Emoji domain not recognised as link

[stophecom]

• I have searched open and closed issues for duplicates
• I am submitting a bug report for existing functionality that does not work as intended
• I have read https://github.com/signalapp/Signal-Android/wiki/Submitting-useful-bug-reports
• This isn't a feature request or a discussion topic

---

Bug description

Links that contain emojis are not recognised as proper links.

Steps to reproduce

• Open Signal
• Go to any chat conversation or Note to Self
• Share a link that contains an emoji. E.g. https://🤫.st

Actual result:
There is no link preview, the URL is not recognised as a proper link (not clickable)

Expected result: Valid domains should be handled as proper links.

It's a bit edge case bug, and might even count as feature request. However I'd be happy for a fix - which most likely goes here somewhere:

Signal-Android/app/src/main/java/org/thoughtcrime/securesms/linkpreview/LinkPreviewUtil.java

Line 35 in dc6dc19

public final class LinkPreviewUtil {

cc @greyson-signal @alan-signal Unfortunately I'm not familiar with Java, otherwise I'd do a PR myself. Let me know if I can help nonetheless.

Screenshots

Device info

Device: OnePlus 6
Android version: 10
Signal version: 5.7.1

Link to debug log

Nothing to log

[hiqua]

With your URL this variable is false:

      boolean validCharacters = ALL_ASCII_PATTERN.matcher(cleanedDomain).matches() ||
                                ALL_NON_ASCII_PATTERN.matcher(cleanedDomain).matches();

because cleanedDomain ends up being 🤫st. Not sure what's the reasoning behind this check though...

[mayjs]

@hiqua I noticed a similar issue with domains containing umlauts (e.g. https://üei.de). Could this be related?

[hiqua]

I would say so.

[stale]

Is this still relevant? If so, what is blocking it? Is there anything you can do to help move it forward?

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.

[stophecom]

Yes, it is this still relevant. Links with emoji or umlauts (ä, ö, ü) are not recognized as proper links.

[stale]

Is this still relevant? If so, what is blocking it? Is there anything you can do to help move it forward?

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs.

[ckujau]

Jup, it's still relevant. Btw, this affects the desktop version as well.

[greyson-signal]

We use the stock android linkifier. I don't anticipate that we'll be making a custom one anytime soon, apologies.

[ckujau]

Hm, both links work in Google Chat (formerly "Hangouts") just fine. And also in other applications (e.g. "Slack").

[greyson-signal]

Ah, I'm sorry, I misspoke. So we do use the default linkifier, but we filter out some results that fail certain rules. Right now, the main rule is that we don't linkify links whose domain has a mix of ascii and non-ascii characters. This is to help prevent homograph attacks. Our rule for that is a little coarse, but honestly given the huge character space we're trying to play it safe here.

So summary:

• If a link is linkified, it's because the default linkifier identified it as a link
• If it's not linkified, it's because the default linkifier missed it or it violates one of the rules we setup to prevent homograph attacks.

That said, I still think the outcome for this ticket is the same: I don't anticipate we'll do more nuanced stuff here just to be safe. But apologies for the previous incorrect reasoning.

[stophecom]

Thanks for the explanation.
I would argue it's a nice trick and this rule definitely helps. But it's also a bit arbitrary. And with newer TLDs this "trick" might become less effective. E.g.

(You hardly notice that his is now within unicode-land)

And what about my emoji domain 😭😂😂😂. So sad.