Trailing closing bracket in URL not considered part of the URL

[Brian Brazil]

Binary package hint: xfce4-terminal

When I'm on IRC, it's fairly common for someone to link to Wikipedia with a URL such as "http://en.wikipedia.org/wiki/Disambiguation_(audio)".

The problem is that the link when you hover or right click doesn't recognise the closing bracket as being part of the URL, instead going to "http://en.wikipedia.org/wiki/Disambiguation_(audio", which obviously doesn't work.

The terminal should recognise the closing bracket as being part of the URL.

I'm on Hardy.

[Brian Brazil (bbrazil)]

Aside: the equivalent bug in Launchpad itself is https://bugs.launchpad.net/ubuntu/+bug/118284

[Charlie Kravetz (cjkgeek)]

Thank you for taking the time to report this bug and helping to make Ubuntu better. Are there any applications that parse the final ")" ? I am not sure this is even possible to be to distinguish the closing brackets for the very reasons given in the launchpad bug referenced. The ones I tried do not.

[Brian Brazil (bbrazil)]

The launchpad bug gives no technical reason why this isn't possible, merely an assertion that this bug arises from the desire to handle a URL inside braces, and that this bug should be fixed in launchpad. Accordingly, I see no reason to mark this bug "Invalid".

[Ikonta (ikonta)]

XFce Terminal suggests to copy url into memory buffer.

If source string looks like:
<a>http://ya.ru/</a>
operation is correct.

But if url doesn't contain tailing '/', terminal copies +1 symbol (doesn't check close tag?).
I.e. if the source string is:
<a>http://ya.ru</a>
In the memory buffer is copied (and pasted) incorrect url string:
'http://ya.ru<'

P.S. If trying to copy url from href argument of 'a' tag both cases (i.e. present and missed tailing '/') works correctly.

[Jshipley-2 (jshipley-2)]

URL recognition seems to have a few problems (and seems to work mostly the same as gnome-terminal which isn't surprising since it looks like both applications use nearly identical regexes to match).

The opening character of a pair ('(', '<', '{', and '[') get added to the URL but terminate the URL. The exception to this is when they directly follow a '/' (not necessarily tailing, "http://www.foo.com/(whatisthis/more" would be recognized as "http://www.foo.com/"). This causes problems for wikipedia style links with '(' and ')' in them (https://bugzilla.xfce.org/show_bug.cgi?id=9800). The closing character of the pair does not get recognized as part of the URL.

Actually, it looks like any single character following a slash will be ignored.

http://www.foo.com/a matches http://www.foo.com/
http://www.foo.com/ab matches http://www.foo.com/ab

If the current regex can't be easily fixed, it may be worth looking into http://daringfireball.net/2010/07/improved_regex_for_matching_urls as a replacement.

I'm going to spend some time with this, and see if I can come up with a good patch.

[Ncopa (ncopa)]

Created attachment 5099
Patch from dev mailing list

this is the patch Jeff sent to mailing list.

It also fixes bug #9800 but I post it only here.

[Jshipley-2 (jshipley-2)]

Created attachment 5100
patch without duplicate USERS define

I recreated the above patch as it duplicated the "#define USERS" line instead of just changing it.

[8-nick (8-nick)]

*** Bug 9800 has been marked as a duplicate of this bug. ***

[8-nick (8-nick)]

Pushed in 2e1c325. Thanks.

[Thaddaeus Tintenfisch (thad-fisch-deactivatedaccount)]

The xfce4-terminal version provided by Ubuntu 14.04 already includes the patch for this bug.

http://git.xfce.org/apps/xfce4-terminal/commit/?id=2e1c32557bcb053c30e5c65eb6fe10929c5683d7