non-quoted attribute support

src/html/HTMLParser.cpp

@@ -249,23 +249,27 @@ void HTMLParser::parseTag(const std::string &element, TagNode &tagNode) const {
     std::string propertyKey;
     for (cursor = 0; cursor < element.length();  cursor++) {
         if (state == 0) {
+            // space or end
             if (element[cursor] == ' ' || element[cursor] == '>') {
+                // set our tag (type / name, i.e. h1)
                 tagNode.tag = element.substr(start, cursor - start);
+                // make sure our tag is lowercase
                 std::transform(tagNode.tag.begin(), tagNode.tag.end(), tagNode.tag.begin(), tolower);
                 start = cursor + 1;
                 state = 1;
             }
         }
-        else if (state == 1) {
+        else if (state == 1) { // attribute search
             if (element[cursor] == ' ') {
                 start = cursor + 1;
             }
             else if (element[cursor] == '=') {
                 propertyKey = element.substr(start, cursor - start);
+                start = cursor + 1; // start for non quotes
                 state = 2;
             }
         }
-        else if (state == 2) {
+        else if (state == 2) { // after = of attribute
             if (element[cursor] == '"') {
                 start = cursor + 1;
                 state = 3;
@@ -273,6 +277,11 @@ void HTMLParser::parseTag(const std::string &element, TagNode &tagNode) const {
             else if (element[cursor] == '\'') {
                 start = cursor + 1;
                 state = 4;
+            } else if (element[cursor] == ' ') {
+                // we just probably found an end of attribute without quotes
+                tagNode.properties.insert(std::pair<std::string, std::string>(propertyKey, element.substr(start, cursor - start)));
+                start = cursor + 1;
+                state = 1;
             }
         }
         else if (state == 3) {
@@ -292,5 +301,15 @@ void HTMLParser::parseTag(const std::string &element, TagNode &tagNode) const {
             }
         }
     }
+    // 2 is attribute without quotes, 3/4 is unclosed quote
+    if (state == 2 || state == 3 || state ==4) {
+        // we were in an atrr=
+        tagNode.properties.insert(std::pair<std::string, std::string>(propertyKey, element.substr(start, cursor - start - 1)));
+    } else {
+        if (state != 1) {
+            // so what's ending on state 0 about (somethin about no atttributes and maybe no tag name/type)
+            std::cout << "HTMLParser::parseTag ending on state " << state << std::endl;
+        }
+    }
 }