Spellcheck and accents in 4.0.x

[cvidal]

Hi!

When I do spellcheck of a text with accents (like in French or Spanish), the accentuated characters are used as separators.

The problem is two-folded: the regexp in aspell.php and the text extracted from the webclient.

The first part is solved as in the WiKi, that is, the regexp used to convert the text to a list of words is:

$words = preg_split('/[^\w\'\xc0-\xfd-]+/', $text);

To test it, I can access the test page directly as:

http://localhost:7780/aspell.php

and fill a text with accents. This works fine.

The problem is the text passed as argument from Zimbra. The WebMail replaces the accents with "?", then, of course, a word like

exécuter

becomes

ex?cuter

which the PHP code correctly splits in the words "ex" and "cuter".

I traced the problem down to AjxStringUtil.convertHtml2Text(), but I don't see anything wrong with the code. Can it be the Element.toString() that is working in a weird way?

Any hints?

Carlos

[pshah]

I believe this is the bug you're seeing..?

The fix for this bug is targeted for the next major release. Regardless, you should vote for it

[claros]

Originally Posted by pshah

I believe this is the bug you're seeing..?

The fix for this bug is targeted for the next major release. Regardless, you should vote for it

Every non-english zimbra user should vote fo this bug...

[romulo]

I voted in it !!!


That is very important to the Zimbra function it here in BRAZIL (and others Latin countries).

These problems of accents would have to be seen with more affection. (It exists problems of accents in other parts of the ZIMBRA, not only in the spellcheck)

Therefore these problems make impracticable, many times the implementation of the ZIMBRA in these countries.

[dkarp]

If you see other accent-related bugs, please check bugzilla. If they're already reported, please vote for them! If they haven't been reported yet, please file a new bug!

[claros]

Originally Posted by dkarp

If you see other accent-related bugs, please check bugzilla. If they're already reported, please vote for them! If they haven't been reported yet, please file a new bug!

I've a question: which LANG setting should be used with zimbra?
Zimbra user have a LANG=C in .bash_profile but the system is also affected by the global LANG settings.
Should we foreign people use "en_US", "en_US.UTF-8" "C" or something like "it_IT" or "it-IT.UTF-8" ?

Claudio

[romulo]

Ok dkarp,

We are preparing the material contend all related BUGS the accents. As soon as it will be ready, will send for you.


Thanks
Rômulo Giordani Boschetti
Interop Informática - HSC

[cvidal]

Originally Posted by claros

I've a question: which LANG setting should be used with zimbra?
Zimbra user have a LANG=C in .bash_profile but the system is also affected by the global LANG settings.
Should we foreign people use "en_US", "en_US.UTF-8" "C" or something like "it_IT" or "it-IT.UTF-8" ?

Claudio

It doesn't matter (AFAIK). Tomcat pumps the data in the locale of the client browser. The spelling works with the locale requested by PHP in the aspell.php. The rest is LANG=C, used by all the little scripts run by cron and other utilities.

Carlos

[cvidal]

Originally Posted by pshah

I believe this is the bug you're seeing..?

The fix for this bug is targeted for the next major release. Regardless, you should vote for it

Yes, it is this bug, I voted too :-)

Carlos

[claros]

Originally Posted by cvidal

It doesn't matter (AFAIK). Tomcat pumps the data in the locale of the client browser. The spelling works with the locale requested by PHP in the aspell.php. The rest is LANG=C, used by all the little scripts run by cron and other utilities.

Carlos

I agree, tomcat does. But zimbra is not just tomcat and there are other programs (and bugs) that can be affected by LANG settings...
For example amavis, saslautd, perdition etc.
With zimbra 3.x it was suggested (at least in the forum) to set LANG=en-US, I'd like to know if there is a suggested setup now as I've found and reported a couple of bugs releted to internationalization end utf support...

Claudio