1 /* The following code was generated by JFlex 1.5.0-SNAPSHOT on 10/3/10 8:05 AM */
3 package org.apache.lucene.analysis.wikipedia;
6 * Licensed to the Apache Software Foundation (ASF) under one or more
7 * contributor license agreements. See the NOTICE file distributed with
8 * this work for additional information regarding copyright ownership.
9 * The ASF licenses this file to You under the Apache License, Version 2.0
10 * (the "License"); you may not use this file except in compliance with
11 * the License. You may obtain a copy of the License at
13 * http://www.apache.org/licenses/LICENSE-2.0
15 * Unless required by applicable law or agreed to in writing, software
16 * distributed under the License is distributed on an "AS IS" BASIS,
17 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
18 * See the License for the specific language governing permissions and
19 * limitations under the License.
22 import org.apache.lucene.analysis.tokenattributes.CharTermAttribute;
26 * This class is a scanner generated by
27 * <a href="http://www.jflex.de/">JFlex</a> 1.5.0-SNAPSHOT
28 * on 10/3/10 8:05 AM from the specification file
29 * <tt>C:/Users/rmuir/workspace/lucene_3xclean/lucene/contrib/analyzers/common/src/java/org/apache/lucene/analysis/wikipedia/WikipediaTokenizerImpl.jflex</tt>
31 class WikipediaTokenizerImpl {
33 /** This character denotes the end of file */
34 public static final int YYEOF = -1;
36 /** initial size of the lookahead buffer */
37 private static final int ZZ_BUFFERSIZE = 16384;
40 public static final int THREE_SINGLE_QUOTES_STATE = 10;
41 public static final int EXTERNAL_LINK_STATE = 6;
42 public static final int DOUBLE_EQUALS_STATE = 14;
43 public static final int INTERNAL_LINK_STATE = 4;
44 public static final int DOUBLE_BRACE_STATE = 16;
45 public static final int CATEGORY_STATE = 2;
46 public static final int YYINITIAL = 0;
47 public static final int STRING = 18;
48 public static final int FIVE_SINGLE_QUOTES_STATE = 12;
49 public static final int TWO_SINGLE_QUOTES_STATE = 8;
52 * ZZ_LEXSTATE[l] is the state in the DFA for the lexical state l
53 * ZZ_LEXSTATE[l+1] is the state in the DFA for the lexical state l
54 * at the beginning of a line
55 * l is of the form l = 2*k, k a non negative integer
57 private static final int ZZ_LEXSTATE[] = {
58 0, 0, 1, 1, 2, 2, 3, 3, 4, 4, 5, 5, 6, 6, 7, 7,
63 * Translates characters to character classes
65 private static final String ZZ_CMAP_PACKED =
66 "\11\0\1\24\1\23\1\0\1\24\1\22\22\0\1\24\1\0\1\12"+
67 "\1\53\2\0\1\3\1\1\4\0\1\14\1\5\1\2\1\10\12\16"+
68 "\1\27\1\0\1\7\1\11\1\13\1\53\1\4\2\15\1\30\5\15"+
69 "\1\41\21\15\1\25\1\0\1\26\1\0\1\6\1\0\1\31\1\43"+
70 "\2\15\1\33\1\40\1\34\1\50\1\41\4\15\1\42\1\35\1\51"+
71 "\1\15\1\36\1\52\1\32\3\15\1\44\1\37\1\15\1\45\1\47"+
72 "\1\46\102\0\27\15\1\0\37\15\1\0\u0568\15\12\17\206\15\12\17"+
73 "\u026c\15\12\17\166\15\12\17\166\15\12\17\166\15\12\17\166\15\12\17"+
74 "\167\15\11\17\166\15\12\17\166\15\12\17\166\15\12\17\340\15\12\17"+
75 "\166\15\12\17\u0166\15\12\17\266\15\u0100\15\u0e00\15\u1040\0\u0150\21\140\0"+
76 "\20\21\u0100\0\200\21\200\0\u19c0\21\100\0\u5200\21\u0c00\0\u2bb0\20\u2150\0"+
77 "\u0200\21\u0465\0\73\21\75\15\43\0";
80 * Translates characters to character classes
82 private static final char [] ZZ_CMAP = zzUnpackCMap(ZZ_CMAP_PACKED);
85 * Translates DFA states to action switch labels.
87 private static final int [] ZZ_ACTION = zzUnpackAction();
89 private static final String ZZ_ACTION_PACKED_0 =
90 "\12\0\4\1\4\2\1\3\1\1\1\4\1\1\2\5"+
91 "\1\6\2\5\1\7\1\5\2\10\1\11\1\12\1\11"+
92 "\1\13\1\14\1\10\1\15\1\16\1\15\1\17\1\20"+
93 "\1\10\1\21\1\10\4\22\1\23\1\22\1\24\1\25"+
94 "\1\26\3\0\1\27\14\0\1\30\1\31\1\32\1\33"+
95 "\1\11\1\0\1\34\1\35\1\0\1\36\1\0\1\37"+
96 "\3\0\1\40\1\41\2\42\1\41\2\43\2\0\1\42"+
97 "\1\0\14\42\1\41\3\0\1\11\1\44\3\0\1\45"+
98 "\1\46\5\0\1\47\4\0\1\47\2\0\2\47\2\0"+
99 "\1\11\5\0\1\31\1\41\1\42\1\50\3\0\1\11"+
100 "\2\0\1\51\30\0\1\52\2\0\1\53\1\54\1\55";
102 private static int [] zzUnpackAction() {
103 int [] result = new int[183];
105 offset = zzUnpackAction(ZZ_ACTION_PACKED_0, offset, result);
109 private static int zzUnpackAction(String packed, int offset, int [] result) {
110 int i = 0; /* index in packed string */
111 int j = offset; /* index in unpacked array */
112 int l = packed.length();
114 int count = packed.charAt(i++);
115 int value = packed.charAt(i++);
116 do result[j++] = value; while (--count > 0);
123 * Translates a state to a row index in the transition table
125 private static final int [] ZZ_ROWMAP = zzUnpackRowMap();
127 private static final String ZZ_ROWMAP_PACKED_0 =
128 "\0\0\0\54\0\130\0\204\0\260\0\334\0\u0108\0\u0134"+
129 "\0\u0160\0\u018c\0\u01b8\0\u01e4\0\u0210\0\u023c\0\u0268\0\u0294"+
130 "\0\u02c0\0\u02ec\0\u01b8\0\u0318\0\u0344\0\u0370\0\u01b8\0\u039c"+
131 "\0\u03c8\0\u03f4\0\u0420\0\u044c\0\u0478\0\u01b8\0\u039c\0\u04a4"+
132 "\0\u01b8\0\u04d0\0\u04fc\0\u0528\0\u0554\0\u0580\0\u05ac\0\u05d8"+
133 "\0\u0604\0\u0630\0\u065c\0\u0688\0\u06b4\0\u01b8\0\u06e0\0\u039c"+
134 "\0\u070c\0\u0738\0\u0764\0\u0790\0\u01b8\0\u01b8\0\u07bc\0\u07e8"+
135 "\0\u0814\0\u01b8\0\u0840\0\u086c\0\u0898\0\u08c4\0\u08f0\0\u091c"+
136 "\0\u0948\0\u0974\0\u09a0\0\u09cc\0\u09f8\0\u0a24\0\u0a50\0\u0a7c"+
137 "\0\u01b8\0\u01b8\0\u0aa8\0\u0ad4\0\u0b00\0\u0b00\0\u0b2c\0\u0b58"+
138 "\0\u0b84\0\u0bb0\0\u0bdc\0\u0c08\0\u0c34\0\u0c60\0\u0c8c\0\u0cb8"+
139 "\0\u0ce4\0\u0d10\0\u0898\0\u0d3c\0\u0d68\0\u0d94\0\u0dc0\0\u0dec"+
140 "\0\u0e18\0\u0e44\0\u0e70\0\u0e9c\0\u0ec8\0\u0ef4\0\u0f20\0\u0f4c"+
141 "\0\u0f78\0\u0fa4\0\u0fd0\0\u0ffc\0\u1028\0\u1054\0\u1080\0\u10ac"+
142 "\0\u10d8\0\u01b8\0\u1104\0\u1130\0\u115c\0\u1188\0\u01b8\0\u11b4"+
143 "\0\u11e0\0\u120c\0\u1238\0\u1264\0\u1290\0\u12bc\0\u12e8\0\u1314"+
144 "\0\u1340\0\u136c\0\u1398\0\u13c4\0\u086c\0\u09f8\0\u13f0\0\u141c"+
145 "\0\u1448\0\u1474\0\u14a0\0\u14cc\0\u14f8\0\u1524\0\u01b8\0\u1550"+
146 "\0\u157c\0\u15a8\0\u15d4\0\u1600\0\u162c\0\u1658\0\u1684\0\u16b0"+
147 "\0\u01b8\0\u16dc\0\u1708\0\u1734\0\u1760\0\u178c\0\u17b8\0\u17e4"+
148 "\0\u1810\0\u183c\0\u1868\0\u1894\0\u18c0\0\u18ec\0\u1918\0\u1944"+
149 "\0\u1970\0\u199c\0\u19c8\0\u19f4\0\u1a20\0\u1a4c\0\u1a78\0\u1aa4"+
150 "\0\u1ad0\0\u1afc\0\u1b28\0\u1b54\0\u01b8\0\u01b8\0\u01b8";
152 private static int [] zzUnpackRowMap() {
153 int [] result = new int[183];
155 offset = zzUnpackRowMap(ZZ_ROWMAP_PACKED_0, offset, result);
159 private static int zzUnpackRowMap(String packed, int offset, int [] result) {
160 int i = 0; /* index in packed string */
161 int j = offset; /* index in unpacked array */
162 int l = packed.length();
164 int high = packed.charAt(i++) << 16;
165 result[j++] = high | packed.charAt(i++);
171 * The transition table of the DFA
173 private static final int [] ZZ_TRANS = zzUnpackTrans();
175 private static final String ZZ_TRANS_PACKED_0 =
176 "\1\13\1\14\5\13\1\15\1\13\1\16\3\13\1\17"+
177 "\1\20\1\21\1\22\1\23\1\24\2\13\1\25\2\13"+
178 "\15\17\1\26\2\13\3\17\1\13\7\27\1\30\5\27"+
179 "\4\31\1\27\1\32\3\27\1\33\1\27\15\31\3\27"+
180 "\3\31\10\27\1\30\5\27\4\34\1\27\1\32\3\27"+
181 "\1\35\1\27\15\34\3\27\3\34\1\27\7\36\1\37"+
182 "\5\36\4\40\1\36\1\32\2\27\1\36\1\41\1\36"+
183 "\15\40\3\36\1\42\2\40\2\36\1\43\5\36\1\37"+
184 "\5\36\4\44\1\36\1\45\2\36\1\46\2\36\15\44"+
185 "\3\36\3\44\10\36\1\37\5\36\4\47\1\36\1\45"+
186 "\2\36\1\46\2\36\15\47\3\36\3\47\10\36\1\37"+
187 "\5\36\4\47\1\36\1\45\2\36\1\50\2\36\15\47"+
188 "\3\36\3\47\10\36\1\37\1\36\1\51\3\36\4\52"+
189 "\1\36\1\45\5\36\15\52\3\36\3\52\10\36\1\53"+
190 "\5\36\4\54\1\36\1\45\5\36\15\54\1\36\1\55"+
191 "\1\36\3\54\1\36\1\56\1\57\5\56\1\60\1\56"+
192 "\1\61\3\56\4\62\1\56\1\63\2\56\1\64\2\56"+
193 "\15\62\2\56\1\65\3\62\1\56\55\0\1\66\62\0"+
194 "\1\67\4\0\4\70\7\0\6\70\1\71\6\70\3\0"+
195 "\3\70\12\0\1\72\43\0\1\73\1\74\1\75\1\76"+
196 "\2\77\1\0\1\100\3\0\1\100\1\17\1\20\1\21"+
197 "\1\22\7\0\15\17\3\0\3\17\3\0\1\101\1\0"+
198 "\1\102\2\103\1\0\1\104\3\0\1\104\3\20\1\22"+
199 "\7\0\15\20\3\0\3\20\2\0\1\73\1\105\1\75"+
200 "\1\76\2\103\1\0\1\104\3\0\1\104\1\21\1\20"+
201 "\1\21\1\22\7\0\15\21\3\0\3\21\3\0\1\106"+
202 "\1\0\1\102\2\77\1\0\1\100\3\0\1\100\4\22"+
203 "\7\0\15\22\3\0\3\22\24\0\1\13\55\0\1\107"+
204 "\73\0\1\110\16\0\1\67\4\0\4\70\7\0\15\70"+
205 "\3\0\3\70\16\0\4\31\7\0\15\31\3\0\3\31"+
206 "\24\0\1\27\56\0\1\111\42\0\4\34\7\0\15\34"+
207 "\3\0\3\34\27\0\1\112\42\0\4\40\7\0\15\40"+
208 "\3\0\3\40\16\0\4\40\7\0\2\40\1\113\12\40"+
209 "\3\0\3\40\2\0\1\114\67\0\4\44\7\0\15\44"+
210 "\3\0\3\44\24\0\1\36\55\0\1\115\43\0\4\47"+
211 "\7\0\15\47\3\0\3\47\26\0\1\116\37\0\1\111"+
212 "\57\0\4\52\7\0\15\52\3\0\3\52\11\0\1\117"+
213 "\4\0\4\70\7\0\15\70\3\0\3\70\16\0\4\54"+
214 "\7\0\15\54\3\0\3\54\47\0\1\111\6\0\1\120"+
215 "\63\0\1\121\57\0\4\62\7\0\15\62\3\0\3\62"+
216 "\24\0\1\56\55\0\1\122\43\0\4\70\7\0\15\70"+
217 "\3\0\3\70\14\0\1\36\1\0\4\123\1\0\3\124"+
218 "\3\0\15\123\3\0\3\123\14\0\1\36\1\0\4\123"+
219 "\1\0\3\124\3\0\3\123\1\125\11\123\3\0\3\123"+
220 "\16\0\1\126\1\0\1\126\10\0\15\126\3\0\3\126"+
221 "\16\0\1\127\1\130\1\131\1\132\7\0\15\127\3\0"+
222 "\3\127\16\0\1\133\1\0\1\133\10\0\15\133\3\0"+
223 "\3\133\16\0\1\134\1\135\1\134\1\135\7\0\15\134"+
224 "\3\0\3\134\16\0\1\136\2\137\1\140\7\0\15\136"+
225 "\3\0\3\136\16\0\1\100\2\141\10\0\15\100\3\0"+
226 "\3\100\16\0\1\142\2\143\1\144\7\0\15\142\3\0"+
227 "\3\142\16\0\4\135\7\0\15\135\3\0\3\135\16\0"+
228 "\1\145\2\146\1\147\7\0\15\145\3\0\3\145\16\0"+
229 "\1\150\2\151\1\152\7\0\15\150\3\0\3\150\16\0"+
230 "\1\153\1\143\1\154\1\144\7\0\15\153\3\0\3\153"+
231 "\16\0\1\155\2\130\1\132\7\0\15\155\3\0\3\155"+
232 "\30\0\1\156\1\157\64\0\1\160\27\0\4\40\7\0"+
233 "\2\40\1\161\12\40\3\0\3\40\2\0\1\162\101\0"+
234 "\1\163\1\164\40\0\4\70\7\0\6\70\1\165\6\70"+
235 "\3\0\3\70\2\0\1\166\63\0\1\167\71\0\1\170"+
236 "\1\171\34\0\1\172\1\0\1\36\1\0\4\123\1\0"+
237 "\3\124\3\0\15\123\3\0\3\123\16\0\4\173\1\0"+
238 "\3\124\3\0\15\173\3\0\3\173\12\0\1\172\1\0"+
239 "\1\36\1\0\4\123\1\0\3\124\3\0\10\123\1\174"+
240 "\4\123\3\0\3\123\2\0\1\73\13\0\1\126\1\0"+
241 "\1\126\10\0\15\126\3\0\3\126\3\0\1\175\1\0"+
242 "\1\102\2\176\6\0\1\127\1\130\1\131\1\132\7\0"+
243 "\15\127\3\0\3\127\3\0\1\177\1\0\1\102\2\200"+
244 "\1\0\1\201\3\0\1\201\3\130\1\132\7\0\15\130"+
245 "\3\0\3\130\3\0\1\202\1\0\1\102\2\200\1\0"+
246 "\1\201\3\0\1\201\1\131\1\130\1\131\1\132\7\0"+
247 "\15\131\3\0\3\131\3\0\1\203\1\0\1\102\2\176"+
248 "\6\0\4\132\7\0\15\132\3\0\3\132\3\0\1\204"+
249 "\2\0\1\204\7\0\1\134\1\135\1\134\1\135\7\0"+
250 "\15\134\3\0\3\134\3\0\1\204\2\0\1\204\7\0"+
251 "\4\135\7\0\15\135\3\0\3\135\3\0\1\176\1\0"+
252 "\1\102\2\176\6\0\1\136\2\137\1\140\7\0\15\136"+
253 "\3\0\3\136\3\0\1\200\1\0\1\102\2\200\1\0"+
254 "\1\201\3\0\1\201\3\137\1\140\7\0\15\137\3\0"+
255 "\3\137\3\0\1\176\1\0\1\102\2\176\6\0\4\140"+
256 "\7\0\15\140\3\0\3\140\3\0\1\201\2\0\2\201"+
257 "\1\0\1\201\3\0\1\201\3\141\10\0\15\141\3\0"+
258 "\3\141\3\0\1\106\1\0\1\102\2\77\1\0\1\100"+
259 "\3\0\1\100\1\142\2\143\1\144\7\0\15\142\3\0"+
260 "\3\142\3\0\1\101\1\0\1\102\2\103\1\0\1\104"+
261 "\3\0\1\104\3\143\1\144\7\0\15\143\3\0\3\143"+
262 "\3\0\1\106\1\0\1\102\2\77\1\0\1\100\3\0"+
263 "\1\100\4\144\7\0\15\144\3\0\3\144\3\0\1\77"+
264 "\1\0\1\102\2\77\1\0\1\100\3\0\1\100\1\145"+
265 "\2\146\1\147\7\0\15\145\3\0\3\145\3\0\1\103"+
266 "\1\0\1\102\2\103\1\0\1\104\3\0\1\104\3\146"+
267 "\1\147\7\0\15\146\3\0\3\146\3\0\1\77\1\0"+
268 "\1\102\2\77\1\0\1\100\3\0\1\100\4\147\7\0"+
269 "\15\147\3\0\3\147\3\0\1\100\2\0\2\100\1\0"+
270 "\1\100\3\0\1\100\1\150\2\151\1\152\7\0\15\150"+
271 "\3\0\3\150\3\0\1\104\2\0\2\104\1\0\1\104"+
272 "\3\0\1\104\3\151\1\152\7\0\15\151\3\0\3\151"+
273 "\3\0\1\100\2\0\2\100\1\0\1\100\3\0\1\100"+
274 "\4\152\7\0\15\152\3\0\3\152\3\0\1\205\1\0"+
275 "\1\102\2\77\1\0\1\100\3\0\1\100\1\153\1\143"+
276 "\1\154\1\144\7\0\15\153\3\0\3\153\3\0\1\206"+
277 "\1\0\1\102\2\103\1\0\1\104\3\0\1\104\1\154"+
278 "\1\143\1\154\1\144\7\0\15\154\3\0\3\154\3\0"+
279 "\1\203\1\0\1\102\2\176\6\0\1\155\2\130\1\132"+
280 "\7\0\15\155\3\0\3\155\31\0\1\157\54\0\1\207"+
281 "\64\0\1\210\26\0\4\40\7\0\15\40\3\0\1\40"+
282 "\1\211\1\40\31\0\1\164\54\0\1\212\35\0\1\36"+
283 "\1\0\4\123\1\0\3\124\3\0\3\123\1\213\11\123"+
284 "\3\0\3\123\2\0\1\214\102\0\1\171\54\0\1\215"+
285 "\34\0\1\216\52\0\1\172\3\0\4\173\7\0\15\173"+
286 "\3\0\3\173\12\0\1\172\1\0\1\217\1\0\4\123"+
287 "\1\0\3\124\3\0\15\123\3\0\3\123\16\0\1\220"+
288 "\1\132\1\220\1\132\7\0\15\220\3\0\3\220\16\0"+
289 "\4\140\7\0\15\140\3\0\3\140\16\0\4\144\7\0"+
290 "\15\144\3\0\3\144\16\0\4\147\7\0\15\147\3\0"+
291 "\3\147\16\0\4\152\7\0\15\152\3\0\3\152\16\0"+
292 "\1\221\1\144\1\221\1\144\7\0\15\221\3\0\3\221"+
293 "\16\0\4\132\7\0\15\132\3\0\3\132\16\0\4\222"+
294 "\7\0\15\222\3\0\3\222\33\0\1\223\61\0\1\224"+
295 "\30\0\4\40\6\0\1\225\15\40\3\0\2\40\1\226"+
296 "\33\0\1\227\32\0\1\172\1\0\1\36\1\0\4\123"+
297 "\1\0\3\124\3\0\10\123\1\230\4\123\3\0\3\123"+
298 "\2\0\1\231\104\0\1\232\36\0\4\233\7\0\15\233"+
299 "\3\0\3\233\3\0\1\175\1\0\1\102\2\176\6\0"+
300 "\1\220\1\132\1\220\1\132\7\0\15\220\3\0\3\220"+
301 "\3\0\1\205\1\0\1\102\2\77\1\0\1\100\3\0"+
302 "\1\100\1\221\1\144\1\221\1\144\7\0\15\221\3\0"+
303 "\3\221\3\0\1\204\2\0\1\204\7\0\4\222\7\0"+
304 "\15\222\3\0\3\222\34\0\1\234\55\0\1\235\26\0"+
305 "\1\236\60\0\4\40\6\0\1\225\15\40\3\0\3\40"+
306 "\34\0\1\237\31\0\1\172\1\0\1\111\1\0\4\123"+
307 "\1\0\3\124\3\0\15\123\3\0\3\123\34\0\1\240"+
308 "\32\0\1\241\2\0\4\233\7\0\15\233\3\0\3\233"+
309 "\35\0\1\242\62\0\1\243\20\0\1\244\77\0\1\245"+
310 "\53\0\1\246\32\0\1\36\1\0\4\173\1\0\3\124"+
311 "\3\0\15\173\3\0\3\173\36\0\1\247\53\0\1\250"+
312 "\33\0\4\251\7\0\15\251\3\0\3\251\36\0\1\252"+
313 "\53\0\1\253\54\0\1\254\61\0\1\255\11\0\1\256"+
314 "\12\0\4\251\7\0\15\251\3\0\3\251\37\0\1\257"+
315 "\53\0\1\260\54\0\1\261\22\0\1\13\62\0\4\262"+
316 "\7\0\15\262\3\0\3\262\40\0\1\263\53\0\1\264"+
317 "\43\0\1\265\26\0\2\262\1\0\2\262\1\0\2\262"+
318 "\2\0\5\262\7\0\15\262\3\0\4\262\27\0\1\266"+
321 private static int [] zzUnpackTrans() {
322 int [] result = new int[7040];
324 offset = zzUnpackTrans(ZZ_TRANS_PACKED_0, offset, result);
328 private static int zzUnpackTrans(String packed, int offset, int [] result) {
329 int i = 0; /* index in packed string */
330 int j = offset; /* index in unpacked array */
331 int l = packed.length();
333 int count = packed.charAt(i++);
334 int value = packed.charAt(i++);
336 do result[j++] = value; while (--count > 0);
343 private static final int ZZ_UNKNOWN_ERROR = 0;
344 private static final int ZZ_NO_MATCH = 1;
345 private static final int ZZ_PUSHBACK_2BIG = 2;
347 /* error messages for the codes above */
348 private static final String ZZ_ERROR_MSG[] = {
349 "Unkown internal scanner error",
350 "Error: could not match input",
351 "Error: pushback value was too large"
355 * ZZ_ATTRIBUTE[aState] contains the attributes of state <code>aState</code>
357 private static final int [] ZZ_ATTRIBUTE = zzUnpackAttribute();
359 private static final String ZZ_ATTRIBUTE_PACKED_0 =
360 "\12\0\1\11\7\1\1\11\3\1\1\11\6\1\1\11"+
361 "\2\1\1\11\14\1\1\11\6\1\2\11\3\0\1\11"+
362 "\14\0\2\1\2\11\1\1\1\0\2\1\1\0\1\1"+
363 "\1\0\1\1\3\0\7\1\2\0\1\1\1\0\15\1"+
364 "\3\0\1\1\1\11\3\0\1\1\1\11\5\0\1\1"+
365 "\4\0\1\1\2\0\2\1\2\0\1\1\5\0\1\11"+
366 "\3\1\3\0\1\1\2\0\1\11\30\0\1\1\2\0"+
369 private static int [] zzUnpackAttribute() {
370 int [] result = new int[183];
372 offset = zzUnpackAttribute(ZZ_ATTRIBUTE_PACKED_0, offset, result);
376 private static int zzUnpackAttribute(String packed, int offset, int [] result) {
377 int i = 0; /* index in packed string */
378 int j = offset; /* index in unpacked array */
379 int l = packed.length();
381 int count = packed.charAt(i++);
382 int value = packed.charAt(i++);
383 do result[j++] = value; while (--count > 0);
388 /** the input device */
389 private java.io.Reader zzReader;
391 /** the current state of the DFA */
394 /** the current lexical state */
395 private int zzLexicalState = YYINITIAL;
397 /** this buffer contains the current text to be matched and is
398 the source of the yytext() string */
399 private char zzBuffer[] = new char[ZZ_BUFFERSIZE];
401 /** the textposition at the last accepting state */
402 private int zzMarkedPos;
404 /** the current text position in the buffer */
405 private int zzCurrentPos;
407 /** startRead marks the beginning of the yytext() string in the buffer */
408 private int zzStartRead;
410 /** endRead marks the last character in the buffer, that has been read
412 private int zzEndRead;
414 /** number of newlines encountered up to the start of the matched text */
417 /** the number of characters up to the start of the matched text */
421 * the number of characters from the last newline up to the start of the
424 private int yycolumn;
427 * zzAtBOL == true <=> the scanner is currently at the beginning of a line
429 private boolean zzAtBOL = true;
431 /** zzAtEOF == true <=> the scanner is at the EOF */
432 private boolean zzAtEOF;
434 /** denotes if the user-EOF-code has already been executed */
435 private boolean zzEOFDone;
439 public static final int ALPHANUM = WikipediaTokenizer.ALPHANUM_ID;
440 public static final int APOSTROPHE = WikipediaTokenizer.APOSTROPHE_ID;
441 public static final int ACRONYM = WikipediaTokenizer.ACRONYM_ID;
442 public static final int COMPANY = WikipediaTokenizer.COMPANY_ID;
443 public static final int EMAIL = WikipediaTokenizer.EMAIL_ID;
444 public static final int HOST = WikipediaTokenizer.HOST_ID;
445 public static final int NUM = WikipediaTokenizer.NUM_ID;
446 public static final int CJ = WikipediaTokenizer.CJ_ID;
447 public static final int INTERNAL_LINK = WikipediaTokenizer.INTERNAL_LINK_ID;
448 public static final int EXTERNAL_LINK = WikipediaTokenizer.EXTERNAL_LINK_ID;
449 public static final int CITATION = WikipediaTokenizer.CITATION_ID;
450 public static final int CATEGORY = WikipediaTokenizer.CATEGORY_ID;
451 public static final int BOLD = WikipediaTokenizer.BOLD_ID;
452 public static final int ITALICS = WikipediaTokenizer.ITALICS_ID;
453 public static final int BOLD_ITALICS = WikipediaTokenizer.BOLD_ITALICS_ID;
454 public static final int HEADING = WikipediaTokenizer.HEADING_ID;
455 public static final int SUB_HEADING = WikipediaTokenizer.SUB_HEADING_ID;
456 public static final int EXTERNAL_LINK_URL = WikipediaTokenizer.EXTERNAL_LINK_URL_ID;
459 private int currentTokType;
460 private int numBalanced = 0;
461 private int positionInc = 1;
462 private int numLinkToks = 0;
463 //Anytime we start a new on a Wiki reserved token (category, link, etc.) this value will be 0, otherwise it will be the number of tokens seen
464 //this can be useful for detecting when a new reserved token is encountered
465 //see https://issues.apache.org/jira/browse/LUCENE-1133
466 private int numWikiTokensSeen = 0;
468 public static final String [] TOKEN_TYPES = WikipediaTokenizer.TOKEN_TYPES;
471 Returns the number of tokens seen inside a category or link, etc.
472 @return the number of tokens seen inside the context of wiki syntax.
474 public final int getNumWikiTokensSeen(){
475 return numWikiTokensSeen;
478 public final int yychar()
483 public final int getPositionIncrement(){
488 * Fills Lucene token with the current token text.
490 final void getText(CharTermAttribute t) {
491 t.copyBuffer(zzBuffer, zzStartRead, zzMarkedPos-zzStartRead);
494 final int setText(StringBuilder buffer){
495 int length = zzMarkedPos - zzStartRead;
496 buffer.append(zzBuffer, zzStartRead, length);
504 * Creates a new scanner
505 * There is also a java.io.InputStream version of this constructor.
507 * @param in the java.io.Reader to read input from.
509 WikipediaTokenizerImpl(java.io.Reader in) {
514 * Creates a new scanner.
515 * There is also java.io.Reader version of this constructor.
517 * @param in the java.io.Inputstream to read input from.
519 WikipediaTokenizerImpl(java.io.InputStream in) {
520 this(new java.io.InputStreamReader(in));
524 * Unpacks the compressed character translation table.
526 * @param packed the packed character translation table
527 * @return the unpacked character translation table
529 private static char [] zzUnpackCMap(String packed) {
530 char [] map = new char[0x10000];
531 int i = 0; /* index in packed string */
532 int j = 0; /* index in unpacked array */
534 int count = packed.charAt(i++);
535 char value = packed.charAt(i++);
536 do map[j++] = value; while (--count > 0);
543 * Refills the input buffer.
545 * @return <code>false</code>, iff there was new input.
547 * @exception java.io.IOException if any I/O-Error occurs
549 private boolean zzRefill() throws java.io.IOException {
551 /* first: make room (if you can) */
552 if (zzStartRead > 0) {
553 System.arraycopy(zzBuffer, zzStartRead,
555 zzEndRead-zzStartRead);
557 /* translate stored positions */
558 zzEndRead-= zzStartRead;
559 zzCurrentPos-= zzStartRead;
560 zzMarkedPos-= zzStartRead;
564 /* is the buffer big enough? */
565 if (zzCurrentPos >= zzBuffer.length) {
566 /* if not: blow it up */
567 char newBuffer[] = new char[zzCurrentPos*2];
568 System.arraycopy(zzBuffer, 0, newBuffer, 0, zzBuffer.length);
569 zzBuffer = newBuffer;
572 /* finally: fill the buffer with new input */
573 int numRead = zzReader.read(zzBuffer, zzEndRead,
574 zzBuffer.length-zzEndRead);
580 // unlikely but not impossible: read 0 characters, but not at end of stream
582 int c = zzReader.read();
586 zzBuffer[zzEndRead++] = (char) c;
597 * Closes the input stream.
599 public final void yyclose() throws java.io.IOException {
600 zzAtEOF = true; /* indicate end of file */
601 zzEndRead = zzStartRead; /* invalidate buffer */
603 if (zzReader != null)
609 * Resets the scanner to read from a new input stream.
610 * Does not close the old reader.
612 * All internal variables are reset, the old input stream
613 * <b>cannot</b> be reused (internal buffer is discarded and lost).
614 * Lexical state is set to <tt>ZZ_INITIAL</tt>.
616 * Internal scan buffer is resized down to its initial length, if it has grown.
618 * @param reader the new input stream
620 public final void yyreset(java.io.Reader reader) {
625 zzEndRead = zzStartRead = 0;
626 zzCurrentPos = zzMarkedPos = 0;
627 yyline = yychar = yycolumn = 0;
628 zzLexicalState = YYINITIAL;
629 if (zzBuffer.length > ZZ_BUFFERSIZE)
630 zzBuffer = new char[ZZ_BUFFERSIZE];
635 * Returns the current lexical state.
637 public final int yystate() {
638 return zzLexicalState;
643 * Enters a new lexical state
645 * @param newState the new lexical state
647 public final void yybegin(int newState) {
648 zzLexicalState = newState;
653 * Returns the text matched by the current regular expression.
655 public final String yytext() {
656 return new String( zzBuffer, zzStartRead, zzMarkedPos-zzStartRead );
661 * Returns the character at position <tt>pos</tt> from the
664 * It is equivalent to yytext().charAt(pos), but faster
666 * @param pos the position of the character to fetch.
667 * A value from 0 to yylength()-1.
669 * @return the character at position pos
671 public final char yycharat(int pos) {
672 return zzBuffer[zzStartRead+pos];
677 * Returns the length of the matched text region.
679 public final int yylength() {
680 return zzMarkedPos-zzStartRead;
685 * Reports an error that occured while scanning.
687 * In a wellformed scanner (no or only correct usage of
688 * yypushback(int) and a match-all fallback rule) this method
689 * will only be called with things that "Can't Possibly Happen".
690 * If this method is called, something is seriously wrong
691 * (e.g. a JFlex bug producing a faulty scanner etc.).
693 * Usual syntax/scanner level error handling should be done
694 * in error fallback rules.
696 * @param errorCode the code of the errormessage to display
698 private void zzScanError(int errorCode) {
701 message = ZZ_ERROR_MSG[errorCode];
703 catch (ArrayIndexOutOfBoundsException e) {
704 message = ZZ_ERROR_MSG[ZZ_UNKNOWN_ERROR];
707 throw new Error(message);
712 * Pushes the specified amount of characters back into the input stream.
714 * They will be read again by then next call of the scanning method
716 * @param number the number of characters to be read again.
717 * This number must not be greater than yylength()!
719 public void yypushback(int number) {
720 if ( number > yylength() )
721 zzScanError(ZZ_PUSHBACK_2BIG);
723 zzMarkedPos -= number;
728 * Resumes scanning until the next regular expression is matched,
729 * the end of input is encountered or an I/O-Error occurs.
731 * @return the next token
732 * @exception java.io.IOException if any I/O-Error occurs
734 public int getNextToken() throws java.io.IOException {
741 int zzEndReadL = zzEndRead;
742 char [] zzBufferL = zzBuffer;
743 char [] zzCMapL = ZZ_CMAP;
745 int [] zzTransL = ZZ_TRANS;
746 int [] zzRowMapL = ZZ_ROWMAP;
747 int [] zzAttrL = ZZ_ATTRIBUTE;
750 zzMarkedPosL = zzMarkedPos;
752 yychar+= zzMarkedPosL-zzStartRead;
756 zzCurrentPosL = zzCurrentPos = zzStartRead = zzMarkedPosL;
758 zzState = ZZ_LEXSTATE[zzLexicalState];
764 if (zzCurrentPosL < zzEndReadL)
765 zzInput = zzBufferL[zzCurrentPosL++];
771 // store back cached positions
772 zzCurrentPos = zzCurrentPosL;
773 zzMarkedPos = zzMarkedPosL;
774 boolean eof = zzRefill();
775 // get translated positions and possibly new buffer
776 zzCurrentPosL = zzCurrentPos;
777 zzMarkedPosL = zzMarkedPos;
778 zzBufferL = zzBuffer;
779 zzEndReadL = zzEndRead;
785 zzInput = zzBufferL[zzCurrentPosL++];
788 int zzNext = zzTransL[ zzRowMapL[zzState] + zzCMapL[zzInput] ];
789 if (zzNext == -1) break zzForAction;
792 int zzAttributes = zzAttrL[zzState];
793 if ( (zzAttributes & 1) == 1 ) {
795 zzMarkedPosL = zzCurrentPosL;
796 if ( (zzAttributes & 8) == 8 ) break zzForAction;
802 // store back cached position
803 zzMarkedPos = zzMarkedPosL;
805 switch (zzAction < 0 ? zzAction : ZZ_ACTION[zzAction]) {
807 { currentTokType = HEADING; yybegin(DOUBLE_EQUALS_STATE); numWikiTokensSeen++; return currentTokType;
811 { positionInc = 1; return ACRONYM;
819 { numBalanced = 0; numWikiTokensSeen = 0; currentTokType = EXTERNAL_LINK;yybegin(EXTERNAL_LINK_STATE);
823 { positionInc = 1; return COMPANY;
827 { numWikiTokensSeen = 0; positionInc = 1; currentTokType = EXTERNAL_LINK_URL; yybegin(EXTERNAL_LINK_STATE);
831 { numWikiTokensSeen = 0; positionInc = 1; currentTokType = CITATION; yybegin(DOUBLE_BRACE_STATE);
835 { numWikiTokensSeen = 0; positionInc = 1; currentTokType = CATEGORY; yybegin(CATEGORY_STATE);
839 { numWikiTokensSeen = 0; positionInc = 1; if (numBalanced == 0){numBalanced++;yybegin(TWO_SINGLE_QUOTES_STATE);} else{numBalanced = 0;}
843 { positionInc = 1; return NUM;
847 { positionInc = 1; return APOSTROPHE;
851 { numWikiTokensSeen = 0; positionInc = 1; yybegin(DOUBLE_EQUALS_STATE);
855 { yybegin(STRING); return currentTokType;/*pipe*/
859 { positionInc = 1; return ALPHANUM;
863 { currentTokType = INTERNAL_LINK; numWikiTokensSeen = 0; yybegin(INTERNAL_LINK_STATE);
867 { yybegin(DOUBLE_BRACE_STATE); numWikiTokensSeen = 0; return currentTokType;
871 { currentTokType = CATEGORY; numWikiTokensSeen = 0; yybegin(CATEGORY_STATE);
875 { yybegin(YYINITIAL);
879 { positionInc = 1; return CJ;
883 { numBalanced = 0;currentTokType = ALPHANUM; yybegin(YYINITIAL);/*end sub header*/
887 { currentTokType = SUB_HEADING; numWikiTokensSeen = 0; yybegin(STRING);
891 { numBalanced = 0;currentTokType = ALPHANUM; yybegin(YYINITIAL);/*end italics*/
895 { yybegin(CATEGORY_STATE); numWikiTokensSeen++; return currentTokType;
903 { yybegin(STRING); numWikiTokensSeen++; return currentTokType;/* STRING ALPHANUM*/
907 { positionInc = 1; numWikiTokensSeen++; yybegin(EXTERNAL_LINK_STATE); return currentTokType;
911 { numLinkToks = 0; yybegin(YYINITIAL);
915 { currentTokType = BOLD; yybegin(THREE_SINGLE_QUOTES_STATE);
919 { currentTokType = EXTERNAL_LINK; numWikiTokensSeen = 0; yybegin(EXTERNAL_LINK_STATE);
923 { yybegin(STRING); numWikiTokensSeen++; return currentTokType;
927 { numBalanced = 0; numWikiTokensSeen = 0; currentTokType = CATEGORY;yybegin(CATEGORY_STATE);
931 { currentTokType = INTERNAL_LINK; numWikiTokensSeen = 0; yybegin(INTERNAL_LINK_STATE);
935 { numBalanced = 0;currentTokType = ALPHANUM;yybegin(YYINITIAL);/*end bold*/
939 { if (numLinkToks == 0){positionInc = 0;} else{positionInc = 1;} numWikiTokensSeen++; currentTokType = EXTERNAL_LINK; yybegin(EXTERNAL_LINK_STATE); numLinkToks++; return currentTokType;
943 { yybegin(INTERNAL_LINK_STATE); numWikiTokensSeen++; return currentTokType;
947 { numWikiTokensSeen = 0; positionInc = 1; currentTokType = INTERNAL_LINK; yybegin(INTERNAL_LINK_STATE);
951 { positionInc = 1; return EMAIL;
955 { numWikiTokensSeen = 0; positionInc = 1;
959 { /* ignore STRING */
963 { currentTokType = BOLD_ITALICS; yybegin(FIVE_SINGLE_QUOTES_STATE);
967 { positionInc = 1; return HOST;
971 { numBalanced = 0; numWikiTokensSeen = 0; currentTokType = INTERNAL_LINK;yybegin(INTERNAL_LINK_STATE);
975 { numBalanced = 0;currentTokType = ALPHANUM; yybegin(YYINITIAL);/*end bold italics*/
979 { currentTokType = ITALICS; numWikiTokensSeen++; yybegin(STRING); return currentTokType;/*italics*/
983 { numLinkToks = 0; positionInc = 0; yybegin(YYINITIAL);
987 if (zzInput == YYEOF && zzStartRead == zzCurrentPos) {
992 zzScanError(ZZ_NO_MATCH);