Revision 48139
Added by Alessia Bardi almost 7 years ago
CleaningXsltFunctions.java | ||
---|---|---|
3 | 3 |
import java.text.Normalizer; |
4 | 4 |
|
5 | 5 |
public class CleaningXsltFunctions { |
6 |
|
|
6 | 7 |
public static String clean(final String s) { |
7 | 8 |
return Normalizer.normalize(s, Normalizer.Form.NFD) |
8 |
.replaceAll("\\(.+\\)", "") |
|
9 |
.replaceAll("(\\W|\\p{InCombiningDiacriticalMarks}|\\p{Punct}|\\n|\\s)+", "") |
|
10 |
.toLowerCase() |
|
11 |
.trim(); |
|
9 |
.replaceAll("\\(.+\\)", "")
|
|
10 |
.replaceAll("(\\W|\\p{InCombiningDiacriticalMarks}|\\p{Punct}|\\n|\\s)+", "")
|
|
11 |
.toLowerCase()
|
|
12 |
.trim();
|
|
12 | 13 |
} |
13 | 14 |
} |
Also available in: Unified diff
integrated (hopefully) all required changes from dnet40