Unix FAQ

This page outlines how to configure Apache to automatically deliver pages in multiple languages. It covers how to do this using Auto ("Transparent") Content negotiation (RFC2295) which is available with both Apache 1.3 and 2.x.

Prerequistes

• Apache version 1.3 or higher.
• mod_negotiation ( this is compiled in by default )
• Server configured to allow 'Options Multiviews' ( see the AllowOverride directive )

Overview

Content negotiation is designed to deliver not just multiple languages but also multiple formats of documents - e.g. GIF or JPG, HTML or PDF. This tends to make the Apache documentation complex and possibly overwhelming for some. This document simply cover how to configure multiple languages automatically. All modern browsers can be configured to negiotiate for several languages, and retrieve from the server the most appriopiate for the user. When setting up your server you should be aware that many users only configure their browsers to accept ONLY their native language. If none of the browser's acceptable languages are available on your server the user will by default see the server's error 406 page - "No acceptable representation". There are two workarounds for this problem, covered below.

You need to identify the ISO639 codes for the languages you intend to support and the naming convention you intend to use. ISO639 codes are in the table below. I recommend you use this file naming convention;
    <resource>.<langCode>.<filetype>
So if you are providing English, French and German throughout your site you will likely have three top level index files named;

Note that whilst ISO639 language codes bear a strong resemblence to, they are not the same as ISO3166 country codes as used within DNS.

Non-Latin languages: If you are providing languages that are NOT based on Latin (European) character set - such as Chinese, Japanese, or Thai, then every such page should have the content-type HTTP header present. So for Thai;
<meta http-equiv="Content-Type" content="text/html; charset=tis-620">

Server Setup

In every directory containing pages in multiple languages you will need an .htaccess that contains at the minimum these two lines;
    Options MultiViews
    DirectoryIndex index
    
The Options directive tells the server to enable content negiotation for this directory. Each time the server processes a non-qualified resource name it will look for the files with the language codes from the browser. By 'Non-qualified' we mean that the GET request does not specify a full path name.
DirectoryIndex index tells the server to look for a resource named 'index' when the request ends with a trailing '/'. It is vital that this does not specify a fully qualified path name.

Writing your HTML

For multi-language support to work seamlessly and transparently across your site you need to be careful about specifying all interpage links. Specifically you must avoid links such as;
   <A HREF="/index.html">Home</a>
If you use the naming convention recommended this link is already broken.

The correct link to the site's home page be specified as
<A HREF="/">Home</a> or <A HREF="/index">Home</a>.

The downside to this is that most HTML authoring tools ( E.g. Dreamweaver ) will flags such links as broken.

Workarounds for Error 406

If you are using version 2.x of Apache you can utilise the ForceLanguagePriority directive to deliver the first language variant specified in LanguagePriority.

The only way of doing something with earlier versions is to trap the error via a ErrorDocument directive and either show the user a specific page or run a CGI-type script. Using the HTTP_REFERER and DOCUMENT_ROOT environment variables it is relatively straight forward to show the user the page in a language of your choosing.

ISO639 Language codes

• BabelFish,
• WorldLingo,
• English -> Thai
• InterTran