1 /* Copyright 2002, 2003 Elliotte Rusty Harold
2 
3    This library is free software; you can redistribute it and/or modify
4    it under the terms of version 2.1 of the GNU Lesser General Public
5    License as published by the Free Software Foundation.
6 
7    This library is distributed in the hope that it will be useful,
8    but WITHOUT ANY WARRANTY; without even the implied warranty of
9    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
10    GNU Lesser General Public License for more details.
11 
12    You should have received a copy of the GNU Lesser General Public
13    License along with this library; if not, write to the
14    Free Software Foundation, Inc., 59 Temple Place, Suite 330,
15    Boston, MA 02111-1307  USA
16 
17    You can contact Elliotte Rusty Harold by sending e-mail to
18    elharo@ibiblio.org. Please include the word "XOM" in the
19    subject line. The XOM home page is located at http://www.xom.nu/
20 */
21 
22 package nu.xom;
23 
24 import java.io.Writer;
25 
26 /**
27  * <p>
28  *   ISO 8859-4, Latin plus the characters needed for Greenlandic,
29  *   Icelandic, and Lappish.
30  * </p>
31  *
32  * @author Elliotte Rusty Harold
33  * @version 1.0
34  *
35  */
36 class Latin4Writer extends TextWriter {
37 
Latin4Writer(Writer out, String encoding)38     Latin4Writer(Writer out, String encoding) {
39         super(out, encoding);
40     }
41 
42     /**
43      * @see nu.xom.TextWriter#needsEscaping(char)
44      */
needsEscaping(char c)45     boolean needsEscaping(char c) {
46         if (c <= 0xA0) return false;
47         switch (c) {
48             case 0x00A1: return true;  // place holder to enable table lookup
49             case 0x00A2: return true;  // place holder to enable table lookup
50             case 0x00A3: return true;  // place holder to enable table lookup
51             case 0x00A4: return false; // CURRENCY SIGN
52             case 0x00A5: return true;  // place holder to enable table lookup
53             case 0x00A6: return true;  // place holder to enable table lookup
54             case 0x00A7: return false; // SECTION SIGN
55             case 0x00A8: return false; // DIAERESIS
56             case 0x00A9: return true;  // place holder to enable table lookup
57             case 0x00AA: return true;  // place holder to enable table lookup
58             case 0x00AB: return true;  // place holder to enable table lookup
59             case 0x00AC: return true;  // place holder to enable table lookup
60             case 0x00AD: return false; // SOFT HYPHEN
61             case 0x00AE: return true;  // place holder to enable table lookup
62             case 0x00AF: return false; // MACRON
63             case 0x00B0: return false; // DEGREE SIGN
64             case 0x00B1: return true;  // place holder to enable table lookup
65             case 0x00B2: return true;  // place holder to enable table lookup
66             case 0x00B3: return true;  // place holder to enable table lookup
67             case 0x00B4: return false; // ACUTE ACCENT
68             case 0x00B5: return true;  // place holder to enable table lookup
69             case 0x00B6: return true;  // place holder to enable table lookup
70             case 0x00B7: return true;  // place holder to enable table lookup
71             case 0x00B8: return false; // CEDILLA
72             case 0x00B9: return true;  // place holder to enable table lookup
73             case 0x00BA: return true;  // place holder to enable table lookup
74             case 0x00BB: return true;  // place holder to enable table lookup
75             case 0x00BC: return true;  // place holder to enable table lookup
76             case 0x00BD: return true;  // place holder to enable table lookup
77             case 0x00BE: return true;  // place holder to enable table lookup
78             case 0x00BF: return true;  // place holder to enable table lookup
79             case 0x00C0: return true;  // place holder to enable table lookup
80             case 0x00C1: return false; // LATIN CAPITAL LETTER A WITH ACUTE
81             case 0x00C2: return false; // LATIN CAPITAL LETTER A WITH CIRCUMFLEX
82             case 0x00C3: return false; // LATIN CAPITAL LETTER A WITH TILDE
83             case 0x00C4: return false; // LATIN CAPITAL LETTER A WITH DIAERESIS
84             case 0x00C5: return false; // LATIN CAPITAL LETTER A WITH RING ABOVE
85             case 0x00C6: return false; // LATIN CAPITAL LETTER AE
86             case 0x00C7: return true;  // place holder to enable table lookup
87             case 0x00C8: return true;  // place holder to enable table lookup
88             case 0x00C9: return false; // LATIN CAPITAL LETTER E WITH ACUTE
89             case 0x00CA: return true;  // place holder to enable table lookup
90             case 0x00CB: return false; // LATIN CAPITAL LETTER E WITH DIAERESIS
91             case 0x00CC: return true;  // place holder to enable table lookup
92             case 0x00CD: return false; // LATIN CAPITAL LETTER I WITH ACUTE
93             case 0x00CE: return false; // LATIN CAPITAL LETTER I WITH CIRCUMFLEX
94             case 0x00CF: return true;  // place holder to enable table lookup
95             case 0x00D0: return true;  // place holder to enable table lookup
96             case 0x00D1: return true;  // place holder to enable table lookup
97             case 0x00D2: return true;  // place holder to enable table lookup
98             case 0x00D3: return true;  // place holder to enable table lookup
99             case 0x00D4: return false; // LATIN CAPITAL LETTER O WITH CIRCUMFLEX
100             case 0x00D5: return false; // LATIN CAPITAL LETTER O WITH TILDE
101             case 0x00D6: return false; // LATIN CAPITAL LETTER O WITH DIAERESIS
102             case 0x00D7: return false; // MULTIPLICATION SIGN
103             case 0x00D8: return false; // LATIN CAPITAL LETTER O WITH STROKE
104             case 0x00D9: return true;  // place holder to enable table lookup
105             case 0x00DA: return false; // LATIN CAPITAL LETTER U WITH ACUTE
106             case 0x00DB: return false; // LATIN CAPITAL LETTER U WITH CIRCUMFLEX
107             case 0x00DC: return false; // LATIN CAPITAL LETTER U WITH DIAERESIS
108             case 0x00DD: return true;  // place holder to enable table lookup
109             case 0x00DE: return true;  // place holder to enable table lookup
110             case 0x00DF: return false; // LATIN SMALL LETTER SHARP S
111             case 0x00E0: return true;  // place holder to enable table lookup
112             case 0x00E1: return false; // LATIN SMALL LETTER A WITH ACUTE
113             case 0x00E2: return false; // LATIN SMALL LETTER A WITH CIRCUMFLEX
114             case 0x00E3: return false; // LATIN SMALL LETTER A WITH TILDE
115             case 0x00E4: return false; // LATIN SMALL LETTER A WITH DIAERESIS
116             case 0x00E5: return false; // LATIN SMALL LETTER A WITH RING ABOVE
117             case 0x00E6: return false; // LATIN SMALL LETTER AE
118             case 0x00E7: return true;  // place holder to enable table lookup
119             case 0x00E8: return true;  // place holder to enable table lookup
120             case 0x00E9: return false; // LATIN SMALL LETTER E WITH ACUTE
121             case 0x00EA: return true;  // place holder to enable table lookup
122             case 0x00EB: return false; // LATIN SMALL LETTER E WITH DIAERESIS
123             case 0x00EC: return true;  // place holder to enable table lookup
124             case 0x00ED: return false; // LATIN SMALL LETTER I WITH ACUTE
125             case 0x00EE: return false; // LATIN SMALL LETTER I WITH CIRCUMFLEX
126             case 0x00EF: return true;  // place holder to enable table lookup
127             case 0x00F0: return true;  // place holder to enable table lookup
128             case 0x00F1: return true;  // place holder to enable table lookup
129             case 0x00F2: return true;  // place holder to enable table lookup
130             case 0x00F3: return true;  // place holder to enable table lookup
131             case 0x00F4: return false; // LATIN SMALL LETTER O WITH CIRCUMFLEX
132             case 0x00F5: return false; // LATIN SMALL LETTER O WITH TILDE
133             case 0x00F6: return false; // LATIN SMALL LETTER O WITH DIAERESIS
134             case 0x00F7: return false; // DIVISION SIGN
135             case 0x00F8: return false; // LATIN SMALL LETTER O WITH STROKE
136             case 0x00F9: return true;  // place holder to enable table lookup
137             case 0x00FA: return false; // LATIN SMALL LETTER U WITH ACUTE
138             case 0x00FB: return false; // LATIN SMALL LETTER U WITH CIRCUMFLEX
139             case 0x00FC: return false; // LATIN SMALL LETTER U WITH DIAERESIS
140             case 0x00FD: return true;  // place holder to enable table lookup
141             case 0x00FE: return true;  // place holder to enable table lookup
142             case 0x00FF: return true;  // place holder to enable table lookup
143             case 0x0100: return false; // LATIN CAPITAL LETTER A WITH MACRON
144             case 0x0101: return false; // LATIN SMALL LETTER A WITH MACRON
145             case 0x0102: return true;  // place holder to enable table lookup
146             case 0x0103: return true;  // place holder to enable table lookup
147             case 0x0104: return false; // LATIN CAPITAL LETTER A WITH OGONEK
148             case 0x0105: return false; // LATIN SMALL LETTER A WITH OGONEK
149             case 0x0106: return true;  // place holder to enable table lookup
150             case 0x0107: return true;  // place holder to enable table lookup
151             case 0x0108: return true;  // place holder to enable table lookup
152             case 0x0109: return true;  // place holder to enable table lookup
153             case 0x010A: return true;  // place holder to enable table lookup
154             case 0x010B: return true;  // place holder to enable table lookup
155             case 0x010C: return false; // LATIN CAPITAL LETTER C WITH CARON
156             case 0x010D: return false; // LATIN SMALL LETTER C WITH CARON
157             case 0x010E: return true;  // place holder to enable table lookup
158             case 0x010F: return true;  // place holder to enable table lookup
159             case 0x0110: return false; // LATIN CAPITAL LETTER D WITH STROKE
160             case 0x0111: return false; // LATIN SMALL LETTER D WITH STROKE
161             case 0x0112: return false; // LATIN CAPITAL LETTER E WITH MACRON
162             case 0x0113: return false; // LATIN SMALL LETTER E WITH MACRON
163             case 0x0114: return true;  // place holder to enable table lookup
164             case 0x0115: return true;  // place holder to enable table lookup
165             case 0x0116: return false; // LATIN CAPITAL LETTER E WITH DOT ABOVE
166             case 0x0117: return false; // LATIN SMALL LETTER E WITH DOT ABOVE
167             case 0x0118: return false; // LATIN CAPITAL LETTER E WITH OGONEK
168             case 0x0119: return false; // LATIN SMALL LETTER E WITH OGONEK
169             case 0x0120: return true;  // place holder to enable table lookup
170             case 0x0121: return true;  // place holder to enable table lookup
171             case 0x0122: return false; // LATIN CAPITAL LETTER G WITH CEDILLA
172             case 0x0123: return false; // LATIN SMALL LETTER G WITH CEDILLA
173             case 0x0124: return true;  // place holder to enable table lookup
174             case 0x0125: return true;  // place holder to enable table lookup
175             case 0x0126: return true;  // place holder to enable table lookup
176             case 0x0127: return true;  // place holder to enable table lookup
177             case 0x0128: return false; // LATIN CAPITAL LETTER I WITH TILDE
178             case 0x0129: return false; // LATIN SMALL LETTER I WITH TILDE
179             case 0x012A: return false; // LATIN CAPITAL LETTER I WITH MACRON
180             case 0x012B: return false; // LATIN SMALL LETTER I WITH MACRON
181             case 0x012C: return true;  // place holder to enable table lookup
182             case 0x012D: return true;  // place holder to enable table lookup
183             case 0x012E: return false; // LATIN CAPITAL LETTER I WITH OGONEK
184             case 0x012F: return false; // LATIN SMALL LETTER I WITH OGONEK
185             case 0x0130: return true;  // place holder to enable table lookup
186             case 0x0131: return true;  // place holder to enable table lookup
187             case 0x0132: return true;  // place holder to enable table lookup
188             case 0x0133: return true;  // place holder to enable table lookup
189             case 0x0134: return true;  // place holder to enable table lookup
190             case 0x0135: return true;  // place holder to enable table lookup
191             case 0x0136: return false; // LATIN CAPITAL LETTER K WITH CEDILLA
192             case 0x0137: return false; // LATIN SMALL LETTER K WITH CEDILLA
193             case 0x0138: return false; // LATIN SMALL LETTER KRA
194             case 0x0139: return true;  // place holder to enable table lookup
195             case 0x013A: return true;  // place holder to enable table lookup
196             case 0x013B: return false; // LATIN CAPITAL LETTER L WITH CEDILLA
197             case 0x013C: return false; // LATIN SMALL LETTER L WITH CEDILLA
198             case 0x013D: return true;  // place holder to enable table lookup
199             case 0x013E: return true;  // place holder to enable table lookup
200             case 0x013F: return true;  // place holder to enable table lookup
201             case 0x0140: return true;  // place holder to enable table lookup
202             case 0x0141: return true;  // place holder to enable table lookup
203             case 0x0142: return true;  // place holder to enable table lookup
204             case 0x0143: return true;  // place holder to enable table lookup
205             case 0x0144: return true;  // place holder to enable table lookup
206             case 0x0145: return false; // LATIN CAPITAL LETTER N WITH CEDILLA
207             case 0x0146: return false; // LATIN SMALL LETTER N WITH CEDILLA
208             case 0x0147: return true;  // place holder to enable table lookup
209             case 0x0148: return true;  // place holder to enable table lookup
210             case 0x0149: return true;  // place holder to enable table lookup
211             case 0x014A: return false; // LATIN CAPITAL LETTER ENG
212             case 0x014B: return false; // LATIN SMALL LETTER ENG
213             case 0x014C: return false; // LATIN CAPITAL LETTER O WITH MACRON
214             case 0x014D: return false; // LATIN SMALL LETTER O WITH MACRON
215             case 0x014E: return true;  // place holder to enable table lookup
216             case 0x014F: return true;  // place holder to enable table lookup
217             case 0x0150: return true;  // place holder to enable table lookup
218             case 0x0151: return true;  // place holder to enable table lookup
219             case 0x0152: return true;  // place holder to enable table lookup
220             case 0x0153: return true;  // place holder to enable table lookup
221             case 0x0154: return true;  // place holder to enable table lookup
222             case 0x0155: return true;  // place holder to enable table lookup
223             case 0x0156: return false; // LATIN CAPITAL LETTER R WITH CEDILLA
224             case 0x0157: return false; // LATIN SMALL LETTER R WITH CEDILLA
225             case 0x0158: return true;  // place holder to enable table lookup
226             case 0x0159: return true;  // place holder to enable table lookup
227             case 0x015A: return true;  // place holder to enable table lookup
228             case 0x015B: return true;  // place holder to enable table lookup
229             case 0x015C: return true;  // place holder to enable table lookup
230             case 0x015D: return true;  // place holder to enable table lookup
231             case 0x015E: return true;  // place holder to enable table lookup
232             case 0x015F: return true;  // place holder to enable table lookup
233             case 0x0160: return false; // LATIN CAPITAL LETTER S WITH CARON
234             case 0x0161: return false; // LATIN SMALL LETTER S WITH CARON
235             case 0x0162: return true;  // place holder to enable table lookup
236             case 0x0163: return true;  // place holder to enable table lookup
237             case 0x0164: return true;  // place holder to enable table lookup
238             case 0x0165: return true;  // place holder to enable table lookup
239             case 0x0166: return false; // LATIN CAPITAL LETTER T WITH STROKE
240             case 0x0167: return false; // LATIN SMALL LETTER T WITH STROKE
241             case 0x0168: return false; // LATIN CAPITAL LETTER U WITH TILDE
242             case 0x0169: return false; // LATIN SMALL LETTER U WITH TILDE
243             case 0x016A: return false; // LATIN CAPITAL LETTER U WITH MACRON
244             case 0x016B: return false; // LATIN SMALL LETTER U WITH MACRON
245             case 0x016C: return true;  // place holder to enable table lookup
246             case 0x016D: return true;  // place holder to enable table lookup
247             case 0x016E: return true;  // place holder to enable table lookup
248             case 0x016F: return true;  // place holder to enable table lookup
249             case 0x0170: return true;  // place holder to enable table lookup
250             case 0x0171: return true;  // place holder to enable table lookup
251             case 0x0172: return false; // LATIN CAPITAL LETTER U WITH OGONEK
252             case 0x0173: return false; // LATIN SMALL LETTER U WITH OGONEK
253             case 0x0174: return true;  // place holder to enable table lookup
254             case 0x0175: return true;  // place holder to enable table lookup
255             case 0x0176: return true;  // place holder to enable table lookup
256             case 0x0177: return true;  // place holder to enable table lookup
257             case 0x0178: return true;  // place holder to enable table lookup
258             case 0x0179: return true;  // place holder to enable table lookup
259             case 0x017A: return true;  // place holder to enable table lookup
260             case 0x017B: return true;  // place holder to enable table lookup
261             case 0x017C: return true;  // place holder to enable table lookup
262             case 0x017D: return false; // LATIN CAPITAL LETTER Z WITH CARON
263             case 0x017E: return false; // LATIN SMALL LETTER Z WITH CARON
264         }
265         switch (c) {
266             case 0x02C7: return false; // CARON
267             case 0x02C8: return true;  // place holder to enable table lookup
268             case 0x02C9: return true;  // place holder to enable table lookup
269             case 0x02CA: return true;  // place holder to enable table lookup
270             case 0x02CB: return true;  // place holder to enable table lookup
271             case 0x02CC: return true;  // place holder to enable table lookup
272             case 0x02CD: return true;  // place holder to enable table lookup
273             case 0x02CE: return true;  // place holder to enable table lookup
274             case 0x02CF: return true;  // place holder to enable table lookup
275             case 0x02D0: return true;  // place holder to enable table lookup
276             case 0x02D1: return true;  // place holder to enable table lookup
277             case 0x02D2: return true;  // place holder to enable table lookup
278             case 0x02D3: return true;  // place holder to enable table lookup
279             case 0x02D4: return true;  // place holder to enable table lookup
280             case 0x02D5: return true;  // place holder to enable table lookup
281             case 0x02D6: return true;  // place holder to enable table lookup
282             case 0x02D7: return true;  // place holder to enable table lookup
283             case 0x02D8: return true;  // place holder to enable table lookup
284             case 0x02D9: return false; // DOT ABOVE
285             case 0x02DA: return true;  // place holder to enable table lookup
286             case 0x02DB: return false; // OGONEK
287         }
288 
289         return true;
290 
291     }
292 
293 }