new serielaze parser (637bf295) · Commits · jan.koester / libhtmlpp

src/html.cpp

+72 −48

Original line number	Diff line number	Diff line
		@@ -368,7 +368,7 @@ NEXTDOCEL:
		goto NEXTDOCEL;
		}
		}
		#include <iostream>

		libhtmlpp::Element *libhtmlpp::HtmlString::_buildTree() {
		DocElements firstEl = nullptr, lastEl = nullptr;

		@@ -421,7 +421,7 @@ libhtmlpp::Element *libhtmlpp::HtmlString::_buildTree() {
		++ii;
		}

		size_t close=ii;
		size_t close=++ii;

		while(ii<_Data.size()){
		if( strncmp(_Data.data()+ii,"</script",8)==0 ) {
		@@ -438,8 +438,7 @@ libhtmlpp::Element *libhtmlpp::HtmlString::_buildTree() {
		while(ii<_Data.size()){
		if(_Data[ii]==HTMLTAG_TERMINATE){
		lastEl->terminator=true;
		}
		if(_Data[ii]==HTMLTAG_CLOSE) {
		}else if(_Data[ii]==HTMLTAG_CLOSE) {
		std::vector<char> tel;
		std::copy(_Data.begin()+start,_Data.begin()+ii,std::back_inserter(tel));
		_serialelize(tel,(HtmlElement*)lastEl->element);
		@@ -448,15 +447,15 @@ libhtmlpp::Element *libhtmlpp::HtmlString::_buildTree() {
		++ii;
		}
		}
		}else if(_Data[ii]!='\r' \|\| _Data[ii]!='\n' \|\| _Data[ii]!=' '){
		}else if(_Data[ii]!='\r' && _Data[ii]!='\n' && _Data[ii]!=' '){
		size_t start=ii;
		while(ii<_Data.size()){
		if(_Data[ii]==HTMLTAG_OPEN) {
		--ii;
		std::vector<char> buf;
		std::copy(_Data.begin()+start,_Data.begin()+ii,std::back_inserter(buf));
		addelement(&firstEl,&lastEl,new TextElement());
		std::copy(buf.begin(),buf.end(),std::back_inserter(((TextElement*)(lastEl->element))->_Text));
		--ii;
		break;
		}
		++ii;
		@@ -464,11 +463,6 @@ libhtmlpp::Element *libhtmlpp::HtmlString::_buildTree() {
		}
		}

		int iii=0;
		for(DocElements *cext=firstEl; cext; cext=cext->nextel){
		std::cout << iii++ << std::endl;
		}

		Element *first = firstEl->element;

		_buildtreenode(firstEl,lastEl);
		@@ -481,7 +475,7 @@ libhtmlpp::Element *libhtmlpp::HtmlString::_buildTree() {
		ft->element->_firstElement=nullptr;
		ft->element->_prevElement=nullptr;
		ft->element->_nextElement=nullptr;
		delete ft->element;
		// delete ft->element;
		}
		delete ft;
		ft=next;
		@@ -519,7 +513,7 @@ void libhtmlpp::HtmlString::_serialelize(std::vector<char> in, libhtmlpp::HtmlEl

		std::vector<char> tag;

		std::copy(in.begin()+st,in.begin()+et,std::insert_iterator<std::vector<char>>(tag,tag.begin()) );
		std::copy(in.begin()+st,in.begin()+et,std::back_inserter(tag));
		tag.push_back('\0');

		out->setTagname(tag.data());
		@@ -529,38 +523,46 @@ void libhtmlpp::HtmlString::_serialelize(std::vector<char> in, libhtmlpp::HtmlEl
		throw excp[HTMLException::Critical] << "no tag in element found!";
		}

		int startpos =-1,vst=-1;
		for(size_t i=et; i<in.size(); ++i){
		bool value=false;
		std::vector<char> key;

		while(et < in.size()) {
		if(in[et]==' ' \|\| in[et]=='>' \|\| in[et]=='=') {
		if(startpos!=-1 && !value){
		std::copy(in.begin()+startpos,in.begin()+et,std::insert_iterator<std::vector<char>>(key,key.begin()) );
		out->setAttribute(key.data(),key.size(),nullptr,0);
		startpos=-1;
		}
		if(in[et]=='='){
		if(in[i]!=' '){
		size_t kstart=i,kend=in.size(),vstart=0,vend=0;
		for(size_t ii=i; ii<in.size(); ++ii){
		if (in[ii]=='='){
		value=true;
		i=ii;
		kend=ii;
		break;
		}
		}else if(in[et]=='\"') {
		if( vst==-1 ){
		vst=et;
		}else if(!key.empty() && value){
		std::vector<char> val;
		++vst;
		std::copy(in.begin()+vst,in.begin()+et,std::insert_iterator<std::vector<char>>(val,val.begin()) );
		out->setAttribute(key.data(),key.size(),val.data(),val.size());
		key.clear();
		vst=-1;
		value=false;
		}
		if(value){
		while(i<in.size()){
		if(in[i]=='\"'){
		if (vstart==0){
		vstart=++i;
		continue;
		}
		vend=i;
		break;
		}
		++i;
		}
		}else{
		if(startpos==-1 && !value){
		startpos=et;
		while(i<in.size()){
		if(in[i] == '>' \|\| in[i] == ' '){
		kend=i;
		break;
		}
		++i;
		}
		}
		std::vector<char> key,val;
		std::copy(in.begin()+kstart,in.begin()+kend,std::back_inserter(key));
		if(vend!=0){
		std::copy(in.begin()+vstart,in.begin()+vend,std::back_inserter(val));
		}
		out->setAttribute(key.data(),key.size(),val.data(),val.size());
		}
		};
		++et;
		}
		}

		@@ -965,6 +967,12 @@ void libhtmlpp::Element::insertBefore(libhtmlpp::Element* el){
		_prevElement->_nextElement=cmt;
		break;
		}
		case ScriptEL:{
		ScriptElement *script= new ScriptElement;
		_copy(script,el);
		_prevElement->_nextElement=script;
		break;
		}
		}
		Element nexel=_prevElement->_nextElement,prev=nullptr;

		@@ -993,6 +1001,10 @@ void libhtmlpp::Element::insertAfter(libhtmlpp::Element* el){
		_nextElement=new CommentElement;;
		break;
		}
		case ScriptEL:{
		_nextElement=new ScriptElement;;
		break;
		}
		}

		_copy(_nextElement,el);
		@@ -1428,12 +1440,24 @@ void libhtmlpp::print(Element* el, HtmlString &output,bool formated) {
		}

		output.append(">");

		if(formated){
		output.append("\r\n");
		for(int i=0; i<lvl+1; ++i){
		output.append(" ");
		}
		}
		output.append(((ScriptElement)el)->_Script.data(),((ScriptElement)el)->_Script.size());

		if(formated){
		output.append("\r\n");
		for(int i=0; i<lvl; ++i){
		output.append(" ");
		}
		}
		output.append("</");
		output.append(((HtmlElement*) el)->getTagname());
		output.append(((ScriptElement*) el)->getTagname());
		output.append(">");
		if(formated)
		output.append("\r\n");
		}break;
		default:
		HTMLException excp;
		@@ -1473,7 +1497,7 @@ libhtmlpp::HtmlElement libhtmlpp::HtmlElement::getElementbyID(const char id) c
		std::stack <Element*> childs;
		const Element *curel=this;
		SEARCHBYID:
		if(curel->getType()==HtmlEl){
		if(curel->getType()==HtmlEl \|\| curel->getType()== ScriptEL){
		if(((HtmlElement*)curel)->_childElement){
		childs.push(((HtmlElement*)curel)->_childElement);
		}
		@@ -1501,7 +1525,7 @@ libhtmlpp::HtmlElement libhtmlpp::HtmlElement::getElementbyTag(const char tag)
		std::stack <Element*> childs;
		const Element *curel=this;
		SEARCHBYTAG:
		if(curel->getType()==HtmlEl){
		if(curel->getType()==HtmlEl \|\| curel->getType()== ScriptEL){
		if(((HtmlElement*)curel)->_childElement){
		childs.push(((HtmlElement*)curel)->_childElement);
		}