java/galago/src/org/tartarus/snowball/ext/russianStemmer.java

   1 // This file was generated automatically by the Snowball to Java compiler
   2
   3 package org.tartarus.snowball.ext;
   4 import org.tartarus.snowball.SnowballProgram;
   5 import org.tartarus.snowball.Among;
   6
   7 /**
   8  * Generated class implementing code defined by a snowball script.
   9  */
  10 public class russianStemmer extends SnowballProgram {
  11
  12         private Among a_0[] = {
  13             new Among ( "\u0432", -1, 1, "", this),
  14             new Among ( "\u0438\u0432", 0, 2, "", this),
  15             new Among ( "\u044B\u0432", 0, 2, "", this),
  16             new Among ( "\u0432\u0448\u0438", -1, 1, "", this),
  17             new Among ( "\u0438\u0432\u0448\u0438", 3, 2, "", this),
  18             new Among ( "\u044B\u0432\u0448\u0438", 3, 2, "", this),
  19             new Among ( "\u0432\u0448\u0438\u0441\u044C", -1, 1, "", this),
  20             new Among ( "\u0438\u0432\u0448\u0438\u0441\u044C", 6, 2, "", this),
  21             new Among ( "\u044B\u0432\u0448\u0438\u0441\u044C", 6, 2, "", this)
  22         };
  23
  24         private Among a_1[] = {
  25             new Among ( "\u0435\u0435", -1, 1, "", this),
  26             new Among ( "\u0438\u0435", -1, 1, "", this),
  27             new Among ( "\u043E\u0435", -1, 1, "", this),
  28             new Among ( "\u044B\u0435", -1, 1, "", this),
  29             new Among ( "\u0438\u043C\u0438", -1, 1, "", this),
  30             new Among ( "\u044B\u043C\u0438", -1, 1, "", this),
  31             new Among ( "\u0435\u0439", -1, 1, "", this),
  32             new Among ( "\u0438\u0439", -1, 1, "", this),
  33             new Among ( "\u043E\u0439", -1, 1, "", this),
  34             new Among ( "\u044B\u0439", -1, 1, "", this),
  35             new Among ( "\u0435\u043C", -1, 1, "", this),
  36             new Among ( "\u0438\u043C", -1, 1, "", this),
  37             new Among ( "\u043E\u043C", -1, 1, "", this),
  38             new Among ( "\u044B\u043C", -1, 1, "", this),
  39             new Among ( "\u0435\u0433\u043E", -1, 1, "", this),
  40             new Among ( "\u043E\u0433\u043E", -1, 1, "", this),
  41             new Among ( "\u0435\u043C\u0443", -1, 1, "", this),
  42             new Among ( "\u043E\u043C\u0443", -1, 1, "", this),
  43             new Among ( "\u0438\u0445", -1, 1, "", this),
  44             new Among ( "\u044B\u0445", -1, 1, "", this),
  45             new Among ( "\u0435\u044E", -1, 1, "", this),
  46             new Among ( "\u043E\u044E", -1, 1, "", this),
  47             new Among ( "\u0443\u044E", -1, 1, "", this),
  48             new Among ( "\u044E\u044E", -1, 1, "", this),
  49             new Among ( "\u0430\u044F", -1, 1, "", this),
  50             new Among ( "\u044F\u044F", -1, 1, "", this)
  51         };
  52
  53         private Among a_2[] = {
  54             new Among ( "\u0435\u043C", -1, 1, "", this),
  55             new Among ( "\u043D\u043D", -1, 1, "", this),
  56             new Among ( "\u0432\u0448", -1, 1, "", this),
  57             new Among ( "\u0438\u0432\u0448", 2, 2, "", this),
  58             new Among ( "\u044B\u0432\u0448", 2, 2, "", this),
  59             new Among ( "\u0449", -1, 1, "", this),
  60             new Among ( "\u044E\u0449", 5, 1, "", this),
  61             new Among ( "\u0443\u044E\u0449", 6, 2, "", this)
  62         };
  63
  64         private Among a_3[] = {
  65             new Among ( "\u0441\u044C", -1, 1, "", this),
  66             new Among ( "\u0441\u044F", -1, 1, "", this)
  67         };
  68
  69         private Among a_4[] = {
  70             new Among ( "\u043B\u0430", -1, 1, "", this),
  71             new Among ( "\u0438\u043B\u0430", 0, 2, "", this),
  72             new Among ( "\u044B\u043B\u0430", 0, 2, "", this),
  73             new Among ( "\u043D\u0430", -1, 1, "", this),
  74             new Among ( "\u0435\u043D\u0430", 3, 2, "", this),
  75             new Among ( "\u0435\u0442\u0435", -1, 1, "", this),
  76             new Among ( "\u0438\u0442\u0435", -1, 2, "", this),
  77             new Among ( "\u0439\u0442\u0435", -1, 1, "", this),
  78             new Among ( "\u0435\u0439\u0442\u0435", 7, 2, "", this),
  79             new Among ( "\u0443\u0439\u0442\u0435", 7, 2, "", this),
  80             new Among ( "\u043B\u0438", -1, 1, "", this),
  81             new Among ( "\u0438\u043B\u0438", 10, 2, "", this),
  82             new Among ( "\u044B\u043B\u0438", 10, 2, "", this),
  83             new Among ( "\u0439", -1, 1, "", this),
  84             new Among ( "\u0435\u0439", 13, 2, "", this),
  85             new Among ( "\u0443\u0439", 13, 2, "", this),
  86             new Among ( "\u043B", -1, 1, "", this),
  87             new Among ( "\u0438\u043B", 16, 2, "", this),
  88             new Among ( "\u044B\u043B", 16, 2, "", this),
  89             new Among ( "\u0435\u043C", -1, 1, "", this),
  90             new Among ( "\u0438\u043C", -1, 2, "", this),
  91             new Among ( "\u044B\u043C", -1, 2, "", this),
  92             new Among ( "\u043D", -1, 1, "", this),
  93             new Among ( "\u0435\u043D", 22, 2, "", this),
  94             new Among ( "\u043B\u043E", -1, 1, "", this),
  95             new Among ( "\u0438\u043B\u043E", 24, 2, "", this),
  96             new Among ( "\u044B\u043B\u043E", 24, 2, "", this),
  97             new Among ( "\u043D\u043E", -1, 1, "", this),
  98             new Among ( "\u0435\u043D\u043E", 27, 2, "", this),
  99             new Among ( "\u043D\u043D\u043E", 27, 1, "", this),
 100             new Among ( "\u0435\u0442", -1, 1, "", this),
 101             new Among ( "\u0443\u0435\u0442", 30, 2, "", this),
 102             new Among ( "\u0438\u0442", -1, 2, "", this),
 103             new Among ( "\u044B\u0442", -1, 2, "", this),
 104             new Among ( "\u044E\u0442", -1, 1, "", this),
 105             new Among ( "\u0443\u044E\u0442", 34, 2, "", this),
 106             new Among ( "\u044F\u0442", -1, 2, "", this),
 107             new Among ( "\u043D\u044B", -1, 1, "", this),
 108             new Among ( "\u0435\u043D\u044B", 37, 2, "", this),
 109             new Among ( "\u0442\u044C", -1, 1, "", this),
 110             new Among ( "\u0438\u0442\u044C", 39, 2, "", this),
 111             new Among ( "\u044B\u0442\u044C", 39, 2, "", this),
 112             new Among ( "\u0435\u0448\u044C", -1, 1, "", this),
 113             new Among ( "\u0438\u0448\u044C", -1, 2, "", this),
 114             new Among ( "\u044E", -1, 2, "", this),
 115             new Among ( "\u0443\u044E", 44, 2, "", this)
 116         };
 117
 118         private Among a_5[] = {
 119             new Among ( "\u0430", -1, 1, "", this),
 120             new Among ( "\u0435\u0432", -1, 1, "", this),
 121             new Among ( "\u043E\u0432", -1, 1, "", this),
 122             new Among ( "\u0435", -1, 1, "", this),
 123             new Among ( "\u0438\u0435", 3, 1, "", this),
 124             new Among ( "\u044C\u0435", 3, 1, "", this),
 125             new Among ( "\u0438", -1, 1, "", this),
 126             new Among ( "\u0435\u0438", 6, 1, "", this),
 127             new Among ( "\u0438\u0438", 6, 1, "", this),
 128             new Among ( "\u0430\u043C\u0438", 6, 1, "", this),
 129             new Among ( "\u044F\u043C\u0438", 6, 1, "", this),
 130             new Among ( "\u0438\u044F\u043C\u0438", 10, 1, "", this),
 131             new Among ( "\u0439", -1, 1, "", this),
 132             new Among ( "\u0435\u0439", 12, 1, "", this),
 133             new Among ( "\u0438\u0435\u0439", 13, 1, "", this),
 134             new Among ( "\u0438\u0439", 12, 1, "", this),
 135             new Among ( "\u043E\u0439", 12, 1, "", this),
 136             new Among ( "\u0430\u043C", -1, 1, "", this),
 137             new Among ( "\u0435\u043C", -1, 1, "", this),
 138             new Among ( "\u0438\u0435\u043C", 18, 1, "", this),
 139             new Among ( "\u043E\u043C", -1, 1, "", this),
 140             new Among ( "\u044F\u043C", -1, 1, "", this),
 141             new Among ( "\u0438\u044F\u043C", 21, 1, "", this),
 142             new Among ( "\u043E", -1, 1, "", this),
 143             new Among ( "\u0443", -1, 1, "", this),
 144             new Among ( "\u0430\u0445", -1, 1, "", this),
 145             new Among ( "\u044F\u0445", -1, 1, "", this),
 146             new Among ( "\u0438\u044F\u0445", 26, 1, "", this),
 147             new Among ( "\u044B", -1, 1, "", this),
 148             new Among ( "\u044C", -1, 1, "", this),
 149             new Among ( "\u044E", -1, 1, "", this),
 150             new Among ( "\u0438\u044E", 30, 1, "", this),
 151             new Among ( "\u044C\u044E", 30, 1, "", this),
 152             new Among ( "\u044F", -1, 1, "", this),
 153             new Among ( "\u0438\u044F", 33, 1, "", this),
 154             new Among ( "\u044C\u044F", 33, 1, "", this)
 155         };
 156
 157         private Among a_6[] = {
 158             new Among ( "\u043E\u0441\u0442", -1, 1, "", this),
 159             new Among ( "\u043E\u0441\u0442\u044C", -1, 1, "", this)
 160         };
 161
 162         private Among a_7[] = {
 163             new Among ( "\u0435\u0439\u0448\u0435", -1, 1, "", this),
 164             new Among ( "\u043D", -1, 2, "", this),
 165             new Among ( "\u0435\u0439\u0448", -1, 1, "", this),
 166             new Among ( "\u044C", -1, 3, "", this)
 167         };
 168
 169         private static final char g_v[] = {33, 65, 8, 232 };
 170
 171         private int I_p2;
 172         private int I_pV;
 173
 174         private void copy_from(russianStemmer other) {
 175             I_p2 = other.I_p2;
 176             I_pV = other.I_pV;
 177             super.copy_from(other);
 178         }
 179
 180         private boolean r_mark_regions() {
 181             int v_1;
 182             // (, line 57
 183             I_pV = limit;
 184             I_p2 = limit;
 185             // do, line 61
 186             v_1 = cursor;
 187             lab0: do {
 188                 // (, line 61
 189                 // gopast, line 62
 190                 golab1: while(true)
 191                 {
 192                     lab2: do {
 193                         if (!(in_grouping(g_v, 1072, 1103)))
 194                         {
 195                             break lab2;
 196                         }
 197                         break golab1;
 198                     } while (false);
 199                     if (cursor >= limit)
 200                     {
 201                         break lab0;
 202                     }
 203                     cursor++;
 204                 }
 205                 // setmark pV, line 62
 206                 I_pV = cursor;
 207                 // gopast, line 62
 208                 golab3: while(true)
 209                 {
 210                     lab4: do {
 211                         if (!(out_grouping(g_v, 1072, 1103)))
 212                         {
 213                             break lab4;
 214                         }
 215                         break golab3;
 216                     } while (false);
 217                     if (cursor >= limit)
 218                     {
 219                         break lab0;
 220                     }
 221                     cursor++;
 222                 }
 223                 // gopast, line 63
 224                 golab5: while(true)
 225                 {
 226                     lab6: do {
 227                         if (!(in_grouping(g_v, 1072, 1103)))
 228                         {
 229                             break lab6;
 230                         }
 231                         break golab5;
 232                     } while (false);
 233                     if (cursor >= limit)
 234                     {
 235                         break lab0;
 236                     }
 237                     cursor++;
 238                 }
 239                 // gopast, line 63
 240                 golab7: while(true)
 241                 {
 242                     lab8: do {
 243                         if (!(out_grouping(g_v, 1072, 1103)))
 244                         {
 245                             break lab8;
 246                         }
 247                         break golab7;
 248                     } while (false);
 249                     if (cursor >= limit)
 250                     {
 251                         break lab0;
 252                     }
 253                     cursor++;
 254                 }
 255                 // setmark p2, line 63
 256                 I_p2 = cursor;
 257             } while (false);
 258             cursor = v_1;
 259             return true;
 260         }
 261
 262         private boolean r_R2() {
 263             if (!(I_p2 <= cursor))
 264             {
 265                 return false;
 266             }
 267             return true;
 268         }
 269
 270         private boolean r_perfective_gerund() {
 271             int among_var;
 272             int v_1;
 273             // (, line 71
 274             // [, line 72
 275             ket = cursor;
 276             // substring, line 72
 277             among_var = find_among_b(a_0, 9);
 278             if (among_var == 0)
 279             {
 280                 return false;
 281             }
 282             // ], line 72
 283             bra = cursor;
 284             switch(among_var) {
 285                 case 0:
 286                     return false;
 287                 case 1:
 288                     // (, line 76
 289                     // or, line 76
 290                     lab0: do {
 291                         v_1 = limit - cursor;
 292                         lab1: do {
 293                             // literal, line 76
 294                             if (!(eq_s_b(1, "\u0430")))
 295                             {
 296                                 break lab1;
 297                             }
 298                             break lab0;
 299                         } while (false);
 300                         cursor = limit - v_1;
 301                         // literal, line 76
 302                         if (!(eq_s_b(1, "\u044F")))
 303                         {
 304                             return false;
 305                         }
 306                     } while (false);
 307                     // delete, line 76
 308                     slice_del();
 309                     break;
 310                 case 2:
 311                     // (, line 83
 312                     // delete, line 83
 313                     slice_del();
 314                     break;
 315             }
 316             return true;
 317         }
 318
 319         private boolean r_adjective() {
 320             int among_var;
 321             // (, line 87
 322             // [, line 88
 323             ket = cursor;
 324             // substring, line 88
 325             among_var = find_among_b(a_1, 26);
 326             if (among_var == 0)
 327             {
 328                 return false;
 329             }
 330             // ], line 88
 331             bra = cursor;
 332             switch(among_var) {
 333                 case 0:
 334                     return false;
 335                 case 1:
 336                     // (, line 97
 337                     // delete, line 97
 338                     slice_del();
 339                     break;
 340             }
 341             return true;
 342         }
 343
 344         private boolean r_adjectival() {
 345             int among_var;
 346             int v_1;
 347             int v_2;
 348             // (, line 101
 349             // call adjective, line 102
 350             if (!r_adjective())
 351             {
 352                 return false;
 353             }
 354             // try, line 109
 355             v_1 = limit - cursor;
 356             lab0: do {
 357                 // (, line 109
 358                 // [, line 110
 359                 ket = cursor;
 360                 // substring, line 110
 361                 among_var = find_among_b(a_2, 8);
 362                 if (among_var == 0)
 363                 {
 364                     cursor = limit - v_1;
 365                     break lab0;
 366                 }
 367                 // ], line 110
 368                 bra = cursor;
 369                 switch(among_var) {
 370                     case 0:
 371                         cursor = limit - v_1;
 372                         break lab0;
 373                     case 1:
 374                         // (, line 115
 375                         // or, line 115
 376                         lab1: do {
 377                             v_2 = limit - cursor;
 378                             lab2: do {
 379                                 // literal, line 115
 380                                 if (!(eq_s_b(1, "\u0430")))
 381                                 {
 382                                     break lab2;
 383                                 }
 384                                 break lab1;
 385                             } while (false);
 386                             cursor = limit - v_2;
 387                             // literal, line 115
 388                             if (!(eq_s_b(1, "\u044F")))
 389                             {
 390                                 cursor = limit - v_1;
 391                                 break lab0;
 392                             }
 393                         } while (false);
 394                         // delete, line 115
 395                         slice_del();
 396                         break;
 397                     case 2:
 398                         // (, line 122
 399                         // delete, line 122
 400                         slice_del();
 401                         break;
 402                 }
 403             } while (false);
 404             return true;
 405         }
 406
 407         private boolean r_reflexive() {
 408             int among_var;
 409             // (, line 128
 410             // [, line 129
 411             ket = cursor;
 412             // substring, line 129
 413             among_var = find_among_b(a_3, 2);
 414             if (among_var == 0)
 415             {
 416                 return false;
 417             }
 418             // ], line 129
 419             bra = cursor;
 420             switch(among_var) {
 421                 case 0:
 422                     return false;
 423                 case 1:
 424                     // (, line 132
 425                     // delete, line 132
 426                     slice_del();
 427                     break;
 428             }
 429             return true;
 430         }
 431
 432         private boolean r_verb() {
 433             int among_var;
 434             int v_1;
 435             // (, line 136
 436             // [, line 137
 437             ket = cursor;
 438             // substring, line 137
 439             among_var = find_among_b(a_4, 46);
 440             if (among_var == 0)
 441             {
 442                 return false;
 443             }
 444             // ], line 137
 445             bra = cursor;
 446             switch(among_var) {
 447                 case 0:
 448                     return false;
 449                 case 1:
 450                     // (, line 143
 451                     // or, line 143
 452                     lab0: do {
 453                         v_1 = limit - cursor;
 454                         lab1: do {
 455                             // literal, line 143
 456                             if (!(eq_s_b(1, "\u0430")))
 457                             {
 458                                 break lab1;
 459                             }
 460                             break lab0;
 461                         } while (false);
 462                         cursor = limit - v_1;
 463                         // literal, line 143
 464                         if (!(eq_s_b(1, "\u044F")))
 465                         {
 466                             return false;
 467                         }
 468                     } while (false);
 469                     // delete, line 143
 470                     slice_del();
 471                     break;
 472                 case 2:
 473                     // (, line 151
 474                     // delete, line 151
 475                     slice_del();
 476                     break;
 477             }
 478             return true;
 479         }
 480
 481         private boolean r_noun() {
 482             int among_var;
 483             // (, line 159
 484             // [, line 160
 485             ket = cursor;
 486             // substring, line 160
 487             among_var = find_among_b(a_5, 36);
 488             if (among_var == 0)
 489             {
 490                 return false;
 491             }
 492             // ], line 160
 493             bra = cursor;
 494             switch(among_var) {
 495                 case 0:
 496                     return false;
 497                 case 1:
 498                     // (, line 167
 499                     // delete, line 167
 500                     slice_del();
 501                     break;
 502             }
 503             return true;
 504         }
 505
 506         private boolean r_derivational() {
 507             int among_var;
 508             // (, line 175
 509             // [, line 176
 510             ket = cursor;
 511             // substring, line 176
 512             among_var = find_among_b(a_6, 2);
 513             if (among_var == 0)
 514             {
 515                 return false;
 516             }
 517             // ], line 176
 518             bra = cursor;
 519             // call R2, line 176
 520             if (!r_R2())
 521             {
 522                 return false;
 523             }
 524             switch(among_var) {
 525                 case 0:
 526                     return false;
 527                 case 1:
 528                     // (, line 179
 529                     // delete, line 179
 530                     slice_del();
 531                     break;
 532             }
 533             return true;
 534         }
 535
 536         private boolean r_tidy_up() {
 537             int among_var;
 538             // (, line 183
 539             // [, line 184
 540             ket = cursor;
 541             // substring, line 184
 542             among_var = find_among_b(a_7, 4);
 543             if (among_var == 0)
 544             {
 545                 return false;
 546             }
 547             // ], line 184
 548             bra = cursor;
 549             switch(among_var) {
 550                 case 0:
 551                     return false;
 552                 case 1:
 553                     // (, line 188
 554                     // delete, line 188
 555                     slice_del();
 556                     // [, line 189
 557                     ket = cursor;
 558                     // literal, line 189
 559                     if (!(eq_s_b(1, "\u043D")))
 560                     {
 561                         return false;
 562                     }
 563                     // ], line 189
 564                     bra = cursor;
 565                     // literal, line 189
 566                     if (!(eq_s_b(1, "\u043D")))
 567                     {
 568                         return false;
 569                     }
 570                     // delete, line 189
 571                     slice_del();
 572                     break;
 573                 case 2:
 574                     // (, line 192
 575                     // literal, line 192
 576                     if (!(eq_s_b(1, "\u043D")))
 577                     {
 578                         return false;
 579                     }
 580                     // delete, line 192
 581                     slice_del();
 582                     break;
 583                 case 3:
 584                     // (, line 194
 585                     // delete, line 194
 586                     slice_del();
 587                     break;
 588             }
 589             return true;
 590         }
 591
 592         public boolean stem() {
 593             int v_1;
 594             int v_2;
 595             int v_3;
 596             int v_4;
 597             int v_5;
 598             int v_6;
 599             int v_7;
 600             int v_8;
 601             int v_9;
 602             int v_10;
 603             // (, line 199
 604             // do, line 201
 605             v_1 = cursor;
 606             lab0: do {
 607                 // call mark_regions, line 201
 608                 if (!r_mark_regions())
 609                 {
 610                     break lab0;
 611                 }
 612             } while (false);
 613             cursor = v_1;
 614             // backwards, line 202
 615             limit_backward = cursor; cursor = limit;
 616             // setlimit, line 202
 617             v_2 = limit - cursor;
 618             // tomark, line 202
 619             if (cursor < I_pV)
 620             {
 621                 return false;
 622             }
 623             cursor = I_pV;
 624             v_3 = limit_backward;
 625             limit_backward = cursor;
 626             cursor = limit - v_2;
 627             // (, line 202
 628             // do, line 203
 629             v_4 = limit - cursor;
 630             lab1: do {
 631                 // (, line 203
 632                 // or, line 204
 633                 lab2: do {
 634                     v_5 = limit - cursor;
 635                     lab3: do {
 636                         // call perfective_gerund, line 204
 637                         if (!r_perfective_gerund())
 638                         {
 639                             break lab3;
 640                         }
 641                         break lab2;
 642                     } while (false);
 643                     cursor = limit - v_5;
 644                     // (, line 205
 645                     // try, line 205
 646                     v_6 = limit - cursor;
 647                     lab4: do {
 648                         // call reflexive, line 205
 649                         if (!r_reflexive())
 650                         {
 651                             cursor = limit - v_6;
 652                             break lab4;
 653                         }
 654                     } while (false);
 655                     // or, line 206
 656                     lab5: do {
 657                         v_7 = limit - cursor;
 658                         lab6: do {
 659                             // call adjectival, line 206
 660                             if (!r_adjectival())
 661                             {
 662                                 break lab6;
 663                             }
 664                             break lab5;
 665                         } while (false);
 666                         cursor = limit - v_7;
 667                         lab7: do {
 668                             // call verb, line 206
 669                             if (!r_verb())
 670                             {
 671                                 break lab7;
 672                             }
 673                             break lab5;
 674                         } while (false);
 675                         cursor = limit - v_7;
 676                         // call noun, line 206
 677                         if (!r_noun())
 678                         {
 679                             break lab1;
 680                         }
 681                     } while (false);
 682                 } while (false);
 683             } while (false);
 684             cursor = limit - v_4;
 685             // try, line 209
 686             v_8 = limit - cursor;
 687             lab8: do {
 688                 // (, line 209
 689                 // [, line 209
 690                 ket = cursor;
 691                 // literal, line 209
 692                 if (!(eq_s_b(1, "\u0438")))
 693                 {
 694                     cursor = limit - v_8;
 695                     break lab8;
 696                 }
 697                 // ], line 209
 698                 bra = cursor;
 699                 // delete, line 209
 700                 slice_del();
 701             } while (false);
 702             // do, line 212
 703             v_9 = limit - cursor;
 704             lab9: do {
 705                 // call derivational, line 212
 706                 if (!r_derivational())
 707                 {
 708                     break lab9;
 709                 }
 710             } while (false);
 711             cursor = limit - v_9;
 712             // do, line 213
 713             v_10 = limit - cursor;
 714             lab10: do {
 715                 // call tidy_up, line 213
 716                 if (!r_tidy_up())
 717                 {
 718                     break lab10;
 719                 }
 720             } while (false);
 721             cursor = limit - v_10;
 722             limit_backward = v_3;
 723             cursor = limit_backward;            return true;
 724         }
 725
 726 }
 727