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: -moz-pre-wrap;
    white-space: pre-wrap;
  }
</style>
<script>
  function update () {
    document.logElement.textContent = '';
    var p = new Parser (new InputStream (document.sourceElement.value));
    var doc = p.doc;
    p.parse ();
    log (dumpTree (doc, ''));
  } // update

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

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

  function Parser (i, doc) {
    this.parseMode = 'pcdata';
    if (!doc) {
      doc = new JSDocument (this);
      doc.manakaiIsHTML = true;
    }
    this.doc = doc;
    this.openElements = [doc];
    this.in = i;
    this.scriptsExecutedAfterParsing = [];
  } // Parser

  Parser.prototype.getNextToken = function () {
    var p = this;
    var i = this.in;
    if (this.parseMode == 'script') {
      var token;
      if (p.insertionPoint <= 0) {
        return {type: 'abort'};
      }
      i.s = i.s.replace (/^([^<]+)/,
      function (s, t) {
        if (0 < p.insertionPoint && p.insertionPoint < t.length) {
          token = {type: 'char', value: t.substring (0, p.insertionPoint)};
          var ip = p.insertionPoint;
          p.insertionPoint = 0;
          return t.substring (ip, t.length);
        }
        token = {type: 'char', value: t};
        p.insertionPoint -= t.length;
        return '';
      });
      if (token) return token;
      i.s = i.s.replace (/^<\/[Ss][Cc][Rr][Ii][Pp][Tt]>/, function (s) {
        if (p.insertionPoint < s.length) {
          token = {type: 'abort'};
          return s;
        }
        token = {type: 'end-tag', value: 'script'};
        p.insertionPoint -= s.length;
        return '';
      });
      if (token) return token;
      var m;
      if ((p.insertionPoint < '</script'.length) &&
          (m = i.s.match (/^<\/([SCRIPTscript]+)/))) {
        var v = m[1].substring (0, p.insertionPoint).toLowerCase ();
        if (v == 'script'.substring (0, p.insertionPoint - '</'.length)) {
          return {type: 'abort'};
        }
      }
      i.s = i.s.replace (/^</,
      function (s) {
        token = {type: 'char', value: s};
        p.insertionPoint -= s.length;
        return '';
      });
      if (token) return token;
      return {type: 'eof'};
    }

    var token;
    i.s = i.s.replace (/^<\/([^>]+)(?:>|$)/, function (s, e) {
      if (p.insertionPoint < s.length ||
          (p.insertionPoint <= s.length &&
           s.substring (s.length - 1, 1) != '>')) {
        token = {type: 'abort'};
        return s;
      }
      token = {type: 'end-tag', value: e.toLowerCase ()};
      p.insertionPoint -= s.length;
      return '';
    });
    if (token) return token;
    i.s = i.s.replace (/^<([^>]+)(?:>|$)/, function (s, e) {
      if (p.insertionPoint < s.length ||
          (p.insertionPoint <= s.length &&
           s.substring (s.length - 1, 1) != '>')) {
        token = {type: 'abort'};
        return s;
      }
      var tagName;
      var attrs = {};
      e = e.replace (/^[\S]+/, function (v) {
        tagName = v.toLowerCase ();
        return '';
      });
      e = e.replace (/^\s*(\S+)\s*(?:=\s*"([^"]*)"|'([^']*)'|([^"']+))?/,
      function (x, attrName, attrValue1, attrValue2, attrValue3) {
        attrs[attrName] = attrValue1 || attrValue2 || attrValue3;
        return '';
      });
      token = {type: 'start-tag', value: tagName, attrs: attrs};
      p.insertionPoint -= s.length;
      return '';
    });
    if (token) return token;
    if (p.insertionPoint <= 0) {
      return {type: 'abort'};
    }
    i.s = i.s.replace (/^[^<]+/, function (s) {
      if (p.insertionPoint < s.length) {
        token = {type: 'char', value: s.substring (0, p.insertionPoint)};
        var ip = p.insertionPoint;
        p.insertionPoint = 0;
        return s.substring (ip, s.length);
      }
      token = {type: 'char', value: s};
      p.insertionPoint -= s.length;
      return '';
    });
    if (token) return token;
    i.s = i.s.replace (/^[\s\S]/, function (s) {
      token = {type: 'char', value: s};
      p.insertionPoint -= s.length;
      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);
          if (token.attrs.async != null) el.async = true;
          if (token.attrs.defer != null) el.defer = true;
          if (token.attrs.src != null) el.src = token.attrs.src;

          // 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
            // until it stops tokenising.
            } else if (token.type == 'eof' ||
                       (token.type == 'end-tag' && token.value == 'script') ||
                       token.type == 'abort') {
              // 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 ...
          var oldInsertionPoint = this.insertionPoint;
          // 9.2. Let the insertion point be just before the next input ...
          this.setInsertionPoint (0);

          // 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 ...
          oldInsertionPoint += this.insertionPoint;
          this.setInsertionPoint (oldInsertionPoint);

          // 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 == 'char') {
        this.openElements[this.openElements.length - 1].manakaiAppendText
            (token.value);
      } else if (token.type == 'eof') {
        break;
      } else if (token.type == 'abort') {
        log ('parse: abort');
        return;
      }
    }

    log ('stop parsing');

    // readyState = 'interactive'

    // "When a script completes loading" rules start applying.

    // TODO: Handles "list of scripts that will execute as soon as possible"
    // and "list of scripts that will execute asynchronously"

    // Handle "list of scripts that will execute when the document has finished
    // parsing".
    var list = this.scriptsExecutedAfterParsing;
    while (list.length > 0) {
      // TODO: break unless completed loading

      // Step 1.
      //

      // Step 2. and Step 3.
      log ('Executing a |defer|red script...');
      executeScript (this.doc, list.shift ());

      // Step 4.
    }

    log ('DOMContentLoaded event fired');

    // "delays tha load event" things has completed:
    // readyState = 'complete'
    log ('load event fired');
  } // parse

  Parser.prototype.setInsertionPoint = function (ip) {
    if (ip == undefined || ip == null || isNaN (ip)) {
      log ('insertion point: set to undefined');
      this.insertionPoint = undefined;
    } else if (ip == this.in.s.length) {
      log ('insertion point: end of file');
      this.insertionPoint = ip;
    } else {
      log ('insertion point: set to ' + ip +
           ' (before "' + this.in.s.substring (0, 10) + '")');
      this.insertionPoint = ip;
    }
  }; // setInsertionPoint

  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 ('Running a script: start');

      var doc = this.ownerDocument || this;
      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) {
        p.scriptsExecutedAfterParsing.push (e);
        log ('Running a script: aborted (defer)');
      } 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 ('Running a script: end');
    }

    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.open = function () {
    // Two or fewer arguments

    // Step 1.
    var type = arguments[0] || 'text/html';
    
    // Step 2.
    var replace = arguments[1] == 'replace';

    // Step 3.
    if (this._parser &&
        !this._parser.scriptCreated &&
        this._parser.in.insertionPoint != undefined) {
      log ('document.open () in parsing mode is ignored');
      return this;
    }

    // Step 4.
    log ('onbeforeunload event fired');
    log ('onunload event fired');

    // Step 5.
    if (this._parser) {
      // Discard the parser.
    }

    // Step 6.
    log ('document cleared by document.open ()');
    this.childNodes = [];

    // Step 7.
    this._parser = new Parser (new InputStream (''), this);
    this._parser.scriptCreated = true;

    // Step 8.
    this.manakaiIsHTML = true;

    // Step 9.
    // If not text/html, ...

    // Step 10.
    if (!replace) {
      // History      
    }

    // Step 11.
    this._parser.setInsertionPoint (this._parser.in.s.length);

    // Step 12.
    return this;
  }; // document.open

  JSDocument.prototype.write = function () {
    var p = this._parser;

    // 1. If the insertion point is undefined, the open() method must be ...
    if (isNaN (p.insertionPoint) || p.insertionPoint == undefined) {
      this.open ();
      p = this._parser;
    }

    // 2. ... inserted into the input stream just before the insertion point.
    var s = Array.join (arguments, '');
    log ('document.write: insert "' + s + '"' +
         ' before "' + p.in.s.substring (p.insertionPoint, p.insertionPoint + 10) + '"');
    p.in.s = p.in.s.substring (0, p.insertionPoint) + s
        + p.in.s.substring (p.insertionPoint, p.in.s.length);
    p.insertionPoint += s.length;

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

    // 4. Process the characters that were inserted, ...
    p.parse ();

    // 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';
        if (node.async) r += '| ' + indent + '  async=""\n';
        if (node.defer) r += '| ' + indent + '  defer=""\n';
        if (node.src) r += '| ' + indent + '  src="' + node.src + '"\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>&lt;script>document.write("cccccc");&lt;/', 'script>bbbbbb');
&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	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			function log (s) {
30			document.logElement.appendChild (document.createTextNode (s + "\n"));
31			} // log
32
33			function InputStream (s) {
34			this.s = s;
35			} // InputStream
36
37	wakaba	1.4	function Parser (i, doc) {
38	wakaba	1.1	this.parseMode = 'pcdata';
39	wakaba	1.4	if (!doc) {
40			doc = new JSDocument (this);
41			doc.manakaiIsHTML = true;
42			}
43			this.doc = doc;
44			this.openElements = [doc];
45	wakaba	1.2	this.in = i;
46	wakaba	1.4	this.scriptsExecutedAfterParsing = [];
47	wakaba	1.1	} // Parser
48
49	wakaba	1.2	Parser.prototype.getNextToken = function () {
50	wakaba	1.3	var p = this;
51	wakaba	1.2	var i = this.in;
52	wakaba	1.1	if (this.parseMode == 'script') {
53			var token;
54	wakaba	1.3	if (p.insertionPoint <= 0) {
55			return {type: 'abort'};
56			}
57	wakaba	1.4	i.s = i.s.replace (/^([^<]+)/,
58	wakaba	1.1	function (s, t) {
59	wakaba	1.3	if (0 < p.insertionPoint && p.insertionPoint < t.length) {
60			token = {type: 'char', value: t.substring (0, p.insertionPoint)};
61			var ip = p.insertionPoint;
62			p.insertionPoint = 0;
63	wakaba	1.4	return t.substring (ip, t.length);
64	wakaba	1.3	}
65	wakaba	1.1	token = {type: 'char', value: t};
66	wakaba	1.4	p.insertionPoint -= t.length;
67			return '';
68	wakaba	1.1	});
69			if (token) return token;
70	wakaba	1.3	i.s = i.s.replace (/^<\/[Ss][Cc][Rr][Ii][Pp][Tt]>/, function (s) {
71	wakaba	1.4	if (p.insertionPoint < s.length) {
72	wakaba	1.3	token = {type: 'abort'};
73			return s;
74			}
75	wakaba	1.1	token = {type: 'end-tag', value: 'script'};
76	wakaba	1.3	p.insertionPoint -= s.length;
77	wakaba	1.1	return '';
78			});
79			if (token) return token;
80	wakaba	1.5	var m;
81			if ((p.insertionPoint < '</script'.length) &&
82			(m = i.s.match (/^<\/([SCRIPTscript]+)/))) {
83			var v = m[1].substring (0, p.insertionPoint).toLowerCase ();
84			if (v == 'script'.substring (0, p.insertionPoint - '</'.length)) {
85			return {type: 'abort'};
86			}
87			}
88	wakaba	1.4	i.s = i.s.replace (/^</,
89			function (s) {
90			token = {type: 'char', value: s};
91			p.insertionPoint -= s.length;
92			return '';
93			});
94			if (token) return token;
95	wakaba	1.1	return {type: 'eof'};
96			}
97
98			var token;
99	wakaba	1.5	i.s = i.s.replace (/^<\/([^>]+)(?:>\|$)/, function (s, e) {
100			if (p.insertionPoint < s.length \|\|
101			(p.insertionPoint <= s.length &&
102			s.substring (s.length - 1, 1) != '>')) {
103	wakaba	1.3	token = {type: 'abort'};
104			return s;
105			}
106	wakaba	1.1	token = {type: 'end-tag', value: e.toLowerCase ()};
107	wakaba	1.3	p.insertionPoint -= s.length;
108	wakaba	1.1	return '';
109			});
110			if (token) return token;
111	wakaba	1.5	i.s = i.s.replace (/^<([^>]+)(?:>\|$)/, function (s, e) {
112			if (p.insertionPoint < s.length \|\|
113			(p.insertionPoint <= s.length &&
114			s.substring (s.length - 1, 1) != '>')) {
115	wakaba	1.3	token = {type: 'abort'};
116			return s;
117			}
118	wakaba	1.4	var tagName;
119			var attrs = {};
120			e = e.replace (/^[\S]+/, function (v) {
121			tagName = v.toLowerCase ();
122			return '';
123			});
124			e = e.replace (/^\s(\S+)\s(?:=\s"([^"])"\|'([^']*)'\|([^"']+))?/,
125			function (x, attrName, attrValue1, attrValue2, attrValue3) {
126			attrs[attrName] = attrValue1 \|\| attrValue2 \|\| attrValue3;
127			return '';
128			});
129			token = {type: 'start-tag', value: tagName, attrs: attrs};
130	wakaba	1.3	p.insertionPoint -= s.length;
131	wakaba	1.1	return '';
132			});
133			if (token) return token;
134	wakaba	1.3	if (p.insertionPoint <= 0) {
135			return {type: 'abort'};
136			}
137	wakaba	1.1	i.s = i.s.replace (/^[^<]+/, function (s) {
138	wakaba	1.3	if (p.insertionPoint < s.length) {
139			token = {type: 'char', value: s.substring (0, p.insertionPoint)};
140			var ip = p.insertionPoint;
141			p.insertionPoint = 0;
142			return s.substring (ip, s.length);
143			}
144	wakaba	1.1	token = {type: 'char', value: s};
145	wakaba	1.3	p.insertionPoint -= s.length;
146	wakaba	1.1	return '';
147			});
148			if (token) return token;
149			i.s = i.s.replace (/^[\s\S]/, function (s) {
150			token = {type: 'char', value: s};
151	wakaba	1.3	p.insertionPoint -= s.length;
152	wakaba	1.1	return '';
153			});
154			if (token) return token;
155			return {type: 'eof'};
156			} // getNextToken
157
158	wakaba	1.2	Parser.prototype.parse = function () {
159	wakaba	1.1	log ('start parsing');
160
161			while (true) {
162	wakaba	1.2	var token = this.getNextToken ();
163	wakaba	1.1	log ('token: ' + token.type + ' "' + token.value + '"');
164
165			if (token.type == 'start-tag') {
166			if (token.value == 'script') {
167	wakaba	1.2	// 1. Create an element for the token in the HTML namespace.
168			var el = new JSElement (this.doc, token.value);
169	wakaba	1.4	if (token.attrs.async != null) el.async = true;
170			if (token.attrs.defer != null) el.defer = true;
171			if (token.attrs.src != null) el.src = token.attrs.src;
172	wakaba	1.2
173			// 2. Mark the element as being "parser-inserted".
174			el.manakaiParserInserted = true;
175
176			// 3. Switch the tokeniser's content model flag to the CDATA state.
177	wakaba	1.1	this.parseMode = 'script';
178
179	wakaba	1.2	// 4.1. Collect all the character tokens.
180	wakaba	1.1	while (true) {
181	wakaba	1.2	var token = this.getNextToken ();
182	wakaba	1.1	log ('token: ' + token.type + ' "' + token.value + '"');
183
184			if (token.type == 'char') {
185	wakaba	1.2	// 5. Append a single Text node to the script element node.
186	wakaba	1.1	el.manakaiAppendText (token.value);
187	wakaba	1.2
188			// 4.2. Until it returns a token that is not a character token, or
189	wakaba	1.3	// until it stops tokenising.
190	wakaba	1.1	} else if (token.type == 'eof' \|\|
191	wakaba	1.3	(token.type == 'end-tag' && token.value == 'script') \|\|
192			token.type == 'abort') {
193	wakaba	1.2	// 6. Switched back to the PCDATA state.
194	wakaba	1.1	this.parseMode = 'pcdata';
195	wakaba	1.2
196			// 7.1. If the next token is not an end tag token with ...
197			if (token.type != 'end-tag') {
198			// 7.2. This is a parse error.
199			log ('Parse error: no </' + 'script>');
200
201			// 7.3. Mark the script element as "already executed".
202			el.manakaiAlreadyExecuted = true;
203			} else {
204			// 7.4. Ignore it.
205			//
206			}
207	wakaba	1.1	break;
208			}
209			}
210
211	wakaba	1.2	// 8.1. If the parser were originally created for the ...
212			if (this.fragmentParsingMode) {
213			// 8.2. Mark the script element as "already executed" and ...
214			el.alreadyExecuted = true;
215			continue;
216			}
217
218			// 9.1. Let the old insertion point have the same value as the ...
219	wakaba	1.3	var oldInsertionPoint = this.insertionPoint;
220	wakaba	1.2	// 9.2. Let the insertion point be just before the next input ...
221	wakaba	1.3	this.setInsertionPoint (0);
222	wakaba	1.2
223			// 10. Append the new element to the current node.
224	wakaba	1.1	this.openElements[this.openElements.length - 1].appendChild (el);
225	wakaba	1.2
226			// 11. Let the insertion point have the value of the old ...
227	wakaba	1.5	oldInsertionPoint += this.insertionPoint;
228	wakaba	1.3	this.setInsertionPoint (oldInsertionPoint);
229	wakaba	1.2
230			// 12. If there is a script that will execute as soon as ...
231
232
233	wakaba	1.1	} else {
234	wakaba	1.2	var el = new JSElement (this.doc, token.value);
235	wakaba	1.1	this.openElements[this.openElements.length - 1].appendChild (el);
236			this.openElements.push (el);
237			}
238			} else if (token.type == 'end-tag') {
239			if (this.openElements[this.openElements.length - 1].localName ==
240			token.value) {
241			this.openElements.pop ();
242			} else {
243			log ('parse error: unmatched end tag: ' + token.value);
244			}
245	wakaba	1.3	} else if (token.type == 'char') {
246			this.openElements[this.openElements.length - 1].manakaiAppendText
247			(token.value);
248	wakaba	1.1	} else if (token.type == 'eof') {
249			break;
250	wakaba	1.3	} else if (token.type == 'abort') {
251			log ('parse: abort');
252			return;
253	wakaba	1.1	}
254			}
255
256			log ('stop parsing');
257	wakaba	1.4
258			// readyState = 'interactive'
259
260			// "When a script completes loading" rules start applying.
261
262			// TODO: Handles "list of scripts that will execute as soon as possible"
263			// and "list of scripts that will execute asynchronously"
264
265			// Handle "list of scripts that will execute when the document has finished
266			// parsing".
267			var list = this.scriptsExecutedAfterParsing;
268			while (list.length > 0) {
269			// TODO: break unless completed loading
270
271			// Step 1.
272			//
273
274			// Step 2. and Step 3.
275			log ('Executing a \|defer\|red script...');
276			executeScript (this.doc, list.shift ());
277
278			// Step 4.
279			}
280
281			log ('DOMContentLoaded event fired');
282
283			// "delays tha load event" things has completed:
284			// readyState = 'complete'
285			log ('load event fired');
286	wakaba	1.1	} // parse
287
288	wakaba	1.3	Parser.prototype.setInsertionPoint = function (ip) {
289			if (ip == undefined \|\| ip == null \|\| isNaN (ip)) {
290			log ('insertion point: set to undefined');
291			this.insertionPoint = undefined;
292	wakaba	1.4	} else if (ip == this.in.s.length) {
293			log ('insertion point: end of file');
294			this.insertionPoint = ip;
295	wakaba	1.3	} else {
296			log ('insertion point: set to ' + ip +
297			' (before "' + this.in.s.substring (0, 10) + '")');
298			this.insertionPoint = ip;
299			}
300			}; // setInsertionPoint
301
302	wakaba	1.2	function JSDocument (p) {
303	wakaba	1.1	this.childNodes = [];
304	wakaba	1.2	this._parser = p;
305	wakaba	1.1	} // JSDocument
306
307	wakaba	1.2	function JSElement (doc, localName) {
308	wakaba	1.1	this.localName = localName;
309	wakaba	1.2	this.ownerDocument = doc;
310	wakaba	1.1	this.childNodes = [];
311			} // JSElement
312
313			JSDocument.prototype.appendChild = JSElement.prototype.appendChild =
314			function (e) {
315			this.childNodes.push (e);
316			e.parentNode = this;
317	wakaba	1.2
318			if (e.localName == 'script') {
319	wakaba	1.4	log ('Running a script: start');
320	wakaba	1.2
321	wakaba	1.3	var doc = this.ownerDocument \|\| this;
322	wakaba	1.2	var p = doc._parser;
323
324			// 1. Script type
325			//
326
327			// 2.1. If scripting is disabled
328			//
329			// 2.2. If the script element was created by an XML ... innerHTML ...
330			//
331			// 2.3. If the user agent does not support the scripting language ...
332			//
333			// 2.4. If the script element has its "already executed" flag set
334			if (e.manakaiAlreadyExecuted) {
335			// 2.5. Abort these steps at this point.
336	wakaba	1.4	log ('Running a script: aborted');
337	wakaba	1.2	return e;
338			}
339
340			// 3. Set the element's "already executed" flag.
341			e.manakaiAlreadyExecuted = true;
342
343			// 4. If the element has a src attribute, then a load for ...
344			// TODO: load an external resource
345
346			// 5. The first of the following options:
347
348			// 5.1.
349			if (/* TODO: If the document is still being parsed && */
350			e.defer && !e.async) {
351	wakaba	1.4	p.scriptsExecutedAfterParsing.push (e);
352			log ('Running a script: aborted (defer)');
353	wakaba	1.2	} else if (e.async && e.src != null) {
354			// TODO
355			} else if (e.async && e.src == null
356			/* && list of scripts that will execute asynchronously is not empty */) {
357			// TODO
358			} else if (e.src != null && e.manakaiParserInserted) {
359			// TODO
360			} else if (e.src != null) {
361			// TODO
362			} else {
363			executeScript (doc, e); // even if other scripts are already executing.
364			}
365
366	wakaba	1.4	log ('Running a script: end');
367	wakaba	1.2	}
368
369	wakaba	1.1	return e;
370			}; // appendChild
371
372	wakaba	1.2	function executeScript (doc, e) {
373			log ('executing a script block: start');
374
375			// If the load resulted in an error, then ... firing an error event ...
376
377			// If the load was successful
378			log ('load event fired at the script element');
379
380			if (true) {
381			// Scripting is enabled, Document.designMode is disabled,
382			// Document is the active document in its browsing context
383
384			var s;
385			if (e.src != null) {
386			// TODO: from external file
387			} else {
388			s = e.text;
389			}
390
391			parseAndRunScript (doc, s);
392			}
393
394			log ('executing a script block: end');
395			} // executeScript
396
397			function parseAndRunScript (doc, s) {
398			while (true) {
399			var matched = false;
400			s = s.replace (/^\sdocument\.write\s\(((?:'[^']'\|"[^"]")\s(?:,\s(?:'[^']'\|"[^"]")))\)\s;\s*/, function (s, t) {
401			matched = true;
402			var args = [];
403			t.replace (/('[^']'\|"[^"]")/g, function (s, v) {
404			args.push (v.substring (1, v.length - 1));
405			return '';
406			});
407			doc.write.apply (doc, args);
408			return '';
409			});
410			if (s == '') break;
411			if (!matched) {
412			log ('Script parse error: "' + s + '"');
413			break;
414			}
415			}
416			} // parseAndRunScript
417
418	wakaba	1.1	function JSText (data) {
419			this.data = data;
420			} // JSText
421
422			JSDocument.prototype.manakaiAppendText =
423			JSElement.prototype.manakaiAppendText =
424			function (s) {
425			if (this.childNodes.length > 0 &&
426			this.childNodes[this.childNodes.length - 1] instanceof JSText) {
427			this.childNodes[this.childNodes.length - 1].data += s;
428			} else {
429			this.childNodes.push (new JSText (s));
430			}
431			}; // manakaiAppendText
432	wakaba	1.2
433	wakaba	1.4	JSDocument.prototype.open = function () {
434			// Two or fewer arguments
435
436			// Step 1.
437			var type = arguments[0] \|\| 'text/html';
438
439			// Step 2.
440			var replace = arguments[1] == 'replace';
441
442			// Step 3.
443			if (this._parser &&
444			!this._parser.scriptCreated &&
445			this._parser.in.insertionPoint != undefined) {
446			log ('document.open () in parsing mode is ignored');
447			return this;
448			}
449
450			// Step 4.
451			log ('onbeforeunload event fired');
452			log ('onunload event fired');
453
454			// Step 5.
455			if (this._parser) {
456			// Discard the parser.
457			}
458
459			// Step 6.
460			log ('document cleared by document.open ()');
461			this.childNodes = [];
462
463			// Step 7.
464			this._parser = new Parser (new InputStream (''), this);
465			this._parser.scriptCreated = true;
466
467			// Step 8.
468			this.manakaiIsHTML = true;
469
470			// Step 9.
471			// If not text/html, ...
472
473			// Step 10.
474			if (!replace) {
475			// History
476			}
477
478			// Step 11.
479			this._parser.setInsertionPoint (this._parser.in.s.length);
480
481			// Step 12.
482			return this;
483			}; // document.open
484
485	wakaba	1.2	JSDocument.prototype.write = function () {
486	wakaba	1.3	var p = this._parser;
487
488	wakaba	1.2	// 1. If the insertion point is undefined, the open() method must be ...
489	wakaba	1.4	if (isNaN (p.insertionPoint) \|\| p.insertionPoint == undefined) {
490			this.open ();
491			p = this._parser;
492	wakaba	1.3	}
493	wakaba	1.2
494			// 2. ... inserted into the input stream just before the insertion point.
495	wakaba	1.3	var s = Array.join (arguments, '');
496			log ('document.write: insert "' + s + '"' +
497			' before "' + p.in.s.substring (p.insertionPoint, p.insertionPoint + 10) + '"');
498			p.in.s = p.in.s.substring (0, p.insertionPoint) + s
499			+ p.in.s.substring (p.insertionPoint, p.in.s.length);
500			p.insertionPoint += s.length;
501	wakaba	1.2
502			// 3. If there is a script that will execute as soon as the parser resumes
503			// TODO
504
505			// 4. Process the characters that were inserted, ...
506	wakaba	1.3	p.parse ();
507	wakaba	1.2
508			// 5. Return
509			log ('document.write: return');
510			return;
511			}; // document.write
512
513			JSElement.prototype.__defineGetter__ ('text', function () {
514			var r = '';
515			for (var i = 0; i < this.childNodes.length; i++) {
516			if (this.childNodes[i] instanceof JSText) {
517			r += this.childNodes[i].data;
518			}
519			}
520			return r;
521			});
522	wakaba	1.1
523			function dumpTree (n, indent) {
524			var r = '';
525			for (var i = 0; i < n.childNodes.length; i++) {
526			var node = n.childNodes[i];
527			if (node instanceof JSElement) {
528			r += '\| ' + indent + node.localName + '\n';
529	wakaba	1.4	if (node.async) r += '\| ' + indent + ' async=""\n';
530			if (node.defer) r += '\| ' + indent + ' defer=""\n';
531			if (node.src) r += '\| ' + indent + ' src="' + node.src + '"\n';
532	wakaba	1.1	r += dumpTree (node, indent + ' ');
533			} else if (node instanceof JSText) {
534			r += '\| ' + indent + '"' + node.data + '"\n';
535			} else {
536			r += '\| ' + indent + node + '\n';
537			}
538			}
539			return r;
540			} // dumpTree
541			</script>
542			</head>
543			<body onload="
544			document.sourceElement = document.getElementsByTagName ('textarea')[0];
545			document.logElement = document.getElementsByTagName ('output')[0];
546			update ();
547			">
548
549			<textarea onchange=" update () "><html>
550			<head></head><body>
551			<p>
552			<script>
553	wakaba	1.3	document.write ('aaaaaaa</p><script>document.write("cccccc");</', 'script>bbbbbb');
554	wakaba	1.1	</script>
555			<p>
556			</textarea>
557
558			<output></output>
559
560			</body>
561			</html>