Parent Directory
| 
Revision Log
<script src> in parsing algorithm is now supported
| 1 | wakaba | 1.1 | <!DOCTYPE HTML> |
| 2 | <html lang=en> | ||
| 3 | <head> | ||
| 4 | <title>Demo of HTML5 Parsing Algorithm with Scripting Enabled</title> | ||
| 5 | <style> | ||
| 6 | textarea { | ||
| 7 | display: block; | ||
| 8 | width: 80%; | ||
| 9 | margin-left: auto; | ||
| 10 | margin-right: auto; | ||
| 11 | min-height: 20em; | ||
| 12 | } | ||
| 13 | output { | ||
| 14 | display: block; | ||
| 15 | font-family: monospace; | ||
| 16 | wakaba | 1.4 | white-space: -moz-pre-wrap; |
| 17 | white-space: pre-wrap; | ||
| 18 | wakaba | 1.1 | } |
| 19 | </style> | ||
| 20 | <script> | ||
| 21 | function update () { | ||
| 22 | document.logElement.textContent = ''; | ||
| 23 | wakaba | 1.2 | var p = new Parser (new InputStream (document.sourceElement.value)); |
| 24 | wakaba | 1.4 | var doc = p.doc; |
| 25 | wakaba | 1.2 | p.parse (); |
| 26 | wakaba | 1.4 | log (dumpTree (doc, '')); |
| 27 | wakaba | 1.1 | } // update |
| 28 | |||
| 29 | wakaba | 1.6 | var logIndentLevel = 0; |
| 30 | wakaba | 1.1 | function log (s) { |
| 31 | wakaba | 1.6 | for (var i = 0; i < logIndentLevel; i++) { |
| 32 | s = ' ' + s; | ||
| 33 | } | ||
| 34 | wakaba | 1.1 | document.logElement.appendChild (document.createTextNode (s + "\n")); |
| 35 | } // log | ||
| 36 | |||
| 37 | function InputStream (s) { | ||
| 38 | this.s = s; | ||
| 39 | } // InputStream | ||
| 40 | |||
| 41 | wakaba | 1.4 | function Parser (i, doc) { |
| 42 | wakaba | 1.1 | this.parseMode = 'pcdata'; |
| 43 | wakaba | 1.4 | if (!doc) { |
| 44 | doc = new JSDocument (this); | ||
| 45 | doc.manakaiIsHTML = true; | ||
| 46 | } | ||
| 47 | this.doc = doc; | ||
| 48 | this.openElements = [doc]; | ||
| 49 | wakaba | 1.2 | this.in = i; |
| 50 | wakaba | 1.4 | this.scriptsExecutedAfterParsing = []; |
| 51 | wakaba | 1.1 | } // Parser |
| 52 | |||
| 53 | wakaba | 1.2 | Parser.prototype.getNextToken = function () { |
| 54 | wakaba | 1.3 | var p = this; |
| 55 | wakaba | 1.2 | var i = this.in; |
| 56 | wakaba | 1.1 | if (this.parseMode == 'script') { |
| 57 | var token; | ||
| 58 | wakaba | 1.3 | if (p.insertionPoint <= 0) { |
| 59 | return {type: 'abort'}; | ||
| 60 | } | ||
| 61 | wakaba | 1.4 | i.s = i.s.replace (/^([^<]+)/, |
| 62 | wakaba | 1.1 | function (s, t) { |
| 63 | wakaba | 1.3 | if (0 < p.insertionPoint && p.insertionPoint < t.length) { |
| 64 | token = {type: 'char', value: t.substring (0, p.insertionPoint)}; | ||
| 65 | var ip = p.insertionPoint; | ||
| 66 | p.insertionPoint = 0; | ||
| 67 | wakaba | 1.4 | return t.substring (ip, t.length); |
| 68 | wakaba | 1.3 | } |
| 69 | wakaba | 1.1 | token = {type: 'char', value: t}; |
| 70 | wakaba | 1.4 | p.insertionPoint -= t.length; |
| 71 | return ''; | ||
| 72 | wakaba | 1.1 | }); |
| 73 | if (token) return token; | ||
| 74 | wakaba | 1.3 | i.s = i.s.replace (/^<\/[Ss][Cc][Rr][Ii][Pp][Tt]>/, function (s) { |
| 75 | wakaba | 1.4 | if (p.insertionPoint < s.length) { |
| 76 | wakaba | 1.3 | token = {type: 'abort'}; |
| 77 | return s; | ||
| 78 | } | ||
| 79 | wakaba | 1.1 | token = {type: 'end-tag', value: 'script'}; |
| 80 | wakaba | 1.3 | p.insertionPoint -= s.length; |
| 81 | wakaba | 1.1 | return ''; |
| 82 | }); | ||
| 83 | if (token) return token; | ||
| 84 | wakaba | 1.5 | var m; |
| 85 | if ((p.insertionPoint < '</script'.length) && | ||
| 86 | (m = i.s.match (/^<\/([SCRIPTscript]+)/))) { | ||
| 87 | var v = m[1].substring (0, p.insertionPoint).toLowerCase (); | ||
| 88 | if (v == 'script'.substring (0, p.insertionPoint - '</'.length)) { | ||
| 89 | return {type: 'abort'}; | ||
| 90 | } | ||
| 91 | } | ||
| 92 | wakaba | 1.4 | i.s = i.s.replace (/^</, |
| 93 | function (s) { | ||
| 94 | token = {type: 'char', value: s}; | ||
| 95 | p.insertionPoint -= s.length; | ||
| 96 | return ''; | ||
| 97 | }); | ||
| 98 | if (token) return token; | ||
| 99 | wakaba | 1.1 | return {type: 'eof'}; |
| 100 | } | ||
| 101 | |||
| 102 | var token; | ||
| 103 | wakaba | 1.5 | i.s = i.s.replace (/^<\/([^>]+)(?:>|$)/, function (s, e) { |
| 104 | if (p.insertionPoint < s.length || | ||
| 105 | (p.insertionPoint <= s.length && | ||
| 106 | s.substring (s.length - 1, 1) != '>')) { | ||
| 107 | wakaba | 1.3 | token = {type: 'abort'}; |
| 108 | return s; | ||
| 109 | } | ||
| 110 | wakaba | 1.1 | token = {type: 'end-tag', value: e.toLowerCase ()}; |
| 111 | wakaba | 1.3 | p.insertionPoint -= s.length; |
| 112 | wakaba | 1.1 | return ''; |
| 113 | }); | ||
| 114 | if (token) return token; | ||
| 115 | wakaba | 1.5 | i.s = i.s.replace (/^<([^>]+)(?:>|$)/, function (s, e) { |
| 116 | if (p.insertionPoint < s.length || | ||
| 117 | (p.insertionPoint <= s.length && | ||
| 118 | s.substring (s.length - 1, 1) != '>')) { | ||
| 119 | wakaba | 1.3 | token = {type: 'abort'}; |
| 120 | return s; | ||
| 121 | } | ||
| 122 | wakaba | 1.4 | var tagName; |
| 123 | var attrs = {}; | ||
| 124 | e = e.replace (/^[\S]+/, function (v) { | ||
| 125 | tagName = v.toLowerCase (); | ||
| 126 | return ''; | ||
| 127 | }); | ||
| 128 | wakaba | 1.6 | e = e.replace (/^\s*([^\s=]+)\s*(?:=\s*(?:"([^"]*)"|'([^']*)'|([^"']+)))?/, |
| 129 | wakaba | 1.4 | function (x, attrName, attrValue1, attrValue2, attrValue3) { |
| 130 | wakaba | 1.6 | v = attrValue1 || attrValue2 || attrValue3; |
| 131 | v = v.replace (/"/g, '"').replace (/'/g, "'") | ||
| 132 | .replace (/&/g, '&'); | ||
| 133 | attrs[attrName.toLowerCase ()] = v; | ||
| 134 | wakaba | 1.4 | return ''; |
| 135 | }); | ||
| 136 | wakaba | 1.6 | if (e.length) { |
| 137 | log ('Broken start tag: "' + e + '"'); | ||
| 138 | } | ||
| 139 | wakaba | 1.4 | token = {type: 'start-tag', value: tagName, attrs: attrs}; |
| 140 | wakaba | 1.3 | p.insertionPoint -= s.length; |
| 141</ |