annotate src/edu/unc/genomics/converters/RomanNumeralize.java @ 2:e16016635b2a

Uploaded
author timpalpant
date Mon, 13 Feb 2012 22:12:06 -0500
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
2
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
1 package edu.unc.genomics.converters;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
2
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
3 import java.io.BufferedReader;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
4 import java.io.BufferedWriter;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
5 import java.io.IOException;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
6 import java.nio.charset.Charset;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
7 import java.nio.file.Files;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
8 import java.nio.file.Path;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
9 import java.util.regex.Matcher;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
10 import java.util.regex.Pattern;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
11
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
12 import org.apache.log4j.Logger;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
13
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
14 import com.beust.jcommander.Parameter;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
15
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
16 import edu.unc.genomics.CommandLineTool;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
17 import edu.unc.genomics.ReadablePathValidator;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
18 import edu.unc.utils.RomanNumeral;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
19
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
20 public class RomanNumeralize extends CommandLineTool {
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
21
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
22 private static final Logger log = Logger.getLogger(RomanNumeralize.class);
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
23
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
24 @Parameter(names = {"-i", "--input"}, description = "Input file", required = true, validateWith = ReadablePathValidator.class)
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
25 public Path inputFile;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
26 @Parameter(names = {"-o", "--output"}, description = "Output file", required = true)
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
27 public Path outputFile;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
28
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
29 /**
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
30 * Pattern for finding "chr12" tokens (will find "chr1" through "chr99")
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
31 */
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
32 Pattern p = Pattern.compile("/chr[\\d]{1,2}/i");
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
33
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
34 @Override
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
35 public void run() throws IOException {
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
36 try (BufferedReader reader = Files.newBufferedReader(inputFile, Charset.defaultCharset());
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
37 BufferedWriter writer = Files.newBufferedWriter(outputFile, Charset.defaultCharset())) {
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
38 log.debug("Copying input to output and replacing with Roman Numerals");
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
39 String line;
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
40 while ((line = reader.readLine()) != null) {
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
41 Matcher m = p.matcher(line);
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
42 StringBuffer converted = new StringBuffer(line.length());
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
43 while (m.find()) {
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
44 String chrNum = line.substring(m.start()+3, m.end());
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
45 int arabic = Integer.parseInt(chrNum);
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
46 String roman = RomanNumeral.int2roman(arabic);
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
47 m.appendReplacement(converted, "chr"+roman);
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
48 }
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
49 m.appendTail(converted);
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
50
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
51 writer.write(converted.toString());
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
52 writer.newLine();
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
53 }
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
54 }
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
55 }
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
56
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
57 public static void main(String[] args) {
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
58 new RomanNumeralize().instanceMain(args);
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
59 }
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
60
e16016635b2a Uploaded
timpalpant
parents:
diff changeset
61 }