Unicode breaks link detection

[kuukunen]

I hope this comes through the internet fine...

If you type something like this:
(　 ﾟ,_ゝﾟ) http://google.fi

It will get the start of the link incorrectly and clicking it will go to http://google.f
And the Problem will be for all links after the unicode string as long as it's on screen.
And adding multiple problematic characters will shift it further, so next it will try to go to: ) http://google.

[kuukunen]

Oh and I'd like to mention this is absolutely infuriating on IRC channels that tend to use silly unicode. Basically none of the links work by clicking. I know people who have switched to mintty+ssh just because of that one single issue.

[kuukunen]

Actually, I think it's that character causing problems: http://www.fileformat.info/info/unicode/char/3000/index.htm
As in "　"

[FauxFaux]

Yeah, I can reproduce this. I've had a quick look and can't see an immediate fix, and will be away for a few weeks...

[incognico]

I can reproduce this also.

Having one line is full of the unicode 0x2500 box drawing chars and every link below the line (no matter where unicode line is) will get 'detected' one line above the actual link. So I can not highlight the link hovering over it but the text in exactly one line above. See http://i.imgur.com/Ovi6a.png - it is really annoying :P

[FauxFaux]

Yeah, that's the same bug. The code is just so wrong that it's funny.

[FauxFaux]

@twpol's fix looks good; will test for a while then release.

[FauxFaux]

Right, apparently "∀" still breaks it. I'm guessing it's a double-width character that chooses to render as single width.
U+0500, Ԁ, Lu, CYRILLIC CAPITAL LETTER KOMI DE too.
益 the other way.

[twpol]

The problem with the original fix was that it worked entirely by chance on the particular Unicode face given; this new fix is wholy better. :)

[Berzerker]

Is there a built exe that someone can link for this? Or is there a link somewhere? Thanks.

[FauxFaux]

This should be fixed as of p0.62-t011. Any more issues, open new bugs!