| 1 | |
|
| 2 | |
|
| 3 | |
package org.galagosearch.core.parse; |
| 4 | |
|
| 5 | |
import java.util.HashMap; |
| 6 | |
|
| 7 | |
|
| 8 | |
|
| 9 | |
|
| 10 | |
|
| 11 | |
|
| 12 | |
|
| 13 | |
|
| 14 | 16 | public class StringPooler { |
| 15 | 16 | HashMap<String, String> pool = new HashMap<String, String>(); |
| 16 | |
|
| 17 | |
|
| 18 | |
|
| 19 | |
|
| 20 | |
|
| 21 | |
|
| 22 | |
|
| 23 | |
public void transform(Document document) { |
| 24 | 92 | for (int i = 0; i < document.terms.size(); i++) { |
| 25 | 80 | String term = document.terms.get(i); |
| 26 | |
|
| 27 | 80 | if (term == null) { |
| 28 | 0 | continue; |
| 29 | |
} |
| 30 | 80 | String cached = pool.get(term); |
| 31 | |
|
| 32 | 80 | if (cached == null) { |
| 33 | 76 | term = new String(term); |
| 34 | 76 | pool.put(term, term); |
| 35 | |
} else { |
| 36 | 4 | term = cached; |
| 37 | |
} |
| 38 | |
|
| 39 | 80 | document.terms.set(i, term); |
| 40 | |
} |
| 41 | |
|
| 42 | |
|
| 43 | 12 | if (pool.size() > 10000) { |
| 44 | 0 | pool.clear(); |
| 45 | |
} |
| 46 | 12 | } |
| 47 | |
} |