src/vendor/html5ever/html5lib-tests/tokenizer/entities.test

   1 {"tests": [
   2
   3 {"description": "Undefined named entity in attribute value ending in semicolon and whose name starts with a known entity name.",
   4 "input":"<h a='&noti;'>",
   5 "output": [["StartTag", "h", {"a": "&noti;"}]]},
   6
   7 {"description": "Entity name followed by the equals sign in an attribute value.",
   8 "input":"<h a='&lang='>",
   9 "output": [["StartTag", "h", {"a": "&lang="}]]},
  10
  11 {"description": "CR as numeric entity",
  12 "input":"&#013;",
  13 "output": ["ParseError", ["Character", "\r"]]},
  14
  15 {"description": "CR as hexadecimal numeric entity",
  16 "input":"&#x00D;",
  17 "output": ["ParseError", ["Character", "\r"]]},
  18
  19 {"description": "Windows-1252 EURO SIGN numeric entity.",
  20 "input":"&#0128;",
  21 "output": ["ParseError", ["Character", "\u20AC"]]},
  22
  23 {"description": "Windows-1252 REPLACEMENT CHAR numeric entity.",
  24 "input":"&#0129;",
  25 "output": ["ParseError", ["Character", "\u0081"]]},
  26
  27 {"description": "Windows-1252 SINGLE LOW-9 QUOTATION MARK numeric entity.",
  28 "input":"&#0130;",
  29 "output": ["ParseError", ["Character", "\u201A"]]},
  30
  31 {"description": "Windows-1252 LATIN SMALL LETTER F WITH HOOK numeric entity.",
  32 "input":"&#0131;",
  33 "output": ["ParseError", ["Character", "\u0192"]]},
  34
  35 {"description": "Windows-1252 DOUBLE LOW-9 QUOTATION MARK numeric entity.",
  36 "input":"&#0132;",
  37 "output": ["ParseError", ["Character", "\u201E"]]},
  38
  39 {"description": "Windows-1252 HORIZONTAL ELLIPSIS numeric entity.",
  40 "input":"&#0133;",
  41 "output": ["ParseError", ["Character", "\u2026"]]},
  42
  43 {"description": "Windows-1252 DAGGER numeric entity.",
  44 "input":"&#0134;",
  45 "output": ["ParseError", ["Character", "\u2020"]]},
  46
  47 {"description": "Windows-1252 DOUBLE DAGGER numeric entity.",
  48 "input":"&#0135;",
  49 "output": ["ParseError", ["Character", "\u2021"]]},
  50
  51 {"description": "Windows-1252 MODIFIER LETTER CIRCUMFLEX ACCENT numeric entity.",
  52 "input":"&#0136;",
  53 "output": ["ParseError", ["Character", "\u02C6"]]},
  54
  55 {"description": "Windows-1252 PER MILLE SIGN numeric entity.",
  56 "input":"&#0137;",
  57 "output": ["ParseError", ["Character", "\u2030"]]},
  58
  59 {"description": "Windows-1252 LATIN CAPITAL LETTER S WITH CARON numeric entity.",
  60 "input":"&#0138;",
  61 "output": ["ParseError", ["Character", "\u0160"]]},
  62
  63 {"description": "Windows-1252 SINGLE LEFT-POINTING ANGLE QUOTATION MARK numeric entity.",
  64 "input":"&#0139;",
  65 "output": ["ParseError", ["Character", "\u2039"]]},
  66
  67 {"description": "Windows-1252 LATIN CAPITAL LIGATURE OE numeric entity.",
  68 "input":"&#0140;",
  69 "output": ["ParseError", ["Character", "\u0152"]]},
  70
  71 {"description": "Windows-1252 REPLACEMENT CHAR numeric entity.",
  72 "input":"&#0141;",
  73 "output": ["ParseError", ["Character", "\u008D"]]},
  74
  75 {"description": "Windows-1252 LATIN CAPITAL LETTER Z WITH CARON numeric entity.",
  76 "input":"&#0142;",
  77 "output": ["ParseError", ["Character", "\u017D"]]},
  78
  79 {"description": "Windows-1252 REPLACEMENT CHAR numeric entity.",
  80 "input":"&#0143;",
  81 "output": ["ParseError", ["Character", "\u008F"]]},
  82
  83 {"description": "Windows-1252 REPLACEMENT CHAR numeric entity.",
  84 "input":"&#0144;",
  85 "output": ["ParseError", ["Character", "\u0090"]]},
  86
  87 {"description": "Windows-1252 LEFT SINGLE QUOTATION MARK numeric entity.",
  88 "input":"&#0145;",
  89 "output": ["ParseError", ["Character", "\u2018"]]},
  90
  91 {"description": "Windows-1252 RIGHT SINGLE QUOTATION MARK numeric entity.",
  92 "input":"&#0146;",
  93 "output": ["ParseError", ["Character", "\u2019"]]},
  94
  95 {"description": "Windows-1252 LEFT DOUBLE QUOTATION MARK numeric entity.",
  96 "input":"&#0147;",
  97 "output": ["ParseError", ["Character", "\u201C"]]},
  98
  99 {"description": "Windows-1252 RIGHT DOUBLE QUOTATION MARK numeric entity.",
 100 "input":"&#0148;",
 101 "output": ["ParseError", ["Character", "\u201D"]]},
 102
 103 {"description": "Windows-1252 BULLET numeric entity.",
 104 "input":"&#0149;",
 105 "output": ["ParseError", ["Character", "\u2022"]]},
 106
 107 {"description": "Windows-1252 EN DASH numeric entity.",
 108 "input":"&#0150;",
 109 "output": ["ParseError", ["Character", "\u2013"]]},
 110
 111 {"description": "Windows-1252 EM DASH numeric entity.",
 112 "input":"&#0151;",
 113 "output": ["ParseError", ["Character", "\u2014"]]},
 114
 115 {"description": "Windows-1252 SMALL TILDE numeric entity.",
 116 "input":"&#0152;",
 117 "output": ["ParseError", ["Character", "\u02DC"]]},
 118
 119 {"description": "Windows-1252 TRADE MARK SIGN numeric entity.",
 120 "input":"&#0153;",
 121 "output": ["ParseError", ["Character", "\u2122"]]},
 122
 123 {"description": "Windows-1252 LATIN SMALL LETTER S WITH CARON numeric entity.",
 124 "input":"&#0154;",
 125 "output": ["ParseError", ["Character", "\u0161"]]},
 126
 127 {"description": "Windows-1252 SINGLE RIGHT-POINTING ANGLE QUOTATION MARK numeric entity.",
 128 "input":"&#0155;",
 129 "output": ["ParseError", ["Character", "\u203A"]]},
 130
 131 {"description": "Windows-1252 LATIN SMALL LIGATURE OE numeric entity.",
 132 "input":"&#0156;",
 133 "output": ["ParseError", ["Character", "\u0153"]]},
 134
 135 {"description": "Windows-1252 REPLACEMENT CHAR numeric entity.",
 136 "input":"&#0157;",
 137 "output": ["ParseError", ["Character", "\u009D"]]},
 138
 139 {"description": "Windows-1252 EURO SIGN hexadecimal numeric entity.",
 140 "input":"&#x080;",
 141 "output": ["ParseError", ["Character", "\u20AC"]]},
 142
 143 {"description": "Windows-1252 REPLACEMENT CHAR hexadecimal numeric entity.",
 144 "input":"&#x081;",
 145 "output": ["ParseError", ["Character", "\u0081"]]},
 146
 147 {"description": "Windows-1252 SINGLE LOW-9 QUOTATION MARK hexadecimal numeric entity.",
 148 "input":"&#x082;",
 149 "output": ["ParseError", ["Character", "\u201A"]]},
 150
 151 {"description": "Windows-1252 LATIN SMALL LETTER F WITH HOOK hexadecimal numeric entity.",
 152 "input":"&#x083;",
 153 "output": ["ParseError", ["Character", "\u0192"]]},
 154
 155 {"description": "Windows-1252 DOUBLE LOW-9 QUOTATION MARK hexadecimal numeric entity.",
 156 "input":"&#x084;",
 157 "output": ["ParseError", ["Character", "\u201E"]]},
 158
 159 {"description": "Windows-1252 HORIZONTAL ELLIPSIS hexadecimal numeric entity.",
 160 "input":"&#x085;",
 161 "output": ["ParseError", ["Character", "\u2026"]]},
 162
 163 {"description": "Windows-1252 DAGGER hexadecimal numeric entity.",
 164 "input":"&#x086;",
 165 "output": ["ParseError", ["Character", "\u2020"]]},
 166
 167 {"description": "Windows-1252 DOUBLE DAGGER hexadecimal numeric entity.",
 168 "input":"&#x087;",
 169 "output": ["ParseError", ["Character", "\u2021"]]},
 170
 171 {"description": "Windows-1252 MODIFIER LETTER CIRCUMFLEX ACCENT hexadecimal numeric entity.",
 172 "input":"&#x088;",
 173 "output": ["ParseError", ["Character", "\u02C6"]]},
 174
 175 {"description": "Windows-1252 PER MILLE SIGN hexadecimal numeric entity.",
 176 "input":"&#x089;",
 177 "output": ["ParseError", ["Character", "\u2030"]]},
 178
 179 {"description": "Windows-1252 LATIN CAPITAL LETTER S WITH CARON hexadecimal numeric entity.",
 180 "input":"&#x08A;",
 181 "output": ["ParseError", ["Character", "\u0160"]]},
 182
 183 {"description": "Windows-1252 SINGLE LEFT-POINTING ANGLE QUOTATION MARK hexadecimal numeric entity.",
 184 "input":"&#x08B;",
 185 "output": ["ParseError", ["Character", "\u2039"]]},
 186
 187 {"description": "Windows-1252 LATIN CAPITAL LIGATURE OE hexadecimal numeric entity.",
 188 "input":"&#x08C;",
 189 "output": ["ParseError", ["Character", "\u0152"]]},
 190
 191 {"description": "Windows-1252 REPLACEMENT CHAR hexadecimal numeric entity.",
 192 "input":"&#x08D;",
 193 "output": ["ParseError", ["Character", "\u008D"]]},
 194
 195 {"description": "Windows-1252 LATIN CAPITAL LETTER Z WITH CARON hexadecimal numeric entity.",
 196 "input":"&#x08E;",
 197 "output": ["ParseError", ["Character", "\u017D"]]},
 198
 199 {"description": "Windows-1252 REPLACEMENT CHAR hexadecimal numeric entity.",
 200 "input":"&#x08F;",
 201 "output": ["ParseError", ["Character", "\u008F"]]},
 202
 203 {"description": "Windows-1252 REPLACEMENT CHAR hexadecimal numeric entity.",
 204 "input":"&#x090;",
 205 "output": ["ParseError", ["Character", "\u0090"]]},
 206
 207 {"description": "Windows-1252 LEFT SINGLE QUOTATION MARK hexadecimal numeric entity.",
 208 "input":"&#x091;",
 209 "output": ["ParseError", ["Character", "\u2018"]]},
 210
 211 {"description": "Windows-1252 RIGHT SINGLE QUOTATION MARK hexadecimal numeric entity.",
 212 "input":"&#x092;",
 213 "output": ["ParseError", ["Character", "\u2019"]]},
 214
 215 {"description": "Windows-1252 LEFT DOUBLE QUOTATION MARK hexadecimal numeric entity.",
 216 "input":"&#x093;",
 217 "output": ["ParseError", ["Character", "\u201C"]]},
 218
 219 {"description": "Windows-1252 RIGHT DOUBLE QUOTATION MARK hexadecimal numeric entity.",
 220 "input":"&#x094;",
 221 "output": ["ParseError", ["Character", "\u201D"]]},
 222
 223 {"description": "Windows-1252 BULLET hexadecimal numeric entity.",
 224 "input":"&#x095;",
 225 "output": ["ParseError", ["Character", "\u2022"]]},
 226
 227 {"description": "Windows-1252 EN DASH hexadecimal numeric entity.",
 228 "input":"&#x096;",
 229 "output": ["ParseError", ["Character", "\u2013"]]},
 230
 231 {"description": "Windows-1252 EM DASH hexadecimal numeric entity.",
 232 "input":"&#x097;",
 233 "output": ["ParseError", ["Character", "\u2014"]]},
 234
 235 {"description": "Windows-1252 SMALL TILDE hexadecimal numeric entity.",
 236 "input":"&#x098;",
 237 "output": ["ParseError", ["Character", "\u02DC"]]},
 238
 239 {"description": "Windows-1252 TRADE MARK SIGN hexadecimal numeric entity.",
 240 "input":"&#x099;",
 241 "output": ["ParseError", ["Character", "\u2122"]]},
 242
 243 {"description": "Windows-1252 LATIN SMALL LETTER S WITH CARON hexadecimal numeric entity.",
 244 "input":"&#x09A;",
 245 "output": ["ParseError", ["Character", "\u0161"]]},
 246
 247 {"description": "Windows-1252 SINGLE RIGHT-POINTING ANGLE QUOTATION MARK hexadecimal numeric entity.",
 248 "input":"&#x09B;",
 249 "output": ["ParseError", ["Character", "\u203A"]]},
 250
 251 {"description": "Windows-1252 LATIN SMALL LIGATURE OE hexadecimal numeric entity.",
 252 "input":"&#x09C;",
 253 "output": ["ParseError", ["Character", "\u0153"]]},
 254
 255 {"description": "Windows-1252 REPLACEMENT CHAR hexadecimal numeric entity.",
 256 "input":"&#x09D;",
 257 "output": ["ParseError", ["Character", "\u009D"]]},
 258
 259 {"description": "Windows-1252 LATIN SMALL LETTER Z WITH CARON hexadecimal numeric entity.",
 260 "input":"&#x09E;",
 261 "output": ["ParseError", ["Character", "\u017E"]]},
 262
 263 {"description": "Windows-1252 LATIN CAPITAL LETTER Y WITH DIAERESIS hexadecimal numeric entity.",
 264 "input":"&#x09F;",
 265 "output": ["ParseError", ["Character", "\u0178"]]},
 266
 267 {"description": "Decimal numeric entity followed by hex character a.",
 268 "input":"&#97a",
 269 "output": ["ParseError", ["Character", "aa"]]},
 270
 271 {"description": "Decimal numeric entity followed by hex character A.",
 272 "input":"&#97A",
 273 "output": ["ParseError", ["Character", "aA"]]},
 274
 275 {"description": "Decimal numeric entity followed by hex character f.",
 276 "input":"&#97f",
 277 "output": ["ParseError", ["Character", "af"]]},
 278
 279 {"description": "Decimal numeric entity followed by hex character A.",
 280 "input":"&#97F",
 281 "output": ["ParseError", ["Character", "aF"]]}
 282
 283 ]}