single tag fixes and implementation into tree, style clean up

src/html/HTMLParser.cpp

@@ -36,11 +36,11 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
     std::vector<unsigned int> starts;
     unsigned int cursor;
     int state = 0;
-    int prependWhiteSpace=false;
+    int prependWhiteSpace = false;
     for (cursor = 0; cursor < html.length(); cursor++) { // TODO handle trying to look ahead past string
         if (state == 0) { // Outside tag
             if (html[cursor] == ' ' || html[cursor] == '\t' || html[cursor] == '\r' || html[cursor] == '\n') {
-                prependWhiteSpace=true;
+                prependWhiteSpace = true;
                 continue;
             }
             else if (html[cursor] == '<') {
@@ -48,15 +48,16 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
                     state = 4;
                 }
                 else if (html[cursor + 1] == '/') {
+                    // start closing tag
                     if (currentNode && currentNode->parent) {
                       currentNode = currentNode->parent;
                     } else {
                       std::cout << "HTMLParser::Parse - currentNode/parent is null - close tag" << std::endl;
                     }
                     state = 1; // ignore closing tags
+                    //starts.push_back(cursor);
                 }
-                // don't need to ignore these tags
-                /*
+                // these don't have closing tags
                 else if (
                     (html[cursor + 1] == 'h' && html[cursor + 2] == 'r') ||
                     (html[cursor + 1] == 'b' && html[cursor + 2] == 'r') ||
@@ -66,9 +67,36 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
                     (html[cursor + 1] == 'm' && html[cursor + 2] == 'e' && html[cursor + 3] == 't' && html[cursor + 4] == 'a') ||
                     (html[cursor + 1] == 'i' && html[cursor + 2] == 'n' && html[cursor + 3] == 'p' && html[cursor + 4] == 'u' && html[cursor + 5] == 't')
                     ) {
-                    state = 1;
+                    std::shared_ptr<TagNode> tagNode = std::make_shared<TagNode>();
+                    if (currentNode) {
+                        currentNode->children.push_back(tagNode);
+                        tagNode->parent = currentNode;
+                    } else {
+                        std::cout << "HTMLParser::Parse - currentNode is null - tagNode" << std::endl;
+                    }
+                    currentNode = tagNode;
+                    size_t closeTagPos = html.substr(cursor + 1).find(">");
+                    //std::cout << "found closeTagPos at " << closeTagPos << std::endl;
+                    if (closeTagPos == std::string::npos) {
+                        std::cout << "HTMLParser::Parse - can't find closing tag for single tag" << std::endl;
+                        cursor ++;
+                    } else {
+                        std::string element = html.substr(cursor, closeTagPos + 2);
+                        //std::cout << "creating element, tag: " << element << std::endl;
+                        parseTag(element, *dynamic_cast<TagNode*>(currentNode.get()));
+                        cursor += 2 + closeTagPos;
+                    }
+                    
+                    // drop back
+                    if (currentNode && currentNode->parent) {
+                        currentNode = currentNode->parent;
+                    } else {
+                        std::cout << "HTMLParser::Parse - currentNode/parent is null - textNode state3" << std::endl;
+                    }
+                    
+                    prependWhiteSpace = false;
+                    state = 0;
                 }
-                 */
                 else {
                     std::shared_ptr<TagNode> tagNode = std::make_shared<TagNode>();
                     if (currentNode) {
@@ -99,15 +127,18 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
         }
         else if (state == 1) { // Skip Over Element
             if (html[cursor] == '>') {
+                //std::string element = html.substr(starts.back(), cursor - starts.back() + 1);
+                //starts.pop_back();
+                //std::cout << "HTMLParser::parse - close tag: " << element << std::endl;
                 state = 0;
-                prependWhiteSpace=false;
+                prependWhiteSpace = false;
             }
         }
         else if (state == 4) { // HTML Comment
             if (html[cursor] == '-' && html[cursor + 1] == '-' && html[cursor + 2] == '>') {
                 state = 0;
                 cursor += 2; // advance cursor to end of comment
-                prependWhiteSpace=false;
+                prependWhiteSpace = false;
             }
         }
         else if (state == 2) { // Tag
@@ -115,8 +146,9 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
                 std::string element = html.substr(starts.back(), cursor - starts.back() + 1);
                 starts.pop_back();
                 parseTag(element, *dynamic_cast<TagNode*>(currentNode.get()));
+                //std::cout << "HTMLParser::parse - close tag: " << element << std::endl;
                 state = 0;
-                prependWhiteSpace=false;
+                prependWhiteSpace = false;
             }
         }
         else if (state == 3) { // Text
@@ -129,7 +161,7 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
                   std::cout << "HTMLParser::Parse - currentNode/parent is null - textNode state3" << std::endl;
                 }
                 state = 0;
-                prependWhiteSpace=false;
+                prependWhiteSpace = false;
             }
         }
     }
@@ -139,6 +171,7 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
 }
 
 void HTMLParser::parseTag(const std::string &element, TagNode &tagNode) const {
+    //std::cout << "HTMLParser::parseTag - element [" << element << "]" << std::endl;
     unsigned int cursor;
     unsigned int start = 1; // skip first <
     int state = 0;