1
|
package eu.dnetlib.data.mapreduce.hbase.lodExport.utils;
|
2
|
|
3
|
import java.net.URLDecoder;
|
4
|
import java.util.ArrayList;
|
5
|
import java.util.Set;
|
6
|
|
7
|
import com.google.common.collect.Multimap;
|
8
|
|
9
|
public class Blocking {
|
10
|
|
11
|
|
12
|
public static ArrayList<String> tokenBlocking(Multimap<String, String >fields){
|
13
|
ArrayList<String> blockingKeys = new ArrayList<>();
|
14
|
|
15
|
Set<String> keys = fields.keySet();
|
16
|
for(String key : keys){
|
17
|
|
18
|
if(key.equals("id")) continue;
|
19
|
|
20
|
for(String value : fields.get(key) ){
|
21
|
try{
|
22
|
value = URLDecoder.decode(value, "UTF-8").replaceAll("_", " ");
|
23
|
} catch (Exception ex) {
|
24
|
value = value.replaceAll("_", " ");
|
25
|
}
|
26
|
String[] tokens = value.split("[\\W_]");
|
27
|
|
28
|
for (int i = 0; i < tokens.length; i++) {
|
29
|
String currentToken = tokens[i].trim();
|
30
|
if (0 < currentToken.length()) {
|
31
|
blockingKeys.add(currentToken);
|
32
|
}
|
33
|
}
|
34
|
}
|
35
|
}
|
36
|
|
37
|
return blockingKeys;
|
38
|
}
|
39
|
|
40
|
}
|