Add support for parsing <!DOCTYPE html>

tildeio · Sep 12, 2019 · 627a023 · 627a023
1 parent 691662c
commit 627a023
Show file tree

Hide file tree

Showing 2 changed files with 27 additions and 0 deletions.
diff --git a/src/types.ts b/src/types.ts
@@ -29,6 +29,11 @@ export interface TokenBase<T extends TokenType> {
   loc?: Location;
 }
 
+export interface DOCTYPE extends TokenBase<TokenType.DOCTYPE> {
+  documentType: string;
+  legacyString: string;
+}
+
 export interface StartTag extends TokenBase<TokenType.StartTag> {
   tagName: string;
   attributes: Attribute[];
@@ -50,6 +55,7 @@ export interface Comment extends TokenBase<TokenType.Comment> {
 export type Token = StartTag | EndTag | Chars | Comment;
 
 export const enum TokenType {
+  DOCTYPE = 'DOCTYPE',
   StartTag = 'StartTag',
   EndTag = 'EndTag',
   Chars = 'Chars',

diff --git a/tests/tokenizer-tests.ts b/tests/tokenizer-tests.ts
@@ -1,5 +1,6 @@
 import {
   tokenize,
+  DOCTYPE,
   StartTag,
   EndTag,
   Comment,
@@ -11,6 +12,18 @@ import {
 
 QUnit.module('simple-html-tokenizer - tokenizer');
 
+QUnit.test('DOCTYPE', function(assert) {
+  let tokens = tokenize('<!DOCTYPE html><html><body></body></html>');
+  assert.deepEqual(tokens, [
+    doctype(),
+    startTag('html'),
+    startTag('body'),
+  ], 'DOCTYPE is included in tokens');
+
+  tokens = tokenize('<!-- comment --><!DOCTYPE html>');
+  assert.deepEqual(tokens, [comment(' comment '), doctype()], 'DOCTYPE after comments is valid');
+});
+
 QUnit.test('Simple content', function(assert) {
   let tokens = tokenize('hello');
   assert.deepEqual(tokens, [chars('hello')]);
@@ -363,6 +376,14 @@ function endTag(tagName: string): EndTag {
   };
 }
 
+function doctype(legacyString: string = ''): DOCTYPE {
+  return {
+    type: 'DOCTYPE',
+    documentType: 'html',
+    legacyString: legacyString,
+  };
+}
+
 function locInfo(
   token: Token,
   startLine: number,