2 **********************************************************************
\r
3 * Copyright (c) 2001-2010, International Business Machines
\r
4 * Corporation and others. All Rights Reserved.
\r
5 **********************************************************************
\r
6 * Date Name Description
\r
7 * 11/29/2001 aliu Creation.
\r
8 * 06/26/2002 aliu Moved to com.ibm.icu.dev.tool.translit
\r
9 **********************************************************************
\r
11 package com.ibm.icu.dev.tool.translit;
\r
12 import java.io.File;
\r
13 import java.io.FileOutputStream;
\r
14 import java.io.IOException;
\r
15 import java.io.OutputStreamWriter;
\r
16 import java.io.PrintWriter;
\r
17 import java.util.Enumeration;
\r
19 import com.ibm.icu.text.Normalizer;
\r
20 import com.ibm.icu.text.Transliterator;
\r
21 import com.ibm.icu.text.UnicodeSet;
\r
24 * Class that generates source set information for a transliterator.
\r
28 * java com.ibm.icu.dev.tool.translit.SourceSet Latin-Katakana NFD lower
\r
30 * Output is produced in the command console, and a file with more detail is also written.
\r
32 * To see if it works, use:
\r
34 * java com.ibm.icu.dev.test.translit.TransliteratorTest -v -nothrow TestIncrementalProgress
\r
38 * java com.ibm.icu.dev.demo.translit.Demo
\r
40 public class SourceSet {
\r
42 public static void main(String[] args) throws IOException {
\r
43 if (args.length == 0) {
\r
44 // Compute and display the source sets for all system
\r
46 for (Enumeration e = Transliterator.getAvailableIDs(); e.hasMoreElements(); ) {
\r
47 String ID = (String) e.nextElement();
\r
48 showSourceSet(ID, Normalizer.NONE, false);
\r
51 // Usage: ID [NFKD | NFD] [lower]
\r
52 Normalizer.Mode m = Normalizer.NONE;
\r
53 boolean lowerFirst = false;
\r
54 if (args.length >= 2) {
\r
55 if (args[1].equalsIgnoreCase("NFD")) {
\r
57 } else if (args[1].equalsIgnoreCase("NFKD")) {
\r
58 m = Normalizer.NFKD;
\r
63 if (args.length >= 3) {
\r
64 if (args[2].equalsIgnoreCase("lower")) {
\r
70 if (args.length > 3) {
\r
73 showSourceSet(args[0], m, lowerFirst);
\r
77 static void showSourceSet(String ID, Normalizer.Mode m, boolean lowerFirst) throws IOException {
\r
78 File f = new File("UnicodeSetClosure.txt");
\r
79 String filename = f.getCanonicalFile().toString();
\r
80 out = new PrintWriter(
\r
81 new OutputStreamWriter(
\r
82 new FileOutputStream(filename), "UTF-8"));
\r
83 out.print('\uFEFF'); // BOM
\r
84 System.out.println();
\r
85 System.out.println("Writing " + filename);
\r
86 Transliterator t = Transliterator.getInstance(ID);
\r
87 showSourceSetAux(t, m, lowerFirst, true);
\r
88 showSourceSetAux(t.getInverse(), m, lowerFirst, false);
\r
92 static PrintWriter out;
\r
94 static void showSourceSetAux(Transliterator t, Normalizer.Mode m, boolean lowerFirst, boolean forward) {
\r
95 UnicodeSet sourceSet = t.getSourceSet();
\r
96 if (m != Normalizer.NONE || lowerFirst) {
\r
97 UnicodeSetClosure.close(sourceSet, m, lowerFirst);
\r
99 System.out.println(t.getID() + ": " +
\r
100 sourceSet.toPattern(true));
\r
101 out.println("# MINIMAL FILTER GENERATED FOR: " + t.getID() + (forward ? "" : " REVERSE"));
\r
103 + (forward ? "" : "( ")
\r
104 + sourceSet.toPattern(true)
\r
105 + (forward ? "" : " )")
\r
107 out.println("# Unicode: " + sourceSet.toPattern(false));
\r
111 static void usage() {
\r
112 System.err.println("Usage: ID [ NFD|NFKD [lower] ]");
\r