issue mentioning two-part / hyphenated names

[ghost]

Hi,

I use humhub ver 1.1.2 and when i try to mention a user with a two part name, i can't find the user.

But i can find him if i start the mentionning with the second part of the name. For instance, if i try to mention Jean-Pierre Smith, typing @Jean-Pierre will show nothing; but typing @pierre will find the correct user. It's the same behaviour with two-part last names.

I did a quick search trough existing issues and saw nothing related to the subject. Is is still an existing issue on new versions? If not, at what point was the problem solved?

Thanks!

[ghost]

Some screenshots from the community humhub...

[Felli]

I think that if you start a @mention with the first name and then don't find your target, adding the last name will always find it. I believe this isn't exactly a bug, just somewhat of a flaw within mentioning. As for Hyphenated names, yes I do see the issue you're talking about.

Examples

First name

Exact name

Last name

Hyphenated name

After words a mentioning hangup happens.

[buddh4]

perhaps we just have to align the regex a bit

[ghost]

I'll give it a try when i have time / when i find the regex.

If i find a fix, it will be for version 1.1.2 but i'll let you know about it and i'm sure it won't need too much work to adapt to the new version.

On the other hand, if you could pinpoint the file and function/lines/bloc concerned, i'd be more than happy. I still spend way too much time to find the right code in the source before doing actual fixes.

Thanks for the quick feedback!

[buddh4]

Hmm I could fix it by adding - to this array in ZendLuceneSearch, will do some further testing tomorrow.

[Felli]

@buddh4 wouldn't it also be a better idea to use
$keyword = str_replace(array('-', '*', '?', '_', '$'), ' ', mb_strtolower($keyword, 'utf8mb4')); instead?

[ghost]

After my first tests, adding '-' to the array seems to work well on version 1.1.2 (yay!)

I haven't tried changing mb_strtolower($keyword, 'utf8mb4') though, i'll wait buddh4's opinions on the subject.

[buddh4]

think utf8mb4 is mysql specific

[Felli]

Using $keyword = str_replace(array('-', '*', '?', '_', '$'), ' ', mb_strtolower($keyword, 'utf-8')); showed no finds when testing in v1.2.2 for me, but switching over to $keyword = str_replace(array('-', '*', '?', '_', '$'), ' ', mb_strtolower($keyword, 'utf8mb4')); picked up searches, maybe make it optional?

[buddh4]

As I said utf8mb4 is mysql specific php supports the following encodings: http://php.net/manual/en/mbstring.supported-encodings.php

[Felli]

Hmm... Then I'll be doing some test to see why utf-8 isn't working on my server...