[bug] `CodeBlock` - mismatched end fence behave differently

[Bistard]

Marked version: 14.1.2

Describe the bug Consider when lexing the following two texts (both have end fence with length 4):

// "```\nmismatched\n````"
// "```\nmismatched\n~~~~"

The following are their tokenized result:

{type:"code", raw:"```\nmismatch\n````", lang:"", text:"mismatch"},
{type:"code", raw:"```\nmismatch\n~~~~", lang:"", text:"mismatch\n~~~~"}

Expected behavior 🤔I was expecting:

{type:"code", raw:"```\nmismatch\n````", lang:"", text:"mismatch"},
{type:"code", raw:"```\nmismatch\n~~~~", lang:"", text:"mismatch"}

or just

{type:"code", raw:"```\nmismatch\n````", lang:"", text:"mismatch\n````"},
{type:"code", raw:"```\nmismatch\n~~~~", lang:"", text:"mismatch\n~~~~"}

[UziTech]

As you can see in these demos the text property is the code that is displayed

demo 1

demo 2

[Bistard]

Given that the ending fences are mismatched in both cases, shouldn't they both be handled (or ignored) within the text property for simplicity and consistency?

This behavior is confusing because:

1. In the case with "~", the ending fence with 4 ~ characters is displayed as part of the text.
2. However, in the case with backticks, the ending fence with 4 backticks does not appear in the text.

Is this behavior defined by CommonMark, or could the behaviors be aligned?

[UziTech]

This is the same behavior as CommonMark

demo 1

demo 2

[Bistard]

OK. That make sense. I should check common mark first. Thanks for the response.