void DocumentBuilder::consumeContent(Node *root){
while(peek() != -1){
consumeText(root);
if (isElement()){
consumeElement(root);
continue;
}
if (isComment()){
consumeComment(root);
continue;
}
if (isCDataSection()){
consumeCDataSection(root);
continue;
}
if(isPI()){
consumePI(root);
continue;
}
if (isCharRef()){
StringBuffer *sb = new StringBuffer(2);
sb->append(consumeCharRef());
appendToLastTextNode(root, sb);
continue;
}
if (isEntityRef()){
String *entext = consumeEntityRef(true);
appendToLastTextNode(root, entext);
continue;
}
if (peek(0) == '<') break;
};
}
void DocumentBuilder::consumeMisc(Node *root){
consumeSpaces();
bool hasTokens = true;
while(hasTokens){
if(isComment()){
consumeComment(root);
consumeSpaces();
}else if(isPI()){
consumePI(root);
consumeSpaces();
}else{
hasTokens = false;
}
}
}
// skip whitespace, then skip a comment, rinse, repeat
void TInputScanner::consumeWhitespaceComment(bool& foundNonSpaceTab)
{
do {
consumeWhiteSpace(foundNonSpaceTab);
// if not starting a comment now, then done
int c = peek();
if (c != '/' || c == EndOfInput)
return;
// skip potential comment
foundNonSpaceTab = true;
if (! consumeComment())
return;
} while (true);
}
void Tokenizer::shift()
{
mpCurrent.reset();
if (mBlockComment)
{
skipEOL();
}
else
{
skipWhiteSpaces();
if (mpInput->eof())
return;
}
mpCurrent.reset(new Token);
mpCurrent->setLine(line());
mpCurrent->setBeginColumn(column());
if (mBlockComment)
{
consumeCStyleBlockComment();
return;
}
auto ch = mpInput->get();
if (isLetter(ch) || isUnderscore(ch))
{
consumeIdentifier(ch);
}
else if (isBiwiseOperatorSymbol(ch))
{
mpCurrent->setType(Token::TYPE_BITWISE_OPERATOR);
absorbed(ch);
mpCurrent->addChar(ch);
mpCurrent->setEndColumn(column());
}
else if (isDot(ch) && consumeDot(ch))
{
// nothing
}
else if ((isDecimalDigit(ch) || isDot(ch) || isSign(ch)) && consumeNumber(ch))
{
// nothing
}
else if (isQuotationMark(ch))
{
if (!consumeString(ch))
shift();
}
else if (isCStyleInitialCommentChar(ch))
{
consumeComment(ch);
}
else if (isArrowSymbol(ch) && consumeArrow(ch))
{
// nothing
}
else if (isBracket(ch))
{
mpCurrent->setType(Token::TYPE_BRACKET);
absorbed(ch);
mpCurrent->addChar(ch);
mpCurrent->setEndColumn(column());
}
else if (isAngleBracket(ch))
{
mpCurrent->setType(Token::TYPE_ANGLE_BRACKET);
absorbed(ch);
mpCurrent->addChar(ch);
mpCurrent->setEndColumn(column());
}
else if (isDelimiter(ch))
{
mpCurrent->setType(Token::TYPE_DELIMITER);
absorbed(ch);
mpCurrent->addChar(ch);
mpCurrent->setEndColumn(column());
}
else if (isOperator(ch) && consumeEqualOperator(ch))
{
// nothing
}
else if (isOperator(ch))
{
mpCurrent->setType(Token::TYPE_OPERATOR);
absorbed(ch);
mpCurrent->addChar(ch);
mpCurrent->setEndColumn(column());
}
else if (isAsterisk(ch))
{
mpCurrent->setType(Token::TYPE_ASTERISK);
absorbed(ch);
mpCurrent->addChar(ch);
mpCurrent->setEndColumn(column());
}
else
{
mpCurrent.reset();
}
}
char* XMLStack::parseXMLStream(char *stream, XMLElement *parent)
{
if(!stream || !parent)
return nullChar;
NSString token;
if(state == XML_FAILED || state == XML_BUFFER_OVERFLOW)
return nullChar;
if(!consumeWhiteSpaces(&stream))
return nullChar;
if(*stream == '<' &&
(!consumeComment(&stream) &&
!consumeXMLHeader(&stream)))
{
stream++;
if(!moveAndCheckProgress()) return nullChar;
getIdentifier(&stream, token);
parent->setName(token);
if(state == XML_FAILED || state == XML_BUFFER_OVERFLOW)
return nullChar;
if(*stream == '>')
{
stream++;
if(!moveAndCheckProgress()) return nullChar;
}
else
{
while((*stream != stx_getOsSeparator() && *(stream + 1) != '>') &&
(*stream != '>'))
{
XMLElement *newTag = new XMLElement();
getIdentifier(&stream, token);
newTag->setName(token);
if(state == XML_FAILED || state == XML_BUFFER_OVERFLOW)
return nullChar;
getStringValue(&stream, token);
newTag->setValue(token);
if(state == XML_FAILED || state == XML_BUFFER_OVERFLOW)
return nullChar;
parent->addChild(newTag);
}
if(*stream == stx_getOsSeparator() && *(stream + 1) == '>')
{
stream +=2;
if(!moveAndCheckProgress(2)) return nullChar;
return stream;
}
stream++;
}
if(parent->getName() != "RawData")
{
getStreamedValue(&stream, token);
parent->setValue(token);
}
else
{
XMLElement *info = parent->getChildByName("length");
bool sucess = (!info) ? fillRawData(&stream, &parent->rawData) :
fillRawData(&stream, &parent->rawData,
info->getValuei());
if(!sucess)
writeFatalLogInfo("\n<!>Failed to load raw data, check your XML file.\n");
if(state == XML_FAILED || state == XML_BUFFER_OVERFLOW)
return nullChar;
}
if(!consumeWhiteSpaces(&stream))
return nullChar;
consumeComment(&stream);
while(*stream == '<' && *(stream + 1) != stx_getOsSeparator())
{
XMLElement *newTag = new XMLElement();
stream = parseXMLStream(stream, newTag);
if(!stream || !consumeWhiteSpaces(&stream))
return nullChar;
parent->addChild(newTag);
}
if(*stream == '<' && *(stream + 1)== stx_getOsSeparator())
{
stream+=2;
if(!moveAndCheckProgress(2)) return nullChar;
getIdentifier(&stream, token);
if(state == XML_FAILED || state == XML_BUFFER_OVERFLOW)
return nullChar;
}
if(token != parent->getName() || *stream != '>')
{
/*
//writeFatalLogInfoList
LOG_PRINT("\n<!>Start tag <%s> & End Tag <%s> do not match,"\
"check your XML file.\n",
parent->getName(), token);
??? */
return nullChar;
}
stream++;
if(!consumeWhiteSpaces(&stream)) return nullChar;
}
return (state == XML_SUCCESS) ? stream : nullChar;
}
void DocumentBuilder::consumeDTD(){
wchar c1 = peek(0);
wchar c2 = peek(1);
if (c1 != '<' || c2 != '!') return;
consume("<!DOCTYPE", 9);
consumeSpaces(1);
delete consumeNCName();
consumeSpaces();
if (peek() == 'S'){
consume("SYSTEM", 6);
consumeSpaces(1);
delete consumeQoutedValue();
}else if (peek() == 'P'){
consume("PUBLIC", 6);
consumeSpaces(1);
delete consumeQoutedValue();
consumeSpaces(1);
delete consumeQoutedValue();
}
consumeSpaces();
//markup decl
if (peek() == '['){
consume("[", 1);
while(peek() != ']'){
if (peek(0) == '<' && peek(1) == '!' && peek(2) == 'E'){
consume("<!ENTITY", 8);
consumeSpaces(1);
String *entityName = consumeNCName();
consumeSpaces(1);
String *entityValue = null;
String *extEntityValue = null;
if (peek() == 'S'){
consume("SYSTEM", 6);
consumeSpaces(1);
extEntityValue = consumeQoutedValue();
}else if (peek() == 'P'){
consume("PUBLIC", 6);
consumeSpaces(1);
delete consumeQoutedValue();
consumeSpaces(1);
extEntityValue = consumeQoutedValue();
}else{
entityValue = consumeQoutedValue();
}
if (entityValue != null){
entitiesHash.put(entityName, entityValue);
}
if (extEntityValue != null){
extEntitiesHash.put(entityName, extEntityValue);
}
delete entityName;
}else if (isComment()){
consumeComment(null);
}
get();
}
consume("]", 1);
consumeSpaces();
}
consume(">", 1);
}