1 /* The following code was generated by JFlex 1.5.0-SNAPSHOT on 9/30/11 1:23 PM */
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 9/30/11 1:23 PM from the specification file
29 * <tt>/lucene/3x/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\36\1\0\1\37\1\0"+
96 "\1\40\3\0\1\41\1\42\2\43\1\42\2\44\2\0"+
97 "\1\43\1\0\14\43\1\42\3\0\1\11\1\45\3\0"+
98 "\1\46\1\47\5\0\1\50\4\0\1\50\2\0\2\50"+
99 "\2\0\1\11\5\0\1\31\1\42\1\43\1\51\3\0"+
100 "\1\11\2\0\1\52\30\0\1\53\2\0\1\54\1\55"+
103 private static int [] zzUnpackAction() {
104 int [] result = new int[184];
106 offset = zzUnpackAction(ZZ_ACTION_PACKED_0, offset, result);
110 private static int zzUnpackAction(String packed, int offset, int [] result) {
111 int i = 0; /* index in packed string */
112 int j = offset; /* index in unpacked array */
113 int l = packed.length();
115 int count = packed.charAt(i++);
116 int value = packed.charAt(i++);
117 do result[j++] = value; while (--count > 0);
124 * Translates a state to a row index in the transition table
126 private static final int [] ZZ_ROWMAP = zzUnpackRowMap();
128 private static final String ZZ_ROWMAP_PACKED_0 =
129 "\0\0\0\54\0\130\0\204\0\260\0\334\0\u0108\0\u0134"+
130 "\0\u0160\0\u018c\0\u01b8\0\u01e4\0\u0210\0\u023c\0\u0268\0\u0294"+
131 "\0\u02c0\0\u02ec\0\u01b8\0\u0318\0\u0344\0\u0370\0\u01b8\0\u039c"+
132 "\0\u03c8\0\u03f4\0\u0420\0\u044c\0\u0478\0\u01b8\0\u039c\0\u04a4"+
133 "\0\u01b8\0\u04d0\0\u04fc\0\u0528\0\u0554\0\u0580\0\u05ac\0\u05d8"+
134 "\0\u0604\0\u0630\0\u065c\0\u0688\0\u06b4\0\u01b8\0\u06e0\0\u039c"+
135 "\0\u070c\0\u0738\0\u0764\0\u0790\0\u01b8\0\u01b8\0\u07bc\0\u07e8"+
136 "\0\u0814\0\u01b8\0\u0840\0\u086c\0\u0898\0\u08c4\0\u08f0\0\u091c"+
137 "\0\u0948\0\u0974\0\u09a0\0\u09cc\0\u09f8\0\u0a24\0\u0a50\0\u0a7c"+
138 "\0\u01b8\0\u01b8\0\u0aa8\0\u0ad4\0\u0b00\0\u0b00\0\u01b8\0\u0b2c"+
139 "\0\u0b58\0\u0b84\0\u0bb0\0\u0bdc\0\u0c08\0\u0c34\0\u0c60\0\u0c8c"+
140 "\0\u0cb8\0\u0ce4\0\u0d10\0\u0898\0\u0d3c\0\u0d68\0\u0d94\0\u0dc0"+
141 "\0\u0dec\0\u0e18\0\u0e44\0\u0e70\0\u0e9c\0\u0ec8\0\u0ef4\0\u0f20"+
142 "\0\u0f4c\0\u0f78\0\u0fa4\0\u0fd0\0\u0ffc\0\u1028\0\u1054\0\u1080"+
143 "\0\u10ac\0\u10d8\0\u01b8\0\u1104\0\u1130\0\u115c\0\u1188\0\u01b8"+
144 "\0\u11b4\0\u11e0\0\u120c\0\u1238\0\u1264\0\u1290\0\u12bc\0\u12e8"+
145 "\0\u1314\0\u1340\0\u136c\0\u1398\0\u13c4\0\u086c\0\u09f8\0\u13f0"+
146 "\0\u141c\0\u1448\0\u1474\0\u14a0\0\u14cc\0\u14f8\0\u1524\0\u01b8"+
147 "\0\u1550\0\u157c\0\u15a8\0\u15d4\0\u1600\0\u162c\0\u1658\0\u1684"+
148 "\0\u16b0\0\u01b8\0\u16dc\0\u1708\0\u1734\0\u1760\0\u178c\0\u17b8"+
149 "\0\u17e4\0\u1810\0\u183c\0\u1868\0\u1894\0\u18c0\0\u18ec\0\u1918"+
150 "\0\u1944\0\u1970\0\u199c\0\u19c8\0\u19f4\0\u1a20\0\u1a4c\0\u1a78"+
151 "\0\u1aa4\0\u1ad0\0\u1afc\0\u1b28\0\u1b54\0\u01b8\0\u01b8\0\u01b8";
153 private static int [] zzUnpackRowMap() {
154 int [] result = new int[184];
156 offset = zzUnpackRowMap(ZZ_ROWMAP_PACKED_0, offset, result);
160 private static int zzUnpackRowMap(String packed, int offset, int [] result) {
161 int i = 0; /* index in packed string */
162 int j = offset; /* index in unpacked array */
163 int l = packed.length();
165 int high = packed.charAt(i++) << 16;
166 result[j++] = high | packed.charAt(i++);
172 * The transition table of the DFA
174 private static final int [] ZZ_TRANS = zzUnpackTrans();
176 private static final String ZZ_TRANS_PACKED_0 =
177 "\1\13\1\14\5\13\1\15\1\13\1\16\3\13\1\17"+
178 "\1\20\1\21\1\22\1\23\1\24\2\13\1\25\2\13"+
179 "\15\17\1\26\2\13\3\17\1\13\7\27\1\30\5\27"+
180 "\4\31\1\27\1\32\3\27\1\33\1\27\15\31\3\27"+
181 "\3\31\10\27\1\30\5\27\4\34\1\27\1\32\3\27"+
182 "\1\35\1\27\15\34\3\27\3\34\1\27\7\36\1\37"+
183 "\5\36\4\40\1\36\1\32\2\27\1\36\1\41\1\36"+
184 "\15\40\3\36\1\42\2\40\2\36\1\43\5\36\1\37"+
185 "\5\36\4\44\1\36\1\45\2\36\1\46\2\36\15\44"+
186 "\3\36\3\44\10\36\1\37\5\36\4\47\1\36\1\45"+
187 "\2\36\1\46\2\36\15\47\3\36\3\47\10\36\1\37"+
188 "\5\36\4\47\1\36\1\45\2\36\1\50\2\36\15\47"+
189 "\3\36\3\47\10\36\1\37\1\36\1\51\3\36\4\52"+
190 "\1\36\1\45\5\36\15\52\3\36\3\52\10\36\1\53"+
191 "\5\36\4\54\1\36\1\45\5\36\15\54\1\36\1\55"+
192 "\1\36\3\54\1\36\1\56\1\57\5\56\1\60\1\56"+
193 "\1\61\3\56\4\62\1\56\1\63\2\56\1\64\2\56"+
194 "\15\62\2\56\1\65\3\62\1\56\55\0\1\66\62\0"+
195 "\1\67\4\0\4\70\7\0\6\70\1\71\6\70\3\0"+
196 "\3\70\12\0\1\72\43\0\1\73\1\74\1\75\1\76"+
197 "\2\77\1\0\1\100\3\0\1\100\1\17\1\20\1\21"+
198 "\1\22\7\0\15\17\3\0\3\17\3\0\1\101\1\0"+
199 "\1\102\2\103\1\0\1\104\3\0\1\104\3\20\1\22"+
200 "\7\0\15\20\3\0\3\20\2\0\1\73\1\105\1\75"+
201 "\1\76\2\103\1\0\1\104\3\0\1\104\1\21\1\20"+
202 "\1\21\1\22\7\0\15\21\3\0\3\21\3\0\1\106"+
203 "\1\0\1\102\2\77\1\0\1\100\3\0\1\100\4\22"+
204 "\7\0\15\22\3\0\3\22\24\0\1\13\55\0\1\107"+
205 "\73\0\1\110\16\0\1\67\4\0\4\70\7\0\15\70"+
206 "\3\0\3\70\16\0\4\31\7\0\15\31\3\0\3\31"+
207 "\24\0\1\27\56\0\1\111\42\0\4\34\7\0\15\34"+
208 "\3\0\3\34\27\0\1\112\42\0\4\40\7\0\15\40"+
209 "\3\0\3\40\16\0\4\40\7\0\2\40\1\113\12\40"+
210 "\3\0\3\40\2\0\1\114\67\0\4\44\7\0\15\44"+
211 "\3\0\3\44\24\0\1\36\55\0\1\115\43\0\4\47"+
212 "\7\0\15\47\3\0\3\47\26\0\1\116\37\0\1\117"+
213 "\57\0\4\52\7\0\15\52\3\0\3\52\11\0\1\120"+
214 "\4\0\4\70\7\0\15\70\3\0\3\70\16\0\4\54"+
215 "\7\0\15\54\3\0\3\54\47\0\1\117\6\0\1\121"+
216 "\63\0\1\122\57\0\4\62\7\0\15\62\3\0\3\62"+
217 "\24\0\1\56\55\0\1\123\43\0\4\70\7\0\15\70"+
218 "\3\0\3\70\14\0\1\36\1\0\4\124\1\0\3\125"+
219 "\3\0\15\124\3\0\3\124\14\0\1\36\1\0\4\124"+
220 "\1\0\3\125\3\0\3\124\1\126\11\124\3\0\3\124"+
221 "\16\0\1\127\1\0\1\127\10\0\15\127\3\0\3\127"+
222 "\16\0\1\130\1\131\1\132\1\133\7\0\15\130\3\0"+
223 "\3\130\16\0\1\134\1\0\1\134\10\0\15\134\3\0"+
224 "\3\134\16\0\1\135\1\136\1\135\1\136\7\0\15\135"+
225 "\3\0\3\135\16\0\1\137\2\140\1\141\7\0\15\137"+
226 "\3\0\3\137\16\0\1\100\2\142\10\0\15\100\3\0"+
227 "\3\100\16\0\1\143\2\144\1\145\7\0\15\143\3\0"+
228 "\3\143\16\0\4\136\7\0\15\136\3\0\3\136\16\0"+
229 "\1\146\2\147\1\150\7\0\15\146\3\0\3\146\16\0"+
230 "\1\151\2\152\1\153\7\0\15\151\3\0\3\151\16\0"+
231 "\1\154\1\144\1\155\1\145\7\0\15\154\3\0\3\154"+
232 "\16\0\1\156\2\131\1\133\7\0\15\156\3\0\3\156"+
233 "\30\0\1\157\1\160\64\0\1\161\27\0\4\40\7\0"+
234 "\2\40\1\162\12\40\3\0\3\40\2\0\1\163\101\0"+
235 "\1\164\1\165\40\0\4\70\7\0\6\70\1\166\6\70"+
236 "\3\0\3\70\2\0\1\167\63\0\1\170\71\0\1\171"+
237 "\1\172\34\0\1\173\1\0\1\36\1\0\4\124\1\0"+
238 "\3\125\3\0\15\124\3\0\3\124\16\0\4\174\1\0"+
239 "\3\125\3\0\15\174\3\0\3\174\12\0\1\173\1\0"+
240 "\1\36\1\0\4\124\1\0\3\125\3\0\10\124\1\175"+
241 "\4\124\3\0\3\124\2\0\1\73\13\0\1\127\1\0"+
242 "\1\127\10\0\15\127\3\0\3\127\3\0\1\176\1\0"+
243 "\1\102\2\177\6\0\1\130\1\131\1\132\1\133\7\0"+
244 "\15\130\3\0\3\130\3\0\1\200\1\0\1\102\2\201"+
245 "\1\0\1\202\3\0\1\202\3\131\1\133\7\0\15\131"+
246 "\3\0\3\131\3\0\1\203\1\0\1\102\2\201\1\0"+
247 "\1\202\3\0\1\202\1\132\1\131\1\132\1\133\7\0"+
248 "\15\132\3\0\3\132\3\0\1\204\1\0\1\102\2\177"+
249 "\6\0\4\133\7\0\15\133\3\0\3\133\3\0\1\205"+
250 "\2\0\1\205\7\0\1\135\1\136\1\135\1\136\7\0"+
251 "\15\135\3\0\3\135\3\0\1\205\2\0\1\205\7\0"+
252 "\4\136\7\0\15\136\3\0\3\136\3\0\1\177\1\0"+
253 "\1\102\2\177\6\0\1\137\2\140\1\141\7\0\15\137"+
254 "\3\0\3\137\3\0\1\201\1\0\1\102\2\201\1\0"+
255 "\1\202\3\0\1\202\3\140\1\141\7\0\15\140\3\0"+
256 "\3\140\3\0\1\177\1\0\1\102\2\177\6\0\4\141"+
257 "\7\0\15\141\3\0\3\141\3\0\1\202\2\0\2\202"+
258 "\1\0\1\202\3\0\1\202\3\142\10\0\15\142\3\0"+
259 "\3\142\3\0\1\106\1\0\1\102\2\77\1\0\1\100"+
260 "\3\0\1\100\1\143\2\144\1\145\7\0\15\143\3\0"+
261 "\3\143\3\0\1\101\1\0\1\102\2\103\1\0\1\104"+
262 "\3\0\1\104\3\144\1\145\7\0\15\144\3\0\3\144"+
263 "\3\0\1\106\1\0\1\102\2\77\1\0\1\100\3\0"+
264 "\1\100\4\145\7\0\15\145\3\0\3\145\3\0\1\77"+
265 "\1\0\1\102\2\77\1\0\1\100\3\0\1\100\1\146"+
266 "\2\147\1\150\7\0\15\146\3\0\3\146\3\0\1\103"+
267 "\1\0\1\102\2\103\1\0\1\104\3\0\1\104\3\147"+
268 "\1\150\7\0\15\147\3\0\3\147\3\0\1\77\1\0"+
269 "\1\102\2\77\1\0\1\100\3\0\1\100\4\150\7\0"+
270 "\15\150\3\0\3\150\3\0\1\100\2\0\2\100\1\0"+
271 "\1\100\3\0\1\100\1\151\2\152\1\153\7\0\15\151"+
272 "\3\0\3\151\3\0\1\104\2\0\2\104\1\0\1\104"+
273 "\3\0\1\104\3\152\1\153\7\0\15\152\3\0\3\152"+
274 "\3\0\1\100\2\0\2\100\1\0\1\100\3\0\1\100"+
275 "\4\153\7\0\15\153\3\0\3\153\3\0\1\206\1\0"+
276 "\1\102\2\77\1\0\1\100\3\0\1\100\1\154\1\144"+
277 "\1\155\1\145\7\0\15\154\3\0\3\154\3\0\1\207"+
278 "\1\0\1\102\2\103\1\0\1\104\3\0\1\104\1\155"+
279 "\1\144\1\155\1\145\7\0\15\155\3\0\3\155\3\0"+
280 "\1\204\1\0\1\102\2\177\6\0\1\156\2\131\1\133"+
281 "\7\0\15\156\3\0\3\156\31\0\1\160\54\0\1\210"+
282 "\64\0\1\211\26\0\4\40\7\0\15\40\3\0\1\40"+
283 "\1\212\1\40\31\0\1\165\54\0\1\213\35\0\1\36"+
284 "\1\0\4\124\1\0\3\125\3\0\3\124\1\214\11\124"+
285 "\3\0\3\124\2\0\1\215\102\0\1\172\54\0\1\216"+
286 "\34\0\1\217\52\0\1\173\3\0\4\174\7\0\15\174"+
287 "\3\0\3\174\12\0\1\173\1\0\1\220\1\0\4\124"+
288 "\1\0\3\125\3\0\15\124\3\0\3\124\16\0\1\221"+
289 "\1\133\1\221\1\133\7\0\15\221\3\0\3\221\16\0"+
290 "\4\141\7\0\15\141\3\0\3\141\16\0\4\145\7\0"+
291 "\15\145\3\0\3\145\16\0\4\150\7\0\15\150\3\0"+
292 "\3\150\16\0\4\153\7\0\15\153\3\0\3\153\16\0"+
293 "\1\222\1\145\1\222\1\145\7\0\15\222\3\0\3\222"+
294 "\16\0\4\133\7\0\15\133\3\0\3\133\16\0\4\223"+
295 "\7\0\15\223\3\0\3\223\33\0\1\224\61\0\1\225"+
296 "\30\0\4\40\6\0\1\226\15\40\3\0\2\40\1\227"+
297 "\33\0\1\230\32\0\1\173\1\0\1\36\1\0\4\124"+
298 "\1\0\3\125\3\0\10\124\1\231\4\124\3\0\3\124"+
299 "\2\0\1\232\104\0\1\233\36\0\4\234\7\0\15\234"+
300 "\3\0\3\234\3\0\1\176\1\0\1\102\2\177\6\0"+
301 "\1\221\1\133\1\221\1\133\7\0\15\221\3\0\3\221"+
302 "\3\0\1\206\1\0\1\102\2\77\1\0\1\100\3\0"+
303 "\1\100\1\222\1\145\1\222\1\145\7\0\15\222\3\0"+
304 "\3\222\3\0\1\205\2\0\1\205\7\0\4\223\7\0"+
305 "\15\223\3\0\3\223\34\0\1\235\55\0\1\236\26\0"+
306 "\1\237\60\0\4\40\6\0\1\226\15\40\3\0\3\40"+
307 "\34\0\1\240\31\0\1\173\1\0\1\117\1\0\4\124"+
308 "\1\0\3\125\3\0\15\124\3\0\3\124\34\0\1\241"+
309 "\32\0\1\242\2\0\4\234\7\0\15\234\3\0\3\234"+
310 "\35\0\1\243\62\0\1\244\20\0\1\245\77\0\1\246"+
311 "\53\0\1\247\32\0\1\36\1\0\4\174\1\0\3\125"+
312 "\3\0\15\174\3\0\3\174\36\0\1\250\53\0\1\251"+
313 "\33\0\4\252\7\0\15\252\3\0\3\252\36\0\1\253"+
314 "\53\0\1\254\54\0\1\255\61\0\1\256\11\0\1\257"+
315 "\12\0\4\252\7\0\15\252\3\0\3\252\37\0\1\260"+
316 "\53\0\1\261\54\0\1\262\22\0\1\13\62\0\4\263"+
317 "\7\0\15\263\3\0\3\263\40\0\1\264\53\0\1\265"+
318 "\43\0\1\266\26\0\2\263\1\0\2\263\1\0\2\263"+
319 "\2\0\5\263\7\0\15\263\3\0\4\263\27\0\1\267"+
322 private static int [] zzUnpackTrans() {
323 int [] result = new int[7040];
325 offset = zzUnpackTrans(ZZ_TRANS_PACKED_0, offset, result);
329 private static int zzUnpackTrans(String packed, int offset, int [] result) {
330 int i = 0; /* index in packed string */
331 int j = offset; /* index in unpacked array */
332 int l = packed.length();
334 int count = packed.charAt(i++);
335 int value = packed.charAt(i++);
337 do result[j++] = value; while (--count > 0);
344 private static final int ZZ_UNKNOWN_ERROR = 0;
345 private static final int ZZ_NO_MATCH = 1;
346 private static final int ZZ_PUSHBACK_2BIG = 2;
348 /* error messages for the codes above */
349 private static final String ZZ_ERROR_MSG[] = {
350 "Unkown internal scanner error",
351 "Error: could not match input",
352 "Error: pushback value was too large"
356 * ZZ_ATTRIBUTE[aState] contains the attributes of state <code>aState</code>
358 private static final int [] ZZ_ATTRIBUTE = zzUnpackAttribute();
360 private static final String ZZ_ATTRIBUTE_PACKED_0 =
361 "\12\0\1\11\7\1\1\11\3\1\1\11\6\1\1\11"+
362 "\2\1\1\11\14\1\1\11\6\1\2\11\3\0\1\11"+
363 "\14\0\2\1\2\11\1\1\1\0\2\1\1\11\1\0"+
364 "\1\1\1\0\1\1\3\0\7\1\2\0\1\1\1\0"+
365 "\15\1\3\0\1\1\1\11\3\0\1\1\1\11\5\0"+
366 "\1\1\4\0\1\1\2\0\2\1\2\0\1\1\5\0"+
367 "\1\11\3\1\3\0\1\1\2\0\1\11\30\0\1\1"+
370 private static int [] zzUnpackAttribute() {
371 int [] result = new int[184];
373 offset = zzUnpackAttribute(ZZ_ATTRIBUTE_PACKED_0, offset, result);
377 private static int zzUnpackAttribute(String packed, int offset, int [] result) {
378 int i = 0; /* index in packed string */
379 int j = offset; /* index in unpacked array */
380 int l = packed.length();
382 int count = packed.charAt(i++);
383 int value = packed.charAt(i++);
384 do result[j++] = value; while (--count > 0);
389 /** the input device */
390 private java.io.Reader zzReader;
392 /** the current state of the DFA */
395 /** the current lexical state */
396 private int zzLexicalState = YYINITIAL;
398 /** this buffer contains the current text to be matched and is
399 the source of the yytext() string */
400 private char zzBuffer[] = new char[ZZ_BUFFERSIZE];
402 /** the textposition at the last accepting state */
403 private int zzMarkedPos;
405 /** the current text position in the buffer */
406 private int zzCurrentPos;
408 /** startRead marks the beginning of the yytext() string in the buffer */
409 private int zzStartRead;
411 /** endRead marks the last character in the buffer, that has been read
413 private int zzEndRead;
415 /** number of newlines encountered up to the start of the matched text */
418 /** the number of characters up to the start of the matched text */
422 * the number of characters from the last newline up to the start of the
425 private int yycolumn;
428 * zzAtBOL == true <=> the scanner is currently at the beginning of a line
430 private boolean zzAtBOL = true;
432 /** zzAtEOF == true <=> the scanner is at the EOF */
433 private boolean zzAtEOF;
435 /** denotes if the user-EOF-code has already been executed */
436 private boolean zzEOFDone;
440 public static final int ALPHANUM = WikipediaTokenizer.ALPHANUM_ID;
441 public static final int APOSTROPHE = WikipediaTokenizer.APOSTROPHE_ID;
442 public static final int ACRONYM = WikipediaTokenizer.ACRONYM_ID;
443 public static final int COMPANY = WikipediaTokenizer.COMPANY_ID;
444 public static final int EMAIL = WikipediaTokenizer.EMAIL_ID;
445 public static final int HOST = WikipediaTokenizer.HOST_ID;
446 public static final int NUM = WikipediaTokenizer.NUM_ID;
447 public static final int CJ = WikipediaTokenizer.CJ_ID;
448 public static final int INTERNAL_LINK = WikipediaTokenizer.INTERNAL_LINK_ID;
449 public static final int EXTERNAL_LINK = WikipediaTokenizer.EXTERNAL_LINK_ID;
450 public static final int CITATION = WikipediaTokenizer.CITATION_ID;
451 public static final int CATEGORY = WikipediaTokenizer.CATEGORY_ID;
452 public static final int BOLD = WikipediaTokenizer.BOLD_ID;
453 public static final int ITALICS = WikipediaTokenizer.ITALICS_ID;
454 public static final int BOLD_ITALICS = WikipediaTokenizer.BOLD_ITALICS_ID;
455 public static final int HEADING = WikipediaTokenizer.HEADING_ID;
456 public static final int SUB_HEADING = WikipediaTokenizer.SUB_HEADING_ID;
457 public static final int EXTERNAL_LINK_URL = WikipediaTokenizer.EXTERNAL_LINK_URL_ID;
460 private int currentTokType;
461 private int numBalanced = 0;
462 private int positionInc = 1;
463 private int numLinkToks = 0;
464 //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
465 //this can be useful for detecting when a new reserved token is encountered
466 //see https://issues.apache.org/jira/browse/LUCENE-1133
467 private int numWikiTokensSeen = 0;
469 public static final String [] TOKEN_TYPES = WikipediaTokenizer.TOKEN_TYPES;
472 Returns the number of tokens seen inside a category or link, etc.
473 @return the number of tokens seen inside the context of wiki syntax.
475 public final int getNumWikiTokensSeen(){
476 return numWikiTokensSeen;
479 public final int yychar()
484 public final int getPositionIncrement(){
489 * Fills Lucene token with the current token text.
491 final void getText(CharTermAttribute t) {
492 t.copyBuffer(zzBuffer, zzStartRead, zzMarkedPos-zzStartRead);
495 final int setText(StringBuilder buffer){
496 int length = zzMarkedPos - zzStartRead;
497 buffer.append(zzBuffer, zzStartRead, length);
505 * Creates a new scanner
506 * There is also a java.io.InputStream version of this constructor.
508 * @param in the java.io.Reader to read input from.
510 WikipediaTokenizerImpl(java.io.Reader in) {
515 * Creates a new scanner.
516 * There is also java.io.Reader version of this constructor.
518 * @param in the java.io.Inputstream to read input from.
520 WikipediaTokenizerImpl(java.io.InputStream in) {
521 this(new java.io.InputStreamReader(in));
525 * Unpacks the compressed character translation table.
527 * @param packed the packed character translation table
528 * @return the unpacked character translation table
530 private static char [] zzUnpackCMap(String packed) {
531 char [] map = new char[0x10000];
532 int i = 0; /* index in packed string */
533 int j = 0; /* index in unpacked array */
535 int count = packed.charAt(i++);
536 char value = packed.charAt(i++);
537 do map[j++] = value; while (--count > 0);
544 * Refills the input buffer.
546 * @return <code>false</code>, iff there was new input.
548 * @exception java.io.IOException if any I/O-Error occurs
550 private boolean zzRefill() throws java.io.IOException {
552 /* first: make room (if you can) */
553 if (zzStartRead > 0) {
554 System.arraycopy(zzBuffer, zzStartRead,
556 zzEndRead-zzStartRead);
558 /* translate stored positions */
559 zzEndRead-= zzStartRead;
560 zzCurrentPos-= zzStartRead;
561 zzMarkedPos-= zzStartRead;
565 /* is the buffer big enough? */
566 if (zzCurrentPos >= zzBuffer.length) {
567 /* if not: blow it up */
568 char newBuffer[] = new char[zzCurrentPos*2];
569 System.arraycopy(zzBuffer, 0, newBuffer, 0, zzBuffer.length);
570 zzBuffer = newBuffer;
573 /* finally: fill the buffer with new input */
574 int numRead = zzReader.read(zzBuffer, zzEndRead,
575 zzBuffer.length-zzEndRead);
581 // unlikely but not impossible: read 0 characters, but not at end of stream
583 int c = zzReader.read();
587 zzBuffer[zzEndRead++] = (char) c;
598 * Closes the input stream.
600 public final void yyclose() throws java.io.IOException {
601 zzAtEOF = true; /* indicate end of file */
602 zzEndRead = zzStartRead; /* invalidate buffer */
604 if (zzReader != null)
610 * Resets the scanner to read from a new input stream.
611 * Does not close the old reader.
613 * All internal variables are reset, the old input stream
614 * <b>cannot</b> be reused (internal buffer is discarded and lost).
615 * Lexical state is set to <tt>ZZ_INITIAL</tt>.
617 * Internal scan buffer is resized down to its initial length, if it has grown.
619 * @param reader the new input stream
621 public final void yyreset(java.io.Reader reader) {
626 zzEndRead = zzStartRead = 0;
627 zzCurrentPos = zzMarkedPos = 0;
628 yyline = yychar = yycolumn = 0;
629 zzLexicalState = YYINITIAL;
630 if (zzBuffer.length > ZZ_BUFFERSIZE)
631 zzBuffer = new char[ZZ_BUFFERSIZE];
636 * Returns the current lexical state.
638 public final int yystate() {
639 return zzLexicalState;
644 * Enters a new lexical state
646 * @param newState the new lexical state
648 public final void yybegin(int newState) {
649 zzLexicalState = newState;
654 * Returns the text matched by the current regular expression.
656 public final String yytext() {
657 return new String( zzBuffer, zzStartRead, zzMarkedPos-zzStartRead );
662 * Returns the character at position <tt>pos</tt> from the
665 * It is equivalent to yytext().charAt(pos), but faster
667 * @param pos the position of the character to fetch.
668 * A value from 0 to yylength()-1.
670 * @return the character at position pos
672 public final char yycharat(int pos) {
673 return zzBuffer[zzStartRead+pos];
678 * Returns the length of the matched text region.
680 public final int yylength() {
681 return zzMarkedPos-zzStartRead;
686 * Reports an error that occured while scanning.
688 * In a wellformed scanner (no or only correct usage of
689 * yypushback(int) and a match-all fallback rule) this method
690 * will only be called with things that "Can't Possibly Happen".
691 * If this method is called, something is seriously wrong
692 * (e.g. a JFlex bug producing a faulty scanner etc.).
694 * Usual syntax/scanner level error handling should be done
695 * in error fallback rules.
697 * @param errorCode the code of the errormessage to display
699 private void zzScanError(int errorCode) {
702 message = ZZ_ERROR_MSG[errorCode];
704 catch (ArrayIndexOutOfBoundsException e) {
705 message = ZZ_ERROR_MSG[ZZ_UNKNOWN_ERROR];
708 throw new Error(message);
713 * Pushes the specified amount of characters back into the input stream.
715 * They will be read again by then next call of the scanning method
717 * @param number the number of characters to be read again.
718 * This number must not be greater than yylength()!
720 public void yypushback(int number) {
721 if ( number > yylength() )
722 zzScanError(ZZ_PUSHBACK_2BIG);
724 zzMarkedPos -= number;
729 * Resumes scanning until the next regular expression is matched,
730 * the end of input is encountered or an I/O-Error occurs.
732 * @return the next token
733 * @exception java.io.IOException if any I/O-Error occurs
735 public int getNextToken() throws java.io.IOException {
742 int zzEndReadL = zzEndRead;
743 char [] zzBufferL = zzBuffer;
744 char [] zzCMapL = ZZ_CMAP;
746 int [] zzTransL = ZZ_TRANS;
747 int [] zzRowMapL = ZZ_ROWMAP;
748 int [] zzAttrL = ZZ_ATTRIBUTE;
751 zzMarkedPosL = zzMarkedPos;
753 yychar+= zzMarkedPosL-zzStartRead;
757 zzCurrentPosL = zzCurrentPos = zzStartRead = zzMarkedPosL;
759 zzState = ZZ_LEXSTATE[zzLexicalState];
761 // set up zzAction for empty match case:
762 int zzAttributes = zzAttrL[zzState];
763 if ( (zzAttributes & 1) == 1 ) {
771 if (zzCurrentPosL < zzEndReadL)
772 zzInput = zzBufferL[zzCurrentPosL++];
778 // store back cached positions
779 zzCurrentPos = zzCurrentPosL;
780 zzMarkedPos = zzMarkedPosL;
781 boolean eof = zzRefill();
782 // get translated positions and possibly new buffer
783 zzCurrentPosL = zzCurrentPos;
784 zzMarkedPosL = zzMarkedPos;
785 zzBufferL = zzBuffer;
786 zzEndReadL = zzEndRead;
792 zzInput = zzBufferL[zzCurrentPosL++];
795 int zzNext = zzTransL[ zzRowMapL[zzState] + zzCMapL[zzInput] ];
796 if (zzNext == -1) break zzForAction;
799 zzAttributes = zzAttrL[zzState];
800 if ( (zzAttributes & 1) == 1 ) {
802 zzMarkedPosL = zzCurrentPosL;
803 if ( (zzAttributes & 8) == 8 ) break zzForAction;
809 // store back cached position
810 zzMarkedPos = zzMarkedPosL;
812 switch (zzAction < 0 ? zzAction : ZZ_ACTION[zzAction]) {
814 { numWikiTokensSeen = 0; positionInc = 1; currentTokType = CATEGORY; yybegin(CATEGORY_STATE);/* Break so we don't hit fall-through warning: */ break;
818 { currentTokType = BOLD_ITALICS; yybegin(FIVE_SINGLE_QUOTES_STATE); /* Break so we don't hit fall-through warning: */ break;
822 { currentTokType = HEADING; yybegin(DOUBLE_EQUALS_STATE); numWikiTokensSeen++; return currentTokType;
826 { numBalanced = 0; numWikiTokensSeen = 0; currentTokType = EXTERNAL_LINK;yybegin(EXTERNAL_LINK_STATE); /* Break so we don't hit fall-through warning: */ break;
830 { positionInc = 1; return ACRONYM;
834 { positionInc = 1; /* Break so we don't hit fall-through warning: */ break;
838 { positionInc = 1; return COMPANY;
842 { numLinkToks = 0; positionInc = 0; yybegin(YYINITIAL); /* Break so we don't hit fall-through warning: */ break;
846 { currentTokType = SUB_HEADING; numWikiTokensSeen = 0; yybegin(STRING); /* Break so we don't hit fall-through warning: */ break;
850 { numWikiTokensSeen = 0; positionInc = 1; if (numBalanced == 0){numBalanced++;yybegin(TWO_SINGLE_QUOTES_STATE);} else{numBalanced = 0;}/* Break so we don't hit fall-through warning: */ break;
854 { positionInc = 1; return NUM;
858 { positionInc = 1; return APOSTROPHE;
862 { yybegin(STRING); return currentTokType;/*pipe*/
866 { /* Break so we don't hit fall-through warning: */ break;/* ignore STRING */
870 { positionInc = 1; return ALPHANUM;
874 { numWikiTokensSeen = 0; positionInc = 1; /* Break so we don't hit fall-through warning: */ break;
878 { yybegin(DOUBLE_BRACE_STATE); numWikiTokensSeen = 0; return currentTokType;
882 { numBalanced = 0;currentTokType = ALPHANUM; yybegin(YYINITIAL); /* Break so we don't hit fall-through warning: */ break;/*end sub header*/
886 { currentTokType = INTERNAL_LINK; numWikiTokensSeen = 0; yybegin(INTERNAL_LINK_STATE); /* Break so we don't hit fall-through warning: */ break;
890 { numBalanced = 0; numWikiTokensSeen = 0; currentTokType = CATEGORY;yybegin(CATEGORY_STATE); /* Break so we don't hit fall-through warning: */ break;
894 { numLinkToks = 0; yybegin(YYINITIAL); /* Break so we don't hit fall-through warning: */ break;
898 { numWikiTokensSeen = 0; positionInc = 1; currentTokType = EXTERNAL_LINK_URL; yybegin(EXTERNAL_LINK_STATE);/* Break so we don't hit fall-through warning: */ break;
902 { numBalanced = 0;currentTokType = ALPHANUM;yybegin(YYINITIAL); /* Break so we don't hit fall-through warning: */ break;/*end bold*/
906 { currentTokType = EXTERNAL_LINK; numWikiTokensSeen = 0; yybegin(EXTERNAL_LINK_STATE); /* Break so we don't hit fall-through warning: */ break;
910 { positionInc = 1; return CJ;
914 { currentTokType = CATEGORY; numWikiTokensSeen = 0; yybegin(CATEGORY_STATE); /* Break so we don't hit fall-through warning: */ break;
918 { yybegin(CATEGORY_STATE); numWikiTokensSeen++; return currentTokType;
922 { currentTokType = BOLD; yybegin(THREE_SINGLE_QUOTES_STATE); /* Break so we don't hit fall-through warning: */ break;
926 { numWikiTokensSeen = 0; positionInc = 1; currentTokType = CITATION; yybegin(DOUBLE_BRACE_STATE);/* Break so we don't hit fall-through warning: */ break;
930 { /* Break so we don't hit fall-through warning: */ break;/* ignore */
934 { yybegin(STRING); numWikiTokensSeen++; return currentTokType;/* STRING ALPHANUM*/
938 { positionInc = 1; numWikiTokensSeen++; yybegin(EXTERNAL_LINK_STATE); return currentTokType;
942 { numBalanced = 0;currentTokType = ALPHANUM; yybegin(YYINITIAL); /* Break so we don't hit fall-through warning: */ break;/*end bold italics*/
946 { yybegin(YYINITIAL); /* Break so we don't hit fall-through warning: */ break;
950 { yybegin(STRING); numWikiTokensSeen++; return currentTokType;
954 { if (numLinkToks == 0){positionInc = 0;} else{positionInc = 1;} numWikiTokensSeen++; currentTokType = EXTERNAL_LINK; yybegin(EXTERNAL_LINK_STATE); numLinkToks++; return currentTokType;
958 { yybegin(INTERNAL_LINK_STATE); numWikiTokensSeen++; return currentTokType;
962 { positionInc = 1; return EMAIL;
966 { currentTokType = INTERNAL_LINK; numWikiTokensSeen = 0; yybegin(INTERNAL_LINK_STATE); /* Break so we don't hit fall-through warning: */ break;
970 { numWikiTokensSeen = 0; positionInc = 1; yybegin(DOUBLE_EQUALS_STATE);/* Break so we don't hit fall-through warning: */ break;
974 { positionInc = 1; return HOST;
978 { numBalanced = 0; numWikiTokensSeen = 0; currentTokType = INTERNAL_LINK;yybegin(INTERNAL_LINK_STATE); /* Break so we don't hit fall-through warning: */ break;
982 { currentTokType = ITALICS; numWikiTokensSeen++; yybegin(STRING); return currentTokType;/*italics*/
986 { numWikiTokensSeen = 0; positionInc = 1; currentTokType = INTERNAL_LINK; yybegin(INTERNAL_LINK_STATE);/* Break so we don't hit fall-through warning: */ break;
990 { numBalanced = 0;currentTokType = ALPHANUM; yybegin(YYINITIAL); /* Break so we don't hit fall-through warning: */ break;/*end italics*/
994 { yybegin(YYINITIAL);/* Break so we don't hit fall-through warning: */ break;
998 if (zzInput == YYEOF && zzStartRead == zzCurrentPos) {
1003 zzScanError(ZZ_NO_MATCH);