errorColumn);
+#endif
 }
 
 /*!
@@ -6315,6 +6365,33 @@ bool QDomDocument::setContent(QXmlInputSource *source, QXmlReader *reader, QStri
 }
 
 /*!
+    \overload
+    \since 5.15
+
+    This function reads the XML document from the QXmlStreamReader \a reader
+    and parses it. Returns \c true if the content was successfully parsed;
+    otherwise returns \c false.
+
+    If \a namespaceProcessing is \c true, the parser recognizes namespaces in the XML
+    file and sets the prefix name, local name and namespace URI to appropriate values.
+    If \a namespaceProcessing is \c false, the parser does no namespace processing when
+    it reads the XML file.
+
+    If a parse error occurs, the error message is placed in \c{*}\a{errorMsg}, the line
+    number in \c{*}\a{errorLine} and the column number in \c{*}\a{errorColumn} (unless
+    the associated pointer is set to 0).
+
+    \sa QXmlStreamReader
+*/
+bool QDomDocument::setContent(QXmlStreamReader *reader, bool namespaceProcessing, QString *errorMsg,
+                              int *errorLine, int *errorColumn)
+{
+    if (!impl)
+        impl = new QDomDocumentPrivate();
+    return IMPL->setContent(reader, namespaceProcessing, errorMsg, errorLine, errorColumn);
+}
+
+/*!
     Converts the parsed document back to its textual representation.
 
     This function uses \a indent as the amount of space to indent
diff --git a/src/xml/dom/qdom.h b/src/xml/dom/qdom.h
index 0a7db7dcd7..1b00a14179 100644
--- a/src/xml/dom/qdom.h
+++ b/src/xml/dom/qdom.h
@@ -91,6 +91,7 @@ class QDomNode;
 class QDomEntity;
 class QDomNotation;
 class QDomCharacterData;
+class QXmlStreamReader;
 
 class Q_XML_EXPORT QDomImplementation
 {
@@ -343,6 +344,8 @@ public:
     bool setContent(const QString& text, QString *errorMsg=nullptr, int *errorLine=nullptr, int *errorColumn=nullptr );
     bool setContent(QIODevice* dev, QString *errorMsg=nullptr, int *errorLine=nullptr, int *errorColumn=nullptr );
     bool setContent(QXmlInputSource *source, QXmlReader *reader, QString *errorMsg=nullptr, int *errorLine=nullptr, int *errorColumn=nullptr );
+    bool setContent(QXmlStreamReader *reader, bool namespaceProcessing, QString *errorMsg = nullptr,
+                    int *errorLine = nullptr, int *errorColumn = nullptr);
 
     // Qt extensions
     QString toString(int = 1) const;
diff --git a/src/xml/dom/qdom_p.h b/src/xml/dom/qdom_p.h
index 06ed0d5f0e..d197e999f1 100644
--- a/src/xml/dom/qdom_p.h
+++ b/src/xml/dom/qdom_p.h
@@ -465,6 +465,8 @@ public:
                     int *errorLine, int *errorColumn);
     bool setContent(QXmlInputSource *source, QXmlReader *reader, QXmlSimpleReader *simpleReader,
                     QString *errorMsg, int *errorLine, int *errorColumn);
+    bool setContent(QXmlStreamReader *reader, bool namespaceProcessing, QString *errorMsg,
+                    int *errorLine, int *errorColumn);
 
     // Attributes
     QDomDocumentTypePrivate *doctype() { return type.data(); }
diff --git a/src/xml/dom/qdomhelpers.cpp b/src/xml/dom/qdomhelpers.cpp
index bda6f8ef87..9399ad3b9b 100644
--- a/src/xml/dom/qdomhelpers.cpp
+++ b/src/xml/dom/qdomhelpers.cpp
@@ -39,6 +39,7 @@
 
 #include "qdomhelpers_p.h"
 #include "qdom_p.h"
+#include "qxmlstream.h"
 #include "private/qxml_p.h"
 
 QT_BEGIN_NAMESPACE
@@ -166,6 +167,18 @@ QDomBuilder::ErrorInfo QDomHandler::errorInfo() const
  *
  **************************************************************/
 
+int QDomDocumentLocator::column() const
+{
+    Q_ASSERT(reader);
+    return static_cast<int>(reader->columnNumber());
+}
+
+int QDomDocumentLocator::line() const
+{
+    Q_ASSERT(reader);
+    return static_cast<int>(reader->lineNumber());
+}
+
 void QSAXDocumentLocator::setLocator(QXmlLocator *l)
 {
     locator = l;
@@ -252,6 +265,44 @@ bool QDomBuilder::startElement(const QString &nsURI, const QString &qName,
     return true;
 }
 
+inline QString stringRefToString(const QStringRef &stringRef)
+{
+    // Calling QStringRef::toString() on a NULL QStringRef in some cases returns
+    // an empty string (i.e. QString("")) instead of a NULL string (i.e. QString()).
+    // QDom implementation differentiates between NULL and empty strings, so
+    // we need this as workaround to keep the current behavior unchanged.
+    return stringRef.isNull() ? QString() : stringRef.toString();
+}
+
+bool QDomBuilder::startElement(const QString &nsURI, const QString &qName,
+                               const QXmlStreamAttributes &atts)
+{
+    QDomNodePrivate *n =
+            nsProcessing ? doc->createElementNS(nsURI, qName) : doc->createElement(qName);
+    if (!n)
+        return false;
+
+    n->setLocation(locator->line(), locator->column());
+
+    node->appendChild(n);
+    node = n;
+
+    // attributes
+    for (const auto &attr : atts) {
+        auto domElement = static_cast<QDomElementPrivate *>(node);
+        if (nsProcessing) {
+            domElement->setAttributeNS(stringRefToString(attr.namespaceUri()),
+                                       stringRefToString(attr.qualifiedName()),
+                                       stringRefToString(attr.value()));
+        } else {
+            domElement->setAttribute(stringRefToString(attr.qualifiedName()),
+                                     stringRefToString(attr.value()));
+        }
+    }
+
+    return true;
+}
+
 bool QDomBuilder::endElement()
 {
     if (!node || node == doc)
@@ -368,4 +419,225 @@ bool QDomBuilder::notationDecl(const QString &name, const QString &publicId,
     return true;
 }
 
+/**************************************************************
+ *
+ * QDomParser
+ *
+ **************************************************************/
+
+QDomParser::QDomParser(QDomDocumentPrivate *d, QXmlStreamReader *r, bool namespaceProcessing)
+    : reader(r), locator(r), domBuilder(d, &locator, namespaceProcessing)
+{
+}
+
+bool QDomParser::parse()
+{
+    return parseProlog() && parseBody();
+}
+
+QDomBuilder::ErrorInfo QDomParser::errorInfo() const
+{
+    return domBuilder.error();
+}
+
+bool QDomParser::parseProlog()
+{
+    Q_ASSERT(reader);
+
+    bool foundDtd = false;
+
+    while (!reader->atEnd()) {
+        reader->readNext();
+
+        if (reader->hasError()) {
+            domBuilder.fatalError(reader->errorString());
+            return false;
+        }
+
+        switch (reader->tokenType()) {
+        case QXmlStreamReader::StartDocument:
+            if (!reader->documentVersion().isEmpty()) {
+                QString value(QLatin1String("version='"));
+                value += reader->documentVersion();
+                value += QLatin1Char('\'');
+                if (!reader->documentEncoding().isEmpty()) {
+                    value += QLatin1String(" encoding='");
+                    value += reader->documentEncoding();
+                    value += QLatin1Char('\'');
+                }
+                if (reader->isStandaloneDocument()) {
+                    value += QLatin1String(" standalone='yes'");
+                } else {
+                    // TODO: Add standalone='no', if 'standalone' is specified. With the current
+                    // QXmlStreamReader there is no way to figure out if it was specified or not.
+                    // QXmlStreamReader needs to be modified for handling that case correctly.
+                }
+
+                if (!domBuilder.processingInstruction(QLatin1String("xml"), value)) {
+                    domBuilder.fatalError(
+                            QDomParser::tr("Error occurred while processing XML declaration"));
+                    return false;
+                }
+            }
+            break;
+        case QXmlStreamReader::DTD:
+            if (foundDtd) {
+                domBuilder.fatalError(QDomParser::tr("Multiple DTD sections are not allowed"));
+                return false;
+            }
+            foundDtd = true;
+
+            if (!domBuilder.startDTD(stringRefToString(reader->dtdName()),
+                                     stringRefToString(reader->dtdPublicId()),
+                                     stringRefToString(reader->dtdSystemId()))) {
+                domBuilder.fatalError(
+                        QDomParser::tr("Error occurred while processing document type declaration"));
+                return false;
+            }
+            if (!parseMarkupDecl())
+                return false;
+            break;
+        case QXmlStreamReader::Comment:
+            if (!domBuilder.comment(reader->text().toString())) {
+                domBuilder.fatalError(QDomParser::tr("Error occurred while processing comment"));
+                return false;
+            }
+            break;
+        case QXmlStreamReader::ProcessingInstruction:
+            if (!domBuilder.processingInstruction(reader->processingInstructionTarget().toString(),
+                                                  reader->processingInstructionData().toString())) {
+                domBuilder.fatalError(
+                        QDomParser::tr("Error occurred while processing a processing instruction"));
+                return false;
+            }
+            break;
+        default:
+            // If the token is none of the above, prolog processing is done.
+            return true;
+        }
+    }
+
+    return true;
+}
+
+bool QDomParser::parseBody()
+{
+    Q_ASSERT(reader);
+
+    std::stack<QStringRef> tagStack;
+    while (!reader->atEnd() && !reader->hasError()) {
+        switch (reader->tokenType()) {
+        case QXmlStreamReader::StartElement:
+            tagStack.push(reader->qualifiedName());
+            if (!domBuilder.startElement(stringRefToString(reader->namespaceUri()),
+                                         stringRefToString(reader->qualifiedName()),
+                                         reader->attributes())) {
+                domBuilder.fatalError(
+                        QDomParser::tr("Error occurred while processing a start element"));
+                return false;
+            }
+            break;
+        case QXmlStreamReader::EndElement:
+            if (tagStack.empty() || reader->qualifiedName() != tagStack.top()) {
+                domBuilder.fatalError(
+                        QDomParser::tr("Unexpected end element '%1'").arg(reader->name()));
+                return false;
+            }
+            tagStack.pop();
+            if (!domBuilder.endElement()) {
+                domBuilder.fatalError(
+                        QDomParser::tr("Error occurred while processing an end element"));
+                return false;
+            }
+            break;
+        case QXmlStreamReader::Characters:
+            if (!reader->isWhitespace()) { // Skip the content consisting of only whitespaces
+                if (!reader->text().toString().trimmed().isEmpty()) {
+                    if (!domBuilder.characters(reader->text().toString(), reader->isCDATA())) {
+                        domBuilder.fatalError(QDomParser::tr(
+                                "Error occurred while processing the element content"));
+                        return false;
+                    }
+                }
+            }
+            break;
+        case QXmlStreamReader::Comment:
+            if (!domBuilder.comment(reader->text().toString())) {
+                domBuilder.fatalError(QDomParser::tr("Error occurred while processing comments"));
+                return false;
+            }
+            break;
+        case QXmlStreamReader::ProcessingInstruction:
+            if (!domBuilder.processingInstruction(reader->processingInstructionTarget().toString(),
+                                                  reader->processingInstructionData().toString())) {
+                domBuilder.fatalError(
+                        QDomParser::tr("Error occurred while processing a processing instruction"));
+                return false;
+            }
+            break;
+        case QXmlStreamReader::EntityReference:
+            if (!domBuilder.skippedEntity(reader->name().toString())) {
+                domBuilder.fatalError(
+                        QDomParser::tr("Error occurred while processing an entity reference"));
+                return false;
+            }
+            break;
+        default:
+            domBuilder.fatalError(QDomParser::tr("Unexpected token"));
+            return false;
+        }
+
+        reader->readNext();
+    }
+
+    if (reader->hasError()) {
+        domBuilder.fatalError(reader->errorString());
+        reader->readNext();
+        return false;
+    }
+
+    if (!tagStack.empty()) {
+        domBuilder.fatalError(QDomParser::tr("Tag mismatch"));
+        return false;
+    }
+
+    return true;
+}
+
+bool QDomParser::parseMarkupDecl()
+{
+    Q_ASSERT(reader);
+
+    const auto entities = reader->entityDeclarations();
+    for (const auto &entityDecl : entities) {
+        // Entity declarations are created only for Extrenal Entities. Internal Entities
+        // are parsed, and QXmlStreamReader handles the parsing itself and returns the
+        // parsed result. So we don't need to do anything for the Internal Entities.
+        if (!entityDecl.publicId().isEmpty() || !entityDecl.systemId().isEmpty()) {
+            // External Entity
+            if (!domBuilder.unparsedEntityDecl(stringRefToString(entityDecl.name()),
+                                               stringRefToString(entityDecl.publicId()),
+                                               stringRefToString(entityDecl.systemId()),
+                                               stringRefToString(entityDecl.notationName()))) {
+                domBuilder.fatalError(
+                        QDomParser::tr("Error occurred while processing entity declaration"));
+                return false;
+            }
+        }
+    }
+
+    const auto notations = reader->notationDeclarations();
+    for (const auto &notationDecl : notations) {
+        if (!domBuilder.notationDecl(stringRefToString(notationDecl.name()),
+                                     stringRefToString(notationDecl.publicId()),
+                                     stringRefToString(notationDecl.systemId()))) {
+            domBuilder.fatalError(
+                    QDomParser::tr("Error occurred while processing notation declaration"));
+            return false;
+        }
+    }
+
+    return true;
+}
+
 QT_END_NAMESPACE
diff --git a/src/xml/dom/qdomhelpers_p.h b/src/xml/dom/qdomhelpers_p.h
index 7647de65d2..f5efd8a42d 100644
--- a/src/xml/dom/qdomhelpers_p.h
+++ b/src/xml/dom/qdomhelpers_p.h
@@ -39,6 +39,7 @@
 #ifndef QDOMHELPERS_P_H
 #define QDOMHELPERS_P_H
 
+#include <qcoreapplication.h>
 #include <qglobal.h>
 #include <qxml.h>
 
@@ -57,6 +58,8 @@ QT_BEGIN_NAMESPACE
 
 class QDomDocumentPrivate;
 class QDomNodePrivate;
+class QXmlStreamReader;
+class QXmlStreamAttributes;
 
 /**************************************************************
  *
@@ -77,6 +80,19 @@ public:
     virtual int line() const = 0;
 };
 
+class QDomDocumentLocator : public QXmlDocumentLocator
+{
+public:
+    QDomDocumentLocator(QXmlStreamReader *r) : reader(r) {}
+    ~QDomDocumentLocator() override = default;
+
+    int column() const override;
+    int line() const override;
+
+private:
+    QXmlStreamReader *reader;
+};
+
 class QSAXDocumentLocator : public QXmlDocumentLocator
 {
 public:
@@ -105,6 +121,7 @@ public:
 
     bool endDocument();
     bool startElement(const QString &nsURI, const QString &qName, const QXmlAttributes &atts);
+    bool startElement(const QString &nsURI, const QString &qName, const QXmlStreamAttributes &atts);
     bool endElement();
     bool characters(const QString &characters, bool cdata = false);
     bool processingInstruction(const QString &target, const QString &data);
@@ -188,6 +205,31 @@ private:
     QDomBuilder domBuilder;
 };
 
+/**************************************************************
+ *
+ * QDomParser
+ *
+ **************************************************************/
+
+class QDomParser
+{
+    Q_DECLARE_TR_FUNCTIONS(QDomParser)
+public:
+    QDomParser(QDomDocumentPrivate *d, QXmlStreamReader *r, bool namespaceProcessing);
+
+    bool parse();
+    QDomBuilder::ErrorInfo errorInfo() const;
+
+private:
+    bool parseProlog();
+    bool parseBody();
+    bool parseMarkupDecl();
+
+    QXmlStreamReader *reader;
+    QDomDocumentLocator locator;
+    QDomBuilder domBuilder;
+};
+
 QT_END_NAMESPACE
 
 #endif // QDOMHELPERS_P_H