KickJava   Java API By Example, From Geeks To Geeks.

Java > Open Source Codes > org > w3c > tidy > EntityTable


1 /*
2  * @(#)EntityTable.java 1.11 2000/08/16
3  *
4  */
5
6 package org.w3c.tidy;
7
8 /**
9  *
10  * Entity hash table
11  *
12  * (c) 1998-2000 (W3C) MIT, INRIA, Keio University
13  * See Tidy.java for the copyright notice.
14  * Derived from <a HREF="http://www.w3.org/People/Raggett/tidy">
15  * HTML Tidy Release 4 Aug 2000</a>
16  *
17  * @author Dave Raggett <dsr@w3.org>
18  * @author Andy Quick <ac.quick@sympatico.ca> (translation to Java)
19  * @version 1.0, 1999/05/22
20  * @version 1.0.1, 1999/05/29
21  * @version 1.1, 1999/06/18 Java Bean
22  * @version 1.2, 1999/07/10 Tidy Release 7 Jul 1999
23  * @version 1.3, 1999/07/30 Tidy Release 26 Jul 1999
24  * @version 1.4, 1999/09/04 DOM support
25  * @version 1.5, 1999/10/23 Tidy Release 27 Sep 1999
26  * @version 1.6, 1999/11/01 Tidy Release 22 Oct 1999
27  * @version 1.7, 1999/12/06 Tidy Release 30 Nov 1999
28  * @version 1.8, 2000/01/22 Tidy Release 13 Jan 2000
29  * @version 1.9, 2000/06/03 Tidy Release 30 Apr 2000
30  * @version 1.10, 2000/07/22 Tidy Release 8 Jul 2000
31  * @version 1.11, 2000/08/16 Tidy Release 4 Aug 2000
32  */
33
34 import java.util.Hashtable JavaDoc;
35 import java.util.Enumeration JavaDoc;
36
37 public class EntityTable {
38
39     public EntityTable()
40     {
41     }
42
43     public Entity lookup( String JavaDoc name )
44     {
45         return (Entity)entityHashtable.get( name );
46     }
47
48     public Entity install( String JavaDoc name, short code )
49     {
50         Entity ent = lookup( name );
51         if ( ent == null ) {
52             ent = new Entity( name, code );
53             entityHashtable.put( name, ent );
54         } else {
55             ent.code = code;
56         }
57         return ent;
58     }
59
60     public Entity install( Entity ent )
61     {
62         return (Entity)entityHashtable.put( ent.name, ent );
63     }
64
65     /* entity starting with "&" returns zero on error */
66     public short entityCode( String JavaDoc name )
67     {
68         int c;
69
70         if (name.length() <= 1)
71             return 0;
72
73         /* numeric entitity: name = "&#" followed by number */
74         if ( name.charAt(1) == '#' ) {
75             c = 0; /* zero on missing/bad number */
76
77             /* 'x' prefix denotes hexadecimal number format */
78             try {
79                 if (name.length() >= 4 && name.charAt(2) == 'x') {
80                     c = Integer.parseInt( name.substring(3), 16 );
81                 } else if (name.length() >= 3) {
82                     c = Integer.parseInt( name.substring(2) );
83                 }
84             }
85             catch ( NumberFormatException JavaDoc e ) {}
86
87             return (short)c;
88         }
89
90         /* Named entity: name ="&" followed by a name */
91         Entity ent = lookup( name.substring(1) );
92         if ( ent != null ) {
93             return ent.code;
94         }
95
96         return 0; /* zero signifies unknown entity name */
97     }
98
99     public String JavaDoc entityName( short code )
100     {
101         String JavaDoc name = null;
102         Entity ent;
103         Enumeration JavaDoc en = entityHashtable.elements();
104         while ( en.hasMoreElements() ) {
105             ent = (Entity)en.nextElement();
106             if ( ent.code == code ) {
107                 name = ent.name;
108                 break;
109             }
110         }
111         return name;
112     }
113
114     private Hashtable JavaDoc entityHashtable = new Hashtable JavaDoc();
115
116     private static EntityTable defaultEntityTable = null;
117
118     private static Entity[] entities = {
119
120     new Entity( "nbsp", 160 ),
121     new Entity( "iexcl", 161 ),
122     new Entity( "cent", 162 ),
123     new Entity( "pound", 163 ),
124     new Entity( "curren", 164 ),
125     new Entity( "yen", 165 ),
126     new Entity( "brvbar", 166 ),
127     new Entity( "sect", 167 ),
128     new Entity( "uml", 168 ),
129     new Entity( "copy", 169 ),
130     new Entity( "ordf", 170 ),
131     new Entity( "laquo", 171 ),
132     new Entity( "not", 172 ),
133     new Entity( "shy", 173 ),
134     new Entity( "reg", 174 ),
135     new Entity( "macr", 175 ),
136     new Entity( "deg", 176 ),
137     new Entity( "plusmn", 177 ),
138     new Entity( "sup2", 178 ),
139     new Entity( "sup3", 179 ),
140     new Entity( "acute", 180 ),
141     new Entity( "micro", 181 ),
142     new Entity( "para", 182 ),
143     new Entity( "middot", 183 ),
144     new Entity( "cedil", 184 ),
145     new Entity( "sup1", 185 ),
146     new Entity( "ordm", 186 ),
147     new Entity( "raquo", 187 ),
148     new Entity( "frac14", 188 ),
149     new Entity( "frac12", 189 ),
150     new Entity( "frac34", 190 ),
151     new Entity( "iquest", 191 ),
152     new Entity( "Agrave", 192 ),
153     new Entity( "Aacute", 193 ),
154     new Entity( "Acirc", 194 ),
155     new Entity( "Atilde", 195 ),
156     new Entity( "Auml", 196 ),
157     new Entity( "Aring", 197 ),
158     new Entity( "AElig", 198 ),
159     new Entity( "Ccedil", 199 ),
160     new Entity( "Egrave", 200 ),
161     new Entity( "Eacute", 201 ),
162     new Entity( "Ecirc", 202 ),
163     new Entity( "Euml", 203 ),
164     new Entity( "Igrave", 204 ),
165     new Entity( "Iacute", 205 ),
166     new Entity( "Icirc", 206 ),
167     new Entity( "Iuml", 207 ),
168     new Entity( "ETH", 208 ),
169     new Entity( "Ntilde", 209 ),
170     new Entity( "Ograve", 210 ),
171     new Entity( "Oacute", 211 ),
172     new Entity( "Ocirc", 212 ),
173     new Entity( "Otilde", 213 ),
174     new Entity( "Ouml", 214 ),
175     new Entity( "times", 215 ),
176     new Entity( "Oslash", 216 ),
177     new Entity( "Ugrave", 217 ),
178     new Entity( "Uacute", 218 ),
179     new Entity( "Ucirc", 219 ),
180     new Entity( "Uuml", 220 ),
181     new Entity( "Yacute", 221 ),
182     new Entity( "THORN", 222 ),
183     new Entity( "szlig", 223 ),
184     new Entity( "agrave", 224 ),
185     new Entity( "aacute", 225 ),
186     new Entity( "acirc", 226 ),
187     new Entity( "atilde", 227 ),
188     new Entity( "auml", 228 ),
189     new Entity( "aring", 229 ),
190     new Entity( "aelig", 230 ),
191     new Entity( "ccedil", 231 ),
192     new Entity( "egrave", 232 ),
193     new Entity( "eacute", 233 ),
194     new Entity( "ecirc", 234 ),
195     new Entity( "euml", 235 ),
196     new Entity( "igrave", 236 ),
197     new Entity( "iacute", 237 ),
198     new Entity( "icirc", 238 ),
199     new Entity( "iuml", 239 ),
200     new Entity( "eth", 240 ),
201     new Entity( "ntilde", 241 ),
202     new Entity( "ograve", 242 ),
203     new Entity( "oacute", 243 ),
204     new Entity( "ocirc", 244 ),
205     new Entity( "otilde", 245 ),
206     new Entity( "ouml", 246 ),
207     new Entity( "divide", 247 ),
208     new Entity( "oslash", 248 ),
209     new Entity( "ugrave", 249 ),
210     new Entity( "uacute", 250 ),
211     new Entity( "ucirc", 251 ),
212     new Entity( "uuml", 252 ),
213     new Entity( "yacute", 253 ),
214     new Entity( "thorn", 254 ),
215     new Entity( "yuml", 255 ),
216     new Entity( "fnof", 402 ),
217     new Entity( "Alpha", 913 ),
218     new Entity( "Beta", 914 ),
219     new Entity( "Gamma", 915 ),
220     new Entity( "Delta", 916 ),
221     new Entity( "Epsilon", 917 ),
222     new Entity( "Zeta", 918 ),
223     new Entity( "Eta", 919 ),
224     new Entity( "Theta", 920 ),
225     new Entity( "Iota", 921 ),
226     new Entity( "Kappa", 922 ),
227     new Entity( "Lambda", 923 ),
228     new Entity( "Mu", 924 ),
229     new Entity( "Nu", 925 ),
230     new Entity( "Xi", 926 ),
231     new Entity( "Omicron", 927 ),
232     new Entity( "Pi", 928 ),
233     new Entity( "Rho", 929 ),
234     new Entity( "Sigma", 931 ),
235     new Entity( "Tau", 932 ),
236     new Entity( "Upsilon", 933 ),
237     new Entity( "Phi", 934 ),
238     new Entity( "Chi", 935 ),
239     new Entity( "Psi", 936 ),
240     new Entity( "Omega", 937 ),
241     new Entity( "alpha", 945 ),
242     new Entity( "beta", 946 ),
243     new Entity( "gamma", 947 ),
244     new Entity( "delta", 948 ),
245     new Entity( "epsilon", 949 ),
246     new Entity( "zeta", 950 ),
247     new Entity( "eta", 951 ),
248     new Entity( "theta", 952 ),
249     new Entity( "iota", 953 ),
250     new Entity( "kappa", 954 ),
251     new Entity( "lambda", 955 ),
252     new Entity( "mu", 956 ),
253     new Entity( "nu", 957 ),
254     new Entity( "xi", 958 ),
255     new Entity( "omicron", 959 ),
256     new Entity( "pi", 960 ),
257     new Entity( "rho", 961 ),
258     new Entity( "sigmaf", 962 ),
259     new Entity( "sigma", 963 ),
260     new Entity( "tau", 964 ),
261     new Entity( "upsilon", 965 ),
262     new Entity( "phi", 966 ),
263     new Entity( "chi", 967 ),
264     new Entity( "psi", 968 ),
265     new Entity( "omega", 969 ),
266     new Entity( "thetasym", 977 ),
267     new Entity( "upsih", 978 ),
268     new Entity( "piv", 982 ),
269     new Entity( "bull", 8226 ),
270     new Entity( "hellip", 8230 ),
271     new Entity( "prime", 8242 ),
272     new Entity( "Prime", 8243 ),
273     new Entity( "oline", 8254 ),
274     new Entity( "frasl", 8260 ),
275     new Entity( "weierp", 8472 ),
276     new Entity( "image", 8465 ),
277     new Entity( "real", 8476 ),
278     new Entity( "trade", 8482 ),
279     new Entity( "alefsym", 8501 ),
280     new Entity( "larr", 8592 ),
281     new Entity( "uarr", 8593 ),
282     new Entity( "rarr", 8594 ),
283     new Entity( "darr", 8595 ),
284     new Entity( "harr", 8596 ),
285     new Entity( "crarr", 8629 ),
286     new Entity( "lArr", 8656 ),
287     new Entity( "uArr", 8657 ),
288     new Entity( "rArr", 8658 ),
289     new Entity( "dArr", 8659 ),
290     new Entity( "hArr", 8660 ),
291     new Entity( "forall", 8704 ),
292     new Entity( "part", 8706 ),
293     new Entity( "exist", 8707 ),
294     new Entity( "empty", 8709 ),
295     new Entity( "nabla", 8711 ),
296     new Entity( "isin", 8712 ),
297     new Entity( "notin", 8713 ),
298     new Entity( "ni", 8715 ),
299     new Entity( "prod", 8719 ),
300     new Entity( "sum", 8721 ),
301     new Entity( "minus", 8722 ),
302     new Entity( "lowast", 8727 ),
303     new Entity( "radic", 8730 ),
304     new Entity( "prop", 8733 ),
305     new Entity( "infin", 8734 ),
306     new Entity( "ang", 8736 ),
307     new Entity( "and", 8743 ),
308     new Entity( "or", 8744 ),
309     new Entity( "cap", 8745 ),
310     new Entity( "cup", 8746 ),
311     new Entity( "int", 8747 ),
312     new Entity( "there4", 8756 ),
313     new Entity( "sim", 8764 ),
314     new Entity( "cong", 8773 ),
315     new Entity( "asymp", 8776 ),
316     new Entity( "ne", 8800 ),
317     new Entity( "equiv", 8801 ),
318     new Entity( "le", 8804 ),
319     new Entity( "ge", 8805 ),
320     new Entity( "sub", 8834 ),
321     new Entity( "sup", 8835 ),
322     new Entity( "nsub", 8836 ),
323     new Entity( "sube", 8838 ),
324     new Entity( "supe", 8839 ),
325     new Entity( "oplus", 8853 ),
326     new Entity( "otimes", 8855 ),
327     new Entity( "perp", 8869 ),
328     new Entity( "sdot", 8901 ),
329     new Entity( "lceil", 8968 ),
330     new Entity( "rceil", 8969 ),
331     new Entity( "lfloor", 8970 ),
332     new Entity( "rfloor", 8971 ),
333     new Entity( "lang", 9001 ),
334     new Entity( "rang", 9002 ),
335     new Entity( "loz", 9674 ),
336     new Entity( "spades", 9824 ),
337     new Entity( "clubs", 9827 ),
338     new Entity( "hearts", 9829 ),
339     new Entity( "diams", 9830 ),
340     new Entity( "quot", 34 ),
341     new Entity( "amp", 38 ),
342     new Entity( "lt", 60 ),
343     new Entity( "gt", 62 ),
344     new Entity( "OElig", 338 ),
345     new Entity( "oelig", 339 ),
346     new Entity( "Scaron", 352 ),
347     new Entity( "scaron", 353 ),
348     new Entity( "Yuml", 376 ),
349     new Entity( "circ", 710 ),
350     new Entity( "tilde", 732 ),
351     new Entity( "ensp", 8194 ),
352     new Entity( "emsp", 8195 ),
353     new Entity( "thinsp", 8201 ),
354     new Entity( "zwnj", 8204 ),
355     new Entity( "zwj", 8205 ),
356     new Entity( "lrm", 8206 ),
357     new Entity( "rlm", 8207 ),
358     new Entity( "ndash", 8211 ),
359     new Entity( "mdash", 8212 ),
360     new Entity( "lsquo", 8216 ),
361     new Entity( "rsquo", 8217 ),
362     new Entity( "sbquo", 8218 ),
363     new Entity( "ldquo", 8220 ),
364     new Entity( "rdquo", 8221 ),
365     new Entity( "bdquo", 8222 ),
366     new Entity( "dagger", 8224 ),
367     new Entity( "Dagger", 8225 ),
368     new Entity( "permil", 8240 ),
369     new Entity( "lsaquo", 8249 ),
370     new Entity( "rsaquo", 8250 ),
371     new Entity( "euro", 8364 )
372
373     };
374
375     public static EntityTable getDefaultEntityTable()
376     {
377         if ( defaultEntityTable == null ) {
378             defaultEntityTable = new EntityTable();
379             for ( int i = 0; i < entities.length; i++ ) {
380                 defaultEntityTable.install( entities[i] );
381             }
382         }
383         return defaultEntityTable;
384     }
385
386 }
387
Popular Tags