htmlentities 3.0.1 → 4.0.0

data/lib/htmlentities/legacy.rb

@@ -0,0 +1,27 @@
+class HTMLEntities
+  class << self
+    
+    #
+    # Legacy compatibility class method allowing direct encoding of XHTML1 entities.
+    # See HTMLEntities#encode for description of parameters.
+    #
+    def encode_entities(*args)
+      xhtml1_entities.encode(*args)
+    end
+    
+    #
+    # Legacy compatibility class method allowing direct decoding of XHTML1 entities.
+    # See HTMLEntities#decode for description of parameters.
+    #
+    def decode_entities(*args)
+      xhtml1_entities.decode(*args)
+    end
+    
+  private
+  
+    def xhtml1_entities
+      @xhtml1_entities ||= new('xhtml1')
+    end
+    
+  end
+end

data/lib/htmlentities/xhtml1.rb

@@ -0,0 +1,258 @@
+class HTMLEntities
+  MAPPINGS = {} unless defined? MAPPINGS
+  MAPPINGS['xhtml1'] = {
+    'Aacute' => 193,
+    'aacute' => 225,
+    'Acirc' => 194,
+    'acirc' => 226,
+    'acute' => 180,
+    'AElig' => 198,
+    'aelig' => 230,
+    'Agrave' => 192,
+    'agrave' => 224,
+    'alefsym' => 8501,
+    'Alpha' => 913,
+    'alpha' => 945,
+    'amp' => 38,
+    'and' => 8743,
+    'ang' => 8736,
+    'apos' => 39,
+    'Aring' => 197,
+    'aring' => 229,
+    'asymp' => 8776,
+    'Atilde' => 195,
+    'atilde' => 227,
+    'Auml' => 196,
+    'auml' => 228,
+    'bdquo' => 8222,
+    'Beta' => 914,
+    'beta' => 946,
+    'brvbar' => 166,
+    'bull' => 8226,
+    'cap' => 8745,
+    'Ccedil' => 199,
+    'ccedil' => 231,
+    'cedil' => 184,
+    'cent' => 162,
+    'Chi' => 935,
+    'chi' => 967,
+    'circ' => 710,
+    'clubs' => 9827,
+    'cong' => 8773,
+    'copy' => 169,
+    'crarr' => 8629,
+    'cup' => 8746,
+    'curren' => 164,
+    'Dagger' => 8225,
+    'dagger' => 8224,
+    'dArr' => 8659,
+    'darr' => 8595,
+    'deg' => 176,
+    'Delta' => 916,
+    'delta' => 948,
+    'diams' => 9830,
+    'divide' => 247,
+    'Eacute' => 201,
+    'eacute' => 233,
+    'Ecirc' => 202,
+    'ecirc' => 234,
+    'Egrave' => 200,
+    'egrave' => 232,
+    'empty' => 8709,
+    'emsp' => 8195,
+    'ensp' => 8194,
+    'Epsilon' => 917,
+    'epsilon' => 949,
+    'equiv' => 8801,
+    'Eta' => 919,
+    'eta' => 951,
+    'ETH' => 208,
+    'eth' => 240,
+    'Euml' => 203,
+    'euml' => 235,
+    'euro' => 8364,
+    'exist' => 8707,
+    'fnof' => 402,
+    'forall' => 8704,
+    'frac12' => 189,
+    'frac14' => 188,
+    'frac34' => 190,
+    'frasl' => 8260,
+    'Gamma' => 915,
+    'gamma' => 947,
+    'ge' => 8805,
+    'gt' => 62,
+    'hArr' => 8660,
+    'harr' => 8596,
+    'hearts' => 9829,
+    'hellip' => 8230,
+    'Iacute' => 205,
+    'iacute' => 237,
+    'Icirc' => 206,
+    'icirc' => 238,
+    'iexcl' => 161,
+    'Igrave' => 204,
+    'igrave' => 236,
+    'image' => 8465,
+    'infin' => 8734,
+    'int' => 8747,
+    'Iota' => 921,
+    'iota' => 953,
+    'iquest' => 191,
+    'isin' => 8712,
+    'Iuml' => 207,
+    'iuml' => 239,
+    'Kappa' => 922,
+    'kappa' => 954,
+    'Lambda' => 923,
+    'lambda' => 955,
+    'lang' => 9001,
+    'laquo' => 171,
+    'lArr' => 8656,
+    'larr' => 8592,
+    'lceil' => 8968,
+    'ldquo' => 8220,
+    'le' => 8804,
+    'lfloor' => 8970,
+    'lowast' => 8727,
+    'loz' => 9674,
+    'lrm' => 8206,
+    'lsaquo' => 8249,
+    'lsquo' => 8216,
+    'lt' => 60,
+    'macr' => 175,
+    'mdash' => 8212,
+    'micro' => 181,
+    'middot' => 183,
+    'minus' => 8722,
+    'Mu' => 924,
+    'mu' => 956,
+    'nabla' => 8711,
+    'nbsp' => 160,
+    'ndash' => 8211,
+    'ne' => 8800,
+    'ni' => 8715,
+    'not' => 172,
+    'notin' => 8713,
+    'nsub' => 8836,
+    'Ntilde' => 209,
+    'ntilde' => 241,
+    'Nu' => 925,
+    'nu' => 957,
+    'Oacute' => 211,
+    'oacute' => 243,
+    'Ocirc' => 212,
+    'ocirc' => 244,
+    'OElig' => 338,
+    'oelig' => 339,
+    'Ograve' => 210,
+    'ograve' => 242,
+    'oline' => 8254,
+    'Omega' => 937,
+    'omega' => 969,
+    'Omicron' => 927,
+    'omicron' => 959,
+    'oplus' => 8853,
+    'or' => 8744,
+    'ordf' => 170,
+    'ordm' => 186,
+    'Oslash' => 216,
+    'oslash' => 248,
+    'Otilde' => 213,
+    'otilde' => 245,
+    'otimes' => 8855,
+    'Ouml' => 214,
+    'ouml' => 246,
+    'para' => 182,
+    'part' => 8706,
+    'permil' => 8240,
+    'perp' => 8869,
+    'Phi' => 934,
+    'phi' => 966,
+    'Pi' => 928,
+    'pi' => 960,
+    'piv' => 982,
+    'plusmn' => 177,
+    'pound' => 163,
+    'Prime' => 8243,
+    'prime' => 8242,
+    'prod' => 8719,
+    'prop' => 8733,
+    'Psi' => 936,
+    'psi' => 968,
+    'quot' => 34,
+    'radic' => 8730,
+    'rang' => 9002,
+    'raquo' => 187,
+    'rArr' => 8658,
+    'rarr' => 8594,
+    'rceil' => 8969,
+    'rdquo' => 8221,
+    'real' => 8476,
+    'reg' => 174,
+    'rfloor' => 8971,
+    'Rho' => 929,
+    'rho' => 961,
+    'rlm' => 8207,
+    'rsaquo' => 8250,
+    'rsquo' => 8217,
+    'sbquo' => 8218,
+    'Scaron' => 352,
+    'scaron' => 353,
+    'sdot' => 8901,
+    'sect' => 167,
+    'shy' => 173,
+    'Sigma' => 931,
+    'sigma' => 963,
+    'sigmaf' => 962,
+    'sim' => 8764,
+    'spades' => 9824,
+    'sub' => 8834,
+    'sube' => 8838,
+    'sum' => 8721,
+    'sup' => 8835,
+    'sup1' => 185,
+    'sup2' => 178,
+    'sup3' => 179,
+    'supe' => 8839,
+    'szlig' => 223,
+    'Tau' => 932,
+    'tau' => 964,
+    'there4' => 8756,
+    'Theta' => 920,
+    'theta' => 952,
+    'thetasym' => 977,
+    'thinsp' => 8201,
+    'THORN' => 222,
+    'thorn' => 254,
+    'tilde' => 732,
+    'times' => 215,
+    'trade' => 8482,
+    'Uacute' => 218,
+    'uacute' => 250,
+    'uArr' => 8657,
+    'uarr' => 8593,
+    'Ucirc' => 219,
+    'ucirc' => 251,
+    'Ugrave' => 217,
+    'ugrave' => 249,
+    'uml' => 168,
+    'upsih' => 978,
+    'Upsilon' => 933,
+    'upsilon' => 965,
+    'Uuml' => 220,
+    'uuml' => 252,
+    'weierp' => 8472,
+    'Xi' => 926,
+    'xi' => 958,
+    'Yacute' => 221,
+    'yacute' => 253,
+    'yen' => 165,
+    'Yuml' => 376,
+    'yuml' => 255,
+    'Zeta' => 918,
+    'zeta' => 950,
+    'zwj' => 8205,
+    'zwnj' => 8204
+  }
+end

data/test/entities_test.rb

@@ -1,74 +1,124 @@
-$: << File.dirname(__FILE__) + '/../lib/'
+$:.unshift(File.dirname(__FILE__) + '/../lib')
 require 'htmlentities'
 require 'test/unit'
 
 $KCODE = 'u'
 
-class TestHTMLEntities < Test::Unit::TestCase
+class HTMLEntities::EntitiesTest < Test::Unit::TestCase
+
+  attr_reader :xhtml1_entities, :html4_entities
+
+  def setup
+    @xhtml1_entities = HTMLEntities.new('xhtml1')
+    @html4_entities = HTMLEntities.new('html4')
+  end
+
+  class PseudoString
+    def initialize(string)
+      @string = string
+    end
+    def to_s
+      @string
+    end
+  end
   
-  def test_basic_decoding
+  def test_should_raise_exception_when_unknown_flavor_specified
+    assert_raises(HTMLEntities::UnknownFlavor) do
+      HTMLEntities.new('foo')
+    end
+  end
+
+  def test_should_allow_symbol_for_flavor
+    assert_nothing_raised do
+      HTMLEntities.new(:xhtml1)
+    end
+  end
+
+  def test_should_allow_upper_case_flavor
+    assert_nothing_raised do
+      HTMLEntities.new('XHTML1')
+    end
+  end
+
+  def test_should_decode_basic_entities
     assert_decode('&', '&amp;')
     assert_decode('<', '&lt;')
     assert_decode('"', '&quot;')
   end
-      
-  def test_basic_encoding
+
+  def test_should_encode_basic_entities
     assert_encode('&amp;', '&', :basic)
     assert_encode('&quot;', '"')
     assert_encode('&lt;', '<', :basic)
     assert_encode('&lt;', '<')
   end
+  
+  def test_should_encode_basic_entities_to_decimal
+    assert_encode('&#38;', '&', :decimal)
+    assert_encode('&#34;', '"', :decimal)
+    assert_encode('&#60;', '<', :decimal)
+    assert_encode('&#62;', '>', :decimal)
+    assert_encode('&#39;', "'", :decimal)
+  end
+
+  def test_should_encode_basic_entities_to_hexadecimal
+    assert_encode('&#x26;', '&', :hexadecimal)
+    assert_encode('&#x22;', '"', :hexadecimal)
+    assert_encode('&#x3c;', '<', :hexadecimal)
+    assert_encode('&#x3e;', '>', :hexadecimal)
+    assert_encode('&#x27;', "'", :hexadecimal)
+  end
 
-  def test_extended_decoding
+  def test_should_decode_extended_named_entities
     assert_decode('±', '&plusmn;')
     assert_decode('ð', '&eth;')
     assert_decode('Œ', '&OElig;')
     assert_decode('œ', '&oelig;')
   end
-      
-  def test_extended_encoding
+
+  def test_should_encode_extended_named_entities
     assert_encode('&plusmn;', '±', :named)
     assert_encode('&eth;', 'ð', :named)
     assert_encode('&OElig;', 'Œ', :named)
     assert_encode('&oelig;', 'œ', :named)
   end
 
-  def test_decimal_decoding
+  def test_should_decode_decimal_entities
     assert_decode('“', '&#8220;')
     assert_decode('…', '&#8230;')
     assert_decode(' ', '&#32;')
   end
-      
-  def test_decimal_encoding
+
+  def test_should_encode_decimal_entities
     assert_encode('&#8220;', '“', :decimal)
     assert_encode('&#8230;', '…', :decimal)
   end
 
-  def test_hexadecimal_decoding
+  def test_should_decode_hexadecimal_entities
     assert_decode('−', '&#x2212;')
     assert_decode('—', '&#x2014;')
     assert_decode('`', '&#x0060;')
     assert_decode('`', '&#x60;')
   end
-      
-  def test_hexadecimal_encoding
+
+  def test_should_encode_hexadecimal_entities
     assert_encode('&#x2212;', '−', :hexadecimal)
     assert_encode('&#x2014;', '—', :hexadecimal)
   end
 
-  def test_mixed_decoding
+  def test_should_decode_text_with_mix_of_entities
     # Just a random headline - I needed something with accented letters.
     assert_decode(
-      'Le tabac pourrait bientôt être banni dans tous les lieux publics en France', 
+      'Le tabac pourrait bientôt être banni dans tous les lieux publics en France',
       'Le tabac pourrait bient&ocirc;t &#234;tre banni dans tous les lieux publics en France'
     )
-    assert_decode(      
+    assert_decode(
       '"bientôt" & 文字',
       '&quot;bient&ocirc;t&quot; &amp; &#25991;&#x5b57;'
     )
   end
 
-  def test_mixed_encoding
+  def test_should_encode_text_using_mix_of_entities
     assert_encode(
       '&quot;bient&ocirc;t&quot; &amp; &#x6587;&#x5b57;',
       '"bientôt" & 文字', :basic, :named, :hexadecimal
@@ -78,8 +128,8 @@ class TestHTMLEntities < Test::Unit::TestCase
       '"bientôt" & 文字', :basic, :named, :decimal
     )
   end
-  
-  def test_mixed_encoding_with_sort
+
+  def test_should_sort_commands_when_encoding_using_mix_of_entities
     assert_encode(
       '&quot;bient&ocirc;t&quot; &amp; &#x6587;&#x5b57;',
       '"bientôt" & 文字', :named, :hexadecimal, :basic
@@ -90,40 +140,67 @@ class TestHTMLEntities < Test::Unit::TestCase
     )
   end
 
-  def test_detect_illegal_encoding_command
+  def test_should_detect_illegal_encoding_command
     assert_raise(HTMLEntities::InstructionError) {
       HTMLEntities.encode_entities('foo', :bar, :baz)
     }
   end
 
-  def test_edge_case_decoding
+  def test_should_decode_empty_string
     assert_decode('', '')
+  end
+
+  def test_should_skip_unknown_entity
     assert_decode('&bogus;', '&bogus;')
+  end
+
+  def test_should_decode_double_encoded_entity_once
     assert_decode('&amp;', '&amp;amp;')
   end
 
-  def test_edge_case_encoding
+  def test_should_not_encode_normal_ASCII
     assert_encode('`', '`')
     assert_encode(' ', ' ')
-    assert_encode('&amp;amp;', '&amp;')
+  end
+
+  def test_should_double_encode_existing_entity
     assert_encode('&amp;amp;', '&amp;')
   end
 
   # Faults found and patched by Moonwolf
-  def test_moonwolf_decoding
-    assert_decode("\x2", '&#2;')
-    assert_decode("\xf", '&#xf;')
+  def test_should_decode_full_hexadecimal_range
+    (0..127).each do |codepoint|
+      assert_decode([codepoint].pack('U'), "&\#x#{codepoint.to_s(16)};")
+    end
+  end
+
+  # Reported by Dallas DeVries and Johan Duflost
+  def test_should_decode_named_entities_reported_as_missing_in_3_0_1
+    assert_decode([178].pack('U'), '&sup2;')
+    assert_decode([8226].pack('U'), '&bull;')
+    assert_decode([948].pack('U'), '&delta;')
+  end
+
+  def test_should_ducktype_parameter_to_string_before_encoding
+    pseudo_string = PseudoString.new('foo')
+    assert_decode('foo', pseudo_string)
+  end
+
+  def test_should_ducktype_parameter_to_string_before_decoding
+    pseudo_string = PseudoString.new('foo')
+    assert_encode('foo', pseudo_string)
   end
-  
-  private
 
   def assert_decode(expected, input)
-    assert_equal(expected, HTMLEntities.decode_entities(input))
+    [xhtml1_entities, html4_entities].each do |coder|
+      assert_equal(expected, coder.decode(input))
+    end
   end
-  
+
   def assert_encode(expected, input, *args)
-    assert_equal(expected, HTMLEntities.encode_entities(input, *args))
+    [xhtml1_entities, html4_entities].each do |coder|
+      assert_equal(expected, coder.encode(input, *args))
+    end
   end
 
 end
-