CP1251CharacterSet


1   package com.icl.saxon.charcode;
2   
3   /**
4   * This class defines properties of the CP1251 Cyrillic character set,
5   * as defined at http://www.microsoft.com/globaldev/reference/sbcs/1251.htm
6   */
7   
8   public class CP1251CharacterSet implements CharacterSet {
9   
10    public final boolean inCharset(int c) {
11      return  ( c <= 0x7f ) ||
12              (c >= 0x0401 && c <= 0x044F) ||
13              (c >= 0x0451 && c <= 0x045f) ||
14              (c == 0x0490) ||
15              (c == 0x0491) ||
16              (c == 0x2013) ||
17              (c == 0x2014) ||
18              (c == 0x2018) ||
19              (c == 0x2019) ||
20              (c == 0x201A) ||
21              (c == 0x201C) ||
22              (c == 0x201D) ||
23              (c == 0x201E) ||
24              (c == 0x2020) ||
25              (c == 0x2021) ||
26              (c == 0x2022) ||
27              (c == 0x2026) ||
28              (c == 0x2030) ||
29              (c == 0x2039) ||
30              (c == 0x203A) ||
31              (c == 0x20AC) ||
32              (c == 0x2116) ||
33              (c == 0x2122); 
34    }
35  
36  }
37  
38  // 80 = U+0402 : CYRILLIC CAPITAL LETTER DJE
39  // 81 = U+0403 : CYRILLIC CAPITAL LETTER GJE
40  // 82 = U+201A : SINGLE LOW-9 QUOTATION MARK
41  // 83 = U+0453 : CYRILLIC SMALL LETTER GJE
42  // 84 = U+201E : DOUBLE LOW-9 QUOTATION MARK
43  // 85 = U+2026 : HORIZONTAL ELLIPSIS
44  // 86 = U+2020 : DAGGER
45  // 87 = U+2021 : DOUBLE DAGGER
46  // 88 = U+20AC : EURO SIGN
47  // 89 = U+2030 : PER MILLE SIGN
48  // 8A = U+0409 : CYRILLIC CAPITAL LETTER LJE
49  // 8B = U+2039 : SINGLE LEFT-POINTING ANGLE QUOTATION MARK
50  // 8C = U+040A : CYRILLIC CAPITAL LETTER NJE
51  // 8D = U+040C : CYRILLIC CAPITAL LETTER KJE
52  // 8E = U+040B : CYRILLIC CAPITAL LETTER TSHE
53  // 8F = U+040F : CYRILLIC CAPITAL LETTER DZHE
54  // 90 = U+0452 : CYRILLIC SMALL LETTER DJE
55  // 91 = U+2018 : LEFT SINGLE QUOTATION MARK
56  // 92 = U+2019 : RIGHT SINGLE QUOTATION MARK
57  // 93 = U+201C : LEFT DOUBLE QUOTATION MARK
58  // 94 = U+201D : RIGHT DOUBLE QUOTATION MARK
59  // 95 = U+2022 : BULLET
60  // 96 = U+2013 : EN DASH
61  // 97 = U+2014 : EM DASH
62  // 99 = U+2122 : TRADE MARK SIGN
63  // 9A = U+0459 : CYRILLIC SMALL LETTER LJE
64  // 9B = U+203A : SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
65  // 9C = U+045A : CYRILLIC SMALL LETTER NJE
66  // 9D = U+045C : CYRILLIC SMALL LETTER KJE
67  // 9E = U+045B : CYRILLIC SMALL LETTER TSHE
68  // 9F = U+045F : CYRILLIC SMALL LETTER DZHE
69  // A0 = U+00A0 : NO-BREAK SPACE
70  // A1 = U+040E : CYRILLIC CAPITAL LETTER SHORT U
71  // A2 = U+045E : CYRILLIC SMALL LETTER SHORT U
72  // A3 = U+0408 : CYRILLIC CAPITAL LETTER JE
73  // A4 = U+00A4 : CURRENCY SIGN
74  // A5 = U+0490 : CYRILLIC CAPITAL LETTER GHE WITH UPTURN
75  // A6 = U+00A6 : BROKEN BAR
76  // A7 = U+00A7 : SECTION SIGN
77  // A8 = U+0401 : CYRILLIC CAPITAL LETTER IO
78  // A9 = U+00A9 : COPYRIGHT SIGN
79  // AA = U+0404 : CYRILLIC CAPITAL LETTER UKRAINIAN IE
80  // AB = U+00AB : LEFT-POINTING DOUBLE ANGLE QUOTATION MARK
81  // AC = U+00AC : NOT SIGN
82  // AD = U+00AD : SOFT HYPHEN
83  // AE = U+00AE : REGISTERED SIGN
84  // AF = U+0407 : CYRILLIC CAPITAL LETTER YI
85  // B0 = U+00B0 : DEGREE SIGN
86  // B1 = U+00B1 : PLUS-MINUS SIGN
87  // B2 = U+0406 : CYRILLIC CAPITAL LETTER BYELORUSSIAN-UKRAINIAN I
88  // B3 = U+0456 : CYRILLIC SMALL LETTER BYELORUSSIAN-UKRAINIAN I
89  // B4 = U+0491 : CYRILLIC SMALL LETTER GHE WITH UPTURN
90  // B5 = U+00B5 : MICRO SIGN
91  // B6 = U+00B6 : PILCROW SIGN
92  // B7 = U+00B7 : MIDDLE DOT
93  // B8 = U+0451 : CYRILLIC SMALL LETTER IO
94  // B9 = U+2116 : NUMERO SIGN
95  // BA = U+0454 : CYRILLIC SMALL LETTER UKRAINIAN IE
96  // BB = U+00BB : RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK
97  // BC = U+0458 : CYRILLIC SMALL LETTER JE
98  // BD = U+0405 : CYRILLIC CAPITAL LETTER DZE
99  // BE = U+0455 : CYRILLIC SMALL LETTER DZE
100 // BF = U+0457 : CYRILLIC SMALL LETTER YI
101 // C0 = U+0410 : CYRILLIC CAPITAL LETTER A
102 // C1 = U+0411 : CYRILLIC CAPITAL LETTER BE
103 // C2 = U+0412 : CYRILLIC CAPITAL LETTER VE
104 // C3 = U+0413 : CYRILLIC CAPITAL LETTER GHE
105 // C4 = U+0414 : CYRILLIC CAPITAL LETTER DE
106 // C5 = U+0415 : CYRILLIC CAPITAL LETTER IE
107 // C6 = U+0416 : CYRILLIC CAPITAL LETTER ZHE
108 // C7 = U+0417 : CYRILLIC CAPITAL LETTER ZE
109 // C8 = U+0418 : CYRILLIC CAPITAL LETTER I
110 // C9 = U+0419 : CYRILLIC CAPITAL LETTER SHORT I
111 // CA = U+041A : CYRILLIC CAPITAL LETTER KA
112 // CB = U+041B : CYRILLIC CAPITAL LETTER EL
113 // CC = U+041C : CYRILLIC CAPITAL LETTER EM
114 // CD = U+041D : CYRILLIC CAPITAL LETTER EN
115 // CE = U+041E : CYRILLIC CAPITAL LETTER O
116 // CF = U+041F : CYRILLIC CAPITAL LETTER PE
117 // D0 = U+0420 : CYRILLIC CAPITAL LETTER ER
118 // D1 = U+0421 : CYRILLIC CAPITAL LETTER ES
119 // D2 = U+0422 : CYRILLIC CAPITAL LETTER TE
120 // D3 = U+0423 : CYRILLIC CAPITAL LETTER U
121 // D4 = U+0424 : CYRILLIC CAPITAL LETTER EF
122 // D5 = U+0425 : CYRILLIC CAPITAL LETTER HA
123 // D6 = U+0426 : CYRILLIC CAPITAL LETTER TSE
124 // D7 = U+0427 : CYRILLIC CAPITAL LETTER CHE
125 // D8 = U+0428 : CYRILLIC CAPITAL LETTER SHA
126 // D9 = U+0429 : CYRILLIC CAPITAL LETTER SHCHA
127 // DA = U+042A : CYRILLIC CAPITAL LETTER HARD SIGN
128 // DB = U+042B : CYRILLIC CAPITAL LETTER YERU
129 // DC = U+042C : CYRILLIC CAPITAL LETTER SOFT SIGN
130 // DD = U+042D : CYRILLIC CAPITAL LETTER E
131 // DE = U+042E : CYRILLIC CAPITAL LETTER YU
132 // DF = U+042F : CYRILLIC CAPITAL LETTER YA
133 // E0 = U+0430 : CYRILLIC SMALL LETTER A
134 // E1 = U+0431 : CYRILLIC SMALL LETTER BE
135 // E2 = U+0432 : CYRILLIC SMALL LETTER VE
136 // E3 = U+0433 : CYRILLIC SMALL LETTER GHE
137 // E4 = U+0434 : CYRILLIC SMALL LETTER DE
138 // E5 = U+0435 : CYRILLIC SMALL LETTER IE
139 // E6 = U+0436 : CYRILLIC SMALL LETTER ZHE
140 // E7 = U+0437 : CYRILLIC SMALL LETTER ZE
141 // E8 = U+0438 : CYRILLIC SMALL LETTER I
142 // E9 = U+0439 : CYRILLIC SMALL LETTER SHORT I
143 // EA = U+043A : CYRILLIC SMALL LETTER KA
144 // EB = U+043B : CYRILLIC SMALL LETTER EL
145 // EC = U+043C : CYRILLIC SMALL LETTER EM
146 // ED = U+043D : CYRILLIC SMALL LETTER EN
147 // EE = U+043E : CYRILLIC SMALL LETTER O
148 // EF = U+043F : CYRILLIC SMALL LETTER PE
149 // F0 = U+0440 : CYRILLIC SMALL LETTER ER
150 // F1 = U+0441 : CYRILLIC SMALL LETTER ES
151 // F2 = U+0442 : CYRILLIC SMALL LETTER TE
152 // F3 = U+0443 : CYRILLIC SMALL LETTER U
153 // F4 = U+0444 : CYRILLIC SMALL LETTER EF
154 // F5 = U+0445 : CYRILLIC SMALL LETTER HA
155 // F6 = U+0446 : CYRILLIC SMALL LETTER TSE
156 // F7 = U+0447 : CYRILLIC SMALL LETTER CHE
157 // F8 = U+0448 : CYRILLIC SMALL LETTER SHA
158 // F9 = U+0449 : CYRILLIC SMALL LETTER SHCHA
159 // FA = U+044A : CYRILLIC SMALL LETTER HARD SIGN
160 // FB = U+044B : CYRILLIC SMALL LETTER YERU
161 // FC = U+044C : CYRILLIC SMALL LETTER SOFT SIGN
162 // FD = U+044D : CYRILLIC SMALL LETTER E
163 // FE = U+044E : CYRILLIC SMALL LETTER YU
164 // FF = U+044F : CYRILLIC SMALL LETTER YA
165 
166 
167 
168 
169 //
170 // The contents of this file are subject to the Mozilla Public License Version 1.0 (the "License");
171 // you may not use this file except in compliance with the License. You may obtain a copy of the
172 // License at http://www.mozilla.org/MPL/ 
173 //
174 // Software distributed under the License is distributed on an "AS IS" basis,
175 // WITHOUT WARRANTY OF ANY KIND, either express or implied.
176 // See the License for the specific language governing rights and limitations under the License. 
177 //
178 // The Original Code is: all this file. 
179 //
180 // The Initial Developer of the Original Code is
181 // Aleksei Makarov [makarov@iitam.omsk.net.ru]
182 //
183 // Portions created by (your name) are Copyright (C) (your legal entity). All Rights Reserved. 
184 //
185 // Contributor(s): none. 
186 //
187
A to Z: JavaDoc & Examples Daily Java News & Articles Open Source Projects Open Source Codes Free Computer Books Remove Frame
Popular Tags