start of doctype parsing

packages/html/charref.js

@@ -2367,7 +2367,7 @@ getCharacterReference = function (scanner, inAttribute, allowedChar) {
     if (! refNumber)
       scanner.fatal("Invalid numerical character reference starting with &#");
     var codepoint;
-    if (refNumber.charAt(0).toLowerCase() === 'x') {
+    if (refNumber.charAt(0) === 'x' || refNumber.charAt(0) === 'X') {
       // hex
       var hex = refNumber.slice(1, -1);
       while (hex.charAt(0) === '0')

packages/html/exports.js

@@ -3,6 +3,7 @@ HTML = {
     // stuff exposed for testing
     Scanner: Scanner,
     getCharacterReference: getCharacterReference,
-    getComment: getComment
+    getComment: getComment,
+    getDoctype: getDoctype
   }
 };

packages/html/tokenize.js

@@ -1,5 +1,11 @@
 
+var HTML_SPACE = /^[\u0009\u000A\u000C\u0020]/;
 
+var asciiLowerCase = function (str) {
+  return str.replace(/[A-Z]/g, function (c) {
+    return String.fromCharCode(c.charCodeAt(0) + 32);
+  });
+};
 
 getComment = function (scanner) {
   if (scanner.rest().slice(0, 4) !== '<!--')
@@ -30,3 +36,41 @@ getComment = function (scanner) {
   return { t: 'Comment',
            v: commentContents };
 };
+
+var skipSpaces = function (scanner) {
+  while (HTML_SPACE.test(scanner.peek()))
+    scanner.pos++;
+};
+
+var requireSpaces = function (scanner) {
+  if (! HTML_SPACE.test(scanner.peek()))
+    scanner.fatal("Expected whitespace");
+  skipSpaces(scanner);
+};
+
+getDoctype = function (scanner) {
+  if (scanner.rest().slice(0, 9) !== '<!DOCTYPE')
+    return null;
+  scanner.pos += 9;
+
+  requireSpaces(scanner);
+
+  var ch = scanner.peek();
+  if ((! ch) || (ch === '>') || (ch === '\u0000'))
+    scanner.fatal('Malformed DOCTYPE');
+  var name = ch;
+
+  while ((ch = scanner.peek()), ! (HTML_SPACE.test(ch) || ch === '>')) {
+    if ((! ch) || (ch === '\u0000'))
+      scanner.fatal('Malformed DOCTYPE');
+    name += ch;
+  }
+  name = asciiLowerCase(name);
+
+  if (ch !== '>') {
+    // XXX
+  }
+
+  return { t: 'Doctype',
+           name: name };
+};