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

  var logIndentLevel = 0;
  function log (s) {
    for (var i = 0; i < logIndentLevel; i++) {
      s = '  ' + 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) {
        v = attrValue1 || attrValue2 || attrValue3;
        v = v.replace (/&quot;/g, '"').replace (/&apos;/g, "'")
            .replace (/&amp;/g, '&');
        attrs[attrName.toLowerCase ()] = v;
        return '';
      });
      if (e.length) {
        log ('Broken start tag: "' + e + '"');
      }
      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 () {
    logIndentLevel++;
    log ('parse: start');

    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 ...
          while (this.scriptExecutedWhenParserResumes) {
            // 12.1. If the tree construction stage is being called reentrantly
            if (this.reentrant) {
              log ('parse: abort (reentrance)');
              logIndentLevel--;
              return;

            // 12.2. Otherwise
            } else {
              // 1.
              var script = this.scriptExecutedWhenParserResumes;
              this.scriptExecutedWhenParserResumes = null;

              // 2. Pause until the script has completed loading.
              // 

              // 3. Let the insertion point to just before the next input char.
              this.setInsertionPoint (0);

              // 4. Execute the script.
              executeScript (this.doc, script);

              // 5. Let the insertion point be undefined again.
              this.setInsertionPoint (undefined);

              // 6. If there is once again a script that will execute ...
              //
            }
          }
        } 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');
        logIndentLevel--;
        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');

    logIndentLevel--;
  } // 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') {
      logIndentLevel++;
      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');
        logIndentLevel--;
        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) {
        if (p.scriptExecutedWhenParserResumes) {
          log ('Error: There is a script that will execute as soon as the parser resumes.');
        }
        p.scriptExecutedWhenParserResumes = e;
        log ('Running a script: aborted (src)');
      } else if (e.src != null) {
        // TODO
      } else {
        executeScript (doc, e); // even if other scripts are already executing.
      }

      log ('Running a script: end');
      logIndentLevel--;
    }

    return e;
  }; // appendChild

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

    var s;
    if (e.src != null) {
      s = getExternalScript (e.src);

      // If the load resulted in an error, then ... firing an error event ...
      if (s == null) {
        log ('error event fired at the script element');
        return;
      }

      log ('External script loaded: "' + s + '"');
    } else {
      s = e.text;
    }

    // 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

      parseAndRunScript (doc, s);
    }

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

  function getExternalScript (uri) {
    if (uri.match (/^javascript:/i)) {
      var m;
      if (m = uri.match (/^javascript:\s*(?:'([^']*)'|"([^"]+)")\s*$/i)) {
        if (m[1]) {
          return m[1];
        } else if (m[2]) {
          return m[2];
        } else {
          return null;
        }
      } else {
        log ('Complex javascript: URI is not supported: <' + uri + '>');
        return null;
      }
    } else {
      log ('URI scheme not supported: <' + uri + '>');
      return null;
    }
  } // getExternalScript

  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 () {
    logIndentLevel++;

    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
    if (p.scriptExecutedAfterParserResumes) {
      log ('document.write: processed later (there is an unprocessed <script src>)');
      logIndentLevel--;
      return;
    }

    // 4. Process the characters that were inserted, ...
    var originalReentrant = p.reentrant;
    p.reentrant = true;
    p.parse ();
    p.reentrant = originalReentrant;
    // TODO: "Abort the processing of any nested invokations of the tokeniser,
    // yielding control back to the caller." (<script> parsing).  Do we need
    // to do something here?

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

    logIndentLevel--;
    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	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	wakaba	1.1	return '';
142			});
143			if (token) return token;
144	wakaba	1.3	if (p.insertionPoint <= 0) {
145			return {type: 'abort'};
146			}
147	wakaba	1.1	i.s = i.s.replace (/^[^<]+/, function (s) {
148	wakaba	1.3	if (p.insertionPoint < s.length) {
149			token = {type: 'char', value: s.substring (0, p.insertionPoint)};
150			var ip = p.insertionPoint;
151			p.insertionPoint = 0;
152			return s.substring (ip, s.length);
153			}
154	wakaba	1.1	token = {type: 'char', value: s};
155	wakaba	1.3	p.insertionPoint -= s.length;
156	wakaba	1.1	return '';
157			});
158			if (token) return token;
159			i.s = i.s.replace (/^[\s\S]/, function (s) {
160			token = {type: 'char', value: s};
161	wakaba	1.3	p.insertionPoint -= s.length;
162	wakaba	1.1	return '';
163			});
164			if (token) return token;
165			return {type: 'eof'};
166			} // getNextToken
167
168	wakaba	1.2	Parser.prototype.parse = function () {
169	wakaba	1.6	logIndentLevel++;
170			log ('parse: start');
171	wakaba	1.1
172			while (true) {
173	wakaba	1.2	var token = this.getNextToken ();
174	wakaba	1.1	log ('token: ' + token.type + ' "' + token.value + '"');
175
176			if (token.type == 'start-tag') {
177			if (token.value == 'script') {
178	wakaba	1.2	// 1. Create an element for the token in the HTML namespace.
179			var el = new JSElement (this.doc, token.value);
180	wakaba	1.4	if (token.attrs.async != null) el.async = true;
181			if (token.attrs.defer != null) el.defer = true;
182			if (token.attrs.src != null) el.src = token.attrs.src;
183	wakaba	1.2
184			// 2. Mark the element as being "parser-inserted".
185			el.manakaiParserInserted = true;
186
187			// 3. Switch the tokeniser's content model flag to the CDATA state.
188	wakaba	1.1	this.parseMode = 'script';
189
190	wakaba	1.2	// 4.1. Collect all the character tokens.
191	wakaba	1.1	while (true) {
192	wakaba	1.2	var token = this.getNextToken ();
193	wakaba	1.1	log ('token: ' + token.type + ' "' + token.value + '"');
194
195			if (token.type == 'char') {
196	wakaba	1.2	// 5. Append a single Text node to the script element node.
197	wakaba	1.1	el.manakaiAppendText (token.value);
198	wakaba	1.2
199			// 4.2. Until it returns a token that is not a character token, or
200	wakaba	1.3	// until it stops tokenising.
201	wakaba	1.1	} else if (token.type == 'eof' \|\|
202	wakaba	1.3	(token.type == 'end-tag' && token.value == 'script') \|\|
203			token.type == 'abort') {
204	wakaba	1.2	// 6. Switched back to the PCDATA state.
205	wakaba	1.1	this.parseMode = 'pcdata';
206	wakaba	1.2
207			// 7.1. If the next token is not an end tag token with ...
208			if (token.type != 'end-tag') {
209			// 7.2. This is a parse error.
210			log ('Parse error: no </' + 'script>');
211
212			// 7.3. Mark the script element as "already executed".
213			el.manakaiAlreadyExecuted = true;
214			} else {
215			// 7.4. Ignore it.
216			//
217			}
218	wakaba	1.1	break;
219			}
220			}
221
222	wakaba	1.2	// 8.1. If the parser were originally created for the ...
223			if (this.fragmentParsingMode) {
224			// 8.2. Mark the script element as "already executed" and ...
225			el.alreadyExecuted = true;
226			continue;
227			}
228
229			// 9.1. Let the old insertion point have the same value as the ...
230	wakaba	1.3	var oldInsertionPoint = this.insertionPoint;
231	wakaba	1.2	// 9.2. Let the insertion point be just before the next input ...
232	wakaba	1.3	this.setInsertionPoint (0);
233	wakaba	1.2
234			// 10. Append the new element to the current node.
235	wakaba	1.1	this.openElements[this.openElements.length - 1].appendChild (el);
236	wakaba	1.2
237			// 11. Let the insertion point have the value of the old ...
238	wakaba	1.5	oldInsertionPoint += this.insertionPoint;
239	wakaba	1.3	this.setInsertionPoint (oldInsertionPoint);
240	wakaba	1.2
241			// 12. If there is a script that will execute as soon as ...
242	wakaba	1.6	while (this.scriptExecutedWhenParserResumes) {
243			// 12.1. If the tree construction stage is being called reentrantly
244			if (this.reentrant) {
245			log ('parse: abort (reentrance)');
246			logIndentLevel--;
247			return;
248
249			// 12.2. Otherwise
250			} else {
251			// 1.
252			var script = this.scriptExecutedWhenParserResumes;
253			this.scriptExecutedWhenParserResumes = null;
254
255			// 2. Pause until the script has completed loading.
256			//
257
258			// 3. Let the insertion point to just before the next input char.
259			this.setInsertionPoint (0);
260
261			// 4. Execute the script.
262			executeScript (this.doc, script);
263
264			// 5. Let the insertion point be undefined again.
265			this.setInsertionPoint (undefined);
266	wakaba	1.2
267	wakaba	1.6	// 6. If there is once again a script that will execute ...
268			//
269			}
270			}
271	wakaba	1.1	} else {
272	wakaba	1.2	var el = new JSElement (this.doc, token.value);
273	wakaba	1.1	this.openElements[this.openElements.length - 1].appendChild (el);
274			this.openElements.push (el);
275			}
276			} else if (token.type == 'end-tag') {
277			if (this.openElements[this.openElements.length - 1].localName ==
278			token.value) {
279			this.openElements.pop ();
280			} else {
281			log ('parse error: unmatched end tag: ' + token.value);
282			}
283	wakaba	1.3	} else if (token.type == 'char') {
284			this.openElements[this.openElements.length - 1].manakaiAppendText
285			(token.value);
286	wakaba	1.1	} else if (token.type == 'eof') {
287			break;
288	wakaba	1.3	} else if (token.type == 'abort') {
289			log ('parse: abort');
290	wakaba	1.6	logIndentLevel--;
291	wakaba	1.3	return;
292	wakaba	1.1	}
293			}
294
295			log ('stop parsing');
296	wakaba	1.4
297			// readyState = 'interactive'
298
299			// "When a script completes loading" rules start applying.
300
301			// TODO: Handles "list of scripts that will execute as soon as possible"
302			// and "list of scripts that will execute asynchronously"
303
304			// Handle "list of scripts that will execute when the document has finished
305			// parsing".
306			var list = this.scriptsExecutedAfterParsing;
307			while (list.length > 0) {
308			// TODO: break unless completed loading
309
310			// Step 1.
311			//
312
313			// Step 2. and Step 3.
314			log ('Executing a \|defer\|red script...');
315			executeScript (this.doc, list.shift ());
316
317			// Step 4.
318			}
319
320			log ('DOMContentLoaded event fired');
321
322			// "delays tha load event" things has completed:
323			// readyState = 'complete'
324			log ('load event fired');
325	wakaba	1.6
326			logIndentLevel--;
327	wakaba	1.1	} // parse
328
329	wakaba	1.3	Parser.prototype.setInsertionPoint = function (ip) {
330			if (ip == undefined \|\| ip == null \|\| isNaN (ip)) {
331			log ('insertion point: set to undefined');
332			this.insertionPoint = undefined;
333	wakaba	1.4	} else if (ip == this.in.s.length) {
334			log ('insertion point: end of file');
335			this.insertionPoint = ip;
336	wakaba	1.3	} else {
337			log ('insertion point: set to ' + ip +
338			' (before "' + this.in.s.substring (0, 10) + '")');
339			this.insertionPoint = ip;
340			}
341			}; // setInsertionPoint
342
343	wakaba	1.2	function JSDocument (p) {
344	wakaba	1.1	this.childNodes = [];
345	wakaba	1.2	this._parser = p;
346	wakaba	1.1	} // JSDocument
347
348	wakaba	1.2	function JSElement (doc, localName) {
349	wakaba	1.1	this.localName = localName;
350	wakaba	1.2	this.ownerDocument = doc;
351	wakaba	1.1	this.childNodes = [];
352			} // JSElement
353
354			JSDocument.prototype.appendChild = JSElement.prototype.appendChild =
355			function (e) {
356			this.childNodes.push (e);
357			e.parentNode = this;
358	wakaba	1.2
359			if (e.localName == 'script') {
360	wakaba	1.6	logIndentLevel++;
361	wakaba	1.4	log ('Running a script: start');
362	wakaba	1.2
363	wakaba	1.3	var doc = this.ownerDocument \|\| this;
364	wakaba	1.2	var p = doc._parser;
365
366			// 1. Script type
367			//
368
369			// 2.1. If scripting is disabled
370			//
371			// 2.2. If the script element was created by an XML ... innerHTML ...
372			//
373			// 2.3. If the user agent does not support the scripting language ...
374			//
375			// 2.4. If the script element has its "already executed" flag set
376			if (e.manakaiAlreadyExecuted) {
377			// 2.5. Abort these steps at this point.
378	wakaba	1.4	log ('Running a script: aborted');
379	wakaba	1.6	logIndentLevel--;
380	wakaba	1.2	return e;
381			}
382
383			// 3. Set the element's "already executed" flag.
384			e.manakaiAlreadyExecuted = true;
385
386			// 4. If the element has a src attribute, then a load for ...
387			// TODO: load an external resource
388
389			// 5. The first of the following options:
390
391			// 5.1.
392			if (/* TODO: If the document is still being parsed && */
393			e.defer && !e.async) {
394	wakaba	1.4	p.scriptsExecutedAfterParsing.push (e);
395			log ('Running a script: aborted (defer)');
396	wakaba	1.2	} else if (e.async && e.src != null) {
397			// TODO
398			} else if (e.async && e.src == null
399			/* && list of scripts that will execute asynchronously is not empty */) {
400			// TODO
401			} else if (e.src != null && e.manakaiParserInserted) {
402	wakaba	1.6	if (p.scriptExecutedWhenParserResumes) {
403			log ('Error: There is a script that will execute as soon as the parser resumes.');
404			}
405			p.scriptExecutedWhenParserResumes = e;
406			log ('Running a script: aborted (src)');
407	wakaba	1.2	} else if (e.src != null) {
408			// TODO
409			} else {
410			executeScript (doc, e); // even if other scripts are already executing.
411			}
412
413	wakaba	1.4	log ('Running a script: end');
414	wakaba	1.6	logIndentLevel--;
415	wakaba	1.2	}
416
417	wakaba	1.1	return e;
418			}; // appendChild
419
420	wakaba	1.2	function executeScript (doc, e) {
421			log ('executing a script block: start');
422
423	wakaba	1.6	var s;
424			if (e.src != null) {
425			s = getExternalScript (e.src);
426
427			// If the load resulted in an error, then ... firing an error event ...
428			if (s == null) {
429			log ('error event fired at the script element');
430			return;
431			}
432
433			log ('External script loaded: "' + s + '"');
434			} else {
435			s = e.text;
436			}
437	wakaba	1.2
438			// If the load was successful
439			log ('load event fired at the script element');
440
441			if (true) {
442			// Scripting is enabled, Document.designMode is disabled,
443			// Document is the active document in its browsing context
444
445			parseAndRunScript (doc, s);
446			}
447
448			log ('executing a script block: end');
449			} // executeScript
450
451	wakaba	1.6	function getExternalScript (uri) {
452			if (uri.match (/^javascript:/i)) {
453			var m;
454			if (m = uri.match (/^javascript:\s(?:'([^'])'\|"([^"]+)")\s*$/i)) {
455			if (m[1]) {
456			return m[1];
457			} else if (m[2]) {
458			return m[2];
459			} else {
460			return null;
461			}
462			} else {
463			log ('Complex javascript: URI is not supported: <' + uri + '>');
464			return null;
465			}
466			} else {
467			log ('URI scheme not supported: <' + uri + '>');
468			return null;
469			}
470			} // getExternalScript
471
472	wakaba	1.2	function parseAndRunScript (doc, s) {
473			while (true) {
474			var matched = false;
475			s = s.replace (/^\sdocument\.write\s\(((?:'[^']'\|"[^"]")\s(?:,\s(?:'[^']'\|"[^"]")))\)\s;\s*/, function (s, t) {
476			matched = true;
477			var args = [];
478			t.replace (/('[^']'\|"[^"]")/g, function (s, v) {
479			args.push (v.substring (1, v.length - 1));
480			return '';
481			});
482			doc.write.apply (doc, args);
483			return '';
484			});
485			if (s == '') break;
486			if (!matched) {
487			log ('Script parse error: "' + s + '"');
488			break;
489			}
490			}
491			} // parseAndRunScript
492
493	wakaba	1.1	function JSText (data) {
494			this.data = data;
495			} // JSText
496
497			JSDocument.prototype.manakaiAppendText =
498			JSElement.prototype.manakaiAppendText =
499			function (s) {
500			if (this.childNodes.length > 0 &&
501			this.childNodes[this.childNodes.length - 1] instanceof JSText) {
502			this.childNodes[this.childNodes.length - 1].data += s;
503			} else {
504			this.childNodes.push (new JSText (s));
505			}
506			}; // manakaiAppendText
507	wakaba	1.2
508	wakaba	1.4	JSDocument.prototype.open = function () {
509			// Two or fewer arguments
510
511			// Step 1.
512			var type = arguments[0] \|\| 'text/html';
513
514			// Step 2.
515			var replace = arguments[1] == 'replace';
516
517			// Step 3.
518			if (this._parser &&
519			!this._parser.scriptCreated &&
520			this._parser.in.insertionPoint != undefined) {
521			log ('document.open () in parsing mode is ignored');
522			return this;
523			}
524
525			// Step 4.
526			log ('onbeforeunload event fired');
527			log ('onunload event fired');
528
529			// Step 5.
530			if (this._parser) {
531			// Discard the parser.
532			}
533
534			// Step 6.
535			log ('document cleared by document.open ()');
536			this.childNodes = [];
537
538			// Step 7.
539			this._parser = new Parser (new InputStream (''), this);
540			this._parser.scriptCreated = true;
541
542			// Step 8.
543			this.manakaiIsHTML = true;
544
545			// Step 9.
546			// If not text/html, ...
547
548			// Step 10.
549			if (!replace) {
550			// History
551			}
552
553			// Step 11.
554			this._parser.setInsertionPoint (this._parser.in.s.length);
555
556			// Step 12.
557			return this;
558			}; // document.open
559
560	wakaba	1.2	JSDocument.prototype.write = function () {
561	wakaba	1.6	logIndentLevel++;
562
563	wakaba	1.3	var p = this._parser;
564
565	wakaba	1.2	// 1. If the insertion point is undefined, the open() method must be ...
566	wakaba	1.4	if (isNaN (p.insertionPoint) \|\| p.insertionPoint == undefined) {
567			this.open ();
568			p = this._parser;
569	wakaba	1.3	}
570	wakaba	1.2
571			// 2. ... inserted into the input stream just before the insertion point.
572	wakaba	1.3	var s = Array.join (arguments, '');
573			log ('document.write: insert "' + s + '"' +
574			' before "' + p.in.s.substring (p.insertionPoint, p.insertionPoint + 10) + '"');
575			p.in.s = p.in.s.substring (0, p.insertionPoint) + s
576			+ p.in.s.substring (p.insertionPoint, p.in.s.length);
577			p.insertionPoint += s.length;
578	wakaba	1.2
579			// 3. If there is a script that will execute as soon as the parser resumes
580	wakaba	1.6	if (p.scriptExecutedAfterParserResumes) {
581			log ('document.write: processed later (there is an unprocessed <script src>)');
582			logIndentLevel--;
583			return;
584			}
585	wakaba	1.2
586			// 4. Process the characters that were inserted, ...
587	wakaba	1.6	var originalReentrant = p.reentrant;
588			p.reentrant = true;
589	wakaba	1.3	p.parse ();
590	wakaba	1.6	p.reentrant = originalReentrant;
591			// TODO: "Abort the processing of any nested invokations of the tokeniser,
592			// yielding control back to the caller." (<script> parsing). Do we need
593			// to do something here?
594	wakaba	1.2
595			// 5. Return
596			log ('document.write: return');
597	wakaba	1.6
598			logIndentLevel--;
599	wakaba	1.2	return;
600			}; // document.write
601
602			JSElement.prototype.__defineGetter__ ('text', function () {
603			var r = '';
604			for (var i = 0; i < this.childNodes.length; i++) {
605			if (this.childNodes[i] instanceof JSText) {
606			r += this.childNodes[i].data;
607			}
608			}
609			return r;
610			});
611	wakaba	1.1
612			function dumpTree (n, indent) {
613			var r = '';
614			for (var i = 0; i < n.childNodes.length; i++) {
615			var node = n.childNodes[i];
616			if (node instanceof JSElement) {
617			r += '\| ' + indent + node.localName + '\n';
618	wakaba	1.4	if (node.async) r += '\| ' + indent + ' async=""\n';
619			if (node.defer) r += '\| ' + indent + ' defer=""\n';
620			if (node.src) r += '\| ' + indent + ' src="' + node.src + '"\n';
621	wakaba	1.1	r += dumpTree (node, indent + ' ');
622			} else if (node instanceof JSText) {
623			r += '\| ' + indent + '"' + node.data + '"\n';
624			} else {
625			r += '\| ' + indent + node + '\n';
626			}
627			}
628			return r;
629			} // dumpTree
630			</script>
631			</head>
632			<body onload="
633			document.sourceElement = document.getElementsByTagName ('textarea')[0];
634			document.logElement = document.getElementsByTagName ('output')[0];
635			update ();
636			">
637
638			<textarea onchange=" update () "><html>
639			<head></head><body>
640			<p>
641			<script>
642	wakaba	1.3	document.write ('aaaaaaa</p><script>document.write("cccccc");</', 'script>bbbbbb');
643	wakaba	1.1	</script>
644			<p>
645			</textarea>
646
647			<output></output>
648
649			</body>
650			</html>