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