Merge branch 'bzr/golem' of /Users/distler/Sites/code/instiki

lib/sanitize.rb

@@ -436,7 +436,7 @@ class String
 	'plussim' => '⨦',
 	'plustwo' => '⨧',
 	'prod' => '∏',
-	'race' => '⧚',
+	'race' => '∽̱',
 	'roplus' => '⨮',
 	'rotimes' => '⨵',
 	'rthree' => '⋌',
@@ -624,7 +624,7 @@ class String
 	'iiota' => '℩',
 	'image' => 'ℑ',
 	'imath' => 'ı',
-	'jmath' => 'j',
+	'jmath' => 'ȷ',
 	'laemptyv' => '⦴',
 	'lltri' => '◺',
 	'lrtri' => '⊿',
@@ -638,7 +638,7 @@ class String
 	'range' => '⦥',
 	'real' => 'ℜ',
 	'tbrk' => '⎴',
-	'trpezium' => '�',
+	'trpezium' => '⏢',
 	'ultri' => '◸',
 	'urtri' => '◹',
 	'vzigzag' => '⦚',
@@ -844,8 +844,10 @@ class String
 	'nu' => 'ν',
 	'Omega' => 'Ω',
 	'omega' => 'ω',
+	'phgr' => 'φ',
 	'Phi' => 'Φ',
-	'phi' => 'ϕ',
+	'phi' => 'φ',
+	'phis' => 'ϕ',
 	'phiv' => 'φ',
 	'Pi' => 'Π',
 	'pi' => 'π',
@@ -1006,7 +1008,7 @@ class String
 	'andv' => '⩚',
 	'angrt' => '∟',
 	'angsph' => '∢',
-	'angst' => 'Å',
+	'angst' => 'Å',
 	'ap' => '≈',
 	'apacir' => '⩯',
 	'awconint' => '∳',
@@ -1036,7 +1038,7 @@ class String
 	'dsol' => '⧶',
 	'dtdot' => '⋱',
 	'dwangle' => '⦦',
-	'elinters' => '�',
+	'elinters' => '⏧',
 	'epar' => '⋕',
 	'eparsl' => '⧣',
 	'equiv' => '≡',
@@ -1063,13 +1065,13 @@ class String
 	'isinsv' => '⋳',
 	'isinv' => '∈',
 	'lagran' => 'ℒ',
-	'Lang' => '《',
-	'lang' => '〈',
+	'Lang' => '⟪',
+	'lang' => '⟨',
 	'lArr' => '⇐',
-	'lbbrk' => '〔',
+	'lbbrk' => '❲',
 	'le' => '≤',
-	'loang' => '〘',
-	'lobrk' => '〚',
+	'loang' => '⟬',
+	'lobrk' => '⟦',
 	'lopar' => '⦅',
 	'lowast' => '∗',
 	'minus' => '−',
@@ -1123,12 +1125,12 @@ class String
 	'qprime' => '⁗',
 	'quatint' => '⨖',
 	'radic' => '√',
-	'Rang' => '》',
-	'rang' => '〉',
+	'Rang' => '⟫',
+	'rang' => '⟩',
 	'rArr' => '⇒',
-	'rbbrk' => '〕',
-	'roang' => '〙',
-	'robrk' => '〛',
+	'rbbrk' => '❳',
+	'roang' => '⟭',
+	'robrk' => '⟧',
 	'ropar' => '⦆',
 	'rppolint' => '⨒',
 	'scpolint' => '⨓',
@@ -1529,7 +1531,7 @@ class String
 	'nbsp' => ' ',
 	'not' => '¬',
 	'num' => '#',
-	'ohm' => 'Ω',
+	'ohm' => 'Ω',
 	'ordf' => 'ª',
 	'ordm' => 'º',
 	'para' => '¶',
@@ -1590,6 +1592,7 @@ class String
 	'fflig' => 'ff',
 	'ffllig' => 'ffl',
 	'filig' => 'fi',
+	'fjlig' => 'fj',
 	'flat' => '♭',
 	'fllig' => 'fl',
 	'frac13' => '⅓',
@@ -1714,9 +1717,9 @@ class String
 	'NotSucceedsTilde' => '≿̸',
 	'oopf' => '𝕠',
 	'OverBar' => '¯',
-	'OverBrace' => '︷',
+	'OverBrace' => '⏞',
 	'OverBracket' => '⎴',
-	'OverParenthesis' => '︵',
+	'OverParenthesis' => '⏜',
 	'planckh' => 'ℎ',
 	'popf' => '𝕡',
 	'Product' => '∏',
@@ -1738,9 +1741,9 @@ class String
 	'ThickSpace' => '   ',
 	'topf' => '𝕥',
 	'UnderBar' => '̲',
-	'UnderBrace' => '︸',
+	'UnderBrace' => '⏟',
 	'UnderBracket' => '⎵',
-	'UnderParenthesis' => '︶',
+	'UnderParenthesis' => '⏝',
 	'uopf' => '𝕦',
 	'UpArrowBar' => '⤒',
 	'Upsilon' => 'Υ',
@@ -1944,11 +1947,11 @@ class String
 	'intprod' => '⨼',
 	'InvisibleComma' => '⁣',
 	'InvisibleTimes' => '⁢',
-	'langle' => '〈',
+	'langle' => '⟨',
 	'Laplacetrf' => 'ℒ',
 	'lbrace' => '{',
 	'lbrack' => '[',
-	'LeftAngleBracket' => '〈',
+	'LeftAngleBracket' => '⟨',
 	'LeftArrow' => '←',
 	'Leftarrow' => '⇐',
 	'leftarrow' => '←',
@@ -1956,7 +1959,7 @@ class String
 	'LeftArrowRightArrow' => '⇆',
 	'leftarrowtail' => '↢',
 	'LeftCeiling' => '⌈',
-	'LeftDoubleBracket' => '〚',
+	'LeftDoubleBracket' => '⟦',
 	'LeftDownVector' => '⇃',
 	'LeftFloor' => '⌊',
 	'leftharpoondown' => '↽',
@@ -2136,7 +2139,7 @@ class String
 	'propto' => '∝',
 	'quaternions' => 'ℍ',
 	'questeq' => '≟',
-	'rangle' => '〉',
+	'rangle' => '⟩',
 	'rationals' => 'ℚ',
 	'rbrace' => '}',
 	'rbrack' => ']',
@@ -2147,7 +2150,7 @@ class String
 	'ReverseElement' => '∋',
 	'ReverseEquilibrium' => '⇋',
 	'ReverseUpEquilibrium' => '⥯',
-	'RightAngleBracket' => '〉',
+	'RightAngleBracket' => '⟩',
 	'RightArrow' => '→',
 	'Rightarrow' => '⇒',
 	'rightarrow' => '→',
@@ -2155,7 +2158,7 @@ class String
 	'RightArrowLeftArrow' => '⇄',
 	'rightarrowtail' => '↣',
 	'RightCeiling' => '⌉',
-	'RightDoubleBracket' => '〛',
+	'RightDoubleBracket' => '⟧',
 	'RightDownVector' => '⇂',
 	'RightFloor' => '⌋',
 	'rightharpoondown' => '⇁',
@@ -2299,7 +2302,14 @@ class String
 	'wedge' => '∧',
 	'wp' => '℘',
 	'wr' => '≀',
-	'zeetrf' => 'ℨ'
+	'zeetrf' => 'ℨ',
+	'AMP' => '&#x0026',
+	'COPY' => '&#x00A9',
+	'GT' => '&#x003E',
+	'LT' => '&#x003C',
+	'QUOT' => '&#x0022',
+	'REG' => '&#x00AE',
+	'TRADE' => '&#x2122'
   } unless const_defined? "MATHML_ENTITIES"
 #:startdoc:
 
@@ -2363,8 +2373,8 @@ class String
     end
 
     def convert_to_utf8 #:nodoc:
-      if self =~ /^(lt|gt|amp|quot|apos)$/
-        self.replace "&" + self + ";"
+      if self =~ /^(lt|gt|amp|quot|apos)$/i
+        self.replace "&" + self.downcase + ";"
       elsif MATHML_ENTITIES.has_key?(self)         
         self.replace MATHML_ENTITIES[self].split(';').collect {|s| s.gsub(/^&#x([A-F0-9]+)$/, '\1').hex }.pack('U*')
       else

lib/stringsupport.rb

@@ -313,7 +313,7 @@ class String
 	'plussim' => '⨦',
 	'plustwo' => '⨧',
 	'prod' => '∏',
-	'race' => '⧚',
+	'race' => '∽̱',
 	'roplus' => '⨮',
 	'rotimes' => '⨵',
 	'rthree' => '⋌',
@@ -721,8 +721,10 @@ class String
 	'nu' => 'ν',
 	'Omega' => 'Ω',
 	'omega' => 'ω',
+	'phgr' => 'φ',
 	'Phi' => 'Φ',
-	'phi' => 'ϕ',
+	'phi' => 'φ',
+	'phis' => 'ϕ',
 	'phiv' => 'φ',
 	'Pi' => 'Π',
 	'pi' => 'π',
@@ -883,7 +885,7 @@ class String
 	'andv' => '⩚',
 	'angrt' => '∟',
 	'angsph' => '∢',
-	'angst' => 'Å',
+	'angst' => 'Å',
 	'ap' => '≈',
 	'apacir' => '⩯',
 	'awconint' => '∳',
@@ -1406,7 +1408,7 @@ class String
 	'nbsp' => ' ',
 	'not' => '¬',
 	'num' => '#',
-	'ohm' => 'Ω',
+	'ohm' => 'Ω',
 	'ordf' => 'ª',
 	'ordm' => 'º',
 	'para' => '¶',
@@ -2177,7 +2179,14 @@ class String
 	'wedge' => '∧',
 	'wp' => '℘',
 	'wr' => '≀',
-	'zeetrf' => 'ℨ'
+	'zeetrf' => 'ℨ',
+	'AMP' => '&#x0026',
+	'COPY' => '&#x00A9',
+	'GT' => '&#x003E',
+	'LT' => '&#x003C',
+	'QUOT' => '&#x0022',
+	'REG' => '&#x00AE',
+	'TRADE' => '&#x2122'
   }
 #:startdoc:
 
@@ -2286,8 +2295,8 @@ class String
     end
 
     def convert_to_utf8 #:nodoc:
-      if self =~ /^(lt|gt|amp|quot|apos)$/
-        self.replace "&" + self + ";"
+      if self =~ /^(lt|gt|amp|quot|apos)$/i
+        self.replace "&" + self.downcase + ";"
       elsif MATHML_ENTITIES.has_key?(self)         
         self.replace MATHML_ENTITIES[self].split(';').collect {|s| s.gsub(/^&#x([A-F0-9]+)$/, '\1').hex }.pack('U*')
       else

test/unit/sanitize_test.rb

@@ -27,13 +27,13 @@ class SanitizeTest < Test::Unit::TestCase
   end
   
   def my_rex(string)
-    sanitize_rexml(rexml_doc(string)).gsub(/\A<div xmlns="http:\/\/www.w3.org\/1999\/xhtml">(.*)<\/div>\Z/m, '\1')
+    sanitize_rexml(rexml_doc(string.to_utf8)).gsub(/\A<div xmlns="http:\/\/www.w3.org\/1999\/xhtml">(.*)<\/div>\Z/m, '\1')
   end
 
   def test_sanitize_named_entities
-    input = '<p>Greek &phi;, double-struck &Aopf;, numeric &#x1D538; &#8279;</p>'
-    output = "<p>Greek \317\225, double-struck \360\235\224\270, numeric \360\235\224\270 \342\201\227</p>"
-    output2 = "<p>Greek \317\225, double-struck \360\235\224\270, numeric &#x1D538; &#8279;</p>"
+    input = '<p>Greek &phis; &phi;, double-struck &Aopf;, numeric &#x1D538; &#8279;, uppercase &TRADE; &LT;</p>'
+    output = "<p>Greek \317\225 \317\206, double-struck \360\235\224\270, numeric \360\235\224\270 \342\201\227, uppercase \342\204\242 &lt;</p>"
+    output2 = "<p>Greek \317\225 \317\206, double-struck \360\235\224\270, numeric &#x1D538; &#8279;, uppercase \342\204\242 &lt;</p>"
     assert_equal(output, sanitize_xhtml(input))
     assert_equal(output, sanitize_html(input))
     assert_equal(output, my_rex(input))

test/unit/sanitizer_test.rb

@@ -22,6 +22,14 @@ class SanitizerTest < Test::Unit::TestCase
     assert_equal xhtmloutput, do_sanitize_xhtml(input)
   end
 
+  def test_sanitize_named_entities
+    input = '<p>Greek &phis; &phi;, double-struck &Aopf;, numeric &#x1D538; &#8279;, uppercase &TRADE; &LT;</p>'
+    output = "<p>Greek \317\225 \317\206, double-struck \360\235\224\270, numeric \360\235\224\270 \342\201\227, uppercase \342\204\242 &lt;</p>"
+    output2 = "<p>Greek \317\225 \317\206, double-struck \360\235\224\270, numeric &#x1D538; &#8279;, uppercase \342\204\242 &lt;</p>"
+    check_sanitization(input, output, output, output)
+    assert_equal(output2, input.to_utf8)
+  end
+
   Sanitizer::ALLOWED_ELEMENTS.each do |tag_name|
     define_method "test_should_allow_#{tag_name}_tag" do
       input       = "<#{tag_name} title='1'>foo <bad>bar</bad> baz</#{tag_name}>"