UNPKG

tiny-html-lexer

Version:

A tiny HTML5 lexer

github.com/alwinb/tiny-html-lexer

alwinb/tiny-html-lexer

158 lines (149 loc) • 6.56 kB

JavaScript

var samples = [ 'charref: named & in data' , 'charref: named non-terminated &amp in data' , 'charref: named non-terminated &ampa in data' , 'charref: hexadecimal ೌ in data' , 'charref: hexadecimal non-terminated &#xccc in data' , 'charref: decimal ф in data' , 'charref: decimal non-terminated &#110 in data' , 'charref: special <input value=asda&not(></input>' , 'charref: special <input value=asda&not-></input>' , 'charref: special <input value=asda&not*=c></input>' , 'charref: special <input value=asda&not=c></input>' , 'charref: special <input value="asda&notit; I tell you"></input>' , 'charref: non-special <input value=asda&notin*=c></input>' , 'charref: non-special <input value=asda&notin=c></input>' , 'charref: non-special <input value=asda∉=c></input>' , 'charref: special &not(' , 'charref: special &not-' , 'charref: special &not*=c in data' , 'charref: special &not=c in data' , 'charref: special &notit; I tell you, in data' , 'charref: special ∉ I tell you, in data' , 'charref: non-special &notin*=c in data' , 'charref: non-special &notin=c in data' , 'charref: non-special ∉=c in data' , 'charref: named <input value="you & me"/> in attribute' , 'charref: named <input value=\'you & me\'/> in attribute' , 'charref: named <input value=you&#12me /> in attribute' , 'charref: named <input value=&me /> in attribute' , 'charref: named <input value=&amp attr=val /> in attribute' , 'charref: named <input value=&ampo attr=val /> in attribute' , 'charref: bogus <input value="you &# am me"/> in attribute' , 'charref: bogus <input value=\'you &# amp me\'/> in attribute' , 'charref: bogus <input value=you&x ampme /> in attribute' , 'charref: ampHash &# such' , '' , 'rcdata <textarea> asdf & & <textareaNot </textarea> and more' , 'rcdata2 <textarea> asdf & & </textarea( and not ending> it' , 'rcdata3 <textarea> asdf & & </textarea/ and ending> it, see <span>yes</span>' , 'rcdata5 <textarea/> asdf & &amp and NOT ending < it, see <span>yes</span>' , 'rcdata4 <textarea> asdf & & </textarea and ending> it' , 'rawtext <script> asdf & <span> </scriptNot </script> and more' , 'rawtext2 <script> asdf & <span> </script( and> not ending it <span>' , 'rawtext3 <script> asdf & <span> </script/ and> ending it <span> see' , 'rawtext4 <script> asdf & <span> </script and ending it <span> see' , 'script <!doctype html>hello <script> and such' , 'comment: <?--> and such' , 'comment:  and such' , 'comment: <!-> and such' , 'comment:  and subsequent data' , 'comment:  and subsequent data' , 'comment:  and subsequent data' , 'comment: <!-- Comment with strange end --!> and subsequent data' , 'bogus comment: <! with end !@> and subsequent data' , 'bogus comment: </ with end !@> and subsequent data' , 'bogus comment: <? with end !@> and subsequent data' , 'bogus comment: <!- with end -> and subsequent data' , '' , 'missing space attribues connected <div name="a"name="b" >' , 'nonalpha attribute weird template tag <div {name="a" name="b" >' , 'normalHtml This is <span class = "s1">html</span> Yeah!' , 'unescaped ampersand data & such' , 'unescaped ampersand Hash data &# such' , 'unescaped ampersand HashEx data &#x such' , 'unescaped ampersand HashExZed data &#xz such' , '' , 'slashes: <span/>' , 'slashes: <span name=foo//>' , 'slashes: <div//>' , 'slashes: <div/foo/bar//>' , 'slashes: <span//>' , 'slashes: <span />' , 'slashes: <span <>' , 'slashes: <span //>' , 'slashes: <span / />' , 'slashes: <span/////>' , 'slashes: <span/////name////=/blabla>' , 'slashes: <span / attr >foo bar</span>' , 'slashes: <span name=/ >asdf' , 'slashes: <span name=/>asdf' , 'slashes: <span name=// />asdf' , 'slashes: <span name= / />asdf' , 'weirdEquals <span attr = / asd >content</span>' , 'weirdEquals2 <span attr = @ asd >content</span>' , 'weirdEquals3 <span attr /= asd >content</span>' , 'weirdEquals4 <span attr @= asd >content</span>' , 'missingValue <span name=>asdf' , 'invalidAttributeValue1 <div class= =at >' , 'invalidAttributeValue2 <div class= <at >' , 'invalidAttributeValue3 <div class= `at >' ] var EOFSamples = [ 'data state eof in da' , 'tagOpen state eof in <' , 'tagName state eof in <d' , 'selfClosingStartTag state in <div /' , 'endTagOpen state in </a' , 'beforeAttributeName state <div ' , 'attributeName state <div at' , 'afterAttributeName state <div attr ' , 'beforeAttributeValue state <div attr =' , 'attributeValueDoubleQuoted state <div attr="te' , 'attributeValueSingleQuoted state <div attr=\'te' , 'attributeValueUnquoted state <div attr=te' , 'afterAttributeValueQuoted state <div attr="test"' , 'markupDeclarationOpen state a markup decl <!' , 'selfClosingTag state An eof after a / <span /' , 'commentStart state a comment start <!--' , 'commentStartDash state a comment start dash <!---' , 'comment state a comment <!-- hello th' , 'commentEndDash state a comment end dash <!-- hello th -' , 'commentEnd state a comment end <!-- hello th --' , 'commentEndBang state a comment end bang <!-- hello th --!' , 'bogusComment state <! bogus comment' , 'charRefIn_ state data &' , 'numericCharRef state data &#' , 'hexadecimalCharRef state data &#x' , 'hexDigits state data &#x1a' , 'decimalCharRef state data &#1' , 'namedCharRef state data &name' , 'namedCharRefInAttr state <span attr="asd&a&b c">text</span>' , 'namedCharRefInData state named charref in data asd&a&b cde' , 'rawtext state eof in raw text <script> funct' , 'plaintext state eof in raw text <plaintext> asdf' , 'rawtextLessThanSign state eof in raw text less than sign <script> if (i<' , 'rawtextEndTagOpen state eof in raw text end tag open <script> asdf </' ] module.exports = { samples: samples , EOFSamples: EOFSamples }