Options.java
2.13 KB
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
package is2.lemmatizer;
import java.io.File;
import java.io.IOException;
import is2.util.OptionsSuper;
public final class Options extends OptionsSuper {
public Options (String[] args) throws IOException {
for(int i = 0; i < args.length; i++) {
if (args[i].equals("--help")) explain();
if (args[i].equals("-normalize")) {
normalize=Boolean.parseBoolean(args[++i]);
} else if (args[i].equals("-features")) {
features= args[i+1]; i++;
} else if (args[i].equals("-hsize")) {
hsize= Integer.parseInt(args[i+1]); i++;
} else if (args[i].equals("-len")) {
maxLen= Integer.parseInt(args[i+1]); i++;
} else if (args[i].equals("-tmp")) {
tmp = args[i+1]; i++;
} else if (args[i].equals("-uc")) {
upper=true;
System.out.println("set uppercase "+upper);
} else super.addOption(args, i);
}
if (trainfile!=null) {
if (tmp!=null) trainforest = File.createTempFile("train", ".tmp", new File(tmp));
else trainforest = File.createTempFile("train", ".tmp"); //,new File("F:\\")
trainforest.deleteOnExit();
}
}
private void explain() {
System.out.println("Usage: ");
System.out.println("java -class mate.jar is2.lemmatizer.Lemmatizer [Options]");
System.out.println();
System.out.println("Options:");
System.out.println("");
System.out.println(" -train <file> the corpus a model is trained on; default "+this.trainfile);
System.out.println(" -test <file> the input corpus for testing; default "+this.testfile);
System.out.println(" -out <file> the output corpus (result) of a test run; default "+this.outfile);
System.out.println(" -model <file> the parsing model for traing the model is stored in the files");
System.out.println(" and for parsing the model is load from this file; default "+this.modelName);
System.out.println(" -i <number> the number of training iterations; good numbers are 10 for smaller corpora and 6 for bigger; default "+this.numIters);
System.out.println(" -count <number> the n first sentences of the corpus are take for the training default "+this.count);
System.exit(0);
}
}