Ticket #91: omega-htmlparser-ignore-javascript-lessthan.patch

htmlparse.h

@@ -30 +28 @@
 class HtmlParser {
     protected:
         void decode_entities(string &s);
+        bool in_script;
         static map<string, unsigned int> named_ents;
     public:
         virtual void process_text(const string &/*text*/) { }

htmlparse.cc

@@ -240 +240 @@
 void
 HtmlParser::parse_html(const string &body)
 {
+    in_script = false;
+
     map<string,string> Param;
     string::const_iterator start = body.begin();
 
@@ -253 +255 @@
             if (p == body.end()) break;
             char ch = *(p + 1);
             // Tag, closing tag, comment (or SGML declaration), or PHP.
-            if (isalpha(ch) || ch == '/' || ch == '!' || ch == '?') break;
+            if ((!in_script && isalpha(ch)) || ch == '/' || ch == '!' || ch == '?') break;
             p++; 
         }
 
@@ -338 +340 @@
                
             if (closing) {
                 closing_tag(tag);
+                if (in_script && tag == "script") in_script = false;
                    
                 /* ignore any bogus parameters on closing tags */
                 p = find(start, body.end(), '>');
@@ -393 +396 @@
                 opening_tag(tag, Param);
                 Param.clear();
 
+                // In <script> tags we ignore opening tags to avoid problems with "a<b".
+                if (tag == "script") in_script = true;
+
                 if (start != body.end() && *start == '>') ++start;
             }
         }