[BUG] Japanese text is misidentified as URL.

[mattn]

Describe the bug
Some Japanese text is unexpectedly misidentified as URL.

To Reproduce
Steps to reproduce the behavior:

1. Open input dialog
2. Type some Japanese characters contains 。 (\u3002)
3. Post the note
4. See the text is misidentified as URL.

Expected behavior
The text should be normal text.

Device (please complete the following information):

• Android Version: 0.37.4

This is related on linkedin/URL-Detector. linkedin/URL-Detector#39

URL-Detector handle 。 as dot. This is not a bug because IDN allow to use 。 as dot. However, most of Japanese text are often misidentified.

[afternooncurry]

Ideographic full stop and full width period are not handled as dot in IDNA2008 while IDNA2003 does. A current recommendation is IDNA2008. linkedin/URL-Detector looks implemented based on IDNA2003 about IDN which may causes the issue.

[mattn]

@vitorpamplona Many users in the East Asian region have been waiting for this fix for a long time.

[vitorpamplona]

We are waiting for the library to be able to support these additional characters. Until that library is fixed, there is not much we can do :(

[mattn]

Thanks. FYI, This issue can be reproduced with English speakers.

[vitorpamplona]

Interesting. That is a different "bug".

We have a separate procedure to linkify all yyy.xxx texts. It requires the domain separator . and should not affect the \u3002 character. Is that an issue for you? I see it here and there during the week but always just ignore it because it works most of the time.

[vitorpamplona]

And now that urls can have any Unicode character, I am not really sure how to solve it. Because the Thanks.<emoji> is a valid URL and could actually exist these days.

[mattn]

I don't make sure what is wrong in the code but this is rendered as URL in only amethyst.

[mattn]

Is this related on this issue?

[mattn]

I confirmed this issue is fixed in #491

Thanks.