From e014e64a2547403393212e06beb32d71a538d061 Mon Sep 17 00:00:00 2001
From: hrb-hub <hrb-hub@users.noreply.github.com>
Date: Tue, 6 May 2025 13:12:22 +0200
Subject: [PATCH] Fix tokenizing of comments starting or ending with more than
 two --

Comments ending with more than two dashes are not terminated correctly,
resulting in markup following these comments to be included.
`<!--Comment---><div>Markup</div>` is tokenized as a comment of
`Comment---><div>Markup</div>`.

Comments starting with three dashes have their first character clipped.
`<!---Comment-->` is tokenized as a comment of `omment`.
---
 src/evented-tokenizer.ts |  6 ++++--
 tests/tokenizer-tests.ts | 37 +++++++++++++++++++++++++++++++++++++
 2 files changed, 41 insertions(+), 2 deletions(-)

diff --git a/src/evented-tokenizer.ts b/src/evented-tokenizer.ts
index d387df7..4227ba4 100644
--- a/src/evented-tokenizer.ts
+++ b/src/evented-tokenizer.ts
@@ -379,7 +379,7 @@ export default class EventedTokenizer {
     commentStart() {
       let char = this.consume();
 
-      if (char === '-') {
+      if (char === '-' && this.peek() === '-') {
         this.transitionTo(TokenizerState.commentStartDash);
       } else if (char === '>') {
         this.delegate.finishComment();
@@ -417,7 +417,9 @@ export default class EventedTokenizer {
     commentEndDash() {
       let char = this.consume();
 
-      if (char === '-') {
+      if (char === '-' && this.peek() === '-') {
+        this.delegate.appendToCommentData(char);
+      } else if (char === '-') {
         this.transitionTo(TokenizerState.commentEnd);
       } else {
         this.delegate.appendToCommentData('-' + char);
diff --git a/tests/tokenizer-tests.ts b/tests/tokenizer-tests.ts
index 58fe989..54f423d 100644
--- a/tests/tokenizer-tests.ts
+++ b/tests/tokenizer-tests.ts
@@ -324,11 +324,48 @@ QUnit.test('A comment that immediately closes', function(assert) {
 QUnit.test('A comment that contains a -', function(assert) {
   let tokens = tokenize('<!-- A perfectly legal - appears -->');
   assert.deepEqual(tokens, [comment(' A perfectly legal - appears ')]);
+
+  tokens = tokenize('<!-- A perfectly legal - -->');
+  assert.deepEqual(tokens, [comment(' A perfectly legal - ')]);
+
+  tokens = tokenize('<!-- A perfectly legal- -->');
+  assert.deepEqual(tokens, [comment(' A perfectly legal- ')]);
 });
 
 QUnit.test('A (buggy) comment that contains two --', function(assert) {
   let tokens = tokenize('<!-- A questionable -- appears -->');
   assert.deepEqual(tokens, [comment(' A questionable -- appears ')]);
+
+  tokens = tokenize('<!-- A questionable -- -->');
+  assert.deepEqual(tokens, [comment(' A questionable -- ')]);
+
+  tokens = tokenize('<!-- A questionable-- -->');
+  assert.deepEqual(tokens, [comment(' A questionable-- ')]);
+});
+
+QUnit.test('A (buggy) comment ending with more than two --', function(assert) {
+  let tokens = tokenize('<!-- A questionable but legal comment --->');
+  assert.deepEqual(tokens, [comment(' A questionable but legal comment -')]);
+
+  tokens = tokenize('<!-- A questionable but legal comment--->');
+  assert.deepEqual(tokens, [comment(' A questionable but legal comment-')]);
+
+  tokens = tokenize('<!-- A questionable but legal comment - --->');
+  assert.deepEqual(tokens, [comment(' A questionable but legal comment - -')]);
+
+  tokens = tokenize('<!-- A questionable but legal comment -- --->');
+  assert.deepEqual(tokens, [comment(' A questionable but legal comment -- -')]);
+
+  tokens = tokenize('<!-- A questionable but legal comment ------>');
+  assert.deepEqual(tokens, [comment(' A questionable but legal comment ----')]);
+});
+
+QUnit.test('A (buggy) comment starting with more than two --', function(assert) {
+  let tokens = tokenize('<!--- Questionable but legal -->');
+  assert.deepEqual(tokens, [comment('- Questionable but legal ')]);
+
+  tokens = tokenize('<!---Questionable but legal -->');
+  assert.deepEqual(tokens, [comment('-Questionable but legal ')]);
 });
 
 QUnit.test('Character references are expanded', function(assert) {