--- markup/html/scripting-parser/parser.html	2008/04/20 07:48:00	1.2
+++ markup/html/scripting-parser/parser.html	2008/04/25 11:40:56	1.5
@@ -13,15 +13,17 @@
   output {
     display: block;
     font-family: monospace;
-    white-space: pre;
+    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 (p.doc, ''));
+    log (dumpTree (doc, ''));
   } // update
 
   function log (s) {
@@ -32,25 +34,61 @@
     this.s = s;
   } // InputStream
 
-  function Parser (i) {
+  function Parser (i, doc) {
     this.parseMode = 'pcdata';
-    this.doc = new JSDocument (this);
-    this.openElements = [this.doc];
+    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;
-      i.s = i.s.replace (/^([\s\S]+?)<\/[Ss][Cc][Rr][Ii][Pp][Tt]>/,
+      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};
-        return '<' + '/script>';
+        p.insertionPoint -= t.length;
+        return '';
       });
       if (token) return token;
-      i.s = i.s.replace (/^<\/[Ss][Cc][Rr][Ii][Pp][Tt]>/, function () {
+      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;
@@ -58,23 +96,59 @@
     }
 
     var token;
-    i.s = i.s.replace (/^<\/([^>]+)>/, function (s, e) {
+    i.s = i.s.replace (/^<\/([^>]+)(?:>|$)/, function (s, e) {
+      if (p.insertionPoint < s.length ||
+          (p.insertionPoint <= s.length &&
+           s.substring (s.length - 1, 1) != '>')) {
+        token = {type: 'abort'};
+        return s;
+      }
       token = {type: 'end-tag', value: e.toLowerCase ()};
+      p.insertionPoint -= s.length;
       return '';
     });
     if (token) return token;
-    i.s = i.s.replace (/^<([^>]+)>/, function (s, e) {
-      token = {type: 'start-tag', value: e.toLowerCase ()};
+    i.s = i.s.replace (/^<([^>]+)(?:>|$)/, 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;
@@ -92,6 +166,9 @@
         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;
@@ -109,9 +186,10 @@
               el.manakaiAppendText (token.value);
 
             // 4.2. Until it returns a token that is not a character token, or
-            // TODO: 4.3. Until it stops tokenising.
+            // until it stops tokenising.
             } else if (token.type == 'eof' ||
-                       (token.type == 'end-tag' && token.value == 'script')) {
+                       (token.type == 'end-tag' && token.value == 'script') ||
+                       token.type == 'abort') {
               // 6. Switched back to the PCDATA state.
               this.parseMode = 'pcdata';
 
@@ -138,13 +216,16 @@
           }
 
           // 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 ...
           
@@ -161,14 +242,63 @@
         } 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;
@@ -186,9 +316,9 @@
     e.parentNode = this;
 
     if (e.localName == 'script') {
-      log ('start running a script');
+      log ('Running a script: start');
 
-      var doc = this.ownerDocument;
+      var doc = this.ownerDocument || this;
       var p = doc._parser;
 
       // 1. Script type
@@ -203,7 +333,7 @@
       // 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');
+        log ('Running a script: aborted');
         return e;
       }
 
@@ -218,7 +348,8 @@
       // 5.1. 
       if (/* TODO: If the document is still being parsed && */
           e.defer && !e.async) {
-        // TODO
+        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
@@ -232,7 +363,7 @@
         executeScript (doc, e); // even if other scripts are already executing.
       }
 
-      log ('end running a script');
+      log ('Running a script: end');
     }
 
     return e;
@@ -299,17 +430,80 @@
     }
   }; // 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.
-    log ('document.write: insert "' + Array.join (arguments, '') + '"');
+    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');
@@ -332,6 +526,9 @@
       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';
@@ -353,7 +550,7 @@
 &lt;head>&lt;/head>&lt;body>
 &lt;p>
 &lt;script>
-document.write ('aaaaaaa&lt;/p>\n&lt;script>\ndocument.write("cccccc")\n&lt;/', 'script>\nbbbbbb');
+document.write ('aaaaaaa&lt;/p>&lt;script>document.write("cccccc");&lt;/', 'script>bbbbbb');
 &lt;/script>
 &lt;p>
 </textarea>