html/scripting-parser/parser.html

<!DOCTYPE HTML>
<html lang=en>
<head>
<title>Demo of HTML5 Parsing Algorithm with Scripting Enabled</title>
<style>
  textarea {
     display: block;
     width: 80%;
     margin-left: auto;
     margin-right: auto;
     min-height: 20em;
  }
  output {
    display: block;
    font-family: monospace;
    white-space: pre;
  }
</style>
<script>
  function update () {
    document.logElement.textContent = '';
    var p = new Parser (new InputStream (document.sourceElement.value));
    p.parse ();
    log (dumpTree (p.doc, ''));
  } // update

  function log (s) {
    document.logElement.appendChild (document.createTextNode (s + "\n"));
  } // log

  function InputStream (s) {
    this.s = s;
  } // InputStream

  function Parser (i) {
    this.parseMode = 'pcdata';
    this.doc = new JSDocument (this);
    this.openElements = [this.doc];
    this.in = i;
  } // Parser

  Parser.prototype.getNextToken = function () {
    var i = this.in;
    if (this.parseMode == 'script') {
      var token;
      i.s = i.s.replace (/^([\s\S]+?)<\/[Ss][Cc][Rr][Ii][Pp][Tt]>/,
      function (s, t) {
        token = {type: 'char', value: t};
        return '<' + '/script>';
      });
      if (token) return token;
      i.s = i.s.replace (/^<\/[Ss][Cc][Rr][Ii][Pp][Tt]>/, function () {
        token = {type: 'end-tag', value: 'script'};
        return '';
      });
      if (token) return token;
      return {type: 'eof'};
    }

    var token;
    i.s = i.s.replace (/^<\/([^>]+)>/, function (s, e) {
      token = {type: 'end-tag', value: e.toLowerCase ()};
      return '';
    });
    if (token) return token;
    i.s = i.s.replace (/^<([^>]+)>/, function (s, e) {
      token = {type: 'start-tag', value: e.toLowerCase ()};
      return '';
    });
    if (token) return token;
    i.s = i.s.replace (/^[^<]+/, function (s) {
      token = {type: 'char', value: s};
      return '';
    });
    if (token) return token;
    i.s = i.s.replace (/^[\s\S]/, function (s) {
      token = {type: 'char', value: s};
      return '';
    });
    if (token) return token;
    return {type: 'eof'};
  } // getNextToken

  Parser.prototype.parse = function () {
    log ('start parsing');

    while (true) {
      var token = this.getNextToken ();
      log ('token: ' + token.type + ' "' + token.value + '"');

      if (token.type == 'start-tag') {
        if (token.value == 'script') {
          // 1. Create an element for the token in the HTML namespace.
          var el = new JSElement (this.doc, token.value);

          // 2. Mark the element as being "parser-inserted".
          el.manakaiParserInserted = true;

          // 3. Switch the tokeniser's content model flag to the CDATA state.
          this.parseMode = 'script';

          // 4.1. Collect all the character tokens.
          while (true) {
            var token = this.getNextToken ();
            log ('token: ' + token.type + ' "' + token.value + '"');

            if (token.type == 'char') {
              // 5. Append a single Text node to the script element node.
              el.manakaiAppendText (token.value);

            // 4.2. Until it returns a token that is not a character token, or
            // TODO: 4.3. Until it stops tokenising.
            } else if (token.type == 'eof' ||
                       (token.type == 'end-tag' && token.value == 'script')) {
              // 6. Switched back to the PCDATA state.
              this.parseMode = 'pcdata';

              // 7.1. If the next token is not an end tag token with ...
              if (token.type != 'end-tag') {
                // 7.2. This is a parse error.
                log ('Parse error: no </' + 'script>');

                // 7.3. Mark the script element as "already executed".
                el.manakaiAlreadyExecuted = true;
              } else {
                // 7.4. Ignore it.
                //
              }
              break;
            }
          }

          // 8.1. If the parser were originally created for the ...
          if (this.fragmentParsingMode) {
            // 8.2. Mark the script element as "already executed" and ...
            el.alreadyExecuted = true;
            continue;
          }

          // 9.1. Let the old insertion point have the same value as the ...

          // 9.2. Let the insertion point be just before the next input ...

          // 10. Append the new element to the current node.
          this.openElements[this.openElements.length - 1].appendChild (el);

          // 11. Let the insertion point have the value of the old ...

          // 12. If there is a script that will execute as soon as ...
          

        } else {
          var el = new JSElement (this.doc, token.value);
          this.openElements[this.openElements.length - 1].appendChild (el);
          this.openElements.push (el);
        }
      } else if (token.type == 'end-tag') {
        if (this.openElements[this.openElements.length - 1].localName ==
            token.value) {
          this.openElements.pop ();
        } else {
          log ('parse error: unmatched end tag: ' + token.value);
        }
      } else if (token.type == 'eof') {
        break;
      }
    }

    log ('stop parsing');
  } // parse

  function JSDocument (p) {
    this.childNodes = [];
    this._parser = p;
  } // JSDocument

  function JSElement (doc, localName) {
    this.localName = localName;
    this.ownerDocument = doc;
    this.childNodes = [];
  } // JSElement

  JSDocument.prototype.appendChild = JSElement.prototype.appendChild =
  function (e) {
    this.childNodes.push (e);
    e.parentNode = this;

    if (e.localName == 'script') {
      log ('start running a script');

      var doc = this.ownerDocument;
      var p = doc._parser;

      // 1. Script type
      //

      // 2.1. If scripting is disabled
      //
      // 2.2. If the script element was created by an XML ... innerHTML ...
      //
      // 2.3. If the user agent does not support the scripting language ...
      //
      // 2.4. If the script element has its "already executed" flag set
      if (e.manakaiAlreadyExecuted) {
        // 2.5. Abort these steps at this point.
        log ('running a script: aborted');
        return e;
      }

      // 3. Set the element's "already executed" flag.
      e.manakaiAlreadyExecuted = true;

      // 4. If the element has a src attribute, then a load for ...
      // TODO: load an external resource

      // 5. The first of the following options:

      // 5.1. 
      if (/* TODO: If the document is still being parsed && */
          e.defer && !e.async) {
        // TODO
      } else if (e.async && e.src != null) {
        // TODO
      } else if (e.async && e.src == null
                 /* && list of scripts that will execute asynchronously is not empty */) {
        // TODO
      } else if (e.src != null && e.manakaiParserInserted) {
        // TODO
      } else if (e.src != null) {
        // TODO
      } else {
        executeScript (doc, e); // even if other scripts are already executing.
      }

      log ('end running a script');
    }

    return e;
  }; // appendChild

  function executeScript (doc, e) {
    log ('executing a script block: start');

    // If the load resulted in an error, then ... firing an error event ...

    // If the load was successful
    log ('load event fired at the script element');

    if (true) {
    // Scripting is enabled, Document.designMode is disabled,
    // Document is the active document in its browsing context

      var s;
      if (e.src != null) {
        // TODO: from external file
      } else {
        s = e.text;
      }

      parseAndRunScript (doc, s);
    }

    log ('executing a script block: end');
  } // executeScript

  function parseAndRunScript (doc, s) {
    while (true) {
      var matched = false;
      s = s.replace (/^\s*document\.write\s*\(((?:'[^']*'|"[^"]*")\s*(?:,\s*(?:'[^']*'|"[^"]*"))*)\)\s*;\s*/, function (s, t) {
        matched = true;
        var args = [];
        t.replace (/('[^']*'|"[^"]*")/g, function (s, v) {
          args.push (v.substring (1, v.length - 1));
          return '';
        });
        doc.write.apply (doc, args);
        return '';
      });
      if (s == '') break;
      if (!matched) {
        log ('Script parse error: "' + s + '"');
        break;
      }
    }
  } // parseAndRunScript

  function JSText (data) {
    this.data = data;
  } // JSText

  JSDocument.prototype.manakaiAppendText =
  JSElement.prototype.manakaiAppendText =
  function (s) {
    if (this.childNodes.length > 0 &&
        this.childNodes[this.childNodes.length - 1] instanceof JSText) {
      this.childNodes[this.childNodes.length - 1].data += s;
    } else {
      this.childNodes.push (new JSText (s));
    }
  }; // manakaiAppendText

  JSDocument.prototype.write = function () {
    // 1. If the insertion point is undefined, the open() method must be ...
    // 

    // 2. ... inserted into the input stream just before the insertion point.
    log ('document.write: insert "' + Array.join (arguments, '') + '"');

    // 3. If there is a script that will execute as soon as the parser resumes
    // TODO

    // 4. Process the characters that were inserted, ...

    // 5. Return
    log ('document.write: return');
    return;
  }; // document.write

  JSElement.prototype.__defineGetter__ ('text', function () {
    var r = '';
    for (var i = 0; i < this.childNodes.length; i++) {
      if (this.childNodes[i] instanceof JSText) {
        r += this.childNodes[i].data;
      }
    }
    return r;
  });

  function dumpTree (n, indent) {
    var r = '';
    for (var i = 0; i < n.childNodes.length; i++) {
      var node = n.childNodes[i];
      if (node instanceof JSElement) {
        r += '| ' + indent + node.localName + '\n';
        r += dumpTree (node, indent + '  ');
      } else if (node instanceof JSText) {
        r += '| ' + indent + '"' + node.data + '"\n';
      } else {
        r += '| ' + indent + node + '\n';
      }
    }
    return r;
  } // dumpTree
</script>
</head>
<body onload="
  document.sourceElement = document.getElementsByTagName ('textarea')[0];
  document.logElement = document.getElementsByTagName ('output')[0];
  update ();
">

<textarea onchange=" update () ">&lt;html>
&lt;head>&lt;/head>&lt;body>
&lt;p>
&lt;script>
document.write ('aaaaaaa&lt;/p>\n&lt;script>\ndocument.write("cccccc")\n&lt;/', 'script>\nbbbbbb');
&lt;/script>
&lt;p>
</textarea>

<output></output>

</body>
</html>
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			var i = this.in;
44	wakaba	1.1	if (this.parseMode == 'script') {
45			var token;
46			i.s = i.s.replace (/^([\s\S]+?)<\/[Ss][Cc][Rr][Ii][Pp][Tt]>/,
47			function (s, t) {
48			token = {type: 'char', value: t};
49			return '<' + '/script>';
50			});
51			if (token) return token;
52			i.s = i.s.replace (/^<\/[Ss][Cc][Rr][Ii][Pp][Tt]>/, function () {
53			token = {type: 'end-tag', value: 'script'};
54			return '';
55			});
56			if (token) return token;
57			return {type: 'eof'};
58			}
59
60			var token;
61			i.s = i.s.replace (/^<\/([^>]+)>/, function (s, e) {
62			token = {type: 'end-tag', value: e.toLowerCase ()};
63			return '';
64			});
65			if (token) return token;
66			i.s = i.s.replace (/^<([^>]+)>/, function (s, e) {
67			token = {type: 'start-tag', value: e.toLowerCase ()};
68			return '';
69			});
70			if (token) return token;
71			i.s = i.s.replace (/^[^<]+/, function (s) {
72			token = {type: 'char', value: s};
73			return '';
74			});
75			if (token) return token;
76			i.s = i.s.replace (/^[\s\S]/, function (s) {
77			token = {type: 'char', value: s};
78			return '';
79			});
80			if (token) return token;
81			return {type: 'eof'};
82			} // getNextToken
83
84	wakaba	1.2	Parser.prototype.parse = function () {
85	wakaba	1.1	log ('start parsing');
86
87			while (true) {
88	wakaba	1.2	var token = this.getNextToken ();
89	wakaba	1.1	log ('token: ' + token.type + ' "' + token.value + '"');
90
91			if (token.type == 'start-tag') {
92			if (token.value == 'script') {
93	wakaba	1.2	// 1. Create an element for the token in the HTML namespace.
94			var el = new JSElement (this.doc, token.value);
95
96			// 2. Mark the element as being "parser-inserted".
97			el.manakaiParserInserted = true;
98
99			// 3. Switch the tokeniser's content model flag to the CDATA state.
100	wakaba	1.1	this.parseMode = 'script';
101
102	wakaba	1.2	// 4.1. Collect all the character tokens.
103	wakaba	1.1	while (true) {
104	wakaba	1.2	var token = this.getNextToken ();
105	wakaba	1.1	log ('token: ' + token.type + ' "' + token.value + '"');
106
107			if (token.type == 'char') {
108	wakaba	1.2	// 5. Append a single Text node to the script element node.
109	wakaba	1.1	el.manakaiAppendText (token.value);
110	wakaba	1.2
111			// 4.2. Until it returns a token that is not a character token, or
112			// TODO: 4.3. Until it stops tokenising.
113	wakaba	1.1	} else if (token.type == 'eof' \|\|
114			(token.type == 'end-tag' && token.value == 'script')) {
115	wakaba	1.2	// 6. Switched back to the PCDATA state.
116	wakaba	1.1	this.parseMode = 'pcdata';
117	wakaba	1.2
118			// 7.1. If the next token is not an end tag token with ...
119			if (token.type != 'end-tag') {
120			// 7.2. This is a parse error.
121			log ('Parse error: no </' + 'script>');
122
123			// 7.3. Mark the script element as "already executed".
124			el.manakaiAlreadyExecuted = true;
125			} else {
126			// 7.4. Ignore it.
127			//
128			}
129	wakaba	1.1	break;
130			}
131			}
132
133	wakaba	1.2	// 8.1. If the parser were originally created for the ...
134			if (this.fragmentParsingMode) {
135			// 8.2. Mark the script element as "already executed" and ...
136			el.alreadyExecuted = true;
137			continue;
138			}
139
140			// 9.1. Let the old insertion point have the same value as the ...
141
142			// 9.2. Let the insertion point be just before the next input ...
143
144			// 10. Append the new element to the current node.
145	wakaba	1.1	this.openElements[this.openElements.length - 1].appendChild (el);
146	wakaba	1.2
147			// 11. Let the insertion point have the value of the old ...
148
149			// 12. If there is a script that will execute as soon as ...
150
151
152	wakaba	1.1	} else {
153	wakaba	1.2	var el = new JSElement (this.doc, token.value);
154	wakaba	1.1	this.openElements[this.openElements.length - 1].appendChild (el);
155			this.openElements.push (el);
156			}
157			} else if (token.type == 'end-tag') {
158			if (this.openElements[this.openElements.length - 1].localName ==
159			token.value) {
160			this.openElements.pop ();
161			} else {
162			log ('parse error: unmatched end tag: ' + token.value);
163			}
164			} else if (token.type == 'eof') {
165			break;
166			}
167			}
168
169			log ('stop parsing');
170			} // parse
171
172	wakaba	1.2	function JSDocument (p) {
173	wakaba	1.1	this.childNodes = [];
174	wakaba	1.2	this._parser = p;
175	wakaba	1.1	} // JSDocument
176
177	wakaba	1.2	function JSElement (doc, localName) {
178	wakaba	1.1	this.localName = localName;
179	wakaba	1.2	this.ownerDocument = doc;
180	wakaba	1.1	this.childNodes = [];
181			} // JSElement
182
183			JSDocument.prototype.appendChild = JSElement.prototype.appendChild =
184			function (e) {
185			this.childNodes.push (e);
186			e.parentNode = this;
187	wakaba	1.2
188			if (e.localName == 'script') {
189			log ('start running a script');
190
191			var doc = this.ownerDocument;
192			var p = doc._parser;
193
194			// 1. Script type
195			//
196
197			// 2.1. If scripting is disabled
198			//
199			// 2.2. If the script element was created by an XML ... innerHTML ...
200			//
201			// 2.3. If the user agent does not support the scripting language ...
202			//
203			// 2.4. If the script element has its "already executed" flag set
204			if (e.manakaiAlreadyExecuted) {
205			// 2.5. Abort these steps at this point.
206			log ('running a script: aborted');
207			return e;
208			}
209
210			// 3. Set the element's "already executed" flag.
211			e.manakaiAlreadyExecuted = true;
212
213			// 4. If the element has a src attribute, then a load for ...
214			// TODO: load an external resource
215
216			// 5. The first of the following options:
217
218			// 5.1.
219			if (/* TODO: If the document is still being parsed && */
220			e.defer && !e.async) {
221			// TODO
222			} else if (e.async && e.src != null) {
223			// TODO
224			} else if (e.async && e.src == null
225			/* && list of scripts that will execute asynchronously is not empty */) {
226			// TODO
227			} else if (e.src != null && e.manakaiParserInserted) {
228			// TODO
229			} else if (e.src != null) {
230			// TODO
231			} else {
232			executeScript (doc, e); // even if other scripts are already executing.
233			}
234
235			log ('end running a script');
236			}
237
238	wakaba	1.1	return e;
239			}; // appendChild
240
241	wakaba	1.2	function executeScript (doc, e) {
242			log ('executing a script block: start');
243
244			// If the load resulted in an error, then ... firing an error event ...
245
246			// If the load was successful
247			log ('load event fired at the script element');
248
249			if (true) {
250			// Scripting is enabled, Document.designMode is disabled,
251			// Document is the active document in its browsing context
252
253			var s;
254			if (e.src != null) {
255			// TODO: from external file
256			} else {
257			s = e.text;
258			}
259
260			parseAndRunScript (doc, s);
261			}
262
263			log ('executing a script block: end');
264			} // executeScript
265
266			function parseAndRunScript (doc, s) {
267			while (true) {
268			var matched = false;
269			s = s.replace (/^\sdocument\.write\s\(((?:'[^']'\|"[^"]")\s(?:,\s(?:'[^']'\|"[^"]")))\)\s;\s*/, function (s, t) {
270			matched = true;
271			var args = [];
272			t.replace (/('[^']'\|"[^"]")/g, function (s, v) {
273			args.push (v.substring (1, v.length - 1));
274			return '';
275			});
276			doc.write.apply (doc, args);
277			return '';
278			});
279			if (s == '') break;
280			if (!matched) {
281			log ('Script parse error: "' + s + '"');
282			break;
283			}
284			}
285			} // parseAndRunScript
286
287	wakaba	1.1	function JSText (data) {
288			this.data = data;
289			} // JSText
290
291			JSDocument.prototype.manakaiAppendText =
292			JSElement.prototype.manakaiAppendText =
293			function (s) {
294			if (this.childNodes.length > 0 &&
295			this.childNodes[this.childNodes.length - 1] instanceof JSText) {
296			this.childNodes[this.childNodes.length - 1].data += s;
297			} else {
298			this.childNodes.push (new JSText (s));
299			}
300			}; // manakaiAppendText
301	wakaba	1.2
302			JSDocument.prototype.write = function () {
303			// 1. If the insertion point is undefined, the open() method must be ...
304			//
305
306			// 2. ... inserted into the input stream just before the insertion point.
307			log ('document.write: insert "' + Array.join (arguments, '') + '"');
308
309			// 3. If there is a script that will execute as soon as the parser resumes
310			// TODO
311
312			// 4. Process the characters that were inserted, ...
313
314			// 5. Return
315			log ('document.write: return');
316			return;
317			}; // document.write
318
319			JSElement.prototype.__defineGetter__ ('text', function () {
320			var r = '';
321			for (var i = 0; i < this.childNodes.length; i++) {
322			if (this.childNodes[i] instanceof JSText) {
323			r += this.childNodes[i].data;
324			}
325			}
326			return r;
327			});
328	wakaba	1.1
329			function dumpTree (n, indent) {
330			var r = '';
331			for (var i = 0; i < n.childNodes.length; i++) {
332			var node = n.childNodes[i];
333			if (node instanceof JSElement) {
334			r += '\| ' + indent + node.localName + '\n';
335			r += dumpTree (node, indent + ' ');
336			} else if (node instanceof JSText) {
337			r += '\| ' + indent + '"' + node.data + '"\n';
338			} else {
339			r += '\| ' + indent + node + '\n';
340			}
341			}
342			return r;
343			} // dumpTree
344			</script>
345			</head>
346			<body onload="
347			document.sourceElement = document.getElementsByTagName ('textarea')[0];
348			document.logElement = document.getElementsByTagName ('output')[0];
349			update ();
350			">
351
352			<textarea onchange=" update () "><html>
353			<head></head><body>
354			<p>
355			<script>
356			document.write ('aaaaaaa</p>\n<script>\ndocument.write("cccccc")\n</', 'script>\nbbbbbb');
357			</script>
358			<p>
359			</textarea>
360
361			<output></output>
362
363			</body>
364			</html>