html/scripting-parser/parser.html

<!DOCTYPE HTML>
<html lang=en>
<head>
<title>Demo of HTML5 Parsing Algorithm with Scripting Enabled</title>
<style>
  h1, h2 {
    margin: 0;
    font-size: 100%;
  }
  p, pre {
    margin: 0;
  }
  textarea {
    width: 100%;
    -width: 99%;
    height: 10em;
  }
  output {
    display: block;
    font-family: monospace;
    white-space: -moz-pre-wrap;
    white-space: pre-wrap;
  }
</style>
<script>
  var delayedUpdater = 0;

  function update () {
    if (delayedUpdater) {
      clearTimeout (delayedUpdater);
      delayedUpdater = 0;
    }
    delayedUpdater = setTimeout (update2, 100);
  } // update

  function update2 () {
    document.logElement.textContent = '';
    var v = document.sourceElement.value;
    var p = new Parser (new InputStream (v));
    var doc = p.doc;
    p.parse ();
    log (dumpTree (doc, ''));

    document.links['permalink'].href
        = location.href + '?s=' + encodeURIComponent (v);
  } // update2

  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 ();
">
<h1>Live Scripting Parser</h1>

<h2>Markup to test
(<a href=data:, id=permalink rel=bookmark>permalink</a>)</h2>
<p>
<textarea onkeydown=" update () " onchange=" update () " oninput=" 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>

<h2>Log</h2>
<p><output></output>

<!-- TODO: short description -->

<!-- TODO: permalink query -> textarea -->

<!-- TODO: multiple attributes are not supported yet -->

</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	wakaba	1.7	h1, h2 {
7			margin: 0;
8			font-size: 100%;
9			}
10			p, pre {
11			margin: 0;
12			}
13	wakaba	1.1	textarea {
14	wakaba	1.7	width: 100%;
15			-width: 99%;
16			height: 10em;
17	wakaba	1.1	}
18			output {
19			display: block;
20			font-family: monospace;
21	wakaba	1.4	white-space: -moz-pre-wrap;
22			white-space: pre-wrap;
23	wakaba	1.1	}
24			</style>
25			<script>
26	wakaba	1.7	var delayedUpdater = 0;
27
28	wakaba	1.1	function update () {
29	wakaba	1.7	if (delayedUpdater) {
30			clearTimeout (delayedUpdater);
31			delayedUpdater = 0;
32			}
33			delayedUpdater = setTimeout (update2, 100);
34			} // update
35
36			function update2 () {
37	wakaba	1.1	document.logElement.textContent = '';
38	wakaba	1.7	var v = document.sourceElement.value;
39			var p = new Parser (new InputStream (v));
40	wakaba	1.4	var doc = p.doc;
41	wakaba	1.2	p.parse ();
42	wakaba	1.4	log (dumpTree (doc, ''));
43	wakaba	1.7
44			document.links['permalink'].href
45			= location.href + '?s=' + encodeURIComponent (v);
46			} // update2
47	wakaba	1.1
48	wakaba	1.6	var logIndentLevel = 0;
49	wakaba	1.1	function log (s) {
50	wakaba	1.6	for (var i = 0; i < logIndentLevel; i++) {
51			s = ' ' + s;
52			}
53	wakaba	1.1	document.logElement.appendChild (document.createTextNode (s + "\n"));
54			} // log
55
56			function InputStream (s) {
57			this.s = s;
58			} // InputStream
59
60	wakaba	1.4	function Parser (i, doc) {
61	wakaba	1.1	this.parseMode = 'pcdata';
62	wakaba	1.4	if (!doc) {
63			doc = new JSDocument (this);
64			doc.manakaiIsHTML = true;
65			}
66			this.doc = doc;
67			this.openElements = [doc];
68	wakaba	1.2	this.in = i;
69	wakaba	1.4	this.scriptsExecutedAfterParsing = [];
70	wakaba	1.1	} // Parser
71
72	wakaba	1.2	Parser.prototype.getNextToken = function () {
73	wakaba	1.3	var p = this;
74	wakaba	1.2	var i = this.in;
75	wakaba	1.1	if (this.parseMode == 'script') {
76			var token;
77	wakaba	1.3	if (p.insertionPoint <= 0) {
78			return {type: 'abort'};
79			}
80	wakaba	1.4	i.s = i.s.replace (/^([^<]+)/,
81	wakaba	1.1	function (s, t) {
82	wakaba	1.3	if (0 < p.insertionPoint && p.insertionPoint < t.length) {
83			token = {type: 'char', value: t.substring (0, p.insertionPoint)};
84			var ip = p.insertionPoint;
85			p.insertionPoint = 0;
86	wakaba	1.4	return t.substring (ip, t.length);
87	wakaba	1.3	}
88	wakaba	1.1	token = {type: 'char', value: t};
89	wakaba	1.4	p.insertionPoint -= t.length;
90			return '';
91	wakaba	1.1	});
92			if (token) return token;
93	wakaba	1.3	i.s = i.s.replace (/^<\/[Ss][Cc][Rr][Ii][Pp][Tt]>/, function (s) {
94	wakaba	1.4	if (p.insertionPoint < s.length) {
95	wakaba	1.3	token = {type: 'abort'};
96			return s;
97			}
98	wakaba	1.1	token = {type: 'end-tag', value: 'script'};
99	wakaba	1.3	p.insertionPoint -= s.length;
100	wakaba	1.1	return '';
101			});
102			if (token) return token;
103	wakaba	1.5	var m;
104			if ((p.insertionPoint < '</script'.length) &&
105			(m = i.s.match (/^<\/([SCRIPTscript]+)/))) {
106			var v = m[1].substring (0, p.insertionPoint).toLowerCase ();
107			if (v == 'script'.substring (0, p.insertionPoint - '</'.length)) {
108			return {type: 'abort'};
109			}
110			}
111	wakaba	1.4	i.s = i.s.replace (/^</,
112			function (s) {
113			token = {type: 'char', value: s};
114			p.insertionPoint -= s.length;
115			return '';
116			});
117			if (token) return token;
118	wakaba	1.1	return {type: 'eof'};
119			}
120
121			var token;
122	wakaba	1.5	i.s = i.s.replace (/^<\/([^>]+)(?:>\|$)/, function (s, e) {
123			if (p.insertionPoint < s.length \|\|
124			(p.insertionPoint <= s.length &&
125			s.substring (s.length - 1, 1) != '>')) {
126	wakaba	1.3	token = {type: 'abort'};
127			return s;
128			}
129	wakaba	1.1	token = {type: 'end-tag', value: e.toLowerCase ()};
130	wakaba	1.3	p.insertionPoint -= s.length;
131	wakaba	1.1	return '';
132			});
133			if (token) return token;
134	wakaba	1.5	i.s = i.s.replace (/^<([^>]+)(?:>\|$)/, function (s, e) {
135			if (p.insertionPoint < s.length \|\|
136			(p.insertionPoint <= s.length &&
137			s.substring (s.length - 1, 1) != '>')) {
138	wakaba	1.3	token = {type: 'abort'};
139			return s;
140			}
141	wakaba	1.4	var tagName;
142			var attrs = {};
143			e = e.replace (/^[\S]+/, function (v) {
144			tagName = v.toLowerCase ();
145			return '';
146			});
147	wakaba	1.6	e = e.replace (/^\s([^\s=]+)\s(?:=\s(?:"([^"])"\|'([^']*)'\|([^"']+)))?/,
148	wakaba	1.4	function (x, attrName, attrValue1, attrValue2, attrValue3) {
149	wakaba	1.6	v = attrValue1 \|\| attrValue2 \|\| attrValue3;
150			v = v.replace (/"/g, '"').replace (/'/g, "'")
151			.replace (/&/g, '&');
152			attrs[attrName.toLowerCase ()] = v;
153	wakaba	1.4	return '';
154			});
155	wakaba	1.6	if (e.length) {
156			log ('Broken start tag: "' + e + '"');
157			}
158	wakaba	1.4	token = {type: 'start-tag', value: tagName, attrs: attrs};
159	wakaba	1.3	p.insertionPoint -= s.length;
160	wakaba	1.1	return '';
161			});
162			if (token) return token;
163	wakaba	1.3	if (p.insertionPoint <= 0) {
164			return {type: 'abort'};
165			}
166	wakaba	1.1	i.s = i.s.replace (/^[^<]+/, function (s) {
167	wakaba	1.3	if (p.insertionPoint < s.length) {
168			token = {type: 'char', value: s.substring (0, p.insertionPoint)};
169			var ip = p.insertionPoint;
170			p.insertionPoint = 0;
171			return s.substring (ip, s.length);
172			}
173	wakaba	1.1	token = {type: 'char', value: s};
174	wakaba	1.3	p.insertionPoint -= s.length;
175	wakaba	1.1	return '';
176			});
177			if (token) return token;
178			i.s = i.s.replace (/^[\s\S]/, function (s) {
179			token = {type: 'char', value: s};
180	wakaba	1.3	p.insertionPoint -= s.length;
181	wakaba	1.1	return '';
182			});
183			if (token) return token;
184			return {type: 'eof'};
185			} // getNextToken
186
187	wakaba	1.2	Parser.prototype.parse = function () {
188	wakaba	1.6	logIndentLevel++;
189			log ('parse: start');
190	wakaba	1.1
191			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 == 'start-tag') {
196			if (token.value == 'script') {
197	wakaba	1.2	// 1. Create an element for the token in the HTML namespace.
198			var el = new JSElement (this.doc, token.value);
199	wakaba	1.4	if (token.attrs.async != null) el.async = true;
200			if (token.attrs.defer != null) el.defer = true;
201			if (token.attrs.src != null) el.src = token.attrs.src;
202	wakaba	1.2
203			// 2. Mark the element as being "parser-inserted".
204			el.manakaiParserInserted = true;
205
206			// 3. Switch the tokeniser's content model flag to the CDATA state.
207	wakaba	1.1	this.parseMode = 'script';
208
209	wakaba	1.2	// 4.1. Collect all the character tokens.
210	wakaba	1.1	while (true) {
211	wakaba	1.2	var token = this.getNextToken ();
212	wakaba	1.1	log ('token: ' + token.type + ' "' + token.value + '"');
213
214			if (token.type == 'char') {
215	wakaba	1.2	// 5. Append a single Text node to the script element node.
216	wakaba	1.1	el.manakaiAppendText (token.value);
217	wakaba	1.2
218			// 4.2. Until it returns a token that is not a character token, or
219	wakaba	1.3	// until it stops tokenising.
220	wakaba	1.1	} else if (token.type == 'eof' \|\|
221	wakaba	1.3	(token.type == 'end-tag' && token.value == 'script') \|\|
222			token.type == 'abort') {
223	wakaba	1.2	// 6. Switched back to the PCDATA state.
224	wakaba	1.1	this.parseMode = 'pcdata';
225	wakaba	1.2
226			// 7.1. If the next token is not an end tag token with ...
227			if (token.type != 'end-tag') {
228			// 7.2. This is a parse error.
229			log ('Parse error: no </' + 'script>');
230
231			// 7.3. Mark the script element as "already executed".
232			el.manakaiAlreadyExecuted = true;
233			} else {
234			// 7.4. Ignore it.
235			//
236			}
237	wakaba	1.1	break;
238			}
239			}
240
241	wakaba	1.2	// 8.1. If the parser were originally created for the ...
242			if (this.fragmentParsingMode) {
243			// 8.2. Mark the script element as "already executed" and ...
244			el.alreadyExecuted = true;
245			continue;
246			}
247
248			// 9.1. Let the old insertion point have the same value as the ...
249	wakaba	1.3	var oldInsertionPoint = this.insertionPoint;
250	wakaba	1.2	// 9.2. Let the insertion point be just before the next input ...
251	wakaba	1.3	this.setInsertionPoint (0);
252	wakaba	1.2
253			// 10. Append the new element to the current node.
254	wakaba	1.1	this.openElements[this.openElements.length - 1].appendChild (el);
255	wakaba	1.2
256			// 11. Let the insertion point have the value of the old ...
257	wakaba	1.7
258	wakaba	1.5	oldInsertionPoint += this.insertionPoint;
259	wakaba	1.3	this.setInsertionPoint (oldInsertionPoint);
260	wakaba	1.2
261			// 12. If there is a script that will execute as soon as ...
262	wakaba	1.6	while (this.scriptExecutedWhenParserResumes) {
263			// 12.1. If the tree construction stage is being called reentrantly
264			if (this.reentrant) {
265			log ('parse: abort (reentrance)');
266			logIndentLevel--;
267			return;
268
269			// 12.2. Otherwise
270			} else {
271			// 1.
272			var script = this.scriptExecutedWhenParserResumes;
273			this.scriptExecutedWhenParserResumes = null;
274
275			// 2. Pause until the script has completed loading.
276			//
277
278			// 3. Let the insertion point to just before the next input char.
279			this.setInsertionPoint (0);
280
281			// 4. Execute the script.
282			executeScript (this.doc, script);
283
284			// 5. Let the insertion point be undefined again.
285			this.setInsertionPoint (undefined);
286	wakaba	1.2
287	wakaba	1.6	// 6. If there is once again a script that will execute ...
288			//
289			}
290			}
291	wakaba	1.1	} else {
292	wakaba	1.2	var el = new JSElement (this.doc, token.value);
293	wakaba	1.1	this.openElements[this.openElements.length - 1].appendChild (el);
294			this.openElements.push (el);
295			}
296			} else if (token.type == 'end-tag') {
297			if (this.openElements[this.openElements.length - 1].localName ==
298			token.value) {
299			this.openElements.pop ();
300			} else {
301			log ('parse error: unmatched end tag: ' + token.value);
302			}
303	wakaba	1.3	} else if (token.type == 'char') {
304			this.openElements[this.openElements.length - 1].manakaiAppendText
305			(token.value);
306	wakaba	1.1	} else if (token.type == 'eof') {
307			break;
308	wakaba	1.3	} else if (token.type == 'abort') {
309			log ('parse: abort');
310	wakaba	1.6	logIndentLevel--;
311	wakaba	1.3	return;
312	wakaba	1.1	}
313			}
314
315			log ('stop parsing');
316	wakaba	1.4
317			// readyState = 'interactive'
318
319			// "When a script completes loading" rules start applying.
320
321			// TODO: Handles "list of scripts that will execute as soon as possible"
322			// and "list of scripts that will execute asynchronously"
323
324			// Handle "list of scripts that will execute when the document has finished
325			// parsing".
326			var list = this.scriptsExecutedAfterParsing;
327			while (list.length > 0) {
328			// TODO: break unless completed loading
329
330			// Step 1.
331			//
332
333			// Step 2. and Step 3.
334			log ('Executing a \|defer\|red script...');
335			executeScript (this.doc, list.shift ());
336
337			// Step 4.
338			}
339
340			log ('DOMContentLoaded event fired');
341
342			// "delays tha load event" things has completed:
343			// readyState = 'complete'
344			log ('load event fired');
345	wakaba	1.6
346			logIndentLevel--;
347	wakaba	1.1	} // parse
348
349	wakaba	1.3	Parser.prototype.setInsertionPoint = function (ip) {
350			if (ip == undefined \|\| ip == null \|\| isNaN (ip)) {
351			log ('insertion point: set to undefined');
352			this.insertionPoint = undefined;
353	wakaba	1.4	} else if (ip == this.in.s.length) {
354			log ('insertion point: end of file');
355			this.insertionPoint = ip;
356	wakaba	1.3	} else {
357			log ('insertion point: set to ' + ip +
358			' (before "' + this.in.s.substring (0, 10) + '")');
359			this.insertionPoint = ip;
360			}
361			}; // setInsertionPoint
362
363	wakaba	1.2	function JSDocument (p) {
364	wakaba	1.1	this.childNodes = [];
365	wakaba	1.2	this._parser = p;
366	wakaba	1.1	} // JSDocument
367
368	wakaba	1.2	function JSElement (doc, localName) {
369	wakaba	1.1	this.localName = localName;
370	wakaba	1.2	this.ownerDocument = doc;
371	wakaba	1.1	this.childNodes = [];
372			} // JSElement
373
374			JSDocument.prototype.appendChild = JSElement.prototype.appendChild =
375			function (e) {
376			this.childNodes.push (e);
377			e.parentNode = this;
378	wakaba	1.2
379			if (e.localName == 'script') {
380	wakaba	1.6	logIndentLevel++;
381	wakaba	1.4	log ('Running a script: start');
382	wakaba	1.2
383	wakaba	1.3	var doc = this.ownerDocument \|\| this;
384	wakaba	1.2	var p = doc._parser;
385
386			// 1. Script type
387			//
388
389			// 2.1. If scripting is disabled
390			//
391			// 2.2. If the script element was created by an XML ... innerHTML ...
392			//
393			// 2.3. If the user agent does not support the scripting language ...
394			//
395			// 2.4. If the script element has its "already executed" flag set
396			if (e.manakaiAlreadyExecuted) {
397			// 2.5. Abort these steps at this point.
398	wakaba	1.4	log ('Running a script: aborted');
399	wakaba	1.6	logIndentLevel--;
400	wakaba	1.2	return e;
401			}
402
403			// 3. Set the element's "already executed" flag.
404			e.manakaiAlreadyExecuted = true;
405
406			// 4. If the element has a src attribute, then a load for ...
407			// TODO: load an external resource
408
409			// 5. The first of the following options:
410
411			// 5.1.
412			if (/* TODO: If the document is still being parsed && */
413			e.defer && !e.async) {
414	wakaba	1.4	p.scriptsExecutedAfterParsing.push (e);
415			log ('Running a script: aborted (defer)');
416	wakaba	1.2	} else if (e.async && e.src != null) {
417			// TODO
418			} else if (e.async && e.src == null
419			/* && list of scripts that will execute asynchronously is not empty */) {
420			// TODO
421			} else if (e.src != null && e.manakaiParserInserted) {
422	wakaba	1.6	if (p.scriptExecutedWhenParserResumes) {
423			log ('Error: There is a script that will execute as soon as the parser resumes.');
424			}
425			p.scriptExecutedWhenParserResumes = e;
426			log ('Running a script: aborted (src)');
427	wakaba	1.2	} else if (e.src != null) {
428			// TODO
429			} else {
430			executeScript (doc, e); // even if other scripts are already executing.
431			}
432
433	wakaba	1.4	log ('Running a script: end');
434	wakaba	1.6	logIndentLevel--;
435	wakaba	1.2	}
436
437	wakaba	1.1	return e;
438			}; // appendChild
439
440	wakaba	1.2	function executeScript (doc, e) {
441			log ('executing a script block: start');
442
443	wakaba	1.6	var s;
444			if (e.src != null) {
445			s = getExternalScript (e.src);
446
447			// If the load resulted in an error, then ... firing an error event ...
448			if (s == null) {
449			log ('error event fired at the script element');
450			return;
451			}
452
453			log ('External script loaded: "' + s + '"');
454			} else {
455			s = e.text;
456			}
457	wakaba	1.2
458			// If the load was successful
459			log ('load event fired at the script element');
460
461			if (true) {
462			// Scripting is enabled, Document.designMode is disabled,
463			// Document is the active document in its browsing context
464
465			parseAndRunScript (doc, s);
466			}
467
468			log ('executing a script block: end');
469			} // executeScript
470
471	wakaba	1.6	function getExternalScript (uri) {
472			if (uri.match (/^javascript:/i)) {
473			var m;
474			if (m = uri.match (/^javascript:\s(?:'([^'])'\|"([^"]+)")\s*$/i)) {
475			if (m[1]) {
476			return m[1];
477			} else if (m[2]) {
478			return m[2];
479			} else {
480			return null;
481			}
482			} else {
483			log ('Complex javascript: URI is not supported: <' + uri + '>');
484			return null;
485			}
486			} else {
487			log ('URI scheme not supported: <' + uri + '>');
488			return null;
489			}
490			} // getExternalScript
491
492	wakaba	1.2	function parseAndRunScript (doc, s) {
493			while (true) {
494			var matched = false;
495			s = s.replace (/^\sdocument\.write\s\(((?:'[^']'\|"[^"]")\s(?:,\s(?:'[^']'\|"[^"]")))\)\s;\s*/, function (s, t) {
496			matched = true;
497			var args = [];
498			t.replace (/('[^']'\|"[^"]")/g, function (s, v) {
499			args.push (v.substring (1, v.length - 1));
500			return '';
501			});
502			doc.write.apply (doc, args);
503			return '';
504			});
505			if (s == '') break;
506			if (!matched) {
507			log ('Script parse error: "' + s + '"');
508			break;
509			}
510			}
511			} // parseAndRunScript
512
513	wakaba	1.1	function JSText (data) {
514			this.data = data;
515			} // JSText
516
517			JSDocument.prototype.manakaiAppendText =
518			JSElement.prototype.manakaiAppendText =
519			function (s) {
520			if (this.childNodes.length > 0 &&
521			this.childNodes[this.childNodes.length - 1] instanceof JSText) {
522			this.childNodes[this.childNodes.length - 1].data += s;
523			} else {
524			this.childNodes.push (new JSText (s));
525			}
526			}; // manakaiAppendText
527	wakaba	1.2
528	wakaba	1.4	JSDocument.prototype.open = function () {
529			// Two or fewer arguments
530
531			// Step 1.
532			var type = arguments[0] \|\| 'text/html';
533
534			// Step 2.
535			var replace = arguments[1] == 'replace';
536
537			// Step 3.
538			if (this._parser &&
539			!this._parser.scriptCreated &&
540			this._parser.in.insertionPoint != undefined) {
541			log ('document.open () in parsing mode is ignored');
542			return this;
543			}
544
545			// Step 4.
546			log ('onbeforeunload event fired');
547			log ('onunload event fired');
548
549			// Step 5.
550			if (this._parser) {
551			// Discard the parser.
552			}
553
554			// Step 6.
555			log ('document cleared by document.open ()');
556			this.childNodes = [];
557
558			// Step 7.
559			this._parser = new Parser (new InputStream (''), this);
560			this._parser.scriptCreated = true;
561
562			// Step 8.
563			this.manakaiIsHTML = true;
564
565			// Step 9.
566			// If not text/html, ...
567
568			// Step 10.
569			if (!replace) {
570			// History
571			}
572
573			// Step 11.
574			this._parser.setInsertionPoint (this._parser.in.s.length);
575
576			// Step 12.
577			return this;
578			}; // document.open
579
580	wakaba	1.2	JSDocument.prototype.write = function () {
581	wakaba	1.6	logIndentLevel++;
582
583	wakaba	1.3	var p = this._parser;
584
585	wakaba	1.2	// 1. If the insertion point is undefined, the open() method must be ...
586	wakaba	1.4	if (isNaN (p.insertionPoint) \|\| p.insertionPoint == undefined) {
587			this.open ();
588			p = this._parser;
589	wakaba	1.3	}
590	wakaba	1.2
591			// 2. ... inserted into the input stream just before the insertion point.
592	wakaba	1.3	var s = Array.join (arguments, '');
593			log ('document.write: insert "' + s + '"' +
594			' before "' + p.in.s.substring (p.insertionPoint, p.insertionPoint + 10) + '"');
595			p.in.s = p.in.s.substring (0, p.insertionPoint) + s
596			+ p.in.s.substring (p.insertionPoint, p.in.s.length);
597			p.insertionPoint += s.length;
598	wakaba	1.2
599			// 3. If there is a script that will execute as soon as the parser resumes
600	wakaba	1.6	if (p.scriptExecutedAfterParserResumes) {
601			log ('document.write: processed later (there is an unprocessed <script src>)');
602			logIndentLevel--;
603			return;
604			}
605	wakaba	1.2
606			// 4. Process the characters that were inserted, ...
607	wakaba	1.6	var originalReentrant = p.reentrant;
608			p.reentrant = true;
609	wakaba	1.3	p.parse ();
610	wakaba	1.6	p.reentrant = originalReentrant;
611			// TODO: "Abort the processing of any nested invokations of the tokeniser,
612			// yielding control back to the caller." (<script> parsing). Do we need
613			// to do something here?
614	wakaba	1.2
615			// 5. Return
616			log ('document.write: return');
617	wakaba	1.6
618			logIndentLevel--;
619	wakaba	1.2	return;
620			}; // document.write
621
622			JSElement.prototype.__defineGetter__ ('text', function () {
623			var r = '';
624			for (var i = 0; i < this.childNodes.length; i++) {
625			if (this.childNodes[i] instanceof JSText) {
626			r += this.childNodes[i].data;
627			}
628			}
629			return r;
630			});
631	wakaba	1.1
632			function dumpTree (n, indent) {
633			var r = '';
634			for (var i = 0; i < n.childNodes.length; i++) {
635			var node = n.childNodes[i];
636			if (node instanceof JSElement) {
637			r += '\| ' + indent + node.localName + '\n';
638	wakaba	1.4	if (node.async) r += '\| ' + indent + ' async=""\n';
639			if (node.defer) r += '\| ' + indent + ' defer=""\n';
640			if (node.src) r += '\| ' + indent + ' src="' + node.src + '"\n';
641	wakaba	1.1	r += dumpTree (node, indent + ' ');
642			} else if (node instanceof JSText) {
643			r += '\| ' + indent + '"' + node.data + '"\n';
644			} else {
645			r += '\| ' + indent + node + '\n';
646			}
647			}
648			return r;
649			} // dumpTree
650			</script>
651			</head>
652			<body onload="
653			document.sourceElement = document.getElementsByTagName ('textarea')[0];
654			document.logElement = document.getElementsByTagName ('output')[0];
655			update ();
656			">
657	wakaba	1.7	<h1>Live Scripting Parser</h1>
658	wakaba	1.1
659	wakaba	1.7	<h2>Markup to test
660			(<a href=data:, id=permalink rel=bookmark>permalink</a>)</h2>
661			<p>
662			<textarea onkeydown=" update () " onchange=" update () " oninput=" update () "><html>
663	wakaba	1.1	<head></head><body>
664			<p>
665			<script>
666	wakaba	1.3	document.write ('aaaaaaa</p><script>document.write("cccccc");</', 'script>bbbbbb');
667	wakaba	1.1	</script>
668			<p>
669			</textarea>
670
671	wakaba	1.7	<h2>Log</h2>
672			<p><output></output>
673
674			<!-- TODO: short description -->
675
676			<!-- TODO: permalink query -> textarea -->
677
678			<!-- TODO: multiple attributes are not supported yet -->
679	wakaba	1.1
680			</body>
681			</html>