FulltextStoplist.php
Go to the documentation of this file.00001 <?php
00002
00003
# This is the MySQL fulltext search stoplist, copied from the
00004
# source file "myisam/ft_static.c" in the MySQL source distribution.
00005
# If you use a new version of MySQL, this might have to be changed.
00006
00007 $wgFulltextStoplist = array(
00008
00009
"a",
00010
"a's",
00011
"able",
00012
"about",
00013
"above",
00014
"according",
00015
"accordingly",
00016
"across",
00017
"actually",
00018
"after",
00019
"afterwards",
00020
"again",
00021
"against",
00022
"ain't",
00023
"all",
00024
"allow",
00025
"allows",
00026
"almost",
00027
"alone",
00028
"along",
00029
"already",
00030
"also",
00031
"although",
00032
"always",
00033
"am",
00034
"among",
00035
"amongst",
00036
"an",
00037
"and",
00038
"another",
00039
"any",
00040
"anybody",
00041
"anyhow",
00042
"anyone",
00043
"anything",
00044
"anyway",
00045
"anyways",
00046
"anywhere",
00047
"apart",
00048
"appear",
00049
"appreciate",
00050
"appropriate",
00051
"are",
00052
"aren't",
00053
"around",
00054
"as",
00055
"aside",
00056
"ask",
00057
"asking",
00058
"associated",
00059
"at",
00060
"available",
00061
"away",
00062
"awfully",
00063
"b",
00064
"be",
00065
"became",
00066
"because",
00067
"become",
00068
"becomes",
00069
"becoming",
00070
"been",
00071
"before",
00072
"beforehand",
00073
"behind",
00074
"being",
00075
"believe",
00076
"below",
00077
"beside",
00078
"besides",
00079
"best",
00080
"better",
00081
"between",
00082
"beyond",
00083
"both",
00084
"brief",
00085
"but",
00086
"by",
00087
"c",
00088
"c'mon",
00089
"c's",
00090
"came",
00091
"can",
00092
"can't",
00093
"cannot",
00094
"cant",
00095
"cause",
00096
"causes",
00097
"certain",
00098
"certainly",
00099
"changes",
00100
"clearly",
00101
"co",
00102
"com",
00103
"come",
00104
"comes",
00105
"concerning",
00106
"consequently",
00107
"consider",
00108
"considering",
00109
"contain",
00110
"containing",
00111
"contains",
00112
"corresponding",
00113
"could",
00114
"couldn't",
00115
"course",
00116
"currently",
00117
"d",
00118
"definitely",
00119
"described",
00120
"despite",
00121
"did",
00122
"didn't",
00123
"different",
00124
"do",
00125
"does",
00126
"doesn't",
00127
"doing",
00128
"don't",
00129
"done",
00130
"down",
00131
"downwards",
00132
"during",
00133
"e",
00134
"each",
00135
"edu",
00136
"eg",
00137
"eight",
00138
"either",
00139
"else",
00140
"elsewhere",
00141
"enough",
00142
"entirely",
00143
"especially",
00144
"et",
00145
"etc",
00146
"even",
00147
"ever",
00148
"every",
00149
"everybody",
00150
"everyone",
00151
"everything",
00152
"everywhere",
00153
"ex",
00154
"exactly",
00155
"example",
00156
"except",
00157
"f",
00158
"far",
00159
"few",
00160
"fifth",
00161
"first",
00162
"five",
00163
"followed",
00164
"following",
00165
"follows",
00166
"for",
00167
"former",
00168
"formerly",
00169
"forth",
00170
"four",
00171
"from",
00172
"further",
00173
"furthermore",
00174
"g",
00175
"get",
00176
"gets",
00177
"getting",
00178
"given",
00179
"gives",
00180
"go",
00181
"goes",
00182
"going",
00183
"gone",
00184
"got",
00185
"gotten",
00186
"greetings",
00187
"h",
00188
"had",
00189
"hadn't",
00190
"happens",
00191
"hardly",
00192
"has",
00193
"hasn't",
00194
"have",
00195
"haven't",
00196
"having",
00197
"he",
00198
"he's",
00199
"hello",
00200
"help",
00201
"hence",
00202
"her",
00203
"here",
00204
"here's",
00205
"hereafter",
00206
"hereby",
00207
"herein",
00208
"hereupon",
00209
"hers",
00210
"herself",
00211
"hi",
00212
"him",
00213
"himself",
00214
"his",
00215
"hither",
00216
"hopefully",
00217
"how",
00218
"howbeit",
00219
"however",
00220
"i",
00221
"i'd",
00222
"i'll",
00223
"i'm",
00224
"i've",
00225
"ie",
00226
"if",
00227
"ignored",
00228
"immediate",
00229
"in",
00230
"inasmuch",
00231
"inc",
00232
"indeed",
00233
"indicate",
00234
"indicated",
00235
"indicates",
00236
"inner",
00237
"insofar",
00238
"instead",
00239
"into",
00240
"inward",
00241
"is",
00242
"isn't",
00243
"it",
00244
"it'd",
00245
"it'll",
00246
"it's",
00247
"its",
00248
"itself",
00249
"j",
00250
"just",
00251
"k",
00252
"keep",
00253
"keeps",
00254
"kept",
00255
"know",
00256
"knows",
00257
"known",
00258
"l",
00259
"last",
00260
"lately",
00261
"later",
00262
"latter",
00263
"latterly",
00264
"least",
00265
"less",
00266
"lest",
00267
"let",
00268
"let's",
00269
"like",
00270
"liked",
00271
"likely",
00272
"little",
00273
"look",
00274
"looking",
00275
"looks",
00276
"ltd",
00277
"m",
00278
"mainly",
00279
"many",
00280
"may",
00281
"maybe",
00282
"me",
00283
"mean",
00284
"meanwhile",
00285
"merely",
00286
"might",
00287
"more",
00288
"moreover",
00289
"most",
00290
"mostly",
00291
"much",
00292
"must",
00293
"my",
00294
"myself",
00295
"n",
00296
"name",
00297
"namely",
00298
"nd",
00299
"near",
00300
"nearly",
00301
"necessary",
00302
"need",
00303
"needs",
00304
"neither",
00305
"never",
00306
"nevertheless",
00307
"new",
00308
"next",
00309
"nine",
00310
"no",
00311
"nobody",
00312
"non",
00313
"none",
00314
"noone",
00315
"nor",
00316
"normally",
00317
"not",
00318
"nothing",
00319
"novel",
00320
"now",
00321
"nowhere",
00322
"o",
00323
"obviously",
00324
"of",
00325
"off",
00326
"often",
00327
"oh",
00328
"ok",
00329
"okay",
00330
"old",
00331
"on",
00332
"once",
00333
"one",
00334
"ones",
00335
"only",
00336
"onto",
00337
"or",
00338
"other",
00339
"others",
00340
"otherwise",
00341
"ought",
00342
"our",
00343
"ours",
00344
"ourselves",
00345
"out",
00346
"outside",
00347
"over",
00348
"overall",
00349
"own",
00350
"p",
00351
"particular",
00352
"particularly",
00353
"per",
00354
"perhaps",
00355
"placed",
00356
"please",
00357
"plus",
00358
"possible",
00359
"presumably",
00360
"probably",
00361
"provides",
00362
"q",
00363
"que",
00364
"quite",
00365
"qv",
00366
"r",
00367
"rather",
00368
"rd",
00369
"re",
00370
"really",
00371
"reasonably",
00372
"regarding",
00373
"regardless",
00374
"regards",
00375
"relatively",
00376
"respectively",
00377
"right",
00378
"s",
00379
"said",
00380
"same",
00381
"saw",
00382
"say",
00383
"saying",
00384
"says",
00385
"second",
00386
"secondly",
00387
"see",
00388
"seeing",
00389
"seem",
00390
"seemed",
00391
"seeming",
00392
"seems",
00393
"seen",
00394
"self",
00395
"selves",
00396
"sensible",
00397
"sent",
00398
"serious",
00399
"seriously",
00400
"seven",
00401
"several",
00402
"shall",
00403
"she",
00404
"should",
00405
"shouldn't",
00406
"since",
00407
"six",
00408
"so",
00409
"some",
00410
"somebody",
00411
"somehow",
00412
"someone",
00413
"something",
00414
"sometime",
00415
"sometimes",
00416
"somewhat",
00417
"somewhere",
00418
"soon",
00419
"sorry",
00420
"specified",
00421
"specify",
00422
"specifying",
00423
"still",
00424
"sub",
00425
"such",
00426
"sup",
00427
"sure",
00428
"t",
00429
"t's",
00430
"take",
00431
"taken",
00432
"tell",
00433
"tends",
00434
"th",
00435
"than",
00436
"thank",
00437
"thanks",
00438
"thanx",
00439
"that",
00440
"that's",
00441
"thats",
00442
"the",
00443
"their",
00444
"theirs",
00445
"them",
00446
"themselves",
00447
"then",
00448
"thence",
00449
"there",
00450
"there's",
00451
"thereafter",
00452
"thereby",
00453
"therefore",
00454
"therein",
00455
"theres",
00456
"thereupon",
00457
"these",
00458
"they",
00459
"they'd",
00460
"they'll",
00461
"they're",
00462
"they've",
00463
"think",
00464
"third",
00465
"this",
00466
"thorough",
00467
"thoroughly",
00468
"those",
00469
"though",
00470
"three",
00471
"through",
00472
"throughout",
00473
"thru",
00474
"thus",
00475
"to",
00476
"together",
00477
"too",
00478
"took",
00479
"toward",
00480
"towards",
00481
"tried",
00482
"tries",
00483
"truly",
00484
"try",
00485
"trying",
00486
"twice",
00487
"two",
00488
"u",
00489
"un",
00490
"under",
00491
"unfortunately",
00492
"unless",
00493
"unlikely",
00494
"until",
00495
"unto",
00496
"up",
00497
"upon",
00498
"us",
00499
"use",
00500
"used",
00501
"useful",
00502
"uses",
00503
"using",
00504
"usually",
00505
"v",
00506
"value",
00507
"various",
00508
"very",
00509
"via",
00510
"viz",
00511
"vs",
00512
"w",
00513
"want",
00514
"wants",
00515
"was",
00516
"wasn't",
00517
"way",
00518
"we",
00519
"we'd",
00520
"we'll",
00521
"we're",
00522
"we've",
00523
"welcome",
00524
"well",
00525
"went",
00526
"were",
00527
"weren't",
00528
"what",
00529
"what's",
00530
"whatever",
00531
"when",
00532
"whence",
00533
"whenever",
00534
"where",
00535
"where's",
00536
"whereafter",
00537
"whereas",
00538
"whereby",
00539
"wherein",
00540
"whereupon",
00541
"wherever",
00542
"whether",
00543
"which",
00544
"while",
00545
"whither",
00546
"who",
00547
"who's",
00548
"whoever",
00549
"whole",
00550
"whom",
00551
"whose",
00552
"why",
00553
"will",
00554
"willing",
00555
"wish",
00556
"with",
00557
"within",
00558
"without",
00559
"won't",
00560
"wonder",
00561
"would",
00562
"would",
00563
"wouldn't",
00564
"x",
00565
"y",
00566
"yes",
00567
"yet",
00568
"you",
00569
"you'd",
00570
"you'll",
00571
"you're",
00572
"you've",
00573
"your",
00574
"yours",
00575
"yourself",
00576
"yourselves",
00577
"z",
00578
"zero" );
00579
00580 class FulltextStoplist {
00581
00582 function
inList( $word )
00583 {
00584 global
$wgFulltextStoplist;
00585
# FIXME: this will fail badly in many circumstances
00586
$w = strtolower( $word );
00587 $w = preg_replace(
"/[^a-z']+/",
"", $w );
00588
return in_array( $w, $wgFulltextStoplist );
00589 }
00590 }
00591
00592 ?>
Generated on Tue Jun 29 23:40:03 2004 for Mediawiki by
1.3.7