--- markup/html/scripting-parser/parser.html	2008/04/29 03:29:41	1.15
+++ markup/html/scripting-parser/parser.html	2008/08/31 09:50:49	1.19
@@ -83,6 +83,7 @@
       doc = new JSDocument (this);
       doc.manakaiIsHTML = true;
     }
+    this.nextToken = [];
     this.doc = doc;
     this.openElements = [doc];
     this.input = i;
@@ -92,6 +93,10 @@
   } // Parser
 
   Parser.prototype.getNextToken = function () {
+    if (this.nextToken.length) {
+      return this.nextToken.shift ();
+    }
+
     var p = this;
     var i = this.input;
     if (this.parseMode == 'cdata') {
@@ -146,7 +151,7 @@
     i.s = i.s.replace (/^<\/([^>]+)(?:>|$)/, function (s, e) {
       if (p.insertionPoint < s.length ||
           (p.insertionPoint <= s.length &&
-           s.substring (s.length - 1, 1) != '>')) {
+           s.substring (s.length - 1, s.length) != '>')) {
         token = {type: 'abort'};
         return s;
       }
@@ -158,7 +163,7 @@
     i.s = i.s.replace (/^<([^>]+)(?:>|$)/, function (s, e) {
       if (p.insertionPoint < s.length ||
           (p.insertionPoint <= s.length &&
-           s.substring (s.length - 1, 1) != '>')) {
+           s.substring (s.length - 1, s.length) != '>')) {
         token = {type: 'abort'};
         return s;
       }
@@ -221,6 +226,24 @@
       var token = this.getNextToken ();
       log ('token: ' + token.type + ' "' + token.value + '"');
 
+      if (this.cdataEndTagRequired) {
+        // Generic CDATA parsing algorithm
+
+        if (token.type != 'abort') {
+          // 7.
+          if (token.type == 'end-tag' && token.value == this.endTagName) {
+            // 7.1. Ignores it.
+            //
+          } else {
+            // 7.2. Parse error.
+            log ('Parse error: no </' + this.endTagName + '>');
+            this.nextToken.unshift (token);
+          }
+          this.cdataEndTagRequired = false;
+          continue;
+        }
+      }
+
       if (token.type == 'start-tag') {
         if (token.value == 'script') {
           // 1. Create an element for the token in the HTML namespace.
@@ -257,6 +280,7 @@
               if (!(token.type == 'end-tag' && token.value == 'script')) {
                 // 7.2. This is a parse error.
                 log ('Parse error: no </' + 'script>');
+                this.nextToken.unshift (token);
 
                 // 7.3. Mark the script element as "already executed".
                 el.manakaiAlreadyExecuted = true;
@@ -288,8 +312,8 @@
           oldInsertionPoint += this.insertionPoint;
           this.setInsertionPoint (oldInsertionPoint);
 
-          // 12. If there is a script that will execute as soon as ...
-          while (this.scriptExecutedWhenParserResumes) {
+          // 12. If there is a pending external script
+          while (this.pendingExternalScript) {
             // 12.1. If the tree construction stage is being called reentrantly
             if (this.reentrant) {
               log ('parse: abort (reentrance)');
@@ -299,8 +323,8 @@
             // 12.2. Otherwise
             } else {
               // 1.
-              var script = this.scriptExecutedWhenParserResumes;
-              this.scriptExecutedWhenParserResumes = null;
+              var script = this.pendingExternalScript;
+              this.pendingExternalScript = null;
 
               // 2. Pause until the script has completed loading.
               // 
@@ -348,11 +372,17 @@
               // 6. Switched back to the PCDATA state.
               this.parseMode = 'pcdata';
 
+              if (token.type == 'abort') {
+                this.cdataEndTagRequired = true;
+                break;
+              }
+
               // 7.1. If the next token is not an end tag token with ...
               if (!(token.type == 'end-tag' &&
                     token.value == this.endTagName)) {
                 // 7.2. This is a parse error.
                 log ('Parse error: no </' + this.endTagName + '>');
+                this.nextToken.unshift (token);
 
                 // 7.3. Mark the script element as "already executed".
                 el.manakaiAlreadyExecuted = true;
@@ -530,10 +560,10 @@
         log ('Running a script: aborted (async)');
         // ISSUE: What is the difference with the case above?
       } 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.');
+        if (p.pendingExternalScript) {
+          log ('Error: There is a pending external script.');
         }
-        p.scriptExecutedWhenParserResumes = e;
+        p.pendingExternalScript = e;
         log ('Running a script: aborted (src parser-inserted)');
       } else if (e.src != null) {
         p.scriptsExecutedSoon.push (e);
@@ -568,7 +598,6 @@
     }
 
     // If the load was successful
-    log ('load event fired at the script element');
 
     if (true) {
     // Scripting is enabled, Document.designMode is disabled,
@@ -577,6 +606,8 @@
       parseAndRunScript (doc, s);
     }
 
+    log ('load event fired at the script element');
+
     log ('executing a script block: end');
   } // executeScript
 
@@ -734,8 +765,8 @@
         + p.input.s.substring (p.insertionPoint, p.input.s.length);
     p.insertionPoint += s.length;
 
-    // 3. If there is a script that will execute as soon as the parser resumes
-    if (p.scriptExecutedAfterParserResumes) {
+    // 3. If there is a pending external script
+    if (p.pendingExternalScript) {
       log ('document.write: processed later (there is an unprocessed <script src>)');
       logIndentLevel--;
       log ('document.write: return');
@@ -866,11 +897,12 @@
 algorithm, and so on.
 <li>Does not raise parse errors for invalid attribute specifications in start 
 or end tags.
-<li>Does not support PCDATA elements (<code>title</code> and
+<li>Does not support RCDATA elements (<code>title</code> and
 <code>textarea</code>).
-<li>Does not strip the first newline in <code>pre</code> elements.
+<li>Does not strip the first newline in <code>pre</code>,
+<code>listing</code>, and <code>textarea</code> elements.
 <li>Does not support <code>&lt;!--</code>..<code>--></code> parsing rule
-in <code>script</code> element.
+in CDATA/RCDATA elements.
 <li>Does not support foreign (SVG or MathML) elements.
 <li>Only supports <code>script</code> <code>type</code>
 <code>text/javascript</code>.  <code>type</code> and <code>language</code>
@@ -912,9 +944,13 @@
 executed.  Currently it does not matter, since we don't allow dynamic
 modification to the |src| content/DOM attribute value yet. -->
 
+<p>See also
+<a href="http://suika.fam.cx/gate/2005/sw/Live%20Scripting%20HTML%20Parser">SuikaWiki:
+Live Scripting HTML Parser</a>.
+
 </body>
 </html>
-<!-- $Date: 2008/04/29 03:29:41 $ -->
+<!-- $Date: 2008/08/31 09:50:49 $ -->
 <!--
 
 Copyright 2008 Wakaba <w@suika.fam.cx>