Best way to achieve multilingual URLs?

Question

I'm developing a niche social networking site that is going multilingual. That means our current URL structure will soon need to start using translated words for slugs like the following:

www.example.com/home becomes www.example.com/inicio

www.example.com/profile becomes www.example.com/perfil

www.example.com/help becomes www.example.com/ayuda

And so on. My question is: what's the best way to support this in a PHP application? For incoming requests, I thought a dictionary like the following in my router.php file would suffice:

<?php
$request = explode("/", trim($_SERVER['REQUEST_URI'], "/"));

// Dictionaries of page slugs.
$slugs = array(
    'es' => array(
        'inicio' => 'home',
        'perfil' => 'profile',
        'ayuda' => 'help',
    )
    // additional languages can be added here
);

// Rewrite any incoming (foreign) requests
if ($host=="www.example.es") { // to be made programmatic
    $lang = "es"; // pick up from locale constant rather being hard-coded
    if (array_key_exists($request[0], $slugs[$lang])) {
        $request[0] = $slugs[$lang][$request[0]];
    }
}

...

Which basically takes URL segments and matches them against an English counter-part if it exists. If not, then it will proceed as normal and most likely cause a 404 as a controller doesn't exist for URL segment.

Although this words, I need it to be backwards-compatible too. For example, when building URLs in my application.

Naturally, as the application is only English at the moment these are just hard-coded. So say, when fetching a User object I do the following:

<?php
class User {

    function __construct($id) {
        // fetch user details
        $this->profile_url = ROOT . "/profile/" . $this->username;
    }
}

What is the best method to then replace instances of "/profile/" being hard-coded to getting the translated version, i.e. "/perfil/" in the Spanish site?

Answer 1

Well, a common pattern I've seen for php is to create a php file for each language and initialize a large dictionary in which the keys are the same for all languages.

Have a session variable called language which can initially be 'en' for english (or whatever you prefer), and then include using command "include(language . '/main.php');" in which you have a folder called 'en' which contains all php files to include for translations. If main gets too large, you can subdivide and include whichever translation serves your needs (for example a /en/forum.php for forum translations and a /en/blob.php for front page translations).

It has the tremendous advantage of being flexible and allowing you to control the language simply by modifying one session variable. You can even do tricks like detect browser language settings and assign language according to that if it hasn't already been defined rather than simply making it english.

Answer 2

I could always be wrong, but here goes...

The standard way to achieve multilingual websites is to use i18n dictionary/template techniques in which you have a separate dictionary for each language, and in some cases different templates.

However, in such cases, I have never seen anybody change the language of their URL's. URL's map a request to files on the server disk (generally speaking), and there for shouldn't change based on language if you can avoid it.

It is common to prefix the 'path' section of your URL with the language you are requesting - ie: http://foo.bar/en-us/foobar.html

To summarize: I wouldn't worry about translating your URLs as it isn't a standard practice (atleast, not that I have seen). Simply prefix the URL 'path' with a language denotation such as in the URL above.