notes and debug

src/html/HTMLParser.cpp

@@ -98,11 +98,13 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
             else if (html[cursor] == '<') {
                 // HTML comments
                 if (html[cursor + 1] == '!' && html[cursor + 2] == '-'  && html[cursor + 3] == '-' ) {
+                    //std::cout << "HTMLParser::Parse - starting HTML comment at " << cursor << std::endl;
                     state = 4;
                 }
                 // close tag
                 else if (html[cursor + 1] == '/') {
                     // start closing tag
+                    //std::cout << "HTMLParser::Parse - starting closing tag at " << html.substr(cursor, 7)  << std::endl;
                     if (currentNode && currentNode->parent) {
                       // we should snap to the level we started at (as we maybe a couple levels deep <ul><li></ul>
                       // but it's the matching part of this tag
@@ -111,7 +113,6 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
                       std::cout << "HTMLParser::Parse - currentNode/parent is null - close tag" << std::endl;
                     }
                     state = 1; // ignore closing tags
-                    //starts.push_back(cursor);
                 }
                 // these have never have a closing tag
                 else if (
@@ -123,6 +124,7 @@ 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')
                     ) {
+                    //std::cout << "HTMLParser::Parse - Starting single tag " << html.substr(cursor, 6)  << std::endl;
                     std::shared_ptr<TagNode> tagNode = std::make_shared<TagNode>();
                     if (currentNode) {
                         currentNode->children.push_back(tagNode);
@@ -138,7 +140,7 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
                         cursor ++;
                     } else {
                         std::string element = html.substr(cursor, closeTagPos + 2);
-                        //std::cout << "creating element, tag: " << element << std::endl;
+                        //std::cout << "HTMLParser::Parse - creating element, tag: " << element << std::endl;
                         parseTag(element, *dynamic_cast<TagNode*>(currentNode.get()));
                         cursor += 2 + closeTagPos;
                     }
@@ -155,6 +157,7 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
                 }
                 // start tag (<bob> <bob part)
                 else {
+                    //std::cout << "HTMLParser::Parse - start oc tag " << html.substr(cursor, 6) << std::endl;
                     std::shared_ptr<TagNode> tagNode = std::make_shared<TagNode>();
                     if (currentNode) {
                       currentNode->children.push_back(tagNode);
@@ -168,6 +171,7 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
                 }
             }
             else { // start text node
+                //std::cout << "HTMLParser::Parse - start text node " << html.substr(cursor, 6) << std::endl;
                 std::shared_ptr<TextNode> textNode = std::make_shared<TextNode>();
                 // not sure why currentNode is null but it is
                 if (currentNode) {
@@ -182,17 +186,16 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
             }
             cursor--;
         }
-        else if (state == 1) { // Skip Over Element
+        else if (state == 1) { // Skip Over Element (used by closing tag)
             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;
+                //std::cout << "HTMLParser::parse - close tag: " << html.substr(starts.back(), cursor - starts.back() + 1) << std::endl;
                 state = 0;
                 prependWhiteSpace = false;
             }
         }
         else if (state == 4) { // HTML Comment
             if (html[cursor] == '-' && html[cursor + 1] == '-' && html[cursor + 2] == '>') {
+                //std::cout << "HTMLParser::Parse - Found end HTML comment at " << html.substr(cursor, 6) << std::endl;
                 state = 0;
                 cursor += 2; // advance cursor to end of comment
                 prependWhiteSpace = false;
@@ -201,7 +204,7 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
         else if (state == 2) { // Search for end tag node
             if (html[cursor] == '>') { // end tag node
                 std::string element = html.substr(starts.back(), cursor - starts.back() + 1);
-                //std::cout << "HTMLParser::parse - close tag: " << element << std::endl;
+                //std::cout << "HTMLParser::parse - end open tag: " << element << std::endl;
                 if (element == "<li>") {
                     // this will close previous li before starting a new one
                     autoCloseTag(currentNode, rootNode, "ul", "li");
@@ -221,6 +224,7 @@ std::shared_ptr<Node> HTMLParser::parse(const std::string &html) const {
         else if (state == 3) { // End text node
             if (html[cursor + 1] == '<') {
                 dynamic_cast<TextNode*>(currentNode.get())->text = (prependWhiteSpace?" ":"") + html.substr(starts.back(), cursor - starts.back() + 1);
+                //std::cout << "HTMLParser::parse - end text node: " << html.substr(starts.back(), cursor - starts.back() + 1) << std::endl;
                 starts.pop_back();
                 if (currentNode && currentNode->parent) {
                   currentNode = currentNode->parent;
@@ -273,6 +277,8 @@ void HTMLParser::parseTag(const std::string &element, TagNode &tagNode) const {
         }
         else if (state == 3) {
             if (element[cursor] == '"') {
+                // was suggested to use tagNode.properties[propertyKey] = element.substr(start, cursor - start);
+                // for better readabiilty
                 tagNode.properties.insert(std::pair<std::string, std::string>(propertyKey, element.substr(start, cursor - start)));
                 start = cursor + 1;
                 state = 1;