xmltokenizerlibxml2.cpp

linux下开源浏览器WebKit的源码,市面上的很多商用浏览器都是移植自WebKit

    if (m_view && !newNode->attached())
        newNode->attach();
}

void XMLTokenizer::comment(const xmlChar* s)
{
    if (m_parserStopped)
        return;

    if (m_parserPaused) {
        m_pendingCallbacks->appendCommentCallback(s);
        return;
    }
    
    exitText();

    RefPtr<Node> newNode = new Comment(m_doc, toString(s));
    m_currentNode->addChild(newNode.get());
    if (m_view && !newNode->attached())
        newNode->attach();
}

void XMLTokenizer::startDocument(const xmlChar* version, const xmlChar* encoding, int standalone)
{
    ExceptionCode ec = 0;

    if (version)
        m_doc->setXMLVersion(toString(version), ec);
    m_doc->setXMLStandalone(standalone == 1, ec); // possible values are 0, 1, and -1
    if (encoding)
        m_doc->setXMLEncoding(toString(encoding));
}

void XMLTokenizer::endDocument()
{
    exitText();
}

void XMLTokenizer::internalSubset(const xmlChar* name, const xmlChar* externalID, const xmlChar* systemID)
{
    if (m_parserStopped)
        return;

    if (m_parserPaused) {
        m_pendingCallbacks->appendInternalSubsetCallback(name, externalID, systemID);
        return;
    }
    
    if (m_doc) {
#if ENABLE(WML)
        String extId = toString(externalID);
        if (isWMLDocument()
            && extId != "-//WAPFORUM//DTD WML 1.3//EN"
            && extId != "-//WAPFORUM//DTD WML 1.2//EN"
            && extId != "-//WAPFORUM//DTD WML 1.1//EN"
            && extId != "-//WAPFORUM//DTD WML 1.0//EN")
            handleError(fatal, "Invalid DTD Public ID", lineNumber(), columnNumber());
#endif

        m_doc->addChild(DocumentType::create(m_doc, toString(name), toString(externalID), toString(systemID)));
    }
}

static inline XMLTokenizer* getTokenizer(void* closure)
{
    xmlParserCtxtPtr ctxt = static_cast<xmlParserCtxtPtr>(closure);
    return static_cast<XMLTokenizer*>(ctxt->_private);
}

// This is a hack around http://bugzilla.gnome.org/show_bug.cgi?id=159219
// Otherwise libxml seems to call all the SAX callbacks twice for any replaced entity.
static inline bool hackAroundLibXMLEntityBug(void* closure)
{
#if LIBXML_VERSION >= 20627
    UNUSED_PARAM(closure);

    // This bug has been fixed in libxml 2.6.27.
    return false;
#else
    return static_cast<xmlParserCtxtPtr>(closure)->node;
#endif
}

static void startElementNsHandler(void* closure, const xmlChar* localname, const xmlChar* prefix, const xmlChar* uri, int nb_namespaces, const xmlChar** namespaces, int nb_attributes, int nb_defaulted, const xmlChar** libxmlAttributes)
{
    if (hackAroundLibXMLEntityBug(closure))
        return;

    getTokenizer(closure)->startElementNs(localname, prefix, uri, nb_namespaces, namespaces, nb_attributes, nb_defaulted, libxmlAttributes);
}

static void endElementNsHandler(void* closure, const xmlChar*, const xmlChar*, const xmlChar*)
{
    if (hackAroundLibXMLEntityBug(closure))
        return;
    
    getTokenizer(closure)->endElementNs();
}

static void charactersHandler(void* closure, const xmlChar* s, int len)
{
    if (hackAroundLibXMLEntityBug(closure))
        return;
    
    getTokenizer(closure)->characters(s, len);
}

static void processingInstructionHandler(void* closure, const xmlChar* target, const xmlChar* data)
{
    if (hackAroundLibXMLEntityBug(closure))
        return;
    
    getTokenizer(closure)->processingInstruction(target, data);
}

static void cdataBlockHandler(void* closure, const xmlChar* s, int len)
{
    if (hackAroundLibXMLEntityBug(closure))
        return;
    
    getTokenizer(closure)->cdataBlock(s, len);
}

static void commentHandler(void* closure, const xmlChar* comment)
{
    if (hackAroundLibXMLEntityBug(closure))
        return;
    
    getTokenizer(closure)->comment(comment);
}

WTF_ATTRIBUTE_PRINTF(2, 3)
static void warningHandler(void* closure, const char* message, ...)
{
    va_list args;
    va_start(args, message);
    getTokenizer(closure)->error(XMLTokenizer::warning, message, args);
    va_end(args);
}

WTF_ATTRIBUTE_PRINTF(2, 3)
static void fatalErrorHandler(void* closure, const char* message, ...)
{
    va_list args;
    va_start(args, message);
    getTokenizer(closure)->error(XMLTokenizer::fatal, message, args);
    va_end(args);
}

WTF_ATTRIBUTE_PRINTF(2, 3)
static void normalErrorHandler(void* closure, const char* message, ...)
{
    va_list args;
    va_start(args, message);
    getTokenizer(closure)->error(XMLTokenizer::nonFatal, message, args);
    va_end(args);
}

// Using a global variable entity and marking it XML_INTERNAL_PREDEFINED_ENTITY is
// a hack to avoid malloc/free. Using a global variable like this could cause trouble
// if libxml implementation details were to change
static xmlChar sharedXHTMLEntityResult[5] = {0,0,0,0,0};
static xmlEntity sharedXHTMLEntity = {
    0, XML_ENTITY_DECL, 0, 0, 0, 0, 0, 0, 0, 
    sharedXHTMLEntityResult, sharedXHTMLEntityResult, 0,
    XML_INTERNAL_PREDEFINED_ENTITY, 0, 0, 0, 0, 0,
#if LIBXML_VERSION >= 20627
    // xmlEntity gained an extra member in 2.6.27.
    1
#endif
};

static xmlEntityPtr getXHTMLEntity(const xmlChar* name)
{
    UChar c = decodeNamedEntity(reinterpret_cast<const char*>(name));
    if (!c)
        return 0;

    CString value = String(&c, 1).utf8();
    ASSERT(value.length() < 5);
    sharedXHTMLEntity.length = value.length();
    sharedXHTMLEntity.name = name;
    memcpy(sharedXHTMLEntityResult, value.data(), sharedXHTMLEntity.length + 1);

    return &sharedXHTMLEntity;
}

static xmlEntityPtr getEntityHandler(void* closure, const xmlChar* name)
{
    xmlParserCtxtPtr ctxt = static_cast<xmlParserCtxtPtr>(closure);
    xmlEntityPtr ent = xmlGetPredefinedEntity(name);
    if (ent) {
        ent->etype = XML_INTERNAL_PREDEFINED_ENTITY;
        return ent;
    }

    ent = xmlGetDocEntity(ctxt->myDoc, name);
    if (!ent && (getTokenizer(closure)->isXHTMLDocument()
#if ENABLE(WML)
                 || getTokenizer(closure)->isWMLDocument()
#endif
       )) {
        ent = getXHTMLEntity(name);
        if (ent)
            ent->etype = XML_INTERNAL_GENERAL_ENTITY;
    }
    
    return ent;
}

static void startDocumentHandler(void* closure)
{
    xmlParserCtxt* ctxt = static_cast<xmlParserCtxt*>(closure);
    getTokenizer(closure)->startDocument(ctxt->version, ctxt->encoding, ctxt->standalone);
    xmlSAX2StartDocument(closure);
}

static void endDocumentHandler(void* closure)
{
    getTokenizer(closure)->endDocument();
    xmlSAX2EndDocument(closure);
}

static void internalSubsetHandler(void* closure, const xmlChar* name, const xmlChar* externalID, const xmlChar* systemID)
{
    getTokenizer(closure)->internalSubset(name, externalID, systemID);
    xmlSAX2InternalSubset(closure, name, externalID, systemID);
}

static void externalSubsetHandler(void* closure, const xmlChar*, const xmlChar* externalId, const xmlChar*)
{
    String extId = toString(externalId);
    if ((extId == "-//W3C//DTD XHTML 1.0 Transitional//EN")
        || (extId == "-//W3C//DTD XHTML 1.1//EN")
        || (extId == "-//W3C//DTD XHTML 1.0 Strict//EN")
        || (extId == "-//W3C//DTD XHTML 1.0 Frameset//EN")
        || (extId == "-//W3C//DTD XHTML Basic 1.0//EN")
        || (extId == "-//W3C//DTD XHTML 1.1 plus MathML 2.0//EN")
        || (extId == "-//W3C//DTD XHTML 1.1 plus MathML 2.0 plus SVG 1.1//EN")
        || (extId == "-//WAPFORUM//DTD XHTML Mobile 1.0//EN"))
        getTokenizer(closure)->setIsXHTMLDocument(true); // controls if we replace entities or not.
}

static void ignorableWhitespaceHandler(void*, const xmlChar*, int)
{
    // nothing to do, but we need this to work around a crasher
    // http://bugzilla.gnome.org/show_bug.cgi?id=172255
    // http://bugs.webkit.org/show_bug.cgi?id=5792
}

void XMLTokenizer::initializeParserContext(const char* chunk)
{
    xmlSAXHandler sax;
    memset(&sax, 0, sizeof(sax));

    sax.error = normalErrorHandler;
    sax.fatalError = fatalErrorHandler;
    sax.characters = charactersHandler;
    sax.processingInstruction = processingInstructionHandler;
    sax.cdataBlock = cdataBlockHandler;
    sax.comment = commentHandler;
    sax.warning = warningHandler;
    sax.startElementNs = startElementNsHandler;
    sax.endElementNs = endElementNsHandler;
    sax.getEntity = getEntityHandler;
    sax.startDocument = startDocumentHandler;
    sax.endDocument = endDocumentHandler;
    sax.internalSubset = internalSubsetHandler;
    sax.externalSubset = externalSubsetHandler;
    sax.ignorableWhitespace = ignorableWhitespaceHandler;
    sax.entityDecl = xmlSAX2EntityDecl;
    sax.initialized = XML_SAX2_MAGIC;
    m_parserStopped = false;
    m_sawError = false;
    m_sawXSLTransform = false;
    m_sawFirstElement = false;

    if (m_parsingFragment)
        m_context = createMemoryParser(&sax, this, chunk);
    else
        m_context = createStringParser(&sax, this);
}

void XMLTokenizer::doEnd()
{
#if ENABLE(XSLT)
    if (m_sawXSLTransform) {
        m_doc->setTransformSource(xmlDocPtrForString(m_doc->docLoader(), m_originalSourceForTransform, m_doc->url().string()));
        
        m_doc->setParsing(false); // Make the doc think it's done, so it will apply xsl sheets.
        m_doc->updateStyleSelector();
        m_doc->setParsing(true);
        m_parserStopped = true;
    }
#endif

    if (m_context) {
        // Tell libxml we're done.
        xmlParseChunk(m_context, 0, 0, 1);
        
        if (m_context->myDoc)
            xmlFreeDoc(m_context->myDoc);
        xmlFreeParserCtxt(m_context);
        m_context = 0;
    }
}

#if ENABLE(XSLT)
void* xmlDocPtrForString(DocLoader* docLoader, const String& source, const String& url)
{
    if (source.isEmpty())
        return 0;

    // Parse in a single chunk into an xmlDocPtr
    // FIXME: Hook up error handlers so that a failure to parse the main document results in
    // good error messages.
    const UChar BOM = 0xFEFF;
    const unsigned char BOMHighByte = *reinterpret_cast<const unsigned char*>(&BOM);

    xmlGenericErrorFunc oldErrorFunc = xmlGenericError;
    void* oldErrorContext = xmlGenericErrorContext;
    
    setLoaderForLibXMLCallbacks(docLoader);        
    xmlSetGenericErrorFunc(0, errorFunc);
    
    xmlDocPtr sourceDoc = xmlReadMemory(reinterpret_cast<const char*>(source.characters()),
                                        source.length() * sizeof(UChar),
                                        url.latin1().data(),
                                        BOMHighByte == 0xFF ? "UTF-16LE" : "UTF-16BE", 
                                        XSLT_PARSE_OPTIONS);
    
    setLoaderForLibXMLCallbacks(0);
    xmlSetGenericErrorFunc(oldErrorContext, oldErrorFunc);
    
    return sourceDoc;
}
#endif

int XMLTokenizer::lineNumber() const
{
    return m_context ? m_context->input->line : 1;
}

int XMLTokenizer::columnNumber() const
{
    return m_context ? m_context->input->col : 1;
}

void XMLTokenizer::stopParsing()
{
    Tokenizer::stopParsing();
    xmlStopParser(m_context);
}

void XMLTokenizer::resumeParsing()
{
    ASSERT(m_parserPaused);
    
    m_parserPaused = false;

    // First, execute any pending callbacks
    while (!m_pendingCallbacks->isEmpty()) {
        m_pendingCallbacks->callAndRemoveFirstCallback(this);
        
        // A callback paused the parser
        if (m_parserPaused)
            return;
    }

    // Then, write any pending data
    SegmentedString rest = m_pendingSrc;
    m_pendingSrc.clear();
    write(rest, false);

    // Finally, if finish() has been called and write() didn't result
    // in any further callbacks being queued, call end()
    if (m_finishCalled && m_pendingCallbacks->isEmpty())
        end();
}

bool parseXMLDocumentFragment(const String& chunk, DocumentFragment* fragment, Element* parent)
{
    if (!chunk.length())
        return true;

    XMLTokenizer tokenizer(fragment, parent);
    
    tokenizer.initializeParserContext(chunk.utf8().data());

    xmlParseContent(tokenizer.m_context);

    tokenizer.endDocument();

    // Check if all the chunk has been processed.
    long bytesProcessed = xmlByteConsumed(tokenizer.m_context);
    if (bytesProcessed == -1 || ((unsigned long)bytesProcessed) == sizeof(UChar) * chunk.length())
        return false;

    // No error if the chunk is well formed or it is not but we have no error.
    return tokenizer.m_context->wellFormed || xmlCtxtGetLastError(tokenizer.m_context) == 0;
}

// --------------------------------

struct AttributeParseState {
    HashMap<String, String> attributes;
    bool gotAttributes;
};

static void attributesStartElementNsHandler(void* closure, const xmlChar* xmlLocalName, const xmlChar* /*xmlPrefix*/,
                                            const xmlChar* /*xmlURI*/, int /*nb_namespaces*/, const xmlChar** /*namespaces*/,
                                            int nb_attributes, int /*nb_defaulted*/, const xmlChar** libxmlAttributes)
{
    if (strcmp(reinterpret_cast<const char*>(xmlLocalName), "attrs") != 0)
        return;
    
    xmlParserCtxtPtr ctxt = static_cast<xmlParserCtxtPtr>(closure);
    AttributeParseState* state = static_cast<AttributeParseState*>(ctxt->_private);
    
    state->gotAttributes = true;
    
    xmlSAX2Attributes* attributes = reinterpret_cast<xmlSAX2Attributes*>(libxmlAttributes);
    for(int i = 0; i < nb_attributes; i++) {
        String attrLocalName = toString(attributes[i].localname);
        int valueLength = (int) (attributes[i].end - attributes[i].value);
        String attrValue = toString(attributes[i].value, valueLength);
        String attrPrefix = toString(attributes[i].prefix);
        String attrQName = attrPrefix.isEmpty() ? attrLocalName : attrPrefix + ":" + attrLocalName;
        
        state->attributes.set(attrQName, attrValue);
    }
}

HashMap<String, String> parseAttributes(const String& string, bool& attrsOK)
{
    AttributeParseState state;
    state.gotAttributes = false;

    xmlSAXHandler sax;
    memset(&sax, 0, sizeof(sax));
    sax.startElementNs = attributesStartElementNsHandler;
    sax.initialized = XML_SAX2_MAGIC;
    xmlParserCtxtPtr parser = createStringParser(&sax, &state);
    String parseString = "<?xml version=\"1.0\"?><attrs " + string + " />";
    xmlParseChunk(parser, reinterpret_cast<const char*>(parseString.characters()), parseString.length() * sizeof(UChar), 1);
    if (parser->myDoc)
        xmlFreeDoc(parser->myDoc);
    xmlFreeParserCtxt(parser);
    attrsOK = state.gotAttributes;
    return state.attributes;
}

}