Removing diacritics (accents) from String in Java

Yesterday I was working on uploading files in my CMS. The file should be downloadable using nice path containing file original file name. The problem is that not all file names can safely be used. For example I wouldn't like to use file name "Výroční zpráva 2007", but I would like to replace it with something like "vyrocni-zprava-2007".

Well how to do it in Java elegantly? I will skip replacing spaces with dashes and lower casing the name as it should be clear and simple. The naive way is to replace all characters containing diacritics using String.replaceAll() method similarly to replacing spaces. The code would look like this:

public static String getPath(String name) {

return name.toLowerCase().replaceAll("[\\s]+", "").

replaceAll("[áàâ]", "a").

replaceAll("[èéêëě]","e").

... a lot of other similar lines ...

}

Well this would essentially work. But when I was think of writing something like that, I was thinking like: "That would look pretty lame and PHP-like and Java is much more advanced language except everything else also with core Unicode support. There must be a way how to write it better."

Except looking lame, the above code has also several disadvantages. The most obvious it that you would have to explicitly name every character with diacritics and that can be difficult as many languages have their own special characters. For example Swedish has characters like 'å', 'ö', 'ä', or Czech (my native language) has characters like 'ě', 'š', 'ř', 'ž', 'ý', 'ú' and even such stupid character as 'ů'. You would have to know languages pretty well to be able to name all the characters.

The other disadvantage is that such code would run pretty slowly as the complexity grows with number of characters you are finding to replace. I don't know a lot of languages, but I guess the number of such characters can even hit 100 mark.

So because of that I decided to look for better solution and the obvious place to start was java.text package. And this is what I got:

// you need to import java.text.*

public static String getPath(String name) {

name = name.toLowerCase().replaceAll("[\\s]+", "");

Normalizer.normalize(name, Normalizer.Form.NFD).

replaceAll("\\p{InCombiningDiacriticalMarks}+", "");

}

So what does the code do? Firstly it replaces spaces and lower-cases the string. The second line decomposes every character with diacritics to more characters where the first is the character without diacritics, followed by characters representing the diacritics. As all characters representing diacritics are from specific Unicode range they can be easily replaced with single and simple replaceAll() method call.

So what are the disadvantages of the code? It works in 1.6 only. As alternative you can use sun.text.Normalizer available also in 1.5, however that doesn't lead to very portable code and would cause problems on Mac OS X for example.