Escape CDATA inserted into script/style tags so they still parse in R…

…EXML and work in browsers.

The output is pretty ugly, but at least these tags will now work without crashing REXML or nuking the script/style. Fixes issue #120.

CHANGELOG.md

@@ -9,6 +9,7 @@
 * More robust table handling.
 * Better handling of lists.
 * Fix the "blahtex" math engine on 1.8.7.
+* "script" and "style" tags now have their generated CDATA tags escaped so the scripts/styles actually work. #120
 
 0.7.0
 -----

lib/maruku/input/html_helper.rb

@@ -93,9 +93,10 @@ def eat_this(line)
         when :inside_cdata
           if @m = CDataEnd.match(@rest)
             my_debug "#{@state}: matched #{@m.to_s.inspect}"
-            @already << @m.pre_match << @m.to_s
-            @rest = @m.post_match
             self.state = %(script style).include?(@tag_stack.last) ? :inside_script_style : :inside_element
+            @already << @m.pre_match
+            @already << @m.to_s unless self.state == :inside_script_style
+            @rest = @m.post_match
           else
             @already << @rest
             @rest = ""
@@ -110,8 +111,8 @@ def eat_this(line)
               my_debug "#{@state}: CDATA: #{@m.to_s.inspect}"
               @already << @m.pre_match << @m.to_s
               @rest = @m.post_match
-              self.state = :inside_cdata
             end
+            self.state = :inside_cdata
           elsif @m = Tag.match(@rest)
             is_closing = !!@m[1]
             tag = @m[2]
@@ -121,7 +122,7 @@ def eat_this(line)
               @rest = @m.post_match
               # This is necessary to properly parse
               # script tags
-              @already << "]]>" unless @already.rstrip.end_with?("]]>")
+              @already << script_style_cdata_end(tag) unless @already.rstrip.end_with?(']]>')
               self.state = :inside_element
               handle_tag false # don't double-add pre_match
             else
@@ -191,7 +192,7 @@ def handle_tag(add_pre_match = true)
         if %w(script style).include?(@tag_stack.last)
           # This is necessary to properly parse
           # script tags
-          @already << "<![CDATA["
+          @already << script_style_cdata_start(@tag_stack.last)
           self.state = :inside_script_style
         end
       end
@@ -218,5 +219,13 @@ def stuff_you_read
     def is_finished?
       (self.state == :inside_element) and @tag_stack.empty?
     end
+
+    def script_style_cdata_start(tag)
+      (tag == 'script') ? "//<![CDATA[\n" : "/*<![CDATA[*/\n"
+    end
+
+    def script_style_cdata_end(tag)
+      (tag == 'script') ? "\n//]]>" : "\n/*]]>*/"
+    end
   end # html helper
 end

spec/block_docs/issue120.md

@@ -0,0 +1,54 @@
+Style tags should be OK with unescaped angle brackets and ampersands. https://github.com/bhollis/maruku/issues/120
+NOTE: Commented CDATA is output because we use XHTML - for HTML mode it should be omitted.
+*** Parameters: ***
+{}
+*** Markdown input: ***
+<style type="text/css">
+  p > .highlight {
+    color: red;
+    background-image: url('/foo?bar&baz');
+  }
+</style>
+
+<style type="text/css">/*<![CDATA[*/
+  p > .highlight {
+    color: red;
+    background-image: url('/foo?bar&baz');
+  }
+/*]]>*/</style>
+
+<style type="text/css">
+/*<![CDATA[*/
+  p > .highlight {
+    color: red;
+    background-image: url('/foo?bar&baz');
+  }
+/*]]>*/
+</style>
+*** Output of inspect ***
+
+*** Output of to_html ***
+<style type='text/css'>/*<![CDATA[*/
+
+  p > .highlight {
+    color: red;
+    background-image: url('/foo?bar&baz');
+  }
+
+/*]]>*/</style><style type='text/css'>/*<![CDATA[*/
+/*<![CDATA[*/
+  p > .highlight {
+    color: red;
+    background-image: url('/foo?bar&baz');
+  }
+/**/
+/*]]>*/</style><style type='text/css'>/*<![CDATA[*/
+
+/*<![CDATA[*/
+  p > .highlight {
+    color: red;
+    background-image: url('/foo?bar&baz');
+  }
+/**/
+
+/*]]>*/</style>

spec/block_docs/issue40.md

@@ -20,21 +20,44 @@ NOTE: CDATA is output because we use XHTML - for HTML mode it should be omitted.
   var x = true && true;
 ]]>
 </script>
+
+<script>//<![CDATA[
+  var x = true && true;
+//]]></script>
+
+<script>
+//<![CDATA[
+  var x = true && true;
+//]]>
+</script>
 *** Output of inspect ***
 
 *** Output of to_html ***
-<script><![CDATA[
+<script>//<![CDATA[
+
   var x = true && true;
-]]></script>
 
-<script><![CDATA[foo && bar]]></script>
+//]]></script><script>//<![CDATA[
+foo && bar
+//]]></script><script>//<![CDATA[
 
-<script><![CDATA[
   var x = true && true;
-]]></script>
 
-<script><![CDATA[foo && bar]]></script>
+//]]></script><script>//<![CDATA[
+foo && bar
+//]]></script><script>//<![CDATA[
 
-<script><![CDATA[
   var x = true && true;
-]]></script>
+
+
+//]]></script><script>//<![CDATA[
+//<![CDATA[
+  var x = true && true;
+//
+//]]></script><script>//<![CDATA[
+
+//<![CDATA[
+  var x = true && true;
+//
+
+//]]></script>