pkgs/html/test/data/tokenizer/unicodeCharsSurrogates.test

{"tests" : [
{"description": "Unicode surrogate (emoji)",
"input": "\uD83D\uDC3C",
"output":[["Character", "\uD83D\uDC3C"]]},

{"description": "Unicode surrogate (emoji) prefixed by characters",
"input": "before\uD83D\uDC3C",
"output":[["Character", "before\uD83D\uDC3C"]]},

{"description": "Unicode surrogate (emoji) suffixed by characters",
"input": "\uD83D\uDC3Cafter",
"output":[["Character", "\uD83D\uDC3Cafter"]]},

{"description":"Quoted attribute with surrogate unicode content",
"generateSpans": true,
"input":"<a href='\uD83D\uDC3C'/>",
"output":[["StartTag","a",{"href":"\uD83D\uDC3C"},true,0,14]]},

{"description":"Surrogate unicode content followed by attribute",
"generateSpans": true,
"input":"\uD83D\uDC3C<a href='b'/>",
"output":[["Character", "\uD83D\uDC3C", 0, 2],["StartTag","a",{"href":"b"},true,2,15]]}
]
}