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 |
|
|
white-space: pre; |
17 |
|
|
} |
18 |
|
|
</style> |
19 |
|
|
<script> |
20 |
|
|
function update () { |
21 |
|
|
document.logElement.textContent = ''; |
22 |
wakaba |
1.2 |
var p = new Parser (new InputStream (document.sourceElement.value)); |
23 |
|
|
p.parse (); |
24 |
wakaba |
1.1 |
log (dumpTree (p.doc, '')); |
25 |
|
|
} // update |
26 |
|
|
|
27 |
|
|
function log (s) { |
28 |
|
|
document.logElement.appendChild (document.createTextNode (s + "\n")); |
29 |
|
|
} // log |
30 |
|
|
|
31 |
|
|
function InputStream (s) { |
32 |
|
|
this.s = s; |
33 |
|
|
} // InputStream |
34 |
|
|
|
35 |
wakaba |
1.2 |
function Parser (i) { |
36 |
wakaba |
1.1 |
this.parseMode = 'pcdata'; |
37 |
wakaba |
1.2 |
this.doc = new JSDocument (this); |
38 |
wakaba |
1.1 |
this.openElements = [this.doc]; |
39 |
wakaba |
1.2 |
this.in = i; |
40 |
wakaba |
1.1 |
} // Parser |
41 |
|
|
|
42 |
wakaba |
1.2 |
Parser.prototype.getNextToken = function () { |
43 |
wakaba |
1.3 |
var p = this; |
44 |
wakaba |
1.2 |
var i = this.in; |
45 |
wakaba |
1.1 |
if (this.parseMode == 'script') { |
46 |
|
|
var token; |
47 |
wakaba |
1.3 |
if (p.insertionPoint <= 0) { |
48 |
|
|
return {type: 'abort'}; |
49 |
|
|
} |
50 |
wakaba |
1.1 |
i.s = i.s.replace (/^([\s\S]+?)<\/[Ss][Cc][Rr][Ii][Pp][Tt]>/, |
51 |
|
|
function (s, t) { |
52 |
wakaba |
1.3 |
if (0 < p.insertionPoint && p.insertionPoint < t.length) { |
53 |
|
|
token = {type: 'char', value: t.substring (0, p.insertionPoint)}; |
54 |
|
|
var ip = p.insertionPoint; |
55 |
|
|
p.insertionPoint = 0; |
56 |
|
|
return t.substring (ip, t.length) + |
57 |
|
|
s.substring (s.length - 9, s.length); |
58 |
|
|
} |
59 |
wakaba |
1.1 |
token = {type: 'char', value: t}; |
60 |
wakaba |
1.3 |
p.insertionPoint -= s.length; |
61 |
wakaba |
1.1 |
return '<' + '/script>'; |
62 |
|
|
}); |
63 |
|
|
if (token) return token; |
64 |
wakaba |
1.3 |
i.s = i.s.replace (/^<\/[Ss][Cc][Rr][Ii][Pp][Tt]>/, function (s) { |
65 |
|
|
if (s.length < p.insertionPoint) { |
66 |
|
|
token = {type: 'abort'}; |
67 |
|
|
return s; |
68 |
|
|
} |
69 |
wakaba |
1.1 |
token = {type: 'end-tag', value: 'script'}; |
70 |
wakaba |
1.3 |
p.insertionPoint -= s.length; |
71 |
wakaba |
1.1 |
return ''; |
72 |
|
|
}); |
73 |
|
|
if (token) return token; |
74 |
|
|
return {type: 'eof'}; |
75 |
|
|
} |
76 |
|
|
|
77 |
|
|
var token; |
78 |
|
|
i.s = i.s.replace (/^<\/([^>]+)>/, function (s, e) { |
79 |
wakaba |
1.3 |
if (p.insertionPoint < s.length) { |
80 |
|
|
token = {type: 'abort'}; |
81 |
|
|
return s; |
82 |
|
|
} |
83 |
wakaba |
1.1 |
token = {type: 'end-tag', value: e.toLowerCase ()}; |
84 |
wakaba |
1.3 |
p.insertionPoint -= s.length; |
85 |
wakaba |
1.1 |
return ''; |
86 |
|
|
}); |
87 |
|
|
if (token) return token; |
88 |
|
|
i.s = i.s.replace (/^<([^>]+)>/, function (s, e) { |
89 |
wakaba |
1.3 |
if (p.insertionPoint < s.length) { |
90 |
|
|
token = {type: 'abort'}; |
91 |
|
|
return s; |
92 |
|
|
} |
93 |
wakaba |
1.1 |
token = {type: 'start-tag', value: e.toLowerCase ()}; |
94 |
wakaba |
1.3 |
p.insertionPoint -= s.length; |
95 |
wakaba |
1.1 |
return ''; |
96 |
|
|
}); |
97 |
|
|
if (token) return token; |
98 |
wakaba |
1.3 |
if (p.insertionPoint <= 0) { |
99 |
|
|
return {type: 'abort'}; |
100 |
|
|
} |
101 |
wakaba |
1.1 |
i.s = i.s.replace (/^[^<]+/, function (s) { |
102 |
wakaba |
1.3 |
if (p.insertionPoint < s.length) { |
103 |
|
|
token = {type: 'char', value: s.substring (0, p.insertionPoint)}; |
104 |
|
|
var ip = p.insertionPoint; |
105 |
|
|
p.insertionPoint = 0; |
106 |
|
|
return s.substring (ip, s.length); |
107 |
|
|
} |
108 |
wakaba |
1.1 |
token = {type: 'char', value: s}; |
109 |
wakaba |
1.3 |
p.insertionPoint -= s.length; |
110 |
wakaba |
1.1 |
return ''; |
111 |
|
|
}); |
112 |
|
|
if (token) return token; |
113 |
|
|
i.s = i.s.replace (/^[\s\S]/, function (s) { |
114 |
|
|
token = {type: 'char', value: s}; |
115 |
wakaba |
1.3 |
p.insertionPoint -= s.length; |
116 |
wakaba |
1.1 |
return ''; |
117 |
|
|
}); |
118 |
|
|
if (token) return token; |
119 |
|
|
return {type: 'eof'}; |
120 |
|
|
} // getNextToken |
121 |
|
|
|
122 |
wakaba |
1.2 |
Parser.prototype.parse = function () { |
123 |
wakaba |
1.1 |
log ('start parsing'); |
124 |
|
|
|
125 |
|
|
while (true) { |
126 |
wakaba |
1.2 |
var token = this.getNextToken (); |
127 |
wakaba |
1.1 |
log ('token: ' + token.type + ' "' + token.value + '"'); |
128 |
|
|
|
129 |
|
|
if (token.type == 'start-tag') { |
130 |
|
|
if (token.value == 'script') { |
131 |
wakaba |
1.2 |
// 1. Create an element for the token in the HTML namespace. |
132 |
|
|
var el = new JSElement (this.doc, token.value); |
133 |
|
|
|
134 |
|
|
// 2. Mark the element as being "parser-inserted". |
135 |
|
|
el.manakaiParserInserted = true; |
136 |
|
|
|
137 |
|
|
// 3. Switch the tokeniser's content model flag to the CDATA state. |
138 |
wakaba |
1.1 |
this.parseMode = 'script'; |
139 |
|
|
|
140 |
wakaba |
1.2 |
// 4.1. Collect all the character tokens. |
141 |
wakaba |
1.1 |
while (true) { |
142 |
wakaba |
1.2 |
var token = this.getNextToken (); |
143 |
wakaba |
1.1 |
log ('token: ' + token.type + ' "' + token.value + '"'); |
144 |
|
|
|
145 |
|
|
if (token.type == 'char') { |
146 |
wakaba |
1.2 |
// 5. Append a single Text node to the script element node. |
147 |
wakaba |
1.1 |
el.manakaiAppendText (token.value); |
148 |
wakaba |
1.2 |
|
149 |
|
|
// 4.2. Until it returns a token that is not a character token, or |
150 |
wakaba |
1.3 |
// until it stops tokenising. |
151 |
wakaba |
1.1 |
} else if (token.type == 'eof' || |
152 |
wakaba |
1.3 |
(token.type == 'end-tag' && token.value == 'script') || |
153 |
|
|
token.type == 'abort') { |
154 |
wakaba |
1.2 |
// 6. Switched back to the PCDATA state. |
155 |
wakaba |
1.1 |
this.parseMode = 'pcdata'; |
156 |
wakaba |
1.2 |
|
157 |
|
|
// 7.1. If the next token is not an end tag token with ... |
158 |
|
|
if (token.type != 'end-tag') { |
159 |
|
|
// 7.2. This is a parse error. |
160 |
|
|
log ('Parse error: no </' + 'script>'); |
161 |
|
|
|
162 |
|
|
// 7.3. Mark the script element as "already executed". |
163 |
|
|
el.manakaiAlreadyExecuted = true; |
164 |
|
|
} else { |
165 |
|
|
// 7.4. Ignore it. |
166 |
|
|
// |
167 |
|
|
} |
168 |
wakaba |
1.1 |
break; |
169 |
|
|
} |
170 |
|
|
} |
171 |
|
|
|
172 |
wakaba |
1.2 |
// 8.1. If the parser were originally created for the ... |
173 |
|
|
if (this.fragmentParsingMode) { |
174 |
|
|
// 8.2. Mark the script element as "already executed" and ... |
175 |
|
|
el.alreadyExecuted = true; |
176 |
|
|
continue; |
177 |
|
|
} |
178 |
|
|
|
179 |
|
|
// 9.1. Let the old insertion point have the same value as the ... |
180 |
wakaba |
1.3 |
var oldInsertionPoint = this.insertionPoint; |
181 |
wakaba |
1.2 |
// 9.2. Let the insertion point be just before the next input ... |
182 |
wakaba |
1.3 |
this.setInsertionPoint (0); |
183 |
wakaba |
1.2 |
|
184 |
|
|
// 10. Append the new element to the current node. |
185 |
wakaba |
1.1 |
this.openElements[this.openElements.length - 1].appendChild (el); |
186 |
wakaba |
1.2 |
|
187 |
|
|
// 11. Let the insertion point have the value of the old ... |
188 |
wakaba |
1.3 |
this.setInsertionPoint (oldInsertionPoint); |
189 |
wakaba |
1.2 |
|
190 |
|
|
// 12. If there is a script that will execute as soon as ... |
191 |
|
|
|
192 |
|
|
|
193 |
wakaba |
1.1 |
} else { |
194 |
wakaba |
1.2 |
var el = new JSElement (this.doc, token.value); |
195 |
wakaba |
1.1 |
this.openElements[this.openElements.length - 1].appendChild (el); |
196 |
|
|
this.openElements.push (el); |
197 |
|
|
} |
198 |
|
|
} else if (token.type == 'end-tag') { |
199 |
|
|
if (this.openElements[this.openElements.length - 1].localName == |
200 |
|
|
token.value) { |
201 |
|
|
this.openElements.pop (); |
202 |
|
|
} else { |
203 |
|
|
log ('parse error: unmatched end tag: ' + token.value); |
204 |
|
|
} |
205 |
wakaba |
1.3 |
} else if (token.type == 'char') { |
206 |
|
|
this.openElements[this.openElements.length - 1].manakaiAppendText |
207 |
|
|
(token.value); |
208 |
wakaba |
1.1 |
} else if (token.type == 'eof') { |
209 |
|
|
break; |
210 |
wakaba |
1.3 |
} else if (token.type == 'abort') { |
211 |
|
|
log ('parse: abort'); |
212 |
|
|
return; |
213 |
wakaba |
1.1 |
} |
214 |
|
|
} |
215 |
|
|
|
216 |
|
|
log ('stop parsing'); |
217 |
|
|
} // parse |
218 |
|
|
|
219 |
wakaba |
1.3 |
Parser.prototype.setInsertionPoint = function (ip) { |
220 |
|
|
if (ip == undefined || ip == null || isNaN (ip)) { |
221 |
|
|
log ('insertion point: set to undefined'); |
222 |
|
|
this.insertionPoint = undefined; |
223 |
|
|
} else { |
224 |
|
|
log ('insertion point: set to ' + ip + |
225 |
|
|
' (before "' + this.in.s.substring (0, 10) + '")'); |
226 |
|
|
this.insertionPoint = ip; |
227 |
|
|
} |
228 |
|
|
}; // setInsertionPoint |
229 |
|
|
|
230 |
wakaba |
1.2 |
function JSDocument (p) { |
231 |
wakaba |
1.1 |
this.childNodes = []; |
232 |
wakaba |
1.2 |
this._parser = p; |
233 |
wakaba |
1.1 |
} // JSDocument |
234 |
|
|
|
235 |
wakaba |
1.2 |
function JSElement (doc, localName) { |
236 |
wakaba |
1.1 |
this.localName = localName; |
237 |
wakaba |
1.2 |
this.ownerDocument = doc; |
238 |
wakaba |
1.1 |
this.childNodes = []; |
239 |
|
|
} // JSElement |
240 |
|
|
|
241 |
|
|
JSDocument.prototype.appendChild = JSElement.prototype.appendChild = |
242 |
|
|
function (e) { |
243 |
|
|
this.childNodes.push (e); |
244 |
|
|
e.parentNode = this; |
245 |
wakaba |
1.2 |
|
246 |
|
|
if (e.localName == 'script') { |
247 |
|
|
log ('start running a script'); |
248 |
|
|
|
249 |
wakaba |
1.3 |
var doc = this.ownerDocument || this; |
250 |
wakaba |
1.2 |
var p = doc._parser; |
251 |
|
|
|
252 |
|
|
// 1. Script type |
253 |
|
|
// |
254 |
|
|
|
255 |
|
|
// 2.1. If scripting is disabled |
256 |
|
|
// |
257 |
|
|
// 2.2. If the script element was created by an XML ... innerHTML ... |
258 |
|
|
// |
259 |
|
|
// 2.3. If the user agent does not support the scripting language ... |
260 |
|
|
// |
261 |
|
|
// 2.4. If the script element has its "already executed" flag set |
262 |
|
|
if (e.manakaiAlreadyExecuted) { |
263 |
|
|
// 2.5. Abort these steps at this point. |
264 |
|
|
log ('running a script: aborted'); |
265 |
|
|
return e; |
266 |
|
|
} |
267 |
|
|
|
268 |
|
|
// 3. Set the element's "already executed" flag. |
269 |
|
|
e.manakaiAlreadyExecuted = true; |
270 |
|
|
|
271 |
|
|
// 4. If the element has a src attribute, then a load for ... |
272 |
|
|
// TODO: load an external resource |
273 |
|
|
|
274 |
|
|
// 5. The first of the following options: |
275 |
|
|
|
276 |
|
|
// 5.1. |
277 |
|
|
if (/* TODO: If the document is still being parsed && */ |
278 |
|
|
e.defer && !e.async) { |
279 |
|
|
// TODO |
280 |
|
|
} else if (e.async && e.src != null) { |
281 |
|
|
// TODO |
282 |
|
|
} else if (e.async && e.src == null |
283 |
|
|
/* && list of scripts that will execute asynchronously is not empty */) { |
284 |
|
|
// TODO |
285 |
|
|
} else if (e.src != null && e.manakaiParserInserted) { |
286 |
|
|
// TODO |
287 |
|
|
} else if (e.src != null) { |
288 |
|
|
// TODO |
289 |
|
|
} else { |
290 |
|
|
executeScript (doc, e); // even if other scripts are already executing. |
291 |
|
|
} |
292 |
|
|
|
293 |
|
|
log ('end running a script'); |
294 |
|
|
} |
295 |
|
|
|
296 |
wakaba |
1.1 |
return e; |
297 |
|
|
}; // appendChild |
298 |
|
|
|
299 |
wakaba |
1.2 |
function executeScript (doc, e) { |
300 |
|
|
log ('executing a script block: start'); |
301 |
|
|
|
302 |
|
|
// If the load resulted in an error, then ... firing an error event ... |
303 |
|
|
|
304 |
|
|
// If the load was successful |
305 |
|
|
log ('load event fired at the script element'); |
306 |
|
|
|
307 |
|
|
if (true) { |
308 |
|
|
// Scripting is enabled, Document.designMode is disabled, |
309 |
|
|
// Document is the active document in its browsing context |
310 |
|
|
|
311 |
|
|
var s; |
312 |
|
|
if (e.src != null) { |
313 |
|
|
// TODO: from external file |
314 |
|
|
} else { |
315 |
|
|
s = e.text; |
316 |
|
|
} |
317 |
|
|
|
318 |
|
|
parseAndRunScript (doc, s); |
319 |
|
|
} |
320 |
|
|
|
321 |
|
|
log ('executing a script block: end'); |
322 |
|
|
} // executeScript |
323 |
|
|
|
324 |
|
|
function parseAndRunScript (doc, s) { |
325 |
|
|
while (true) { |
326 |
|
|
var matched = false; |
327 |
|
|
s = s.replace (/^\s*document\.write\s*\(((?:'[^']*'|"[^"]*")\s*(?:,\s*(?:'[^']*'|"[^"]*"))*)\)\s*;\s*/, function (s, t) { |
328 |
|
|
matched = true; |
329 |
|
|
var args = []; |
330 |
|
|
t.replace (/('[^']*'|"[^"]*")/g, function (s, v) { |
331 |
|
|
args.push (v.substring (1, v.length - 1)); |
332 |
|
|
return ''; |
333 |
|
|
}); |
334 |
|
|
doc.write.apply (doc, args); |
335 |
|
|
return ''; |
336 |
|
|
}); |
337 |
|
|
if (s == '') break; |
338 |
|
|
if (!matched) { |
339 |
|
|
log ('Script parse error: "' + s + '"'); |
340 |
|
|
break; |
341 |
|
|
} |
342 |
|
|
} |
343 |
|
|
} // parseAndRunScript |
344 |
|
|
|
345 |
wakaba |
1.1 |
function JSText (data) { |
346 |
|
|
this.data = data; |
347 |
|
|
} // JSText |
348 |
|
|
|
349 |
|
|
JSDocument.prototype.manakaiAppendText = |
350 |
|
|
JSElement.prototype.manakaiAppendText = |
351 |
|
|
function (s) { |
352 |
|
|
if (this.childNodes.length > 0 && |
353 |
|
|
this.childNodes[this.childNodes.length - 1] instanceof JSText) { |
354 |
|
|
this.childNodes[this.childNodes.length - 1].data += s; |
355 |
|
|
} else { |
356 |
|
|
this.childNodes.push (new JSText (s)); |
357 |
|
|
} |
358 |
|
|
}; // manakaiAppendText |
359 |
wakaba |
1.2 |
|
360 |
|
|
JSDocument.prototype.write = function () { |
361 |
wakaba |
1.3 |
var p = this._parser; |
362 |
|
|
|
363 |
wakaba |
1.2 |
// 1. If the insertion point is undefined, the open() method must be ... |
364 |
wakaba |
1.3 |
if (p.insertionPoint == NaN || p.insertionPoint == undefined) { |
365 |
|
|
// TODO: open () |
366 |
|
|
} |
367 |
wakaba |
1.2 |
|
368 |
|
|
// 2. ... inserted into the input stream just before the insertion point. |
369 |
wakaba |
1.3 |
var s = Array.join (arguments, ''); |
370 |
|
|
log ('document.write: insert "' + s + '"' + |
371 |
|
|
' before "' + p.in.s.substring (p.insertionPoint, p.insertionPoint + 10) + '"'); |
372 |
|
|
p.in.s = p.in.s.substring (0, p.insertionPoint) + s |
373 |
|
|
+ p.in.s.substring (p.insertionPoint, p.in.s.length); |
374 |
|
|
p.insertionPoint += s.length; |
375 |
wakaba |
1.2 |
|
376 |
|
|
// 3. If there is a script that will execute as soon as the parser resumes |
377 |
|
|
// TODO |
378 |
|
|
|
379 |
|
|
// 4. Process the characters that were inserted, ... |
380 |
wakaba |
1.3 |
p.parse (); |
381 |
wakaba |
1.2 |
|
382 |
|
|
// 5. Return |
383 |
|
|
log ('document.write: return'); |
384 |
|
|
return; |
385 |
|
|
}; // document.write |
386 |
|
|
|
387 |
|
|
JSElement.prototype.__defineGetter__ ('text', function () { |
388 |
|
|
var r = ''; |
389 |
|
|
for (var i = 0; i < this.childNodes.length; i++) { |
390 |
|
|
if (this.childNodes[i] instanceof JSText) { |
391 |
|
|
r += this.childNodes[i].data; |
392 |
|
|
} |
393 |
|
|
} |
394 |
|
|
return r; |
395 |
|
|
}); |
396 |
wakaba |
1.1 |
|
397 |
|
|
function dumpTree (n, indent) { |
398 |
|
|
var r = ''; |
399 |
|
|
for (var i = 0; i < n.childNodes.length; i++) { |
400 |
|
|
var node = n.childNodes[i]; |
401 |
|
|
if (node instanceof JSElement) { |
402 |
|
|
r += '| ' + indent + node.localName + '\n'; |
403 |
|
|
r += dumpTree (node, indent + ' '); |
404 |
|
|
} else if (node instanceof JSText) { |
405 |
|
|
r += '| ' + indent + '"' + node.data + '"\n'; |
406 |
|
|
} else { |
407 |
|
|
r += '| ' + indent + node + '\n'; |
408 |
|
|
} |
409 |
|
|
} |
410 |
|
|
return r; |
411 |
|
|
} // dumpTree |
412 |
|
|
</script> |
413 |
|
|
</head> |
414 |
|
|
<body onload=" |
415 |
|
|
document.sourceElement = document.getElementsByTagName ('textarea')[0]; |
416 |
|
|
document.logElement = document.getElementsByTagName ('output')[0]; |
417 |
|
|
update (); |
418 |
|
|
"> |
419 |
|
|
|
420 |
|
|
<textarea onchange=" update () "><html> |
421 |
|
|
<head></head><body> |
422 |
|
|
<p> |
423 |
|
|
<script> |
424 |
wakaba |
1.3 |
document.write ('aaaaaaa</p><script>document.write("cccccc");</', 'script>bbbbbb'); |
425 |
wakaba |
1.1 |
</script> |
426 |
|
|
<p> |
427 |
|
|
</textarea> |
428 |
|
|
|
429 |
|
|
<output></output> |
430 |
|
|
|
431 |
|
|
</body> |
432 |
|
|
</html> |