Replace Unicode Control Characters

Cyril Gandon picture Cyril Gandon · Aug 9, 2010 · Viewed 13.5k times · Source

I need to replace all special control character in a string in Java.

I want to ask the Google maps API v3, and Google doesn't seems to like these characters.

Example: http://www.google.com/maps/api/geocode/json?sensor=false&address=NEW%20YORK%C2%8F

This URL contains this character: http://www.fileformat.info/info/unicode/char/008f/index.htm

So I receive some data, and I need to geocode this data. I know some character would not pass the geocoding, but I don't know the exact list.

I was not able to find any documentation about this issue, so I think the list of characters that Google doesn't like is this one: http://www.fileformat.info/info/unicode/category/Cc/list.htm

Is there any already built function to get rid of these characters, or do I have to build a new one, with a replace one by one?

Or is there a good regexp to do the job done?

And does somebody know which exact list of characters Google doesn't like?

Edit : Google have create a webpage for this :

https://developers.google.com/maps/documentation/webservices/?hl=fr#BuildingURLs

Answer

polygenelubricants picture polygenelubricants · Aug 9, 2010

If you want to delete all characters in Other/Control Unicode category, you can do something like this:

    System.out.println(
        "a\u0000b\u0007c\u008fd".replaceAll("\\p{Cc}", "")
    ); // abcd

Note that this actually removes (among others) '\u008f' Unicode character from the string, not the escaped form "%8F" string.

If the blacklist is not nicely captured by one Unicode block/category, Java does have a powerful character class arithmetics featuring intersection, subtraction, etc that you can use. Alternatively you can also use a negated whitelist approach, i.e. instead of explicitly specifying what characters are illegal, you specify what are legal, and everything else then becomes illegal.

API links


Examples

Here's a subtraction example:

    System.out.println(
        "regular expressions: now you have two problems!!"
            .replaceAll("[a-z&&[^aeiou]]", "_")
    );
    //   _e_u_a_ e___e__io__: _o_ _ou _a_e __o __o__e__!!

The […] is a character class. Something like [aeiou] matches one of any of the lowercase vowels. [^…] is a negated character class. [^aeiou] matches one of anything but the lowercase vowels.

[a-z&&[^aeiou]] matches [a-z] subtracted by [aeiou], i.e. all lowercase consonants.

The next example shows the negated whitelist approach:

    System.out.println(
        "regular expressions: now you have two problems!!"
            .replaceAll("[^a-z]", "_")
    );
    //   regular_expressions__now_you_have_two_problems__

Only lowercase letters a-z are legal; everything else is illegal.