From ce8578d2d0fb4e529c6ee456acd37cd27f4ad560 Mon Sep 17 00:00:00 2001
From: Jacques Distler <distler@golem.ph.utexas.edu>
Date: Sat, 19 Jun 2010 03:02:15 -0500
Subject: [PATCH] Some Maruku Regexp Refactoring

---
 lib/instiki_stringsupport.rb                  |  2 +-
 .../maruku/ext/math/mathml_engines/none.rb    |  6 +++---
 .../maruku/lib/maruku/output/s5/to_s5.rb      |  9 +--------
 .../maruku/lib/maruku/output/to_html.rb       | 14 ++++---------
 .../plugins/maruku/lib/maruku/string_utils.rb | 20 +++++++++++++++++++
 5 files changed, 29 insertions(+), 22 deletions(-)

diff --git a/lib/instiki_stringsupport.rb b/lib/instiki_stringsupport.rb
index 09010202..5ea52a55 100644
--- a/lib/instiki_stringsupport.rb
+++ b/lib/instiki_stringsupport.rb
@@ -2313,7 +2313,7 @@ end
       "'" => '&#39;',
       '"' => '&quot;',
     }
-    TO_ESCAPE_PATTERN = Regexp.union(TO_ESCAPE.keys)
+    TO_ESCAPE_PATTERN = Regexp.union(*TO_ESCAPE.keys)
     
     def escapeHTML
       self.gsub(TO_ESCAPE_PATTERN){|m| TO_ESCAPE[m]}
diff --git a/vendor/plugins/maruku/lib/maruku/ext/math/mathml_engines/none.rb b/vendor/plugins/maruku/lib/maruku/ext/math/mathml_engines/none.rb
index bec8283b..c427002b 100644
--- a/vendor/plugins/maruku/lib/maruku/ext/math/mathml_engines/none.rb
+++ b/vendor/plugins/maruku/lib/maruku/ext/math/mathml_engines/none.rb
@@ -1,14 +1,14 @@
 module MaRuKu; module Out; module HTML
 
+  require 'maruku/string_utils'
+
 	def convert_to_mathml_none(kind, tex)
 		# You can: either return a REXML::Element
 		#    return Element.new 'div'    
 		# or return an empty array on error
 		#    return []  
 		# or have a string parsed by REXML:
-		mathml = "<code>#{tex.gsub( /&/, "&amp;" ).
-		  gsub( /</, "&lt;" ).gsub( />/, "&gt;"  ).
-		  gsub(/'/, "&#39;" ).gsub(/"/, "&quot;" )}</code>"
+		mathml = "<code>#{html_escape(tex)}</code>"
 		return Document.new(mathml).root
 	end
 
diff --git a/vendor/plugins/maruku/lib/maruku/output/s5/to_s5.rb b/vendor/plugins/maruku/lib/maruku/output/s5/to_s5.rb
index d2dd9f74..36d8d18a 100644
--- a/vendor/plugins/maruku/lib/maruku/output/s5/to_s5.rb
+++ b/vendor/plugins/maruku/lib/maruku/output/s5/to_s5.rb
@@ -8,20 +8,13 @@ begin
 rescue LoadError
         $rexml_new_version = false      
 end
+  require 'maruku/string_utils'
 	 
 	class MDDocument
 
 	def s5_theme
 	  html_escape(self.attributes[:slide_theme] || "default")
 	end
-
-	def  html_escape(string)
-	  string.gsub( /&/, "&amp;" ).
-	         gsub( /</, "&lt;" ).
-	         gsub( />/, "&gt;" ).
-	         gsub( /'/, "&#39;" ).
-	         gsub( /"/, "&quot;" )
-	end
 		
 	# Render as an HTML fragment (no head, just the content of BODY). (returns a string)
 	def to_s5(context={})
diff --git a/vendor/plugins/maruku/lib/maruku/output/to_html.rb b/vendor/plugins/maruku/lib/maruku/output/to_html.rb
index b5f27aae..ebbdbad1 100644
--- a/vendor/plugins/maruku/lib/maruku/output/to_html.rb
+++ b/vendor/plugins/maruku/lib/maruku/output/to_html.rb
@@ -74,9 +74,7 @@ module MaRuKu; module Out; module HTML
 			div.write(xml,indent,transitive=true,ie_hack)
 		end
 
-		xml.gsub!(/\A<dummy>\s*/,'')
-		xml.gsub!(/\s*<\/dummy>\Z/,'')
-		xml.gsub!(/\A<dummy\s*\/>/,'')
+		xml.gsub!(/\A<dummy>\s*|\s*<\/dummy>\Z|\A<dummy\s*\/>/,'')
 		xml
 	end
 	
@@ -507,8 +505,7 @@ by Maruku, to have the same results in both HTML and LaTeX.
 	def source2html(source)
 #		source = source.gsub(/&/,'&amp;')
 		source = Text.normalize(source)
-		source = source.gsub(/\&apos;/,'&#39;') # IE bug
-		source = source.gsub(/'/,'&#39;') # IE bug
+		source.gsub!(/\&apos;|'/,'&#39;') # IE bug
 		Text.new(source, true, nil, true )
 	end
 		
@@ -571,8 +568,7 @@ and
 				source = source.gsub(/\n*\Z/,'')
 				
 				html = convertor.convert( source )
-				html = html.gsub(/\&apos;/,'&#39;') # IE bug
-				html = html.gsub(/'/,'&#39;') # IE bug
+				html.gsub!(/\&apos;|'/,'&#39;') # IE bug
 	#			html = html.gsub(/&/,'&amp;') 
 				
 				code = Document.new(html, {:respect_whitespace =>:all}).root
@@ -633,9 +629,7 @@ of the form `#ff00ff`.
 		s = source
 		
 #		s  = s.gsub(/&/,'&amp;')
-		s = Text.normalize(s)
-		s  = s.gsub(/\&apos;/,'&#39;') # IE bug
-		s  = s.gsub(/'/,'&#39;') # IE bug
+		s = Text.normalize(s).gsub(/\&apos;|'/,'&#39;') # IE bug
 
 		if get_setting(:code_show_spaces) 
 			# 187 = raquo
diff --git a/vendor/plugins/maruku/lib/maruku/string_utils.rb b/vendor/plugins/maruku/lib/maruku/string_utils.rb
index 6b3b1d70..832f7fae 100644
--- a/vendor/plugins/maruku/lib/maruku/string_utils.rb
+++ b/vendor/plugins/maruku/lib/maruku/string_utils.rb
@@ -141,6 +141,26 @@ module MaRuKu
       return s
     end
 
+#--
+    MARUKU_HTML_ESCAPE = {
+      '&' => '&amp;',
+      '<' => '&lt;',
+      '>' => '&gt;',
+      "'" => '&#39;',
+      '"' => '&quot;',
+    }
+    MARUKU_HTML_ESCAPE_PATTERN = Regexp.union(*MARUKU_HTML_ESCAPE.keys)
+#++
+
+    # HTML-escapes a string.
+    #
+    # @param str [String]
+    # @return [String]
+    
+    def html_escape(string)
+      string.gsub(MARUKU_HTML_ESCAPE_PATTERN){|m| MARUKU_HTML_ESCAPE[m]}
+    end
+
     # Escapes a string so that it can be safely used in a Bourne shell command line.
     #
     # Note that a resulted string should be used unquoted