No-ascii characters should be supported in AutoLinkInline #278

New Issue

claunia · 2026-01-29T14:32:36Z

claunia commented

2026-01-29 14:32:36 +00:00

Originally created by @OpportunityLiu on GitHub (Mar 7, 2019).

https://babelmark.github.io/?text=%3Chttp%3A%2F%2F%E2%98%83.net%3F%E2%98%83%3E

Input:

<http://☃.net?☃>

Expacted:

<p>
  <a href="http://xn--n3h.net?%E2%98%83">http://☃.net?☃</a>
</p>

Originally created by @OpportunityLiu on GitHub (Mar 7, 2019). https://babelmark.github.io/?text=%3Chttp%3A%2F%2F%E2%98%83.net%3F%E2%98%83%3E Input: ```md <http://☃.net?☃> ``` Expacted: ```html <p> <a href="http://xn--n3h.net?%E2%98%83">http://☃.net?☃</a> </p> ```

claunia added the bug PR Welcome! labels 2026-01-29 14:32:36 +00:00

claunia closed this issue

2026-01-29 14:32:36 +00:00

claunia commented

2026-01-29 14:32:36 +00:00

@OpportunityLiu commented on GitHub (Mar 7, 2019):

as well as AutoLink

https://babelmark.github.io/?text=http%3A%2F%2F%E2%98%83.net%3F%E2%98%83+https%3A%2F%2Fcommonmark.org%2F

@OpportunityLiu commented on GitHub (Mar 7, 2019): as well as AutoLink https://babelmark.github.io/?text=http%3A%2F%2F%E2%98%83.net%3F%E2%98%83+https%3A%2F%2Fcommonmark.org%2F

claunia commented

2026-01-29 14:32:37 +00:00

@xoofx commented on GitHub (Mar 7, 2019):

Good catch, PR welcome!

@xoofx commented on GitHub (Mar 7, 2019): Good catch, PR welcome!

claunia commented

2026-01-29 14:32:38 +00:00

@MihaZupan commented on GitHub (Mar 7, 2019):

I implemented the IsValidDomain function according to the GFM spec.

A valid domain consists of alphanumeric characters, underscores (_), hyphens (-) and periods (.). There must be at least one period, and no underscores may be present in the last two segments of the domain.

That said, it seems that even GFM parses domain names like http://☃.net?☃ into <a href="http://%E2%98%83.net?%E2%98%83" rel="nofollow">http://☃.net?☃</a>.

@xoofx In reference to #252, should the test be changed from alphanumeric characters into is not ascii punctuation (with exceptions), as for example the snowman emoji falls under "other symbols", not alphanumerics?

Also as far as encoding is concerned, do you prefer we encode such domain names using IDNA ("xn--") or percent escaped (the majority of other parsers seem to be using percent escaping, like Markdig does now)?

@MihaZupan commented on GitHub (Mar 7, 2019): I implemented the `IsValidDomain` function according to [the GFM spec](https://github.github.com/gfm/#extended-www-autolink). > A valid domain consists of alphanumeric characters, underscores (_), hyphens (-) and periods (.). There must be at least one period, and no underscores may be present in the last two segments of the domain. That said, it seems that even GFM parses domain names like `http://☃.net?☃` into `<a href="http://%E2%98%83.net?%E2%98%83" rel="nofollow">http://☃.net?☃</a>`. @xoofx In reference to #252, should the test be changed from `alphanumeric characters` into `is not ascii punctuation (with exceptions)`, as for example the snowman emoji falls under "other symbols", not alphanumerics? Also as far as encoding is concerned, do you prefer we encode such domain names using IDNA ("xn--") or percent escaped (the majority of other parsers seem to be using percent escaping, like Markdig does now)?

claunia commented

2026-01-29 14:32:38 +00:00

@OpportunityLiu commented on GitHub (Mar 8, 2019):

Although most of browsers could handle it, the correct result is <a href="http://xn--n3h.net?%E2%98%83">http://☃.net?☃</a>

https://en.wikipedia.org/wiki/Punycode

@OpportunityLiu commented on GitHub (Mar 8, 2019): Although most of browsers could handle it, the correct result is `<a href="http://xn--n3h.net?%E2%98%83">http://☃.net?☃</a>` https://en.wikipedia.org/wiki/Punycode

claunia commented

2026-01-29 14:32:39 +00:00

@MihaZupan commented on GitHub (Mar 9, 2019):

I've implemented the changes needed for this, there is one minor problem tho.
IdnMapping class is not available for netstandard1.1 or for portable.
Currently I've put it in a # if region, where those two versions revert to percent-escaping.
@xoofx Would you be okay with non-consistent behavior for those legacy platforms in this case?

@MihaZupan commented on GitHub (Mar 9, 2019): I've implemented the changes needed for this, there is one minor problem tho. `IdnMapping` class is not available for netstandard1.1 or for portable. Currently I've put it in a `# if` region, where those two versions revert to percent-escaping. @xoofx Would you be okay with non-consistent behavior for those legacy platforms in this case?

Sign in to join this conversation.

Branches Tags

1 Participants

Notifications

Due Date

No due date set.

Dependencies

No dependencies set.

Reference: starred/markdig#278