tes (9dc72a9e) · Commits · jan.koester / libhtmlpp

src/html.cpp

+71 −14

Original line number	Diff line number	Diff line
		@@ -100,9 +100,6 @@ libhtmlpp::HtmlString::HtmlString(const std::string& str) : HtmlString(){
		}

		libhtmlpp::HtmlString::~HtmlString(){
		while(!_Childs.empty()){
		_Childs.pop();
		}
		}

		libhtmlpp::HtmlString::HtmlString(const libhtmlpp::HtmlString& str) : HtmlString() {
		@@ -128,10 +125,6 @@ void libhtmlpp::HtmlString::insert(size_t pos, char src){
		}

		void libhtmlpp::HtmlString::clear(){
		while(!_Childs.empty()){
		_Childs.pop();
		}
		_Data.clear();
		}

		bool libhtmlpp::HtmlString::empty(){
		@@ -235,8 +228,68 @@ const std::vector<char>& libhtmlpp::HtmlString::data() {

		libhtmlpp::Element &libhtmlpp::HtmlString::parse() {
		HTMLException excp;
		Element &el=_buildTree();
		return el;
		_buildTree();
		return *rootEl;
		}

		void printElement(libhtmlpp::Element* el, int depth, std::ostream& out) {
		if (!el) {
		return;
		}

		// 1. Print Indentation
		for (int i = 0; i < depth; ++i) {
		out << "\| "; // Use a pipe and spaces for visual depth
		}

		// 2. Print Node Information
		if (el->isHtmlElement()) {
		libhtmlpp::HtmlElement* html_el = static_cast<libhtmlpp::HtmlElement*>(el);
		out << "├── <" << html_el->getTag() << ">";

		// Print attributes (optional, but helpful for debugging)
		const auto& attrs = html_el->getAttributeList();
		if (!attrs.empty()) {
		out << " [Attrs: " << attrs.size() << "]";
		}
		out << std::endl;

		} else if (el->isTextElement()) {
		libhtmlpp::TextElement* text_el = static_cast<libhtmlpp::TextElement*>(el);
		// Print a snippet of text content, trimming whitespace
		std::string content = text_el->getContent();
		// Remove leading/trailing whitespace for clean output
		size_t first = content.find_first_not_of(" \t\n\r");
		size_t last = content.find_last_not_of(" \t\n\r");
		if (first != std::string::npos) {
		content = content.substr(first, (last - first + 1));
		}

		out << "├── #TEXT: \"";
		if (content.length() > 30) {
		out << content.substr(0, 27) << "...\"";
		} else if (!content.empty()) {
		out << content << "\"";
		} else {
		out << "(whitespace only)\"";
		}
		out << std::endl;

		} else {
		out << "├── Unknown Element Type" << std::endl;
		}

		// 3. Recursive Traversal

		// A. Recurse into the first child
		if (el->_childElement) {
		printElement(el->_childElement.get(), depth + 1, out);
		}

		// B. Move to the next sibling
		if (el->_nextElement) {
		printElement(el->_nextElement.get(), depth, out);
		}
		}

		void libhtmlpp::HtmlString::_buildtreenode(std::unique_ptr<libhtmlpp::DocElements> &firstel,libhtmlpp::DocElements *lastel){
		@@ -328,10 +381,13 @@ NEXTDOCEL:
		childel.start=next->nextel.get();
		childel.end=parent;
		cpylist.push(childel);
		next=parent;
		next=parent->nextel.get();
		}

		while(next){
		if(next && next->element)
		std::cout << next->element->getType() << std::endl;

		while(next && next->element){
		if(!next->terminator){
		start->element->_nextElement=std::move(next->element);
		prev=start;
		@@ -356,10 +412,11 @@ NEXTDOCEL:
		}
		}

		libhtmlpp::Element &libhtmlpp::HtmlString::_buildTree() {
		void libhtmlpp::HtmlString::_buildTree() {
		DocElements *lastEl = nullptr;
		std::unique_ptr<DocElements> firstEl=nullptr;

		auto addelement = [this](DocElements **last){
		auto addelement = [&firstEl](DocElements **last){
		if (!firstEl.get()) {
		firstEl = std::make_unique<DocElements>();
		*last = firstEl.get();
		@@ -499,7 +556,7 @@ BUILDTREE_STARTLOOP:

		_buildtreenode(firstEl,lastEl);

		return *firstEl->element;
		rootEl=std::move(firstEl->element);
		}

		void libhtmlpp::HtmlString::_serialelize(std::vector<char> in, libhtmlpp::HtmlElement *out) {

src/html.h

+2 −3

Original line number	Diff line number	Diff line
		@@ -288,13 +288,12 @@ namespace libhtmlpp {
		libhtmlpp::Element &parse();

		private:
		std::unique_ptr<DocElements> firstEl;
		std::unique_ptr<Element> rootEl;
		void _serialelize(std::vector<char> in, HtmlElement* out);
		Element &_buildTree();
		void _buildTree();
		void _buildtreenode(std::unique_ptr<DocElements> &firstel,DocElements *lastel);
		std::vector<char> _Data;
		std::string _Str;
		std::stack<Element*> _Childs;
		friend void HtmlEncode(const std::string &input,HtmlString *output);
		friend class HtmlPage;
		};