annotate software/mpdl-services/mpiwg-mpdl-lt/src/de/mpg/mpiwg/berlin/mpdl/lt/text/norm/Normalizer.java @ 19:4a3641ae14d2

Erstellung
author Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
date Wed, 09 Nov 2011 15:32:05 +0100
parents
children 7d6d969b10cf
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
19
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1 package de.mpg.mpiwg.berlin.mpdl.lt.text.norm;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
2
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
3 import java.io.IOException;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
4 import java.io.StringReader;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
5 import java.util.ArrayList;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
6
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
7 import de.mpg.mpiwg.berlin.mpdl.exception.ApplicationException;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
8 import de.mpg.mpiwg.berlin.mpdl.lt.general.Language;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
9 import de.mpg.mpiwg.berlin.mpdl.lt.text.norm.lang.MpdlNormalizerLexAR;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
10 import de.mpg.mpiwg.berlin.mpdl.lt.text.norm.lang.MpdlNormalizerLexDE;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
11 import de.mpg.mpiwg.berlin.mpdl.lt.text.norm.lang.MpdlNormalizerLexEL;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
12 import de.mpg.mpiwg.berlin.mpdl.lt.text.norm.lang.MpdlNormalizerLexEN;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
13 import de.mpg.mpiwg.berlin.mpdl.lt.text.norm.lang.MpdlNormalizerLexFR;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
14 import de.mpg.mpiwg.berlin.mpdl.lt.text.norm.lang.MpdlNormalizerLexIT;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
15 import de.mpg.mpiwg.berlin.mpdl.lt.text.norm.lang.MpdlNormalizerLexLA;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
16 import de.mpg.mpiwg.berlin.mpdl.lt.text.norm.lang.MpdlNormalizerLexNL;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
17 import de.mpg.mpiwg.berlin.mpdl.lt.text.norm.lang.MpdlNormalizerLexZH;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
18 import de.mpg.mpiwg.berlin.mpdl.lt.text.reg.Regularization;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
19 import de.mpg.mpiwg.berlin.mpdl.lt.text.reg.RegularizationManager;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
20
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
21 public class Normalizer {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
22 public static int DISPLAY = 1; // normalization in DISPLAY mode
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
23 public static int DICTIONARY = 2; // normalization in DICTIONARY mode
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
24 public static int SEARCH = 3; // normalization in SEARCH mode; never used so far in indexing because it does not support the morph. lexicons such as CELEX (e.g. eingeschränkt would not be stemmed to eingeschraenkt)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
25 private int normMode = DISPLAY; // Default e.g. for indexing and querying
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
26 private String[] normFunctions = {"norm"}; // default is to use the norm function
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
27 private String language;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
28 private int[] offsets;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
29
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
30 public Normalizer(String[] normFunctions, String lang) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
31 this.normFunctions = normFunctions;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
32 String language = Language.getInstance().getLanguageId(lang);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
33 this.language = language;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
34 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
35
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
36 public Normalizer(String language) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
37 this.language = language;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
38 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
39
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
40 public String getLanguage() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
41 return language;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
42 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
43
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
44 public void setNormMode(int normMode) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
45 this.normMode = normMode;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
46 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
47
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
48 /**
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
49 * Applies the normalization rules in <code>language</code> to
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
50 * <code>s</code>, without offset tracking.
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
51 *
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
52 * @param s source string
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
53 * @return normalized string
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
54 */
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
55 public String normalize(String s) throws ApplicationException {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
56 String normStr = s;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
57 if (useSpecialNormFunction())
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
58 normStr = removeSpecialNWDMarks(normStr);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
59 if (useRegFunction()) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
60 // try to regularize the string to the norm form over predefined regularizations
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
61 RegularizationManager regManager = RegularizationManager.getInstance();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
62 ArrayList<Regularization> regs = regManager.findRegsByOrig(language, s);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
63 if (regs != null && regs.size() > 0) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
64 Regularization reg = regs.get(0); // only one: the first one
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
65 String regNormStr = reg.getNorm();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
66 normStr = regNormStr;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
67 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
68 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
69 if (useNormFunction()) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
70 // normalize the string by string replacements
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
71 if (normMode == DICTIONARY) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
72 normStr = normalize(normStr, DICTIONARY);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
73 } else if (normMode == DISPLAY) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
74 normStr = normalize(normStr, DISPLAY);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
75 } else if (normMode == SEARCH) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
76 normStr = normalize(normStr, SEARCH);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
77 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
78 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
79 if (useSpecialNormFunction())
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
80 normStr = insertSpecialNWDMarks(normStr);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
81 return normStr;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
82 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
83
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
84 private boolean useRegFunction() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
85 boolean useReg = false;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
86 for (int i=0; i< normFunctions.length; i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
87 String function = normFunctions[i];
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
88 if (function.equals("reg"))
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
89 return true;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
90 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
91 return useReg;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
92 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
93
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
94 private boolean useNormFunction() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
95 boolean useNorm = false;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
96 for (int i=0; i< normFunctions.length; i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
97 String function = normFunctions[i];
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
98 if (function.equals("norm") || function.equals("specialNorm"))
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
99 return true;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
100 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
101 return useNorm;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
102 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
103
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
104 private boolean useSpecialNormFunction() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
105 boolean useNorm = false;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
106 for (int i=0; i< normFunctions.length; i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
107 String function = normFunctions[i];
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
108 if (function.equals("specialNorm"))
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
109 return true;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
110 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
111 return useNorm;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
112 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
113
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
114 private String normalize(String s, int mode) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
115 String inputStr = s;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
116 StringReader strReader = new StringReader(inputStr + "\n");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
117 String retStr = "";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
118 String token = "";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
119 try {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
120 if (Language.getInstance().isLatin(language)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
121 MpdlNormalizerLexLA mpdlNormalizerLex = new MpdlNormalizerLexLA(strReader);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
122 if (mode == DISPLAY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
123 mpdlNormalizerLex.yybegin(MpdlNormalizerLexLA.DISP);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
124 else if (mode == DICTIONARY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
125 mpdlNormalizerLex.yybegin(MpdlNormalizerLexLA.DICT);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
126 else if (mode == SEARCH)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
127 mpdlNormalizerLex.yybegin(MpdlNormalizerLexLA.SEARCH);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
128 while (token != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
129 token = mpdlNormalizerLex.yylex();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
130 if (token != null)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
131 retStr += token;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
132 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
133 } else if (Language.getInstance().isArabic(language)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
134 MpdlNormalizerLexAR mpdlNormalizerLex = new MpdlNormalizerLexAR(strReader);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
135 if (mode == DISPLAY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
136 mpdlNormalizerLex.yybegin(MpdlNormalizerLexAR.DISP);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
137 else if (mode == DICTIONARY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
138 mpdlNormalizerLex.yybegin(MpdlNormalizerLexAR.DICT);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
139 else if (mode == SEARCH)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
140 mpdlNormalizerLex.yybegin(MpdlNormalizerLexAR.SEARCH);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
141 while (token != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
142 token = mpdlNormalizerLex.yylex();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
143 if (token != null)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
144 retStr += token;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
145 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
146 } else if (Language.getInstance().isGerman(language)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
147 MpdlNormalizerLexDE mpdlNormalizerLex = new MpdlNormalizerLexDE(strReader);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
148 if (mode == DISPLAY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
149 mpdlNormalizerLex.yybegin(MpdlNormalizerLexDE.DISP);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
150 else if (mode == DICTIONARY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
151 mpdlNormalizerLex.yybegin(MpdlNormalizerLexDE.CELEX);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
152 else if (mode == SEARCH)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
153 mpdlNormalizerLex.yybegin(MpdlNormalizerLexDE.SEARCH);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
154 while (token != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
155 token = mpdlNormalizerLex.yylex();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
156 if (token != null)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
157 retStr += token;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
158 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
159 } else if (Language.getInstance().isGreek(language)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
160 MpdlNormalizerLexEL mpdlNormalizerLex = new MpdlNormalizerLexEL(strReader);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
161 if (mode == DISPLAY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
162 mpdlNormalizerLex.yybegin(MpdlNormalizerLexEL.DISP);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
163 else if (mode == DICTIONARY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
164 mpdlNormalizerLex.yybegin(MpdlNormalizerLexEL.SIGMA);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
165 else if (mode == SEARCH)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
166 mpdlNormalizerLex.yybegin(MpdlNormalizerLexEL.SEARCH);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
167 while (token != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
168 token = mpdlNormalizerLex.yylex();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
169 if (token != null)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
170 retStr += token;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
171 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
172 } else if (Language.getInstance().isEnglish(language)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
173 MpdlNormalizerLexEN mpdlNormalizerLex = new MpdlNormalizerLexEN(strReader);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
174 if (mode == DISPLAY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
175 mpdlNormalizerLex.yybegin(MpdlNormalizerLexEN.DISP);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
176 else if (mode == DICTIONARY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
177 mpdlNormalizerLex.yybegin(MpdlNormalizerLexEN.DICT);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
178 else if (mode == SEARCH)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
179 mpdlNormalizerLex.yybegin(MpdlNormalizerLexEN.SEARCH);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
180 while (token != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
181 token = mpdlNormalizerLex.yylex();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
182 if (token != null)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
183 retStr += token;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
184 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
185 } else if (Language.getInstance().isFrench(language)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
186 MpdlNormalizerLexFR mpdlNormalizerLex = new MpdlNormalizerLexFR(strReader);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
187 if (mode == DISPLAY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
188 mpdlNormalizerLex.yybegin(MpdlNormalizerLexFR.DISP);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
189 else if (mode == DICTIONARY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
190 mpdlNormalizerLex.yybegin(MpdlNormalizerLexFR.DICT_ASCII);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
191 else if (mode == SEARCH)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
192 mpdlNormalizerLex.yybegin(MpdlNormalizerLexFR.SEARCH);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
193 while (token != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
194 token = mpdlNormalizerLex.yylex();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
195 if (token != null)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
196 retStr += token;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
197 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
198 } else if (Language.getInstance().isItalian(language)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
199 MpdlNormalizerLexIT mpdlNormalizerLex = new MpdlNormalizerLexIT(strReader);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
200 if (mode == DISPLAY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
201 mpdlNormalizerLex.yybegin(MpdlNormalizerLexIT.DISP);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
202 else if (mode == DICTIONARY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
203 mpdlNormalizerLex.yybegin(MpdlNormalizerLexIT.DICT);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
204 else if (mode == SEARCH)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
205 mpdlNormalizerLex.yybegin(MpdlNormalizerLexIT.SEARCH);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
206 while (token != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
207 token = mpdlNormalizerLex.yylex();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
208 if (token != null)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
209 retStr += token;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
210 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
211 } else if (Language.getInstance().isDutch(language)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
212 MpdlNormalizerLexNL mpdlNormalizerLex = new MpdlNormalizerLexNL(strReader);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
213 if (mode == DISPLAY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
214 mpdlNormalizerLex.yybegin(MpdlNormalizerLexNL.DISP);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
215 else if (mode == DICTIONARY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
216 mpdlNormalizerLex.yybegin(MpdlNormalizerLexNL.DICT);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
217 else if (mode == SEARCH)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
218 mpdlNormalizerLex.yybegin(MpdlNormalizerLexNL.SEARCH);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
219 while (token != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
220 token = mpdlNormalizerLex.yylex();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
221 if (token != null)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
222 retStr += token;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
223 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
224 } else if (Language.getInstance().isChinese(language)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
225 MpdlNormalizerLexZH mpdlNormalizerLex = new MpdlNormalizerLexZH(strReader);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
226 if (mode == DISPLAY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
227 mpdlNormalizerLex.yybegin(MpdlNormalizerLexZH.DISP);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
228 else if (mode == DICTIONARY)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
229 mpdlNormalizerLex.yybegin(MpdlNormalizerLexZH.DICT);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
230 else if (mode == SEARCH)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
231 mpdlNormalizerLex.yybegin(MpdlNormalizerLexZH.SEARCH);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
232 while (token != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
233 token = mpdlNormalizerLex.yylex();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
234 if (token != null)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
235 retStr += token;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
236 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
237 } else {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
238 retStr = s; // return the string unchanged
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
239 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
240 } catch (IOException e ) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
241 // nothing cause IOException is not needed for a StringReader
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
242 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
243 return retStr;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
244 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
245
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
246
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
247 // used only in XmlTokenizerContentHandler // TODO make it better
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
248 private String removeSpecialNWDMarks(String inputString) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
249 String COMPLEX_ELEMENT_NWD_MARK = new Character('\u2424').toString(); // not word delimiting element
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
250 String cleanedWord = inputString;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
251 boolean startsWithNWDMark = cleanedWord.startsWith(COMPLEX_ELEMENT_NWD_MARK);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
252 if (startsWithNWDMark)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
253 cleanedWord = cleanedWord.substring(1);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
254 int countNWDMarks = cleanedWord.length() - cleanedWord.replaceAll(COMPLEX_ELEMENT_NWD_MARK, "").length();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
255 if (countNWDMarks > 1)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
256 cleanedWord = cleanedWord.replaceAll(COMPLEX_ELEMENT_NWD_MARK + "+", COMPLEX_ELEMENT_NWD_MARK);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
257 // boolean notHyphenPlusNWD = cleanedWord.matches(".*[^-]+" + COMPLEX_ELEMENT_NWD_MARK + "+.*"); // e.g. "praebi ta"
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
258 // if (notHyphenPlusNWD)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
259 // cleanedWord = cleanedWord.replaceAll("([^-]+)" + COMPLEX_ELEMENT_NWD_MARK + "+", "$1-" + COMPLEX_ELEMENT_NWD_MARK); // e.g. "praebi ta" is replaced by "praebi- ta"
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
260 cleanedWord = cleanedWord.replaceAll(COMPLEX_ELEMENT_NWD_MARK, " ");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
261 return cleanedWord;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
262 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
263
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
264 private String insertSpecialNWDMarks(String inputString) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
265 String COMPLEX_ELEMENT_NWD_MARK = new Character('\u2424').toString(); // not word delimiting element
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
266 String retStr = inputString;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
267 boolean startsWithNWDMark = retStr.startsWith(COMPLEX_ELEMENT_NWD_MARK);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
268 int countNWDMarks = retStr.length() - retStr.replaceAll(COMPLEX_ELEMENT_NWD_MARK, "").length();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
269 retStr = retStr.replaceAll(" ", COMPLEX_ELEMENT_NWD_MARK);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
270 // if (notHyphenPlusNWD)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
271 // normalizedWordStr = normalizedWordStr.replaceAll("-" + COMPLEX_ELEMENT_NWD_MARK, COMPLEX_ELEMENT_NWD_MARK); // e.g. "praebi- ta" is replaced by "praebi ta"
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
272 if (countNWDMarks > 1) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
273 String nwdStr = "";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
274 for (int i=0; i<countNWDMarks; i++)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
275 nwdStr += COMPLEX_ELEMENT_NWD_MARK;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
276 retStr = retStr.replaceAll(COMPLEX_ELEMENT_NWD_MARK, nwdStr);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
277 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
278 if (startsWithNWDMark)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
279 retStr = COMPLEX_ELEMENT_NWD_MARK + retStr;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
280 return retStr;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
281 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
282
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
283 /**
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
284 * Old code from Arboreal (Malcolm Hyman)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
285 * Applies the normalization rules in <code>language</code> to
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
286 * <code>s</code>, with offset tracking.<p>
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
287 *
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
288 * <strong>WARNING:</strong>
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
289 * Arboreal will not work properly if a normalization substitution
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
290 * replaces a source character with more than two target characters!
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
291 * This is simply a BUG, and should be fixed. Fortunately, however,
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
292 * one does not often need such a replacement.<p>
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
293 *
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
294 * @param s source string
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
295 * @param offsets character offset table
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
296 * @return normalized string
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
297 */
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
298 private String normalize4Lexica(String s, int[] offsets) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
299 this.offsets = offsets;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
300 if (language.equals("la") || language.equals("lat")) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
301 StringBuffer buf = new StringBuffer();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
302 int n = 0;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
303 for (int i = 0; i < s.length(); i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
304 char c = s.charAt(i);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
305 String replace = new String();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
306 switch (c) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
307 case 'j': replace = "i"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
308 case 'v': replace = "u"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
309 /*
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
310 * Linguistic note: /u/ and /v/ are rarely phonemic
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
311 * in Latin, as in alui 's/he nourished' vs.
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
312 * alvi 'of a belly', volui 's/he wished' or 'it rolled'
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
313 * vs. volvi 'to be rolled', (in)seruit 's/he joined
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
314 * together' vs. (in)servit 's/he serves'.
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
315 */
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
316 case 'q':
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
317 if ((i < s.length() - 1) && (s.charAt(i + 1) == ';'))
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
318 replace = "qu";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
319 else
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
320 replace = "q";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
321 break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
322 case ';':
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
323 if ((i > 0) && (s.charAt(i - 1) == 'q'))
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
324 replace = "e";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
325 else if ((i == 0) || ! Character.isLetter(s.charAt(i - 1)))
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
326 replace = ";";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
327 else
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
328 replace = "";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
329 break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
330 case '\u0300': replace = ""; break; // COMBINING GRAVE ACCENT
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
331 case '\u0301': replace = ""; break; // COMBINING ACCUTE ACCENT
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
332 case '\u0302': replace = ""; break; // COMBINING CIRCUMFLEX ACCENT
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
333
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
334 case '\u00c0': replace = "A"; break; // LATIN CAPITAL LETTER A GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
335 case '\u00c1': replace = "A"; break; // LATIN CAPITAL LETTER A ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
336 case '\u00c2': replace = "A"; break; // LATIN CAPITAL LETTER A CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
337 case '\u00c4': replace = "A"; break; // LATIN CAPITAL LETTER A DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
338 case '\u00c6': replace = "Ae"; break; // LATIN CAPITAL LETTER A E
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
339 case '\u00c7': replace = "C"; break; // LATIN CAPITAL LETTER C CEDILLA
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
340 case '\u00c8': replace = "E"; break; // LATIN CAPITAL LETTER E GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
341 case '\u00c9': replace = "E"; break; // LATIN CAPITAL LETTER E ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
342 case '\u00ca': replace = "E"; break; // LATIN CAPITAL LETTER E CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
343 case '\u00cb': replace = "E"; break; // LATIN CAPITAL LETTER E DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
344 case '\u00cc': replace = "I"; break; // LATIN CAPITAL LETTER I GRAVE;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
345 case '\u00cd': replace = "I"; break; // LATIN CAPITAL LETTER I ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
346 case '\u00ce': replace = "I"; break; // LATIN CAPITAL LETTER I CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
347 case '\u00cf': replace = "I"; break; // LATIN CAPITAL LETTER I DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
348 case '\u00d2': replace = "O"; break; // LATIN CAPITAL LETTER O GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
349 case '\u00d3': replace = "O"; break; // LATIN CAPITAL LETTER O ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
350 case '\u00d4': replace = "O"; break; // LATIN CAPITAL LETTER O CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
351 case '\u00d6': replace = "O"; break; // LATIN CAPITAL LETTER O DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
352 case '\u00d9': replace = "U"; break; // LATIN CAPITAL LETTER U GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
353 case '\u00da': replace = "U"; break; // LATIN CAPITAL LETTER U ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
354 case '\u00db': replace = "U"; break; // LATIN CAPITAL LETTER U CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
355 case '\u00dc': replace = "U"; break; // LATIN CAPITAL LETTER U DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
356 case '\u00e0': replace = "a"; break; // LATIN SMALL LETTER A GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
357 case '\u00e1': replace = "a"; break; // LATIN SMALL LETTER A ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
358 case '\u00e2': replace = "a"; break; // LATIN SMALL LETTER A CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
359 case '\u00e4': replace = "a"; break; // LATIN SMALL LETTER A DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
360 case '\u00e6': replace = "ae"; break; // LATIN SMALL LETTER A E
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
361 case '\u00e7': replace = "c"; break; // LATIN SMALL LETTER C CEDILLA
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
362 case '\u00e8': replace = "e"; break; // LATIN SMALL LETTER E GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
363 case '\u00e9': replace = "e"; break; // LATIN SMALL LETTER E ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
364 case '\u00ea': replace = "e"; break; // LATIN SMALL LETTER E CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
365 case '\u00eb': replace = "e"; break; // LATIN SMALL LETTER E DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
366 case '\u00ec': replace = "i"; break; // LATIN SMALL LETTER I GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
367 case '\u00ed': replace = "i"; break; // LATIN SMALL LETTER I ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
368 case '\u00ee': replace = "i"; break; // LATIN SMALL LETTER I CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
369 case '\u00ef': replace = "i"; break; // LATIN SMALL LETTER I DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
370 case '\u00f2': replace = "o"; break; // LATIN SMALL LETTER O GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
371 case '\u00f3': replace = "o"; break; // LATIN SMALL LETTER O ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
372 case '\u00f4': replace = "o"; break; // LATIN SMALL LETTER O CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
373 case '\u00f6': replace = "o"; break; // LATIN SMALL LETTER O DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
374 case '\u00f9': replace = "u"; break; // LATIN SMALL LETTER U GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
375 case '\u00fa': replace = "u"; break; // LATIN SMALL LETTER U ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
376 case '\u00fb': replace = "u"; break; // LATIN SMALL LETTER U CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
377 case '\u00fc': replace = "u"; break; // LATIN SMALL LETTER U DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
378 case '\u0100': replace = "A"; break; // LATIN CAPITAL LETTER A MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
379 case '\u0101': replace = "a"; break; // LATIN SMALL LETTER A MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
380 case '\u0102': replace = "A"; break; // LATIN CAPITAL LETTER A BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
381 case '\u0103': replace = "a"; break; // LATIN SMALL LETTER A BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
382 case '\u0112': replace = "E"; break; // LATIN CAPITAL LETTER E MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
383 case '\u0113': replace = "e"; break; // LATIN SMALL LETTER E MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
384 case '\u0114': replace = "E"; break; // LATIN CAPITAL LETTER E BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
385 case '\u0115': replace = "e"; break; // LATIN SMALL LETTER E BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
386 case '\u0118': replace = "Ae"; break; // LATIN CAPITAL LETTER E OGONEK
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
387 case '\u0119': replace = "ae"; break; // LATIN SMALL LETTER E OGONEK
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
388 case '\u012a': replace = "I"; break; // LATIN CAPITAL LETTER I MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
389 case '\u012b': replace = "i"; break; // LATIN SMALL LETTER I MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
390 case '\u012c': replace = "I"; break; // LATIN CAPITAL LETTER I BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
391 case '\u012d': replace = "i"; break; // LATIN SMALL LETTER I BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
392 case '\u014c': replace = "O"; break; // LATIN CAPITAL LETTER O MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
393 case '\u014d': replace = "o"; break; // LATIN SMALL LETTER O MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
394 case '\u014e': replace = "O"; break; // LATIN CAPITAL LETTER O BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
395 case '\u014f': replace = "o"; break; // LATIN SMALL LETTER O BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
396 case '\u0152': replace = "Oe"; break; // LATIN CAPITAL LETTER O E
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
397 case '\u0153': replace = "oe"; break; // LATIN SMALL LETTER O E
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
398 case '\u016a': replace = "U"; break; // LATIN CAPITAL LETTER U MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
399 case '\u016b': replace = "u"; break; // LATIN SMALL LETTER U MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
400 case '\u016c': replace = "U"; break; // LATIN CAPITAL LETTER U BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
401 case '\u016d': replace = "u"; break; // LATIN SMALL LETTER U BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
402 case '\u017f': replace = "s"; break; // LATIN SMALL LETTER LONG S
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
403 case '\u00df': replace = "ss"; break; // LATIN SMALL LETTER SHARP S
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
404 case '\u00ad': break; // soft hyphen
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
405 // new in MPDL project by J. Willenborg
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
406 case '\u1e14': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
407 case '\u1e15': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
408 case '\u1e16': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
409 case '\u1e17': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
410 case '\u1e18': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
411 case '\u1e19': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
412 case '\u1e1a': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
413 case '\u1e1b': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
414 case '\u1e1c': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
415 case '\u1e1d': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
416 case '\u1eb8': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
417 case '\u1eb9': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
418 case '\u1eba': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
419 case '\u1ebb': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
420 case '\u1ebc': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
421 case '\u1ebd': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
422 case '\u1ebe': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
423 case '\u1ebf': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
424 case '\u1ec0': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
425 case '\u1ec1': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
426 case '\u1ec2': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
427 case '\u1ec3': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
428 case '\u1ec4': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
429 case '\u1ec5': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
430 case '\u1ec6': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
431 case '\u1ec7': replace = "e"; break; // LATIN ... LETTER E WITH ...
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
432 // by Malcolm
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
433 case '\u2329': break; // BRA
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
434 case '\u232a': break; // KET
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
435 default: replace += c; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
436 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
437 buf.append(replace);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
438 // update offsets if replacement is a different length
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
439 if (offsets != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
440 int r = replace.length();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
441 if (r == 0)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
442 this.offsets = arrayKill(this.offsets, i - n);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
443 else if (r == 2)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
444 this.offsets = arrayInsert(this.offsets, i - n + 1, this.offsets[i - n], r - 1);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
445 n += 1 - r;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
446 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
447 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
448 return buf.toString();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
449 } else if (language.equals("it")) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
450 // new Mpdl code: added by J. Willenborg: some of Malcolms code did not work without errors so it has to be taken away, also all latin stuff is imported
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
451 StringBuffer buf = new StringBuffer();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
452 int n = 0;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
453 for (int i = 0; i < s.length(); i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
454 char c = s.charAt(i);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
455 String replace = new String();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
456 switch (c) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
457 case '\u00c0': replace = "A"; break; // LATIN CAPITAL LETTER A GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
458 case '\u00c1': replace = "A"; break; // LATIN CAPITAL LETTER A ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
459 case '\u00c2': replace = "A"; break; // LATIN CAPITAL LETTER A CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
460 case '\u00c4': replace = "A"; break; // LATIN CAPITAL LETTER A DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
461 case '\u00c6': replace = "Ae"; break; // LATIN CAPITAL LETTER A E
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
462 case '\u00c7': replace = "C"; break; // LATIN CAPITAL LETTER C CEDILLA
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
463 case '\u00c8': replace = "E"; break; // LATIN CAPITAL LETTER E GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
464 case '\u00c9': replace = "E"; break; // LATIN CAPITAL LETTER E ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
465 case '\u00ca': replace = "E"; break; // LATIN CAPITAL LETTER E CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
466 case '\u00cb': replace = "E"; break; // LATIN CAPITAL LETTER E DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
467 case '\u00cc': replace = "I"; break; // LATIN CAPITAL LETTER I GRAVE;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
468 case '\u00cd': replace = "I"; break; // LATIN CAPITAL LETTER I ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
469 case '\u00ce': replace = "I"; break; // LATIN CAPITAL LETTER I CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
470 case '\u00cf': replace = "I"; break; // LATIN CAPITAL LETTER I DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
471 case '\u00d2': replace = "O"; break; // LATIN CAPITAL LETTER O GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
472 case '\u00d3': replace = "O"; break; // LATIN CAPITAL LETTER O ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
473 case '\u00d4': replace = "O"; break; // LATIN CAPITAL LETTER O CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
474 case '\u00d6': replace = "O"; break; // LATIN CAPITAL LETTER O DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
475 case '\u00d9': replace = "U"; break; // LATIN CAPITAL LETTER U GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
476 case '\u00da': replace = "U"; break; // LATIN CAPITAL LETTER U ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
477 case '\u00db': replace = "U"; break; // LATIN CAPITAL LETTER U CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
478 case '\u00dc': replace = "U"; break; // LATIN CAPITAL LETTER U DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
479 case '\u00e0': replace = "a"; break; // LATIN SMALL LETTER A GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
480 case '\u00e1': replace = "a"; break; // LATIN SMALL LETTER A ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
481 case '\u00e2': replace = "a"; break; // LATIN SMALL LETTER A CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
482 case '\u00e4': replace = "a"; break; // LATIN SMALL LETTER A DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
483 case '\u00e6': replace = "ae"; break; // LATIN SMALL LETTER A E
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
484 case '\u00e7': replace = "c"; break; // LATIN SMALL LETTER C CEDILLA
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
485 case '\u00e8': replace = "e"; break; // LATIN SMALL LETTER E GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
486 case '\u00e9': replace = "e"; break; // LATIN SMALL LETTER E ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
487 case '\u00ea': replace = "e"; break; // LATIN SMALL LETTER E CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
488 case '\u00eb': replace = "e"; break; // LATIN SMALL LETTER E DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
489 case '\u00ec': replace = "i"; break; // LATIN SMALL LETTER I GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
490 case '\u00ed': replace = "i"; break; // LATIN SMALL LETTER I ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
491 case '\u00ee': replace = "i"; break; // LATIN SMALL LETTER I CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
492 case '\u00ef': replace = "i"; break; // LATIN SMALL LETTER I DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
493 case '\u00f2': replace = "o"; break; // LATIN SMALL LETTER O GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
494 case '\u00f3': replace = "o"; break; // LATIN SMALL LETTER O ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
495 case '\u00f4': replace = "o"; break; // LATIN SMALL LETTER O CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
496 case '\u00f6': replace = "o"; break; // LATIN SMALL LETTER O DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
497 case '\u00f9': replace = "u"; break; // LATIN SMALL LETTER U GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
498 case '\u00fa': replace = "u"; break; // LATIN SMALL LETTER U ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
499 case '\u00fb': replace = "u"; break; // LATIN SMALL LETTER U CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
500 case '\u00fc': replace = "u"; break; // LATIN SMALL LETTER U DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
501 case '\u0100': replace = "A"; break; // LATIN CAPITAL LETTER A MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
502 case '\u0101': replace = "a"; break; // LATIN SMALL LETTER A MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
503 case '\u0102': replace = "A"; break; // LATIN CAPITAL LETTER A BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
504 case '\u0103': replace = "a"; break; // LATIN SMALL LETTER A BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
505 case '\u0112': replace = "E"; break; // LATIN CAPITAL LETTER E MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
506 case '\u0113': replace = "e"; break; // LATIN SMALL LETTER E MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
507 case '\u0114': replace = "E"; break; // LATIN CAPITAL LETTER E BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
508 case '\u0115': replace = "e"; break; // LATIN SMALL LETTER E BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
509 case '\u0118': replace = "Ae"; break; // LATIN CAPITAL LETTER E OGONEK
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
510 case '\u0119': replace = "ae"; break; // LATIN SMALL LETTER E OGONEK
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
511 case '\u012a': replace = "I"; break; // LATIN CAPITAL LETTER I MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
512 case '\u012b': replace = "i"; break; // LATIN SMALL LETTER I MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
513 case '\u012c': replace = "I"; break; // LATIN CAPITAL LETTER I BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
514 case '\u012d': replace = "i"; break; // LATIN SMALL LETTER I BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
515 case '\u014c': replace = "O"; break; // LATIN CAPITAL LETTER O MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
516 case '\u014d': replace = "o"; break; // LATIN SMALL LETTER O MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
517 case '\u014e': replace = "O"; break; // LATIN CAPITAL LETTER O BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
518 case '\u014f': replace = "o"; break; // LATIN SMALL LETTER O BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
519 case '\u0152': replace = "Oe"; break; // LATIN CAPITAL LETTER O E
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
520 case '\u0153': replace = "oe"; break; // LATIN SMALL LETTER O E
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
521 case '\u016a': replace = "U"; break; // LATIN CAPITAL LETTER U MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
522 case '\u016b': replace = "u"; break; // LATIN SMALL LETTER U MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
523 case '\u016c': replace = "U"; break; // LATIN CAPITAL LETTER U BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
524 case '\u016d': replace = "u"; break; // LATIN SMALL LETTER U BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
525 case '\u017f': replace = "s"; break; // LATIN SMALL LETTER LONG S
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
526 case '\u00df': replace = "ss"; break; // LATIN SMALL LETTER SHARP S
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
527 // new in MPDL project by J. Willenborg
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
528 case '\u1e8d': replace = "e"; break; // LATIN SMALL LETTER E WITH TILDE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
529 default: replace += c; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
530 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
531 buf.append(replace);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
532 // update offsets if replacement is a different length
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
533 if (offsets != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
534 int r = replace.length();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
535 if (r == 0) this.offsets = arrayKill(this.offsets, i - n);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
536 else if (r == 2)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
537 this.offsets = arrayInsert(this.offsets, i - n + 1, this.offsets[i - n], r - 1);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
538 n += 1 - r;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
539 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
540 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
541 return buf.toString();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
542 // new Mpdl code: added by J. Willenborg: most of the latin replacements also in english
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
543 } else if (language.equals("en")) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
544 StringBuffer buf = new StringBuffer();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
545 int n = 0;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
546 for (int i = 0; i < s.length(); i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
547 char c = s.charAt(i);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
548 String replace = new String();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
549 switch (c) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
550 case '\u0300': replace = ""; break; // COMBINING GRAVE ACCENT
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
551 case '\u0301': replace = ""; break; // COMBINING ACCUTE ACCENT
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
552 case '\u0302': replace = ""; break; // COMBINING CIRCUMFLEX ACCENT
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
553
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
554 case '\u00c0': replace = "A"; break; // LATIN CAPITAL LETTER A GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
555 case '\u00c1': replace = "A"; break; // LATIN CAPITAL LETTER A ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
556 case '\u00c2': replace = "A"; break; // LATIN CAPITAL LETTER A CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
557 case '\u00c4': replace = "A"; break; // LATIN CAPITAL LETTER A DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
558 case '\u00c6': replace = "Ae"; break; // LATIN CAPITAL LETTER A E
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
559 case '\u00c7': replace = "C"; break; // LATIN CAPITAL LETTER C CEDILLA
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
560 case '\u00c8': replace = "E"; break; // LATIN CAPITAL LETTER E GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
561 case '\u00c9': replace = "E"; break; // LATIN CAPITAL LETTER E ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
562 case '\u00ca': replace = "E"; break; // LATIN CAPITAL LETTER E CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
563 case '\u00cb': replace = "E"; break; // LATIN CAPITAL LETTER E DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
564 case '\u00cc': replace = "I"; break; // LATIN CAPITAL LETTER I GRAVE;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
565 case '\u00cd': replace = "I"; break; // LATIN CAPITAL LETTER I ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
566 case '\u00ce': replace = "I"; break; // LATIN CAPITAL LETTER I CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
567 case '\u00cf': replace = "I"; break; // LATIN CAPITAL LETTER I DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
568 case '\u00d2': replace = "O"; break; // LATIN CAPITAL LETTER O GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
569 case '\u00d3': replace = "O"; break; // LATIN CAPITAL LETTER O ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
570 case '\u00d4': replace = "O"; break; // LATIN CAPITAL LETTER O CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
571 case '\u00d6': replace = "O"; break; // LATIN CAPITAL LETTER O DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
572 case '\u00d9': replace = "U"; break; // LATIN CAPITAL LETTER U GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
573 case '\u00da': replace = "U"; break; // LATIN CAPITAL LETTER U ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
574 case '\u00db': replace = "U"; break; // LATIN CAPITAL LETTER U CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
575 case '\u00dc': replace = "U"; break; // LATIN CAPITAL LETTER U DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
576 case '\u00e0': replace = "a"; break; // LATIN SMALL LETTER A GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
577 case '\u00e1': replace = "a"; break; // LATIN SMALL LETTER A ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
578 case '\u00e2': replace = "a"; break; // LATIN SMALL LETTER A CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
579 case '\u00e4': replace = "a"; break; // LATIN SMALL LETTER A DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
580 case '\u00e6': replace = "ae"; break; // LATIN SMALL LETTER A E
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
581 case '\u00e7': replace = "c"; break; // LATIN SMALL LETTER C CEDILLA
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
582 case '\u00e8': replace = "e"; break; // LATIN SMALL LETTER E GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
583 case '\u00e9': replace = "e"; break; // LATIN SMALL LETTER E ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
584 case '\u00ea': replace = "e"; break; // LATIN SMALL LETTER E CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
585 case '\u00eb': replace = "e"; break; // LATIN SMALL LETTER E DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
586 case '\u00ec': replace = "i"; break; // LATIN SMALL LETTER I GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
587 case '\u00ed': replace = "i"; break; // LATIN SMALL LETTER I ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
588 case '\u00ee': replace = "i"; break; // LATIN SMALL LETTER I CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
589 case '\u00ef': replace = "i"; break; // LATIN SMALL LETTER I DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
590 case '\u00f2': replace = "o"; break; // LATIN SMALL LETTER O GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
591 case '\u00f3': replace = "o"; break; // LATIN SMALL LETTER O ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
592 case '\u00f4': replace = "o"; break; // LATIN SMALL LETTER O CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
593 case '\u00f6': replace = "o"; break; // LATIN SMALL LETTER O DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
594 case '\u00f9': replace = "u"; break; // LATIN SMALL LETTER U GRAVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
595 case '\u00fa': replace = "u"; break; // LATIN SMALL LETTER U ACUTE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
596 case '\u00fb': replace = "u"; break; // LATIN SMALL LETTER U CIRCUMFLEX
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
597 case '\u00fc': replace = "u"; break; // LATIN SMALL LETTER U DIAERESIS
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
598 case '\u0100': replace = "A"; break; // LATIN CAPITAL LETTER A MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
599 case '\u0101': replace = "a"; break; // LATIN SMALL LETTER A MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
600 case '\u0102': replace = "A"; break; // LATIN CAPITAL LETTER A BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
601 case '\u0103': replace = "a"; break; // LATIN SMALL LETTER A BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
602 case '\u0112': replace = "E"; break; // LATIN CAPITAL LETTER E MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
603 case '\u0113': replace = "e"; break; // LATIN SMALL LETTER E MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
604 case '\u0114': replace = "E"; break; // LATIN CAPITAL LETTER E BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
605 case '\u0115': replace = "e"; break; // LATIN SMALL LETTER E BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
606 case '\u0118': replace = "Ae"; break; // LATIN CAPITAL LETTER E OGONEK
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
607 case '\u0119': replace = "ae"; break; // LATIN SMALL LETTER E OGONEK
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
608 case '\u012a': replace = "I"; break; // LATIN CAPITAL LETTER I MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
609 case '\u012b': replace = "i"; break; // LATIN SMALL LETTER I MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
610 case '\u012c': replace = "I"; break; // LATIN CAPITAL LETTER I BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
611 case '\u012d': replace = "i"; break; // LATIN SMALL LETTER I BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
612 case '\u014c': replace = "O"; break; // LATIN CAPITAL LETTER O MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
613 case '\u014d': replace = "o"; break; // LATIN SMALL LETTER O MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
614 case '\u014e': replace = "O"; break; // LATIN CAPITAL LETTER O BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
615 case '\u014f': replace = "o"; break; // LATIN SMALL LETTER O BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
616 case '\u0152': replace = "Oe"; break; // LATIN CAPITAL LETTER O E
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
617 case '\u0153': replace = "oe"; break; // LATIN SMALL LETTER O E
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
618 case '\u016a': replace = "U"; break; // LATIN CAPITAL LETTER U MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
619 case '\u016b': replace = "u"; break; // LATIN SMALL LETTER U MACRON
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
620 case '\u016c': replace = "U"; break; // LATIN CAPITAL LETTER U BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
621 case '\u016d': replace = "u"; break; // LATIN SMALL LETTER U BREVE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
622 case '\u017f': replace = "s"; break; // LATIN SMALL LETTER LONG S
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
623 case '\u00df': replace = "ss"; break; // LATIN SMALL LETTER SHARP S
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
624 // new in MPDL project by J. Willenborg
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
625 case '\u1e8d': replace = "e"; break; // LATIN SMALL LETTER E WITH TILDE
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
626 // by Malcolm
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
627 case '\u00ad': break; // soft hyphen
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
628 case '\u2329': break; // BRA
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
629 case '\u232a': break; // KET
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
630 default: replace += c; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
631 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
632 buf.append(replace);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
633 // update offsets if replacement is a different length
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
634 if (offsets != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
635 int r = replace.length();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
636 if (r == 0)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
637 this.offsets = arrayKill(this.offsets, i - n);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
638 else if (r == 2)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
639 this.offsets = arrayInsert(this.offsets, i - n + 1, this.offsets[i - n], r - 1);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
640 n += 1 - r;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
641 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
642 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
643 return buf.toString();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
644 } else if (language.equals("fr")) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
645 // new Mpdl code: added by J. Willenborg: some of Malcolms code did not work without errors so it has to be taken away
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
646 StringBuffer buf = new StringBuffer();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
647 int n = 0;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
648 for (int i = 0; i < s.length(); i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
649 char c = s.charAt(i);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
650 String replace = new String();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
651 switch (c) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
652 case '\u00e6': replace = "ae"; break; // LATIN SMALL LETTER A E
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
653 case '\u017f': replace = "s"; break; // LATIN SMALL LETTER LONG S
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
654 case '\u00df': replace = "ss"; break; // LATIN SMALL LETTER SHARP S
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
655 case '\u00ad': break; // soft hyphen
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
656 case '-': break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
657 default: replace += c; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
658 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
659 buf.append(replace);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
660 // update offsets if replacement is a different length
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
661 if (offsets != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
662 int r = replace.length();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
663 if (r == 0)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
664 this.offsets = arrayKill(this.offsets, i - n);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
665 else if (r == 2)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
666 this.offsets = arrayInsert(this.offsets, i - n + 1, this.offsets[i - n], r - 1);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
667 n += 1 - r;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
668 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
669 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
670 return buf.toString();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
671 } else if (language.equals("de")) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
672 StringBuffer buf = new StringBuffer();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
673 int n = 0;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
674 for (int i = 0; i < s.length(); i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
675 char c = s.charAt(i);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
676 String replace = new String();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
677 switch (c) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
678 case '\u00c4': replace = "Ae"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
679 case '\u00d6': replace = "Oe"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
680 case '\u00dc': replace = "Ue"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
681 case '\u00df': replace = "ss"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
682 case '\u00e4': replace = "ae"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
683 case '\u00f6': replace = "oe"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
684 case '\u00fc': replace = "ue"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
685 case '\u00ad': break; // soft hyphen
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
686 case '\u00e9': replace = "e"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
687 // new in MPDL project by J. Willenborg
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
688 case '\u017f': replace = "s"; break; // LATIN SMALL LETTER LONG S
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
689 // case '-': break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
690 default: replace += c; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
691 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
692 buf.append(replace);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
693 // update offsets if replacement is a different length
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
694 if (offsets != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
695 int r = replace.length();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
696 if (r == 0)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
697 this.offsets = arrayKill(this.offsets, i - n);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
698 else if (r == 2)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
699 this.offsets = arrayInsert(this.offsets, i - n + 1, this.offsets[i - n], r - 1);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
700 n += 1 - r;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
701 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
702 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
703 return buf.toString();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
704 } else if (language.equals("zh")) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
705 StringBuffer buf = new StringBuffer();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
706 int n = 0;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
707 for (int i = 0; i < s.length(); i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
708 char c = s.charAt(i);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
709 String replace = new String();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
710 switch (c) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
711 case '\u00b9': replace = "1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
712 case '\u00b2': replace = "2"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
713 case '\u00b3': replace = "3"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
714 case '\u2074': replace = "4"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
715 case '\u2075': replace = "5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
716 // original by Malcolm Hyman: with the following replacements
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
717 // case '\u3000': replace = " "; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
718 // case '\u3001': replace = ","; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
719 // case '\u3002': replace = "."; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
720 // case '\u200b': break; // BREAKS EVERYTHING!
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
721 default: replace += c; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
722 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
723 buf.append(replace);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
724 // update offsets if replacement is a different length
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
725 if (offsets != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
726 int r = replace.length();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
727 if (r == 0)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
728 this.offsets = arrayKill(this.offsets, i - n);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
729 else if (r == 2)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
730 this.offsets = arrayInsert(this.offsets, i - n + 1, this.offsets[i - n], r - 1);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
731 n += 1 - r;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
732 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
733 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
734 return buf.toString();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
735 } else if (language.equals("akk") ||
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
736 language.equals("qam") ||
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
737 language.equals("qpc") ||
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
738 language.equals("elx") ||
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
739 language.equals("sux") ||
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
740 language.equals("hit") ||
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
741 language.equals("qhu") ||
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
742 language.equals("peo") ||
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
743 language.equals("uga") ||
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
744 language.equals("ura") ||
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
745 language.equals("qcu")) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
746 StringBuffer buf = new StringBuffer();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
747 int n = 0;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
748 char last = '\u0000';
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
749 for (int i = 0; i < s.length(); i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
750 char c = s.charAt(i);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
751 c = Character.toLowerCase(c);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
752 String replace = new String();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
753 switch (c) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
754 case '{': replace += "-"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
755 case '}': replace += "-"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
756 // These are from PSD::ATF::Unicode by Steve Tinney
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
757 case '\u0161': replace += "sz"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
758 case '\u1e63': replace += "s,"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
759 case '\u1e6d': replace += "t,"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
760 case '\u014b': replace += "j"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
761 case '\u015b': replace += "s'"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
762 case '\u2080': replace += "0"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
763 case '\u2081': replace += "1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
764 case '\u2082': replace += "2"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
765 case '\u2083': replace += "3"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
766 case '\u2084': replace += "4"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
767 case '\u2085': replace += "5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
768 case '\u2086': replace += "6"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
769 case '\u2087': replace += "7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
770 case '\u2088': replace += "8"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
771 case '\u2089': replace += "9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
772
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
773 case 'c': // shin (except where used as modifier)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
774 if ((i > 0) && ((last == '~') || (last == '@')))
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
775 replace += "c";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
776 else replace += "sz";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
777 break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
778 default: replace += c; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
779 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
780 // suppress grapheme boundary before or after word boundary
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
781 if (replace.equals("-")) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
782 if ((i + 1 == s.length()) || (s.charAt(i + 1) == ' ') || (i == 0) || (s.charAt(i - 1) == ' '))
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
783 replace = "";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
784 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
785 last = c;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
786 buf.append(replace);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
787 // update offsets if replacement is a different length
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
788 if (offsets != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
789 int r = replace.length();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
790 if (r == 0)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
791 this.offsets = arrayKill(this.offsets, i - n);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
792 else if (r == 2)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
793 this.offsets = arrayInsert(this.offsets, i - n + 1, this.offsets[i - n], r - 1);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
794 n += 1 - r;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
795 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
796 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
797 return buf.toString();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
798 } else if (language.equals("el") || language.equals("grc")) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
799 StringBuffer buf = new StringBuffer();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
800 int n = 0;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
801 for (int i = 0; i < s.length(); i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
802 char c = s.charAt(i);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
803 String replace = new String();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
804 switch (c) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
805 case '\u03c2': replace = "\u03c3"; break; // GREEK SMALL LETTER FINAL SIGMA
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
806 case '<': break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
807 case '>': break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
808 case '[': break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
809 case ']': break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
810 case '1': break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
811 case '2': break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
812 case '\u03ac': replace = "\u1f71"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
813 case '\u03ad': replace = "\u1f73"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
814 case '\u03ae': replace = "\u1f75"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
815 case '\u03af': replace = "\u1f77"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
816 case '\u03cc': replace = "\u1f79"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
817 case '\u03cd': replace = "\u1f7b"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
818 case '\u03ce': replace = "\u1f7d"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
819 case '-': break; // same treatment as soft hyphen
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
820 case '\u00ad': break; // soft hyphen
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
821 default: replace += c; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
822 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
823 buf.append(replace);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
824 // update offsets if replacement is a different length
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
825 if (offsets != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
826 int r = replace.length();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
827 if (r == 0)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
828 this.offsets = arrayKill(this.offsets, i - n);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
829 else if (r == 2)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
830 this.offsets = arrayInsert(this.offsets, i - n + 1, this.offsets[i - n], r - 1);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
831 n += 1 - r;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
832 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
833 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
834 return buf.toString();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
835 } else if (language.equals("el_atonic")) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
836 StringBuffer buf = new StringBuffer();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
837 int n = 0;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
838 for (int i = 0; i < s.length(); i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
839 char c = s.charAt(i);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
840 String replace = new String();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
841 switch (c) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
842 case '\u03c2': replace = "\u03c3"; break; // GREEK SMALL LETTER FINAL SIGMA
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
843 // map characters with diacritics to their plain equivalent
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
844 // cf. <code>BetaCode.java</code>
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
845 case '\u03aa': replace = "\u0399"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
846 case '\u03ab': replace = "\u03a5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
847 case '\u03ac': replace = "\u0381"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
848 case '\u03ad': replace = "\u0385"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
849 case '\u03ae': replace = "\u0387"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
850 case '\u03af': replace = "\u0389"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
851 case '\u03ca': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
852 case '\u03cb': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
853 case '\u03cc': replace = "\u03bf"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
854 case '\u03cd': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
855 case '\u03ce': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
856 case '\u1f00': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
857 case '\u1f01': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
858 case '\u1f02': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
859 case '\u1f03': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
860 case '\u1f04': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
861 case '\u1f05': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
862 case '\u1f06': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
863 case '\u1f07': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
864 case '\u1f08': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
865 case '\u1f09': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
866 case '\u1f0a': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
867 case '\u1f0b': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
868 case '\u1f0c': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
869 case '\u1f0d': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
870 case '\u1f0e': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
871 case '\u1f0f': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
872 case '\u1f10': replace = "\u03b5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
873 case '\u1f11': replace = "\u03b5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
874 case '\u1f12': replace = "\u03b5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
875 case '\u1f13': replace = "\u03b5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
876 case '\u1f14': replace = "\u03b5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
877 case '\u1f15': replace = "\u03b5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
878 case '\u1f18': replace = "\u0395"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
879 case '\u1f19': replace = "\u0395"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
880 case '\u1f1a': replace = "\u0395"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
881 case '\u1f1b': replace = "\u0395"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
882 case '\u1f1c': replace = "\u0395"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
883 case '\u1f1d': replace = "\u0395"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
884 case '\u1f20': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
885 case '\u1f21': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
886 case '\u1f22': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
887 case '\u1f23': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
888 case '\u1f24': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
889 case '\u1f25': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
890 case '\u1f26': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
891 case '\u1f27': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
892 case '\u1f28': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
893 case '\u1f29': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
894 case '\u1f2a': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
895 case '\u1f2b': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
896 case '\u1f2c': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
897 case '\u1f2d': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
898 case '\u1f2e': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
899 case '\u1f2f': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
900 case '\u1f30': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
901 case '\u1f31': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
902 case '\u1f32': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
903 case '\u1f33': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
904 case '\u1f34': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
905 case '\u1f35': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
906 case '\u1f36': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
907 case '\u1f37': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
908 case '\u1f38': replace = "\u0399"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
909 case '\u1f39': replace = "\u0399"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
910 case '\u1f3a': replace = "\u0399"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
911 case '\u1f3b': replace = "\u0399"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
912 case '\u1f3c': replace = "\u0399"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
913 case '\u1f3d': replace = "\u0399"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
914 case '\u1f3e': replace = "\u0399"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
915 case '\u1f3f': replace = "\u0399"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
916 case '\u1f40': replace = "\u03bf"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
917 case '\u1f41': replace = "\u03bf"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
918 case '\u1f42': replace = "\u03bf"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
919 case '\u1f43': replace = "\u03bf"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
920 case '\u1f44': replace = "\u03bf"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
921 case '\u1f45': replace = "\u03bf"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
922 case '\u1f48': replace = "\u039f"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
923 case '\u1f49': replace = "\u039f"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
924 case '\u1f4a': replace = "\u039f"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
925 case '\u1f4b': replace = "\u039f"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
926 case '\u1f4c': replace = "\u039f"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
927 case '\u1f4d': replace = "\u039f"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
928 case '\u1f50': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
929 case '\u1f51': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
930 case '\u1f52': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
931 case '\u1f53': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
932 case '\u1f54': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
933 case '\u1f55': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
934 case '\u1f56': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
935 case '\u1f57': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
936 case '\u1f58': replace = "\u03a5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
937 case '\u1f59': replace = "\u03a5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
938 case '\u1f5a': replace = "\u03a5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
939 case '\u1f5b': replace = "\u03a5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
940 case '\u1f5c': replace = "\u03a5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
941 case '\u1f5d': replace = "\u03a5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
942 case '\u1f5e': replace = "\u03a5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
943 case '\u1f5f': replace = "\u03a5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
944 case '\u1f60': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
945 case '\u1f61': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
946 case '\u1f62': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
947 case '\u1f63': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
948 case '\u1f64': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
949 case '\u1f65': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
950 case '\u1f66': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
951 case '\u1f67': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
952 case '\u1f68': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
953 case '\u1f69': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
954 case '\u1f6a': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
955 case '\u1f6b': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
956 case '\u1f6c': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
957 case '\u1f6d': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
958 case '\u1f6e': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
959 case '\u1f6f': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
960 case '\u1f70': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
961 case '\u1f71': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
962 case '\u1f72': replace = "\u03b5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
963 case '\u1f73': replace = "\u03b5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
964 case '\u1f74': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
965 case '\u1f75': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
966 case '\u1f76': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
967 case '\u1f77': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
968 case '\u1f78': replace = "\u03bf"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
969 case '\u1f79': replace = "\u03bf"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
970 case '\u1f7a': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
971 case '\u1f7b': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
972 case '\u1f7c': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
973 case '\u1f7d': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
974 case '\u1f80': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
975 case '\u1f81': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
976 case '\u1f82': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
977 case '\u1f83': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
978 case '\u1f84': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
979 case '\u1f85': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
980 case '\u1f86': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
981 case '\u1f87': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
982 case '\u1f88': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
983 case '\u1f89': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
984 case '\u1f8a': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
985 case '\u1f8b': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
986 case '\u1f8c': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
987 case '\u1f8d': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
988 case '\u1f8e': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
989 case '\u1f8f': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
990 case '\u1f90': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
991 case '\u1f91': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
992 case '\u1f92': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
993 case '\u1f93': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
994 case '\u1f94': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
995 case '\u1f95': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
996 case '\u1f96': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
997 case '\u1f97': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
998 case '\u1f98': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
999 case '\u1f99': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1000 case '\u1f9a': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1001 case '\u1f9b': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1002 case '\u1f9c': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1003 case '\u1f9d': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1004 case '\u1f9e': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1005 case '\u1f9f': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1006 case '\u1fa0': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1007 case '\u1fa1': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1008 case '\u1fa2': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1009 case '\u1fa3': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1010 case '\u1fa4': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1011 case '\u1fa5': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1012 case '\u1fa6': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1013 case '\u1fa7': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1014 case '\u1fa8': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1015 case '\u1fa9': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1016 case '\u1faa': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1017 case '\u1fab': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1018 case '\u1fac': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1019 case '\u1fad': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1020 case '\u1fae': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1021 case '\u1faf': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1022 case '\u1fb2': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1023 case '\u1fb3': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1024 case '\u1fb4': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1025 case '\u1fb6': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1026 case '\u1fb7': replace = "\u03b1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1027 case '\u1fba': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1028 case '\u1fbb': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1029 case '\u1fbc': replace = "\u0391"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1030 case '\u1fc2': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1031 case '\u1fc3': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1032 case '\u1fc4': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1033 case '\u1fc6': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1034 case '\u1fc7': replace = "\u03b7"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1035 case '\u1fca': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1036 case '\u1fcb': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1037 case '\u1fcc': replace = "\u0397"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1038 case '\u1fd2': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1039 case '\u1fd3': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1040 case '\u1fd6': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1041 case '\u1fd7': replace = "\u03b9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1042 case '\u1fda': replace = "\u0399"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1043 case '\u1fdb': replace = "\u039f"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1044 case '\u1fe2': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1045 case '\u1fe3': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1046 case '\u1fe4': replace = "\u03c1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1047 case '\u1fe5': replace = "\u03c1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1048 case '\u1fe6': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1049 case '\u1fe7': replace = "\u03c5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1050 case '\u1fea': replace = "\u03a5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1051 case '\u1feb': replace = "\u03a5"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1052 case '\u1fec': replace = "\u03a1"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1053 case '\u1ff2': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1054 case '\u1ff3': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1055 case '\u1ff4': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1056 case '\u1ff6': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1057 case '\u1ff7': replace = "\u03c9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1058 case '\u1ff8': replace = "\u039f"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1059 case '\u1ff9': replace = "\u039f"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1060 case '\u1ffa': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1061 case '\u1ffb': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1062 case '\u1ffc': replace = "\u03a9"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1063
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1064 case '<': break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1065 case '>': break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1066 case '-': break; // same treatment as soft hyphen
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1067 case '\u00ad': break; // soft hyphen
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1068 default: replace += c; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1069 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1070 buf.append(replace);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1071 // update offsets if replacement is a different length
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1072 if (offsets != null) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1073 int r = replace.length();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1074 if (r == 0)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1075 this.offsets = arrayKill(this.offsets, i - n);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1076 else if (r == 2)
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1077 this.offsets = arrayInsert(this.offsets, i - n + 1, this.offsets[i - n], r - 1);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1078 n += 1 - r;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1079 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1080 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1081 return buf.toString();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1082 } else { // unknown or no language
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1083 return s;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1084 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1085 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1086
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1087 /*
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1088 // explicit words
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1089 normStr = normStr.replaceAll("aliàs", "alias");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1090 normStr = normStr.replaceAll("hîc", "hic");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1091 normStr = normStr.replaceAll("quòd", "quod");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1092 normStr = normStr.replaceAll("Quòd", "Quod");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1093 normStr = normStr.replaceAll("QVòd", "Quod");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1094 normStr = normStr.replaceAll("Cùmque", "Cumque");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1095 normStr = normStr.replaceAll("aër", "aer");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1096 // ij
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1097 normStr = normStr.replaceAll("ij", "ii");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1098 // qu/qv
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1099 normStr = normStr.replaceAll("qv", "qu");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1100 // normStr = normStr.replaceAll("qV", "qU");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1101 normStr = normStr.replaceAll("Qv", "Qu");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1102 normStr = normStr.replaceAll("QV", "QU");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1103 // u/v
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1104 String vowels = getVowels();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1105 String consonants = getConsonants();
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1106 normStr = normStr.replaceAll("([" + vowels + "])([-]*)u([" + vowels +"])", "$1$2v$3"); // vowel + u + vowel --> vowel + v + vowel
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1107 normStr = normStr.replaceAll("([" + vowels + "])([-]*)U([" + vowels +"])", "$1$2V$3"); // vowel + U + vowel --> vowel + V + vowel
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1108 normStr = normStr.replaceAll("([" + consonants + "])([-]*)v([" + consonants +"])", "$1$2u$3"); // consonant + v + consonant --> consonant + u + consonant
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1109 normStr = normStr.replaceAll("([" + consonants + "])([-]*)V([" + consonants +"])", "$1$2U$3"); // consonant + V + consonant --> consonant + U + consonant
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1110 normStr = normStr.replaceAll("^v([" + consonants +"])", "u$1"); // v + consonant --> u + consonant
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1111 normStr = normStr.replaceAll("^V([" + consonants +"])", "U$1"); // V + consonant --> U + consonant
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1112 // end of word: diacritica
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1113 normStr = normStr.replaceAll("à$", "a");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1114 normStr = normStr.replaceAll("è$", "e");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1115 normStr = normStr.replaceAll("ò$", "o");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1116 normStr = normStr.replaceAll("àm$", "am");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1117 normStr = normStr.replaceAll("ùm$", "um");
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1118 String normStrTmp = normStr;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1119 normStr = "";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1120 for (int i = 0; i < normStrTmp.length(); i++) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1121 char c = normStrTmp.charAt(i);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1122 String replace = "";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1123 switch (c) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1124 case 'ſ': replace = "s"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1125 case 'ß': replace = "ss"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1126 case 'æ': replace = "ae"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1127 case 'Æ': replace = "AE"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1128 case 'ę': replace = "ae"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1129 case 'œ': replace = "oe"; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1130 default: replace += c; break;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1131 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1132 normStr = normStr + replace;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1133 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1134
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1135
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1136 private String getVowels() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1137 String retStr = null;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1138 if (Language.getInstance().isItalian(language)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1139 retStr = "AEIOUaeiou" +
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1140 "\u00c6\u00e6" + // AE ligatures
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1141 "\u0152\u0153"; // OE ligatures
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1142 } else if (Language.getInstance().isLatin(language)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1143 retStr = "AEIOUaeiouÆœęàèòù";
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1144 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1145 return retStr;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1146 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1147
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1148 private String getConsonants() {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1149 String retStr = null;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1150 if (Language.getInstance().isItalian(language)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1151 retStr = "BCDFGHKLMNPQRSTVWXZ" +
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1152 "bcdfghklmnpqrstvwxz" +
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1153 "ſß"; // long/sharp S
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1154 } else if (Language.getInstance().isLatin(language)) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1155 retStr = "BCDFGHKLMNPQRSTVWXZ" +
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1156 "bcdfghklmnpqrstvwxz" +
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1157 "ſß"; // long/sharp S
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1158 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1159 return retStr;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1160 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1161
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1162
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1163
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1164
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1165
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1166 *
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1167 *
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1168 *
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1169 *
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1170 */
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1171
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1172
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1173
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1174
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1175
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1176
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1177 /**
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1178 * Returns a copy of an integer array with the element at
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1179 * <code>index</code> removed ("killed").
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1180 *
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1181 * @param array integer array
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1182 * @param index index of element to remove
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1183 */
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1184 private int[] arrayKill(int[] array, int index) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1185 int[] newArray = new int[array.length - 1];
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1186 System.arraycopy(array, 0, newArray, 0, index);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1187 System.arraycopy(array, index + 1, newArray, index, array.length - index - 1);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1188 return newArray;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1189 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1190
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1191 /**
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1192 * Returns a copy of an integer array with <code>count</code> elements
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1193 * inserted at <code>index</code>.
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1194 *
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1195 * @param array integer array
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1196 * @param index index to insert new elements
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1197 * @param value value to insert into new slots
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1198 * @param count number of new slots to insert
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1199 */
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1200 private int[] arrayInsert(int[] array, int index, int value, int count) {
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1201 int[] newArray = new int[array.length + count];
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1202 System.arraycopy(array, 0, newArray, 0, index);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1203 for (int i = 0; i < count; i++) newArray[index + i] = value;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1204 System.arraycopy(array, index, newArray, index + count, array.length - index);
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1205 return newArray;
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1206 }
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1207
4a3641ae14d2 Erstellung
Josef Willenborg <jwillenborg@mpiwg-berlin.mpg.de>
parents:
diff changeset
1208 }