html/testdata/html5lib-tests/tokenizer/domjs.test

{
    "tests": [
        {
            "description":"CR in bogus comment state",
            "input":"<?\u000d",
            "output":[["Comment", "?\u000a"]],
            "errors":[
                { "code": "unexpected-question-mark-instead-of-tag-name", "line": 1, "col": 2 }
            ]
        },
        {
            "description":"CRLF in bogus comment state",
            "input":"<?\u000d\u000a",
            "output":[["Comment", "?\u000a"]],
            "errors":[
                { "code": "unexpected-question-mark-instead-of-tag-name", "line": 1, "col": 2 }
            ]
        },
        {
            "description":"CRLFLF in bogus comment state",
            "input":"<?\u000d\u000a\u000a",
            "output":[["Comment", "?\u000a\u000a"]],
            "errors":[
                { "code": "unexpected-question-mark-instead-of-tag-name", "line": 1, "col": 2 }
            ]
        },
        {
            "description":"Raw NUL replacement",
            "doubleEscaped":true,
            "initialStates":["RCDATA state", "RAWTEXT state", "PLAINTEXT state", "Script data state"],
            "input":"\\u0000",
            "output":[["Character", "\\uFFFD"]],
            "errors":[
                { "code": "unexpected-null-character", "line": 1, "col": 1 }
            ]
        },
        {
            "description":"NUL in CDATA section",
            "doubleEscaped":true,
            "initialStates":["CDATA section state"],
            "input":"\\u0000]]>",
            "output":[["Character", "\\u0000"]]
        },
        {
           "description":"NUL in script HTML comment",
           "doubleEscaped":true,
           "initialStates":["Script data state"],
           "input":"<!--test\\u0000--><!--test-\\u0000--><!--test--\\u0000-->",
           "output":[["Character", "<!--test\\uFFFD--><!--test-\\uFFFD--><!--test--\\uFFFD-->"]],
           "errors":[
               { "code": "unexpected-null-character", "line": 1, "col": 9 },
               { "code": "unexpected-null-character", "line": 1, "col": 22 },
               { "code": "unexpected-null-character", "line": 1, "col": 36 }
           ]
        },
        {
           "description":"NUL in script HTML comment - double escaped",
           "doubleEscaped":true,
           "initialStates":["Script data state"],
           "input":"<!--<script>\\u0000--><!--<script>-\\u0000--><!--<script>--\\u0000-->",
           "output":[["Character", "<!--<script>\\uFFFD--><!--<script>-\\uFFFD--><!--<script>--\\uFFFD-->"]],
           "errors":[
                { "code": "unexpected-null-character", "line": 1, "col": 13 },
                { "code": "unexpected-null-character", "line": 1, "col": 30 },
                { "code": "unexpected-null-character", "line": 1, "col": 48 }
           ]
        },
        {
           "description":"EOF in script HTML comment",
           "initialStates":["Script data state"],
           "input":"<!--test",
           "output":[["Character", "<!--test"]],
           "errors":[
               { "code": "eof-in-script-html-comment-like-text", "line": 1, "col": 9 }
           ]
        },
        {
           "description":"EOF in script HTML comment after dash",
           "initialStates":["Script data state"],
           "input":"<!--test-",
           "output":[["Character", "<!--test-"]],
           "errors":[
               { "code": "eof-in-script-html-comment-like-text", "line": 1, "col": 10 }
           ]
        },
        {
           "description":"EOF in script HTML comment after dash dash",
           "initialStates":["Script data state"],
           "input":"<!--test--",
           "output":[["Character", "<!--test--"]],
           "errors":[
               { "code": "eof-in-script-html-comment-like-text", "line": 1, "col": 11 }
           ]
        },
        {
           "description":"EOF in script HTML comment double escaped after dash",
           "initialStates":["Script data state"],
           "input":"<!--<script>-",
           "output":[["Character", "<!--<script>-"]],
           "errors":[
               { "code": "eof-in-script-html-comment-like-text", "line": 1, "col": 14 }
           ]
        },
        {
           "description":"EOF in script HTML comment double escaped after dash dash",
           "initialStates":["Script data state"],
           "input":"<!--<script>--",
           "output":[["Character", "<!--<script>--"]],
           "errors":[
               { "code": "eof-in-script-html-comment-like-text", "line": 1, "col": 15 }
           ]
        },
        {
           "description":"EOF in script HTML comment - double escaped",
           "initialStates":["Script data state"],
           "input":"<!--<script>",
           "output":[["Character", "<!--<script>"]],
           "errors":[
               { "code": "eof-in-script-html-comment-like-text", "line": 1, "col": 13 }
           ]
        },
        {
            "description":"Dash in script HTML comment",
            "initialStates":["Script data state"],
            "input":"<!-- - -->",
            "output":[["Character", "<!-- - -->"]]
        },
        {
            "description":"Dash less-than in script HTML comment",
            "initialStates":["Script data state"],
            "input":"<!-- -< -->",
            "output":[["Character", "<!-- -< -->"]]
        },
        {
            "description":"Dash at end of script HTML comment",
            "initialStates":["Script data state"],
            "input":"<!--test--->",
            "output":[["Character", "<!--test--->"]]
        },
        {
            "description":"</script> in script HTML comment",
            "initialStates":["Script data state"],
            "lastStartTag":"script",
            "input":"<!-- </script> --></script>",
            "output":[["Character", "<!-- "], ["EndTag", "script"], ["Character", " -->"], ["EndTag", "script"]]
        },
        {
            "description":"</script> in script HTML comment - double escaped",
            "initialStates":["Script data state"],
            "lastStartTag":"script",
            "input":"<!-- <script></script> --></script>",
            "output":[["Character", "<!-- <script></script> -->"], ["EndTag", "script"]]
        },
        {
            "description":"</script> in script HTML comment - double escaped with nested <script>",
            "initialStates":["Script data state"],
            "lastStartTag":"script",
            "input":"<!-- <script><script></script></script> --></script>",
            "output":[["Character", "<!-- <script><script></script>"], ["EndTag", "script"], ["Character", " -->"], ["EndTag", "script"]]
        },
        {
            "description":"</script> in script HTML comment - double escaped with abrupt end",
            "initialStates":["Script data state"],
            "lastStartTag":"script",
            "input":"<!-- <script>--></script> --></script>",
            "output":[["Character", "<!-- <script>-->"], ["EndTag", "script"], ["Character", " -->"], ["EndTag", "script"]]
        },
        {
            "description":"Incomplete start tag in script HTML comment double escaped",
            "initialStates":["Script data state"],
            "lastStartTag":"script",
            "input":"<!--<scrip></script>-->",
            "output":[["Character", "<!--<scrip>"], ["EndTag", "script"], ["Character", "-->"]]
        },
        {
            "description":"Unclosed start tag in script HTML comment double escaped",
            "initialStates":["Script data state"],
            "lastStartTag":"script",
            "input":"<!--<script</script>-->",
            "output":[["Character", "<!--<script"], ["EndTag", "script"], ["Character", "-->"]]
        },
        {
            "description":"Incomplete end tag in script HTML comment double escaped",
            "initialStates":["Script data state"],
            "lastStartTag":"script",
            "input":"<!--<script></scrip>-->",
            "output":[["Character", "<!--<script></scrip>-->"]]
        },
        {
            "description":"Unclosed end tag in script HTML comment double escaped",
            "initialStates":["Script data state"],
            "lastStartTag":"script",
            "input":"<!--<script></script-->",
            "output":[["Character", "<!--<script></script-->"]]
        },
        {
            "description":"leading U+FEFF must pass through",
            "initialStates":["Data state", "RCDATA state", "RAWTEXT state", "Script data state"],
            "doubleEscaped":true,
            "input":"\\uFEFFfoo\\uFEFFbar",
            "output":[["Character", "\\uFEFFfoo\\uFEFFbar"]]
        },
        {
            "description":"Non BMP-charref in RCDATA",
            "initialStates":["RCDATA state"],
            "input":"&NotEqualTilde;",
            "output":[["Character", "\u2242\u0338"]]
        },
        {
            "description":"Bad charref in RCDATA",
            "initialStates":["RCDATA state"],
            "input":"&NotEqualTild;",
            "output":[["Character", "&NotEqualTild;"]],
            "errors":[
               { "code": "unknown-named-character-reference", "line": 1, "col": 14 }
            ]
        },
        {
            "description":"lowercase endtags",
            "initialStates":["RCDATA state", "RAWTEXT state", "Script data state"],
            "lastStartTag":"xmp",
            "input":"</XMP>",
            "output":[["EndTag","xmp"]]
        },
        {
            "description":"bad endtag (space before name)",
            "initialStates":["RCDATA state", "RAWTEXT state", "Script data state"],
            "lastStartTag":"xmp",
            "input":"</ XMP>",
            "output":[["Character","</ XMP>"]]
        },
        {
            "description":"bad endtag (not matching last start tag)",
            "initialStates":["RCDATA state", "RAWTEXT state", "Script data state"],
            "lastStartTag":"xmp",
            "input":"</xm>",
            "output":[["Character","</xm>"]]
        },
        {
            "description":"bad endtag (without close bracket)",
            "initialStates":["RCDATA state", "RAWTEXT state", "Script data state"],
            "lastStartTag":"xmp",
            "input":"</xm ",
            "output":[["Character","</xm "]]
        },
        {
            "description":"bad endtag (trailing solidus)",
            "initialStates":["RCDATA state", "RAWTEXT state", "Script data state"],
            "lastStartTag":"xmp",
            "input":"</xm/",
            "output":[["Character","</xm/"]]
        },
        {
            "description":"Non BMP-charref in attribute",
            "input":"<p id=\"&NotEqualTilde;\">",
            "output":[["StartTag", "p", {"id":"\u2242\u0338"}]]
        },
        {
            "description":"--!NUL in comment ",
            "doubleEscaped":true,
            "input":"<!----!\\u0000-->",
            "output":[["Comment", "--!\\uFFFD"]],
            "errors":[
                { "code": "unexpected-null-character", "line": 1, "col": 8 }
            ]
        },
        {
            "description":"space EOF after doctype ",
            "input":"<!DOCTYPE html ",
            "output":[["DOCTYPE", "html", null, null , false]],
            "errors":[
                { "code": "eof-in-doctype", "line": 1, "col": 16 }
            ]
        },
        {
            "description":"CDATA in HTML content",
            "input":"<![CDATA[foo]]>",
            "output":[["Comment", "[CDATA[foo]]"]],
            "errors":[
                { "code": "cdata-in-html-content", "line": 1, "col": 9 }
            ]
        },
        {
            "description":"CDATA content",
            "input":"foo&#32;]]>",
            "initialStates":["CDATA section state"],
            "output":[["Character", "foo&#32;"]]
        },
        {
            "description":"CDATA followed by HTML content",
            "input":"foo&#32;]]>&#32;",
            "initialStates":["CDATA section state"],
            "output":[["Character", "foo&#32; "]]
        },
        {
            "description":"CDATA with extra bracket",
            "input":"foo]]]>",
            "initialStates":["CDATA section state"],
            "output":[["Character", "foo]"]]
        },
        {
            "description":"CDATA without end marker",
            "input":"foo",
            "initialStates":["CDATA section state"],
            "output":[["Character", "foo"]],
            "errors":[
                { "code": "eof-in-cdata", "line": 1, "col": 4 }
            ]
        },
        {
            "description":"CDATA with single bracket ending",
            "input":"foo]",
            "initialStates":["CDATA section state"],
            "output":[["Character", "foo]"]],
            "errors":[
                { "code": "eof-in-cdata", "line": 1, "col": 5 }
            ]
        },
        {
            "description":"CDATA with two brackets ending",
            "input":"foo]]",
            "initialStates":["CDATA section state"],
            "output":[["Character", "foo]]"]],
            "errors":[
                { "code": "eof-in-cdata", "line": 1, "col": 6 }
            ]
        },
        {
            "description": "HTML tag in script data",
            "input": "<b>hello world</b>",
            "initialStates": ["Script data state"],
            "output": [["Character", "<b>hello world</b>"]]
        }
    ]
}