Different Languages in PHP

Scott · Posted: Sat Mar 08, 2008 8:53 pm Post subject: Different Languages in PHP

I am working on making my script multi-lingual, it currently is only in English but people are also requesting to translate it into other languages.

First of all (although it is not a PHP issue) do I need to change the charset or something in the <head> for it to work with different languages?

Second, for the regex I currently have it just checks for a-zA-Z (when checking the letters). Although, if I use \w (word character) to check, will that return correctly for letters from another language?
_________________
Tutorial Management Script - Version 1.3 Released
TutorialToday - Up and running, submit your tutorials!
Linux Tutorials - Coming Soon

krt · ... Joined: 11 Jan 2005 Posts: 4607 Location: Australia

I'd use Unicode and send a header and use a meta tag for the best cross browser compatibility. Some don't deal with just the one.

In PHP: header("Content-Type: text/html; charset: utf-8");
(obviously changing text/html if needed)

Then mimicking this in HTML:

Scott · Posted: Sun Mar 09, 2008 8:39 am Post subject:

I would I go about validating titles with regex if they are in a different language.
_________________
Tutorial Management Script - Version 1.3 Released
TutorialToday - Up and running, submit your tutorials!
Linux Tutorials - Coming Soon

Pie32 · Posted: Sun Mar 09, 2008 8:52 am Post subject:

I'm not sure what version of PHP 6 we have on L2P servers, but that's something you may want to look into. One of the big features they are working on for it are support for languages with non-Latin characters.
_________________
[img]http://luneknight.com.ru/counter.jpg[/img]
Random Battle: [img]http://luneknight.com.ru/l.jpg[/img] vs. [img]http://luneknight.com.ru/r.jpg[/img]

Scott · Posted: Sun Mar 09, 2008 9:19 am Post subject:

LP-SolidRaven · Posted: Tue Mar 11, 2008 7:43 am Post subject:

You could use \xhexhere
Replace hexhere with the hex code of the character.
_________________

leontius · Novice Poster Joined: 26 Mar 2008 Posts: 2

'\w' in regex corresponds to the locale set in the server. PHP 5 manual said that

ClickFanatic · Est. 2005 Joined: 18 Jan 2005 Posts: 3857

The problem with locales is that they have to be present on the server. Some servers simply have the en_US locale and nothing else.

But as I don't really see alternatives (unless you want to manually create a regex for every language, using \x## patterns) it's probably worth the risk.
_________________
Captain Jell-O Buster from the Future
[img]http://feeds.feedburner.com/sparepencil.1.gif[/img]

linuxdoctor · Posted: Wed Mar 26, 2008 8:45 am Post subject:

the E107 content management system has an interesting way to handle different languages. It's written entirely in PHP so perhaps this is something you might want to look at.

http://e107.org
http://e107coders.org

To quickly describe what they do for different languages.

First they put all of the text that is to be output in strings in different language files.

LP-SolidRaven · Posted: Wed Mar 26, 2008 10:45 am Post subject:

That's a good idea until you start with an extensive template system...
_________________

ClickFanatic · Est. 2005 Joined: 18 Jan 2005 Posts: 3857

Defining constants is quite a good solution, but not very elegant in my opinion. Especially when the number of different phrases becomes really big (consider this constant USER_PASSWORD_CHARACTER_MAXLENGTH_EXCEEDED).

Also, if the translations are outdated, or certain phrases remain untranslated, what will happen? You will see the constant (which is undefined) in plain text. Not really nice.

A better solution is the PHP implementation of gettext: http://savannah.nongnu.org/projects/php-gettext/

It is used in WordPress, among others. It basically works like this:
Throughout the code the programmer can use the default language to write messages. For example: Your password exceeds the maximum length

To make it possible to translate this text, it will be wrapped in a function.

linuxdoctor · Posted: Wed Mar 26, 2008 4:29 pm Post subject:

That could be a good way. The particular gettext you pointed to is not the GNU gettext but something different. PHP supports the standard GNU gettext.

I'm not that familiar with WordPress but if it does use gettext it would most likely be the GNU version.

The biggest disadvantage is the requirement of the use the .po files. These files have a fairly complex structure that will require a learning curve which might be somewhat of a deterrent for people not familar with Unix conventions.

Another disadvantage is speed. Gettext in PHP is slow. Everything in PHP is slow. PHP is an interpreted language. First the Zend compiler reads all the source files in and converts it into an intermediate form before executing it. The macro implementation is very fast since the translation is done in situ by the Zend compiler as the source is read in. Using gettext would not cause the translations to actually be done until the execution phase and the net result might be a noticable slowness in response or a jerkiness in rendering pages.

While gettext provide the most general solution, it was designed for compiled to object programmes. In PHP it is cumbersome and slow. In my view, that hardly qualifies it as elegant.
_________________
Misanthrope: someone who realizes that humans really are as stupid as they appear.

If you think I'm 'politically' incorrect you have the wrong politics.

LP-SolidRaven · Posted: Thu Mar 27, 2008 3:05 am Post subject:

There is a fast way to use GNU Gettext in php.
You might want to read this. I've used it before and it's quite fast. The only issue is that most servers don't have it compiled. Haven't checked if LLP has it. And if it doesn't I don't think Trel would mind installing it. Oh yeah keep in mind the documentation of this module is terrible in several ways. If I find some time I might write a small tutorial on it.

And if the site is intended to work under extreme high loads you could always use a squid cache server combined with an accelerator.
_________________

linuxdoctor · Posted: Thu Mar 27, 2008 7:21 am Post subject:

You're sort of making my point for me. I agree with you that gettext is a very general solution to the problem of multi-lingual translations for applications. Gettext is widely used in Linux apps and there is a lot to say for it.

However, there are also weaknesses and we've pointed them out. Another weakness with a lot of GNU software, including gettext, is that in its endeavour to be general it also becomes extremely complex. Some critics have said needlessly complex. On the other hand, being less complex would also mean being less general and less complete necessitating the developer to add other libraries and modules to add additional functionality that would be missing from a more general solution. In the end, the application becomes even more complex than it otherwise would have been and and almost certainly much larger and slower.

What I suggested was a simple approach. Certainly it's not the best but it's simple and it works. While one programmer is spending a lot of time trying to get gettext to work another programmer could be up in running with my proposed solution.
_________________
Misanthrope: someone who realizes that humans really are as stupid as they appear.

If you think I'm 'politically' incorrect you have the wrong politics.

ClickFanatic · Est. 2005 Joined: 18 Jan 2005 Posts: 3857

The PHP gettext that I referred to is indeed not the GNU gettext extension that may be installed on servers.
Portability would be a valid reason to use the PHP gettext implementation instead of the GNU gettext extension for PHP. WordPress doesn't rely on the PHP extension, but uses a modified version of PHP gettext.
_________________
Captain Jell-O Buster from the Future
[img]http://feeds.feedburner.com/sparepencil.1.gif[/img]

	Lifelesspeople.com Forum Index -> Web Architects' Abode	All times are GMT - 6 Hours
Page 1 of 1