parser otimized (88593bdb) · Commits · jan.koester / libhtmlpp

src/html.cpp

+63 −62

Changes for src/html.cpp: 63 added lines, 62 removed lines.
Original line number	Diff line number	Diff line
		@@ -52,12 +52,6 @@ libhtmlpp::HtmlString::HtmlString(){
		_InitString();
		}

		libhtmlpp::HtmlString::HtmlString(char *header){
		_InitString();
		_HtmlHeader=header;

		}

		libhtmlpp::HtmlString::~HtmlString(){
		clear();
		}
		@@ -140,22 +134,23 @@ libhtmlpp::HtmlString &libhtmlpp::HtmlString::operator<<(char src){
		return *this;
		}

		void libhtmlpp::HtmlString::_printHtml(HtmlElement* child, HtmlElement* Parent) {
		void libhtmlpp::HtmlString::_printHtml(HtmlElement* child) {
		for (HtmlElement* cur = child; cur; cur = cur->_nextElement) {
		if (cur->_Child) {
		_printHtml(cur->_Child);
		_Cstr.append("</");
		_Cstr.append(cur->_TagName.c_str());
		_printHtml(nullptr, cur->_Child);
		} else if(Parent){
		_Cstr.append(Parent->_TagName.c_str());
		} else {
		_Cstr.append(cur->_TagName.c_str());
		_Cstr.append(">");
		}
		_Cstr.append("<");
		_Cstr.append(cur->_TagName.c_str());
		_Cstr.append(">");
		}
		}

		const char *libhtmlpp::HtmlString::c_str() {
		_Cstr.clear();
		_printHtml(_HtmlRootNode,nullptr);
		_printHtml(_HtmlRootNode);
		sys::cout << _Cstr << sys::endl;
		return _Data.c_str();
		}
		@@ -172,24 +167,31 @@ size_t libhtmlpp::HtmlString::length() {
		void libhtmlpp::HtmlString::parse(){
		HTMLException excp;
		_parseTree();
		_HtmlHeader.clear();
		delete _HtmlRootNode;
		ssize_t pos = 0;
		_HtmlRootNode = _buildTree(pos);
		}

		libhtmlpp::HtmlElement *libhtmlpp::HtmlString::_buildTree(ssize_t &pos){

		HtmlElement *firsthel=nullptr;
		HtmlElement *lasthel= nullptr;

		struct el {
		sys::array<char> data;
		struct el* terminator;
		HtmlElement* elhtml;
		bool terminator;
		struct el* nextel;
		struct el* prevel;

		}firstEl=nullptr,lastEl=nullptr;
		el() {
		nextel = nullptr;
		prevel = nullptr;
		elhtml = nullptr;
		terminator = false;
		}

		~el() {
		delete nextel;
		}

		}firstEl=nullptr,lastEl=nullptr,*before=nullptr;

		int startpos=0;

		@@ -197,66 +199,65 @@ libhtmlpp::HtmlElement *libhtmlpp::HtmlString::_buildTree(ssize_t &pos){
		if (!firstEl) {
		firstEl = new el;
		lastEl = firstEl;
		lastEl->prevel = nullptr;
		}else {

		lastEl->nextel = new el;
		lastEl->nextel->prevel = lastEl;
		lastEl = lastEl->nextel;
		}
		lastEl->data = _Data.substr((_HTable[i][0]+1),((_HTable[i][2]-_HTable[i][0])-1));
		lastEl->nextel = nullptr;
		lastEl->terminator = nullptr;

		lastEl->data = _Data.substr((_HTable[i][0]+1),(_HTable[i][2]-(_HTable[i][0]+1)));

		if (lastEl->data[0] == '/')
		lastEl->terminator = true;

		_serialelize(lastEl->data, &lastEl->elhtml);
		}

		int lvl = 0;
		HtmlElement end,cur,*before;
		for (el* curel = lastEl; curel; curel=curel->prevel) {
		if (curel->data[0]=='/') {
		_serialelize(curel->data.substr(1,(curel->data.length()-1)), &end);

		for (el* fcurel = firstEl; fcurel; fcurel = fcurel->nextel) {
		HtmlElement* firsthel = nullptr, * lasthel = nullptr;

		_serialelize(fcurel->data,&cur);
		for (el* curel = firstEl; curel; curel = curel->nextel) {
		if (curel->terminator) {
		for (el* begin = curel->prevel; begin; begin=begin->prevel) {
		if (curel && curel!=begin && (curel->elhtml->_TagName == begin->elhtml->_TagName) ) {

		if (end && end->_TagName == cur->_TagName) {
		delete end;
		end = nullptr;
		cur->_Child = before;
		} else if(end){
		el* nexel = curel->nextel;
		curel->elhtml->_nextElement = nullptr;
		curel->nextel = nullptr;
		delete curel;
		curel = nexel;
		begin->elhtml->_Child = curel->elhtml;
		break;
		}
		}
		}
		if (lasthel) {
		lasthel->_nextElement = cur;
		lasthel->_nextElement->_prevElement = lasthel;
		lasthel->_nextElement = curel->elhtml;
		lasthel = lasthel->_nextElement;
		lasthel->_prevElement = before->elhtml;
		} else {
		firsthel = cur;
		firsthel->_nextElement = nullptr;
		firsthel->_prevElement = nullptr;
		firsthel = curel->elhtml;
		lasthel = firsthel;
		}
		} else {
		HTMLException excp;
		throw excp[HTMLException::Critical] << "htmltree wrong!";
		}
		before = cur;
		}
		before = curel;
		}
		}

		_HtmlRootNode = firsthel;

		return nullptr;
		return firsthel;
		}

		void libhtmlpp::HtmlString::_serialelize(sys::array<char> in, HtmlElement **out) {
		int i;
		int i,s=0;

		for (i = 0; i < in.length(); ++i) {
		if (in[i] == ' ') {
		*out = new HtmlElement(in.substr(0, i).c_str());
		if(in[i] == '/'){
		++s;
		continue;
		}else if (in[i] == ' ') {
		*out = new HtmlElement(in.substr(s, i-s).c_str());
		break;
		}else if (i == (in.length()-1)){
		*out = new HtmlElement(in.substr(0, i).c_str());
		*out = new HtmlElement(in.substr(s, i-s).c_str());
		break;
		}
		}
		@@ -416,19 +417,19 @@ void libhtmlpp::HtmlPage::loadFile(const char* path){
		throw excp[HTMLException::Critical] << e.what();
		}

		int length = fs.getsize();

		size_t readed = 0;
		while (readed < fs.getsize()) {
		char buf[HTML_BLOCKSIZE+1];
		if(size_t rd = fs.read(buf,HTML_BLOCKSIZE)){
		int rd = fs.read(buf,HTML_BLOCKSIZE);
		readed += rd;
		if (rd > 0) {
		buf[rd] = '\0';
		char *dest;
		size_t cdd=cleannewline(buf,rd,&dest);
		_HtmlDocument->assign(dest,cdd);
		delete[] dest;
		_HtmlDocument->assign(buf,rd);
		} else {
		HTMLException excp;
		throw excp[HTMLException::Critical] << "Could not read file";
		}
		}
		_HtmlDocument->parse();
		}

src/html.h

+2 −3

Changes for src/html.h: 2 added lines, 3 removed lines.
Original line number	Diff line number	Diff line
		@@ -59,6 +59,7 @@ namespace libhtmlpp {

		private:
		sys::array<char> _TagName;
		sys::array<char> _Text;
		HtmlAttributes *_firstAttr;
		HtmlAttributes *_lastAttr;

		@@ -68,7 +69,6 @@ namespace libhtmlpp {
		class HtmlString {
		public:
		HtmlString();
		HtmlString(char *header);
		~HtmlString();

		void assign(const char *src,size_t srcsize);
		@@ -97,14 +97,13 @@ namespace libhtmlpp {
		private:
		void _InitString();
		void _parseTree();
		void _printHtml(HtmlElement* child, HtmlElement* Parent);
		void _printHtml(HtmlElement* child);
		void _serialelize(sys::array<char> in,HtmlElement **out);
		HtmlElement *_buildTree(ssize_t &pos);
		sys::array<char> _Data;
		sys::array<char> _Cstr;
		ssize_t **_HTable;
		size_t _HTableSize;
		sys::array<char> _HtmlHeader;
		HtmlElement *_HtmlRootNode;
		};

test/htmlfiles/right.html

+1 −0

Changes for test/htmlfiles/right.html: 1 added line, 0 removed lines.
Original line number	Diff line number	Diff line
		@@ -7,6 +7,7 @@
		<script src="script.js"></script>
		</head>
		<body>
		<span>test text</span>
		<!-- page content -->
		</body>
		</html>