在PHP webapp中模板化不同的语言

Question

I'm writing a webapp in german, so all buttons,text,tooltips etc. are in german for now. But I want to use some kind of template file for the webapp so I can quick change to another language if needed. I thought about textfile that I explode with "\\n" and load into a sessionvariable to have always all text the user will need in his session. An other approach would be to parse such a file ie a XML document like this:

<?xml version="1.0"?>
<phrase>
    <placeholder></placeholder>
    <value></value>
</phrase>

where every field has it's own name/value that represents a textsnippet or button or whatever on the website, and then cast it into an object an cache it for everyone. I think the second approach is the best for working with multiple languages for a webapp. Anybody perhaps some pointers what I could do even better, or just post how you did this kind of languagetemplating for mutlinational webpages/webapps in the past.

Answer 1

Take a look at GNU Gettext , its very handy for multilanguage support.

The main idea is that you just wrap your words or phrases into a function, like

echo _g('Hello');

so you do not have any engine changes. You will have to add translation files for each language you are using.

Answer 2

Since you are looking for a translation solution, I understand you don't use a framework to develop your site, since most of them provide you with solutions to handle translations.

Most frameworks and apps I've seen in php use arrays, where the original sentence is the key and the translation is the value. So, to make easier to translate it to several languages, the key is in english.

In case you use gettext as suggested, or another aproach, it'll be useful also to parse your code to catch all strings to be translated automatically, since it can be a mess doing it manually when the base code grows and you want to keep up to date your translations.

Answer 3

You've come up with 2 solutions for storing the data, but I suggest you need to think further about the architecture and take a more complete view of the lifecycle of each request.

Regarding architecture: neither solution scales up to describe an extensive vocabulary very well - although for one or 2 pages it will suffice. The alternative approach, to manage a translation database (such as gettext) which might be overkill - and performs less optimally with small numbers of pages but importantly performance does not deteriorate significantly with large/multiple dictionaries. A compromise solution might be to have a dataset for each URL/language (which might be extracted from a consolidated database).

If it were me, I would not use either method you proposed for storing the data: parsing XML creates a sginficant overhead for each page request : using \\n as a delimiter precludes the use of \\n within a translation. Using a serialized PHP array seems to be the least expensive solution.