1 /* 2 * Copyright (C) 2008,2009 OMRON SOFTWARE Co., Ltd. 3 * 4 * Licensed under the Apache License, Version 2.0 (the "License"); 5 * you may not use this file except in compliance with the License. 6 * You may obtain a copy of the License at 7 * 8 * http://www.apache.org/licenses/LICENSE-2.0 9 * 10 * Unless required by applicable law or agreed to in writing, software 11 * distributed under the License is distributed on an "AS IS" BASIS, 12 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. 13 * See the License for the specific language governing permissions and 14 * limitations under the License. 15 */ 16 17 #include "nj_lib.h" 18 #include "nj_err.h" 19 #include "nj_ext.h" 20 #include "nj_dic.h" 21 #include "njd.h" 22 #ifdef NJ_LEARN_MUHENKAN_DEBUG 23 #include <stdio.h> 24 #include <def_mojicode.h> 25 #endif 26 #ifdef NJ_AWNN22_DEBUG 27 #include <stdio.h> 28 #include <def_mojicode.h> 29 #endif 30 31 #define QUE_TYPE_EMPTY 0 32 #define QUE_TYPE_NEXT 0 33 #define QUE_TYPE_JIRI 1 34 #define QUE_TYPE_FZK 2 35 #define POS_DATA_OFFSET 0x20 36 #define POS_LEARN_WORD 0x24 37 #define POS_MAX_WORD 0x28 38 #define POS_QUE_SIZE 0x2C 39 #define POS_NEXT_QUE 0x30 40 #define POS_WRITE_FLG 0x34 41 #define POS_INDEX_OFFSET 0x3C 42 #define POS_INDEX_OFFSET2 0x40 43 44 #define LEARN_INDEX_TOP_ADDR(x) ((x) + (NJ_INT32_READ((x) + POS_INDEX_OFFSET))) 45 #define LEARN_INDEX_TOP_ADDR2(x) ((x) + (NJ_INT32_READ((x) + POS_INDEX_OFFSET2))) 46 #define LEARN_DATA_TOP_ADDR(x) ((x) + (NJ_INT32_READ((x) + POS_DATA_OFFSET))) 47 48 #define LEARN_INDEX_BOTTOM_ADDR(x) (LEARN_DATA_TOP_ADDR(x) - 1) 49 50 #define LEARN_QUE_STRING_OFFSET 5 51 52 #define ADDRESS_TO_POS(x,adr) (((adr) - LEARN_DATA_TOP_ADDR(x)) / QUE_SIZE(x)) 53 #define POS_TO_ADDRESS(x,pos) (LEARN_DATA_TOP_ADDR(x) + QUE_SIZE(x) * (pos)) 54 55 #define GET_UINT16(ptr) ((((NJ_UINT16)(*(ptr))) << 8) | (*((ptr) + 1) & 0x00ff)) 56 57 #define GET_FPOS_FROM_DATA(x) ((NJ_UINT16)NJ_INT16_READ((x)+1) >> 7) 58 #define GET_YSIZE_FROM_DATA(x) ((NJ_UINT8)((NJ_UINT16)NJ_INT16_READ((x)+1) & 0x7F)) 59 #define GET_BPOS_FROM_DATA(x) ((NJ_UINT16)NJ_INT16_READ((x)+3) >> 7) 60 #define GET_KSIZE_FROM_DATA(x) ((NJ_UINT8)((NJ_UINT16)NJ_INT16_READ((x)+3) & 0x7F)) 61 #define GET_BPOS_FROM_EXT_DATA(x) ((NJ_UINT16)NJ_INT16_READ(x) >> 7) 62 #define GET_YSIZE_FROM_EXT_DATA(x) ((NJ_UINT8)((NJ_UINT16)NJ_INT16_READ(x) & 0x7F)) 63 64 #define SET_BPOS_AND_YSIZE(x,bpos,ysize) \ 65 NJ_INT16_WRITE((x), ((NJ_UINT16)((bpos) << 7) | ((ysize) & 0x7F))) 66 #define SET_FPOS_AND_YSIZE(x,fpos,ysize) \ 67 NJ_INT16_WRITE(((x)+1), ((NJ_UINT16)((fpos) << 7) | ((ysize) & 0x7F))) 68 #define SET_BPOS_AND_KSIZE(x,bpos,ksize) \ 69 NJ_INT16_WRITE(((x)+3), ((NJ_UINT16)((bpos) << 7) | ((ksize) & 0x7F))) 70 71 #define GET_TYPE_FROM_DATA(x) (*(x) & 0x03) 72 #define GET_UFLG_FROM_DATA(x) (*(x) >> 7) 73 #define GET_FFLG_FROM_DATA(x) ((*(x) >> 6) & 0x01) 74 #define GET_MFLG_FROM_DATA(x) (*(x) & 0x10) 75 76 #define SET_TYPE_UFLG_FFLG(x,type,u,f) \ 77 (*(x) = (NJ_UINT8)(((type) & 0x03) | \ 78 (((u) & 0x01) << 7) | (((f) & 0x01) << 6))) 79 #define SET_TYPE_ALLFLG(x,type,u,f,m) \ 80 (*(x) = (NJ_UINT8)(((type) & 0x03) | \ 81 (((u) & 0x01) << 7) | (((f) & 0x01) << 6) | (((m) & 0x01) << 4))) 82 83 #define RESET_FFLG(x) (*(x) &= 0xbf) 84 85 #define STATE_COPY(to, from) \ 86 { ((NJ_UINT8*)(to))[0] = ((NJ_UINT8*)(from))[0]; \ 87 ((NJ_UINT8*)(to))[1] = ((NJ_UINT8*)(from))[1]; \ 88 ((NJ_UINT8*)(to))[2] = ((NJ_UINT8*)(from))[2]; \ 89 ((NJ_UINT8*)(to))[3] = ((NJ_UINT8*)(from))[3]; } 90 91 #define USE_QUE_NUM(que_size, str_size) \ 92 ( (((str_size) % ((que_size) - 1)) == 0) \ 93 ? ((str_size) / ((que_size) - 1)) \ 94 : ((str_size) / ((que_size) - 1) + 1) ) 95 96 #define NEXT_QUE(que, max) ( ((que) < ((max) - 1)) ? ((que) + 1) : 0 ) 97 98 #define PREV_QUE(que, max) ( ((que) == 0) ? ((max) - 1) : ((que) - 1) ) 99 100 #define COPY_QUE(handle, src, dst) \ 101 nj_memcpy(POS_TO_ADDRESS((handle), (dst)), POS_TO_ADDRESS((handle), (src)), QUE_SIZE(handle)) 102 103 104 #define INIT_HINDO (-10000) 105 106 #define LOC_CURRENT_NO_ENTRY 0xffffffffU 107 108 109 110 static NJ_WQUE *get_que(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT16 que_id); 111 static NJ_INT16 is_continued(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT16 que_id); 112 static NJ_UINT16 search_next_que(NJ_DIC_HANDLE handle, NJ_UINT16 que_id); 113 static NJ_INT16 que_strcmp_complete_with_hyouki(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT16 que_id, NJ_CHAR *yomi, NJ_UINT16 yomi_len, NJ_CHAR *hyouki, NJ_UINT8 multi_flg); 114 static NJ_CHAR *get_string(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT16 que_id, NJ_UINT8 *slen); 115 static NJ_CHAR *get_hyouki(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT16 que_id, NJ_UINT8 *slen); 116 static NJ_INT16 get_cand_by_sequential(NJ_CLASS *iwnn, NJ_SEARCH_CONDITION *cond, NJ_SEARCH_LOCATION_SET *loctset, NJ_UINT8 search_pattern, NJ_UINT8 comp_flg); 117 static NJ_INT16 get_cand_by_evaluate(NJ_CLASS *iwnn, NJ_SEARCH_CONDITION *cond, NJ_SEARCH_LOCATION_SET *loctset, NJ_UINT8 search_pattern); 118 static NJ_INT16 get_cand_by_evaluate2(NJ_CLASS *iwnn, NJ_SEARCH_CONDITION *cond, NJ_SEARCH_LOCATION_SET *loctset, NJ_UINT8 search_pattern, NJ_UINT16 hIdx); 119 static NJ_INT16 search_range_by_yomi(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT8 op, NJ_CHAR *yomi, NJ_UINT16 ylen, NJ_UINT16 *from, NJ_UINT16 *to, NJ_UINT8 *forward_flag); 120 static NJ_INT16 search_range_by_yomi2(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT8 op, NJ_CHAR *yomi, NJ_UINT16 ylen, NJ_UINT16 sfrom, NJ_UINT16 sto, NJ_UINT16 *from, NJ_UINT16 *to, 121 NJ_UINT8 *forward_flag); 122 static NJ_INT16 search_range_by_yomi_multi(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_CHAR *yomi, NJ_UINT16 ylen, NJ_UINT16 *from, NJ_UINT16 *to); 123 static NJ_INT16 str_que_cmp(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_CHAR *yomi, NJ_UINT16 yomiLen, NJ_UINT16 que_id, NJ_UINT8 mode); 124 static NJ_WQUE *get_que_type_and_next(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT16 que_id); 125 static NJ_WQUE *get_que_allHinsi(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT16 que_id); 126 static NJ_WQUE *get_que_yomiLen_and_hyoukiLen(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT16 que_id); 127 static NJ_INT16 continue_cnt(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT16 que_id); 128 129 static NJ_UINT8 *get_search_index_address(NJ_DIC_HANDLE handle, NJ_UINT8 search_pattern); 130 131 static NJ_HINDO get_hindo(NJ_CLASS *iwnn, NJ_SEARCH_LOCATION_SET *loctset, NJ_UINT8 search_pattern); 132 133 static NJ_HINDO calculate_hindo(NJ_DIC_HANDLE handle, NJ_INT32 freq, NJ_DIC_FREQ *dic_freq, NJ_INT16 freq_max, NJ_INT16 freq_min); 134 static NJ_INT16 que_strcmp_include(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT16 que_id, NJ_CHAR *yomi); 135 136 #define GET_LEARN_MAX_WORD_COUNT(h) ((NJ_UINT16)NJ_INT32_READ((h) + POS_MAX_WORD)) 137 138 #define GET_LEARN_WORD_COUNT(h) \ 139 ((NJ_UINT16)NJ_INT32_READ((h) + POS_LEARN_WORD)) 140 #define SET_LEARN_WORD_COUNT(h, n) \ 141 NJ_INT32_WRITE((h)+POS_LEARN_WORD, (NJ_UINT32)(n)) 142 #define GET_LEARN_NEXT_WORD_POS(h) \ 143 ((NJ_UINT16)NJ_INT32_READ((h) + POS_NEXT_QUE)) 144 #define SET_LEARN_NEXT_WORD_POS(h, id) \ 145 NJ_INT32_WRITE((h)+POS_NEXT_QUE, (NJ_UINT32)(id)) 146 #define QUE_SIZE(h) ((NJ_UINT16)NJ_INT32_READ((h) + POS_QUE_SIZE)) 147 148 #define COPY_UINT16(dst,src) (*(NJ_UINT16 *)(dst) = *(NJ_UINT16 *)(src)) 149 150 static NJ_UINT8 *get_search_index_address(NJ_DIC_HANDLE handle, NJ_UINT8 search_pattern) { 151 152 153 154 return LEARN_INDEX_TOP_ADDR(handle); 155 } 156 157 NJ_INT16 njd_l_search_word(NJ_CLASS *iwnn, NJ_SEARCH_CONDITION *con, 158 NJ_SEARCH_LOCATION_SET *loctset, 159 NJ_UINT8 comp_flg) { 160 161 NJ_UINT16 word_count; 162 NJ_UINT32 type; 163 NJ_DIC_INFO *pdicinfo; 164 NJ_UINT16 hIdx; 165 NJ_INT16 ret; 166 167 168 word_count = GET_LEARN_WORD_COUNT(loctset->loct.handle); 169 if (word_count == 0) { 170 171 loctset->loct.status = NJ_ST_SEARCH_END_EXT; 172 return 0; 173 } 174 175 type = NJ_GET_DIC_TYPE_EX(loctset->loct.type, loctset->loct.handle); 176 177 if (type == NJ_DIC_TYPE_CUSTOM_INCOMPRESS) { 178 if ((con->operation == NJ_CUR_OP_COMP) || 179 (con->operation == NJ_CUR_OP_FORE)){ 180 181 if (con->ylen > NJ_GET_MAX_YLEN(loctset->loct.handle)) { 182 loctset->loct.status = NJ_ST_SEARCH_END_EXT; 183 return 0; 184 } 185 } 186 } 187 188 189 switch (con->operation) { 190 case NJ_CUR_OP_COMP: 191 if (con->mode != NJ_CUR_MODE_FREQ) { 192 193 loctset->loct.status = NJ_ST_SEARCH_END_EXT; 194 break; 195 } 196 197 198 return get_cand_by_sequential(iwnn, con, loctset, con->operation, comp_flg); 199 200 case NJ_CUR_OP_FORE: 201 202 if (con->mode == NJ_CUR_MODE_YOMI) { 203 204 return get_cand_by_sequential(iwnn, con, loctset, con->operation, 0); 205 } else { 206 207 208 pdicinfo = con->ds->dic; 209 for (hIdx = 0; (hIdx < NJ_MAX_DIC) && (pdicinfo->handle != loctset->loct.handle); hIdx++) { 210 pdicinfo++; 211 } 212 213 if (hIdx == NJ_MAX_DIC) { 214 215 loctset->loct.status = NJ_ST_SEARCH_END; 216 return 0; 217 } 218 219 220 221 if ((con->ds->dic[hIdx].srhCache == NULL) || (con->ylen == 0) || 222 !(con->ds->mode & 0x0001)) { 223 return get_cand_by_evaluate(iwnn, con, loctset, con->operation); 224 } else { 225 ret = get_cand_by_evaluate2(iwnn, con, loctset, con->operation, hIdx); 226 if (ret == NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_CACHE_NOT_ENOUGH)) { 227 228 NJ_SET_CACHEOVER_TO_SCACHE(con->ds->dic[hIdx].srhCache); 229 ret = get_cand_by_evaluate2(iwnn, con, loctset, con->operation, hIdx); 230 } 231 return ret; 232 } 233 } 234 235 case NJ_CUR_OP_LINK: 236 237 if (NJ_GET_DIC_TYPE_EX(loctset->loct.type, loctset->loct.handle) == NJ_DIC_TYPE_USER) { 238 239 loctset->loct.status = NJ_ST_SEARCH_END_EXT; 240 break; 241 } 242 if (con->mode != NJ_CUR_MODE_FREQ) { 243 244 loctset->loct.status = NJ_ST_SEARCH_END_EXT; 245 break; 246 } 247 248 if (comp_flg == 0) { 249 250 return get_cand_by_sequential(iwnn, con, loctset, con->operation, 0); 251 } else { 252 253 return get_cand_by_evaluate(iwnn, con, loctset, con->operation); 254 } 255 256 default: 257 loctset->loct.status = NJ_ST_SEARCH_END_EXT; 258 } 259 260 return 0; 261 } 262 263 static NJ_WQUE *get_que_type_and_next(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, 264 NJ_UINT16 que_id) { 265 NJ_UINT8 *ptr; 266 NJ_WQUE *que = &(iwnn->que_tmp); 267 268 269 if (que_id >= GET_LEARN_MAX_WORD_COUNT(handle)) { 270 return NULL; 271 } 272 273 ptr = POS_TO_ADDRESS(handle, que_id); 274 275 que->type = GET_TYPE_FROM_DATA(ptr); 276 que->next_flag = GET_FFLG_FROM_DATA(ptr); 277 278 switch (que->type) { 279 case QUE_TYPE_EMPTY: 280 case QUE_TYPE_JIRI: 281 case QUE_TYPE_FZK: 282 return que; 283 default: 284 break; 285 } 286 #ifdef LEARN_DEBUG 287 printf("FATAL : Illegal que was gotten (que_id=%d)\n", que_id); 288 #endif 289 return NULL; 290 } 291 292 static NJ_WQUE *get_que_yomiLen_and_hyoukiLen(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, 293 NJ_UINT16 que_id) { 294 NJ_UINT8 *ptr; 295 NJ_WQUE *que = &(iwnn->que_tmp); 296 297 298 if (que_id >= GET_LEARN_MAX_WORD_COUNT(handle)) { 299 return NULL; 300 } 301 302 ptr = POS_TO_ADDRESS(handle, que_id); 303 304 que->type = GET_TYPE_FROM_DATA(ptr); 305 que->yomi_byte = GET_YSIZE_FROM_DATA(ptr); 306 que->yomi_len = que->yomi_byte / sizeof(NJ_CHAR); 307 que->hyouki_byte = GET_KSIZE_FROM_DATA(ptr); 308 que->hyouki_len = que->hyouki_byte / sizeof(NJ_CHAR); 309 310 switch (que->type) { 311 case QUE_TYPE_JIRI: 312 case QUE_TYPE_FZK: 313 return que; 314 default: 315 break; 316 } 317 #ifdef LEARN_DEBUG 318 printf("FATAL : Illegal que was gotten (que_id=%d)\n", que_id); 319 #endif 320 return NULL; 321 } 322 323 static NJ_WQUE *get_que_allHinsi(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, 324 NJ_UINT16 que_id) { 325 NJ_UINT8 *ptr; 326 NJ_WQUE *que = &(iwnn->que_tmp); 327 328 329 if (que_id >= GET_LEARN_MAX_WORD_COUNT(handle)) { 330 return NULL; 331 } 332 333 ptr = POS_TO_ADDRESS(handle, que_id); 334 335 que->type = GET_TYPE_FROM_DATA(ptr); 336 que->mae_hinsi = GET_FPOS_FROM_DATA(ptr); 337 que->ato_hinsi = GET_BPOS_FROM_DATA(ptr); 338 339 switch (que->type) { 340 case QUE_TYPE_JIRI: 341 case QUE_TYPE_FZK: 342 return que; 343 default: 344 break; 345 } 346 #ifdef LEARN_DEBUG 347 printf("FATAL : Illegal que was gotten (que_id=%d)\n", que_id); 348 #endif 349 return NULL; 350 } 351 352 static NJ_WQUE *get_que(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT16 que_id) { 353 NJ_UINT8 *ptr; 354 NJ_WQUE *que = &(iwnn->que_tmp); 355 356 357 if (que_id >= GET_LEARN_MAX_WORD_COUNT(handle)) { 358 return NULL; 359 } 360 361 ptr = POS_TO_ADDRESS(handle, que_id); 362 363 que->entry = que_id; 364 que->type = GET_TYPE_FROM_DATA(ptr); 365 que->mae_hinsi = GET_FPOS_FROM_DATA(ptr); 366 que->ato_hinsi = GET_BPOS_FROM_DATA(ptr); 367 que->yomi_byte = GET_YSIZE_FROM_DATA(ptr); 368 que->yomi_len = que->yomi_byte / sizeof(NJ_CHAR); 369 que->hyouki_byte= GET_KSIZE_FROM_DATA(ptr); 370 que->hyouki_len = que->hyouki_byte / sizeof(NJ_CHAR); 371 que->next_flag = GET_FFLG_FROM_DATA(ptr); 372 373 switch (que->type) { 374 case QUE_TYPE_JIRI: 375 case QUE_TYPE_FZK: 376 return que; 377 default: 378 break; 379 } 380 #ifdef LEARN_DEBUG 381 printf("FATAL : Illegal que was gotten (que_id=%d)\n", que_id); 382 #endif 383 return NULL; 384 } 385 386 static NJ_INT16 is_continued(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT16 que_id) { 387 NJ_WQUE *que; 388 NJ_UINT16 i; 389 NJ_UINT16 max, end; 390 391 392 max = GET_LEARN_MAX_WORD_COUNT(handle); 393 end = GET_LEARN_NEXT_WORD_POS(handle); 394 395 for (i = 0; i < max; i++) { 396 que_id++; 397 if (que_id >= GET_LEARN_MAX_WORD_COUNT(handle)) { 398 399 que_id = 0; 400 } 401 402 403 if (que_id == end) { 404 405 return 0; 406 } 407 408 que = get_que_type_and_next(iwnn, handle, que_id); 409 #ifdef IWNN_ERR_CHECK 410 if (iwnn->err_check_flg == 1) { 411 que = NULL; 412 } 413 #endif 414 if (que == NULL) { 415 return NJ_SET_ERR_VAL(NJ_FUNC_IS_CONTINUED, NJ_ERR_DIC_BROKEN); 416 } 417 if (que->type != QUE_TYPE_EMPTY) { 418 419 if (que->next_flag != 0) { 420 421 return 1; 422 } else { 423 424 return 0; 425 } 426 } 427 } 428 429 430 return 0; 431 } 432 433 static NJ_INT16 continue_cnt(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT16 que_id) { 434 NJ_WQUE *que; 435 NJ_UINT16 i; 436 NJ_UINT16 max, end; 437 NJ_INT16 cnt = 0; 438 439 440 max = GET_LEARN_MAX_WORD_COUNT(handle); 441 end = GET_LEARN_NEXT_WORD_POS(handle); 442 443 for (i = 0; i < max; i++) { 444 que_id++; 445 if (que_id >= max) { 446 447 que_id = 0; 448 } 449 450 451 if (que_id == end) { 452 453 return cnt; 454 } 455 456 que = get_que_type_and_next(iwnn, handle, que_id); 457 if (que == NULL) { 458 return NJ_SET_ERR_VAL(NJ_FUNC_CONTINUE_CNT, NJ_ERR_DIC_BROKEN); 459 } 460 if (que->type != QUE_TYPE_EMPTY) { 461 462 if (que->next_flag != 0) { 463 464 cnt++; 465 466 467 if (cnt >= (NJD_MAX_CONNECT_CNT - 1)) { 468 return cnt; 469 } 470 } else { 471 472 return cnt; 473 } 474 } 475 } 476 477 478 return 0; 479 } 480 481 static NJ_UINT16 search_next_que(NJ_DIC_HANDLE handle, NJ_UINT16 que_id) { 482 NJ_UINT16 max; 483 NJ_UINT16 i; 484 485 486 max = GET_LEARN_MAX_WORD_COUNT(handle); 487 488 for (i = 0; i < max; i++) { 489 que_id++; 490 if (que_id >= max) { 491 492 que_id = 0; 493 } 494 495 if (GET_TYPE_FROM_DATA(POS_TO_ADDRESS(handle, que_id)) != QUE_TYPE_EMPTY) { 496 497 return que_id; 498 } 499 } 500 501 502 return 0; 503 } 504 505 static NJ_INT16 que_strcmp_complete_with_hyouki(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, 506 NJ_UINT16 que_id, NJ_CHAR *yomi, NJ_UINT16 yomi_len, NJ_CHAR *hyouki, 507 NJ_UINT8 multi_flg) { 508 NJ_CHAR *str; 509 NJ_INT16 ret; 510 NJ_UINT8 slen; 511 NJ_UINT16 hyouki_len; 512 NJ_UINT16 que_yomilen, que_hyoukilen; 513 NJ_INT16 que_count = 1; 514 NJ_INT16 cnt = 0; 515 516 517 518 hyouki_len = nj_strlen(hyouki); 519 520 if (multi_flg == 0) { 521 522 cnt = 1; 523 } else { 524 525 526 cnt = GET_LEARN_WORD_COUNT(handle); 527 } 528 529 while (cnt--) { 530 str = get_string(iwnn, handle, que_id, &slen); 531 if (str == NULL) { 532 return NJ_SET_ERR_VAL(NJ_FUNC_QUE_STRCMP_COMPLETE_WITH_HYOUKI, 533 NJ_ERR_DIC_BROKEN); 534 } 535 que_yomilen = slen; 536 537 ret = nj_strncmp(yomi, str, que_yomilen); 538 if (ret != 0) { 539 540 return 0; 541 } 542 543 str = get_hyouki(iwnn, handle, que_id, &slen); 544 if (str == NULL) { 545 return NJ_SET_ERR_VAL(NJ_FUNC_QUE_STRCMP_COMPLETE_WITH_HYOUKI, 546 NJ_ERR_DIC_BROKEN); 547 } 548 que_hyoukilen = slen; 549 550 ret = nj_strncmp(hyouki, str, que_hyoukilen); 551 if (ret != 0) { 552 553 return 0; 554 } 555 556 if ((yomi_len == que_yomilen) && 557 (hyouki_len == que_hyoukilen)) { 558 559 return que_count; 560 } 561 562 if ((que_yomilen > yomi_len) || 563 (que_hyoukilen > hyouki_len)) { 564 565 return 0; 566 } 567 568 ret = is_continued(iwnn, handle, que_id); 569 if (ret <= 0) { 570 571 return ret; 572 } 573 574 575 if (que_count >= (NJD_MAX_CONNECT_CNT - 1)) { 576 577 return 0; 578 } 579 580 yomi_len -= que_yomilen; 581 yomi += que_yomilen; 582 583 hyouki_len -= que_hyoukilen; 584 hyouki += que_hyoukilen; 585 586 587 que_id = search_next_que(handle, que_id); 588 que_count++; 589 } 590 return 0; 591 } 592 593 static NJ_INT16 que_strcmp_include(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, 594 NJ_UINT16 que_id, NJ_CHAR *yomi) { 595 NJ_CHAR *str; 596 NJ_UINT16 que_len; 597 NJ_UINT16 yomi_len; 598 NJ_INT16 ret; 599 NJ_INT16 que_count = 1; 600 NJ_UINT16 i = 0; 601 NJ_UINT8 slen; 602 603 604 #ifdef LEARN_DEBUG 605 printf("que_strcmp_include(que_id=%d, yomi=[%s])\n", que_id, yomi); 606 #endif 607 yomi_len = nj_strlen(yomi); 608 if (yomi_len == 0) { 609 return que_count; 610 } 611 612 i = GET_LEARN_WORD_COUNT(handle); 613 614 while (--i) { 615 616 617 ret = is_continued(iwnn, handle, que_id); 618 if (ret < 0) { 619 620 return ret; 621 } else if (ret == 0) { 622 623 return que_count; 624 } 625 626 627 que_id = search_next_que(handle, que_id); 628 629 str = get_string(iwnn, handle, que_id, &slen); 630 #ifdef IWNN_ERR_CHECK 631 if (iwnn->err_check_flg == 2) { 632 str = NULL; 633 } 634 #endif 635 if (str == NULL) { 636 return NJ_SET_ERR_VAL(NJ_FUNC_QUE_STRCMP_INCLUDE, NJ_ERR_DIC_BROKEN); 637 } 638 que_len = slen; 639 640 641 if (que_len > yomi_len) { 642 #ifdef LEARN_DEBUG 643 printf(" >> mismatch [%s] (que_len > yomi_len)\n", str); 644 #endif 645 return que_count; 646 } 647 648 649 ret = nj_strncmp(yomi, str, que_len); 650 if (ret != 0) { 651 #ifdef LEARN_DEBUG 652 printf(" >> mismatch [%s]\n", str); 653 #endif 654 655 return que_count; 656 } 657 658 659 if (que_len == yomi_len) { 660 #ifdef LEARN_DEBUG 661 printf(" >> match! [%s](%d)\n", str, que_count); 662 #endif 663 return (que_count + 1); 664 } 665 666 que_count++; 667 if (que_count >= NJD_MAX_CONNECT_CNT) { 668 669 return que_count; 670 } 671 672 673 yomi_len -= que_len; 674 yomi += que_len; 675 } 676 677 return que_count; 678 } 679 680 static NJ_CHAR *get_string(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, 681 NJ_UINT16 que_id, NJ_UINT8 *slen) { 682 NJ_UINT8 *src, *dst; 683 NJ_UINT8 copy_size, size; 684 NJ_UINT8 i; 685 NJ_UINT8 *top_addr; 686 NJ_UINT8 *bottom_addr; 687 NJ_UINT16 que_size; 688 689 690 src = POS_TO_ADDRESS(handle, que_id); 691 switch (GET_TYPE_FROM_DATA(src)) { 692 case QUE_TYPE_JIRI: 693 case QUE_TYPE_FZK: 694 size = GET_YSIZE_FROM_DATA(src); 695 *slen = (NJ_UINT8)(size / sizeof(NJ_CHAR)); 696 break; 697 698 default: 699 #ifdef LEARN_DEBUG 700 printf("get_string(handle=%p, que_id=%d) : broken que\n", handle, que_id); 701 #endif 702 return NULL; 703 } 704 705 if (NJ_GET_DIC_TYPE(handle) == NJ_DIC_TYPE_USER) { 706 if (*slen > NJ_MAX_USER_LEN) { 707 return NULL; 708 } 709 } else { 710 if (*slen > NJ_MAX_LEN) { 711 return NULL; 712 } 713 } 714 715 716 src += LEARN_QUE_STRING_OFFSET; 717 718 que_size = QUE_SIZE(handle); 719 720 721 copy_size = (NJ_UINT8)que_size - LEARN_QUE_STRING_OFFSET; 722 dst = (NJ_UINT8*)&(iwnn->learn_string_tmp[0]); 723 if (copy_size > size) { 724 725 copy_size = size; 726 } 727 for (i = 0; i < copy_size; i++) { 728 *dst++ = *src++; 729 } 730 731 732 top_addr = LEARN_DATA_TOP_ADDR(handle); 733 bottom_addr = top_addr; 734 bottom_addr += que_size * GET_LEARN_MAX_WORD_COUNT(handle) - 1; 735 736 while (size -= copy_size) { 737 738 if (src >= bottom_addr) { 739 src = top_addr; 740 } 741 742 743 if (*src != QUE_TYPE_NEXT) { 744 #ifdef LEARN_DEBUG 745 printf("FATAL: src que was broken(not QUE_TYPE_NEXT) [src=%x]\n", src); 746 #endif 747 return NULL; 748 } 749 750 src++; 751 if (size < que_size) { 752 753 copy_size = size; 754 } else { 755 copy_size = (NJ_UINT8)(que_size - 1); 756 } 757 for (i = 0; i < copy_size; i++) { 758 *dst++ = *src++; 759 } 760 } 761 iwnn->learn_string_tmp[*slen] = NJ_CHAR_NUL; 762 763 return &(iwnn->learn_string_tmp[0]); 764 } 765 766 static NJ_CHAR *get_hyouki(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, 767 NJ_UINT16 que_id, NJ_UINT8 *slen) { 768 NJ_UINT8 *src, *dst; 769 NJ_WQUE *que; 770 NJ_UINT8 copy_size, size; 771 NJ_UINT8 i; 772 NJ_UINT8 *top_addr; 773 NJ_UINT8 *bottom_addr; 774 NJ_CHAR *hira; 775 NJ_UINT16 que_size; 776 NJ_UINT32 dictype; 777 778 779 que = get_que_yomiLen_and_hyoukiLen(iwnn, handle, que_id); 780 if (que == NULL) { 781 return NULL; 782 } 783 784 dictype = NJ_GET_DIC_TYPE(handle); 785 if (dictype == NJ_DIC_TYPE_USER) { 786 if (que->yomi_len > NJ_MAX_USER_LEN) { 787 return NULL; 788 } 789 if (que->hyouki_len > NJ_MAX_USER_KOUHO_LEN) { 790 return NULL; 791 } 792 } else { 793 if (que->yomi_len > NJ_MAX_LEN) { 794 return NULL; 795 } 796 if (que->hyouki_len > NJ_MAX_RESULT_LEN) { 797 return NULL; 798 } 799 } 800 801 src = POS_TO_ADDRESS(handle, que_id); 802 803 if (que->hyouki_len == 0) { 804 hira = get_string(iwnn, handle, que_id, slen); 805 if (hira == NULL) { 806 return NULL; 807 } 808 809 if (GET_MFLG_FROM_DATA(src) != 0) { 810 *slen = (NJ_UINT8)nje_convert_hira_to_kata(hira, &(iwnn->muhenkan_tmp[0]), *slen); 811 return &(iwnn->muhenkan_tmp[0]); 812 } else { 813 return hira; 814 } 815 } 816 817 src += LEARN_QUE_STRING_OFFSET; 818 819 que_size = QUE_SIZE(handle); 820 821 822 size = que->yomi_byte; 823 copy_size = (NJ_UINT8)que_size - LEARN_QUE_STRING_OFFSET; 824 dst = (NJ_UINT8*)&(iwnn->learn_string_tmp[0]); 825 if (copy_size > size) { 826 827 copy_size = size; 828 } 829 830 831 top_addr = LEARN_DATA_TOP_ADDR(handle); 832 bottom_addr = top_addr; 833 bottom_addr += que_size * GET_LEARN_MAX_WORD_COUNT(handle) - 1; 834 835 src += copy_size; 836 while (size -= copy_size) { 837 838 839 if (src >= bottom_addr) { 840 src = top_addr; 841 } 842 843 844 if (*src != QUE_TYPE_NEXT) { 845 #ifdef LEARN_DEBUG 846 printf("FATAL: src que was broken(not QUE_TYPE_NEXT) [src=%x]\n", src); 847 #endif 848 return NULL; 849 } 850 851 src++; 852 if (size < que_size) { 853 854 copy_size = size; 855 } else { 856 copy_size = (NJ_UINT8)(que_size - 1); 857 } 858 src += copy_size; 859 } 860 861 862 863 if (((src - top_addr) % que_size) == 0) { 864 865 if (src >= bottom_addr) { 866 src = top_addr; 867 } 868 869 if (*src++ != QUE_TYPE_NEXT) { 870 #ifdef LEARN_DEBUG 871 printf("FATAL: src que was broken(QUE_TYPE_NEXT) [src=%x]\n", src - 1); 872 #endif 873 return NULL; 874 } 875 } 876 877 size = que->hyouki_byte; 878 879 880 copy_size = (NJ_UINT8)(que_size); 881 copy_size -= (NJ_UINT8)((src - top_addr) % que_size); 882 if (copy_size > size) { 883 884 copy_size = size; 885 } 886 for (i = 0; i < copy_size; i++) { 887 *dst++ = *src++; 888 } 889 890 while (size -= copy_size) { 891 892 893 if (src >= bottom_addr) { 894 src = top_addr; 895 } 896 897 898 if (*src != QUE_TYPE_NEXT) { 899 #ifdef LEARN_DEBUG 900 printf("FATAL: src que was broken(not QUE_TYPE_NEXT) [src=%x]\n", src); 901 #endif 902 return NULL; 903 } 904 905 src++; 906 if (size < que_size) { 907 908 copy_size = size; 909 } else { 910 copy_size = (NJ_UINT8)(que_size - 1); 911 } 912 913 for (i = 0; i < copy_size; i++) { 914 *dst++ = *src++; 915 } 916 } 917 918 *slen = que->hyouki_len; 919 iwnn->learn_string_tmp[*slen] = NJ_CHAR_NUL; 920 921 return &(iwnn->learn_string_tmp[0]); 922 } 923 924 static NJ_INT16 get_cand_by_sequential(NJ_CLASS *iwnn, NJ_SEARCH_CONDITION *cond, 925 NJ_SEARCH_LOCATION_SET *loctset, NJ_UINT8 search_pattern, 926 NJ_UINT8 comp_flg) { 927 NJ_UINT16 current, from, to; 928 NJ_UINT16 que_id; 929 NJ_UINT8 *ptr, *p; 930 NJ_INT16 ret, num_count; 931 NJ_CHAR *yomi; 932 NJ_WQUE *que; 933 NJ_UINT8 forward_flag = 0; 934 935 936 937 if (GET_LOCATION_STATUS(loctset->loct.status) == NJ_ST_SEARCH_NO_INIT) { 938 939 ret = search_range_by_yomi(iwnn, loctset->loct.handle, search_pattern, 940 cond->yomi, cond->ylen, &from, &to, &forward_flag); 941 if (ret < 0) { 942 return ret; 943 } 944 if (ret == 0) { 945 if (forward_flag) { 946 loctset->loct.status = NJ_ST_SEARCH_END; 947 } else { 948 loctset->loct.status = NJ_ST_SEARCH_END_EXT; 949 } 950 return 0; 951 } 952 loctset->loct.top = from; 953 loctset->loct.bottom = to; 954 current = from; 955 } else if (GET_LOCATION_STATUS(loctset->loct.status) == NJ_ST_SEARCH_READY) { 956 957 current = (NJ_UINT16)(loctset->loct.current + 1); 958 } else { 959 loctset->loct.status = NJ_ST_SEARCH_END; 960 return 0; 961 } 962 963 964 ptr = get_search_index_address(loctset->loct.handle, cond->operation); 965 p = ptr + (current * NJ_INDEX_SIZE); 966 967 while (current <= loctset->loct.bottom) { 968 que_id = GET_UINT16(p); 969 if (search_pattern == NJ_CUR_OP_COMP) { 970 971 ret = str_que_cmp(iwnn, loctset->loct.handle, cond->yomi, cond->ylen, que_id, 1); 972 973 974 if (ret == 2) { 975 ret = 0; 976 } 977 } else if (search_pattern == NJ_CUR_OP_FORE) { 978 979 ret = str_que_cmp(iwnn, loctset->loct.handle, cond->yomi, cond->ylen, que_id, 2); 980 981 982 if (ret == 2) { 983 ret = 0; 984 } 985 } else { 986 987 988 989 ret = que_strcmp_complete_with_hyouki(iwnn, loctset->loct.handle, que_id, 990 cond->yomi, cond->ylen, cond->kanji, 0); 991 } 992 993 if (ret < 0) { 994 return ret; 995 } 996 if (ret > 0) { 997 if (search_pattern == NJ_CUR_OP_LINK) { 998 999 1000 num_count = continue_cnt(iwnn, loctset->loct.handle, que_id); 1001 if (num_count < 0) { 1002 1003 return num_count; 1004 } 1005 1006 1007 if (num_count >= ret) { 1008 1009 loctset->loct.current_info = (NJ_UINT8)(((num_count + 1) << 4) | ret); 1010 loctset->loct.current = current; 1011 loctset->loct.status = NJ_ST_SEARCH_READY; 1012 loctset->cache_freq = get_hindo(iwnn, loctset, search_pattern); 1013 return 1; 1014 } 1015 } else { 1016 1017 1018 1019 1020 1021 1022 que = get_que_allHinsi(iwnn, loctset->loct.handle, que_id); 1023 if (njd_connect_test(cond, que->mae_hinsi, que->ato_hinsi)) { 1024 1025 1026 switch (NJ_GET_DIC_TYPE_EX(loctset->loct.type, loctset->loct.handle)) { 1027 case NJ_DIC_TYPE_CUSTOM_INCOMPRESS: 1028 if ((search_pattern == NJ_CUR_OP_COMP) && (comp_flg == 1)) { 1029 yomi = cond->yomi + cond->ylen; 1030 ret = que_strcmp_include(iwnn, loctset->loct.handle, que_id, yomi); 1031 if (ret < 0) { 1032 return ret; 1033 } 1034 } 1035 break; 1036 default: 1037 break; 1038 } 1039 loctset->loct.current = current; 1040 loctset->loct.status = NJ_ST_SEARCH_READY; 1041 1042 loctset->loct.current_info = (ret & 0x0f) << 4; 1043 loctset->cache_freq = get_hindo(iwnn, loctset, search_pattern); 1044 return 1; 1045 } 1046 } 1047 } 1048 p += NJ_INDEX_SIZE; 1049 current++; 1050 } 1051 1052 1053 loctset->loct.status = NJ_ST_SEARCH_END; 1054 return 0; 1055 } 1056 1057 static NJ_INT16 get_cand_by_evaluate(NJ_CLASS *iwnn, NJ_SEARCH_CONDITION *cond, 1058 NJ_SEARCH_LOCATION_SET *loctset, NJ_UINT8 search_pattern) { 1059 NJ_UINT16 from, to, i; 1060 NJ_UINT16 que_id, oldest; 1061 NJ_UINT32 max_value, eval, current; 1062 NJ_UINT8 *ptr, *p; 1063 NJ_WQUE *que; 1064 NJ_INT16 ret, num_count; 1065 NJ_INT32 found = 0; 1066 NJ_UINT8 forward_flag = 0; 1067 NJ_INT32 is_first_search, is_better_freq; 1068 1069 1070 1071 ptr = get_search_index_address(loctset->loct.handle, cond->operation); 1072 1073 1074 oldest = GET_LEARN_NEXT_WORD_POS(loctset->loct.handle); 1075 1076 1077 current = 0; 1078 if (GET_LOCATION_STATUS(loctset->loct.status) == NJ_ST_SEARCH_NO_INIT) { 1079 if (search_pattern == NJ_CUR_OP_LINK) { 1080 1081 1082 1083 ret = search_range_by_yomi_multi(iwnn, loctset->loct.handle, 1084 cond->yomi, cond->ylen, &from, &to); 1085 } else { 1086 1087 1088 ret = search_range_by_yomi(iwnn, loctset->loct.handle, search_pattern, 1089 cond->yomi, cond->ylen, &from, &to, &forward_flag); 1090 } 1091 if (ret <= 0) { 1092 loctset->loct.status = NJ_ST_SEARCH_END; 1093 return ret; 1094 } 1095 loctset->loct.top = from; 1096 loctset->loct.bottom = to; 1097 is_first_search = 1; 1098 } else if (GET_LOCATION_STATUS(loctset->loct.status) == NJ_ST_SEARCH_READY) { 1099 current = GET_UINT16(ptr + (loctset->loct.current * NJ_INDEX_SIZE)); 1100 if (current < oldest) { 1101 current += GET_LEARN_MAX_WORD_COUNT(loctset->loct.handle); 1102 } 1103 is_first_search = 0; 1104 } else { 1105 loctset->loct.status = NJ_ST_SEARCH_END; 1106 return 0; 1107 } 1108 1109 1110 max_value = oldest; 1111 1112 p = ptr + (loctset->loct.top * NJ_INDEX_SIZE); 1113 eval = current; 1114 for (i = (NJ_UINT16)loctset->loct.top; i <= (NJ_UINT16)loctset->loct.bottom; i++) { 1115 que_id = GET_UINT16(p); 1116 if (que_id < oldest) { 1117 eval = que_id + GET_LEARN_MAX_WORD_COUNT(loctset->loct.handle); 1118 } else { 1119 eval = que_id; 1120 } 1121 #ifdef LEARN_DEBUG 1122 printf("que(%d) : eval = %d\n", que_id, eval); 1123 #endif 1124 is_better_freq = ((eval >= max_value) && ((is_first_search) || (eval < current))) ? 1 : 0; 1125 1126 if (is_better_freq) { 1127 1128 if (search_pattern == NJ_CUR_OP_LINK) { 1129 1130 ret = que_strcmp_complete_with_hyouki(iwnn, loctset->loct.handle, que_id, 1131 cond->yomi, cond->ylen, cond->kanji, 1); 1132 } else { 1133 1134 ret = str_que_cmp(iwnn, loctset->loct.handle, cond->yomi, cond->ylen, que_id, 2); 1135 1136 if (ret == 2) { 1137 ret = 0; 1138 } 1139 } 1140 if (ret < 0) { 1141 return ret; 1142 } 1143 if (ret >= 1) { 1144 if (search_pattern == NJ_CUR_OP_LINK) { 1145 1146 1147 num_count = continue_cnt(iwnn, loctset->loct.handle, que_id); 1148 if (num_count < 0) { 1149 1150 return num_count; 1151 } 1152 1153 1154 if (num_count >= ret) { 1155 1156 loctset->loct.current_info = (NJ_UINT8)(((num_count + 1) << 4) | ret); 1157 loctset->loct.current = i; 1158 max_value = eval; 1159 found = 1; 1160 } 1161 } else { 1162 1163 1164 1165 1166 1167 que = get_que_allHinsi(iwnn, loctset->loct.handle, que_id); 1168 if (njd_connect_test(cond, que->mae_hinsi, que->ato_hinsi)) { 1169 1170 loctset->loct.current_info = (NJ_UINT8)0x10; 1171 loctset->loct.current = i; 1172 max_value = eval; 1173 found = 1; 1174 #ifdef LEARN_DEBUG 1175 printf("---keep."); 1176 #endif 1177 } 1178 } 1179 } 1180 } 1181 p += NJ_INDEX_SIZE; 1182 } 1183 1184 1185 if (found == 0) { 1186 loctset->loct.status = NJ_ST_SEARCH_END; 1187 return 0; 1188 } else { 1189 loctset->loct.status = NJ_ST_SEARCH_READY; 1190 loctset->cache_freq = get_hindo(iwnn, loctset, search_pattern); 1191 return 1; 1192 } 1193 1194 } 1195 1196 static NJ_INT16 search_range_by_yomi(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT8 op, 1197 NJ_CHAR *yomi, NJ_UINT16 len, NJ_UINT16 *from, NJ_UINT16 *to, 1198 NJ_UINT8 *forward_flag) { 1199 NJ_UINT16 right, mid = 0, left, max; 1200 NJ_UINT16 que_id; 1201 NJ_UINT8 *ptr, *p; 1202 NJ_CHAR *str; 1203 NJ_INT16 ret = 0; 1204 NJ_INT32 found = 0; 1205 NJ_UINT8 slen; 1206 NJ_INT32 cmp; 1207 1208 1209 1210 ptr = get_search_index_address(handle, op); 1211 1212 max = GET_LEARN_WORD_COUNT(handle); 1213 1214 right = max - 1; 1215 left = 0; 1216 1217 #ifdef LEARN_DEBUG 1218 printf("src:[%s]\n", yomi); 1219 #endif 1220 1221 *forward_flag = 0; 1222 1223 1224 switch (op) { 1225 case NJ_CUR_OP_COMP: 1226 case NJ_CUR_OP_LINK: 1227 case NJ_CUR_OP_FORE: 1228 1229 1230 1231 break; 1232 default: 1233 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_PARAM_OPERATION); 1234 } 1235 1236 while (left <= right) { 1237 mid = left + ((right - left) / 2); 1238 p = ptr + (mid * NJ_INDEX_SIZE); 1239 que_id = GET_UINT16(p); 1240 str = get_string(iwnn, handle, que_id, &slen); 1241 1242 #ifdef IWNN_ERR_CHECK 1243 if (iwnn->err_check_flg == 3) { 1244 str = NULL; 1245 } 1246 #endif 1247 if (str == NULL) { 1248 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_DIC_BROKEN); 1249 } 1250 1251 ret = nj_strncmp(yomi, str, len); 1252 if (op != NJ_CUR_OP_FORE) { 1253 1254 1255 if (ret == 0) { 1256 if ((*forward_flag == 0) && (len <= (NJ_UINT16)slen)) { 1257 1258 *forward_flag = 1; 1259 } 1260 if (len > (NJ_UINT16)slen) { 1261 ret = 1; 1262 } else if (len < (NJ_UINT16)slen) { 1263 ret = -1; 1264 } 1265 } 1266 } 1267 #ifdef LEARN_DEBUG 1268 printf(" [%d][%d][%d]COMPARE:[%s] = %d\n", left, mid, right, str, ret); 1269 #endif 1270 if (ret == 0) { 1271 1272 found = 1; 1273 break; 1274 } else if (ret < 0) { 1275 1276 right = mid - 1; 1277 if (mid == 0) { 1278 break; 1279 } 1280 } else { 1281 1282 left = mid + 1; 1283 } 1284 } 1285 1286 if (!found) { 1287 return 0; 1288 } 1289 1290 if (mid == 0) { 1291 *from = mid; 1292 } else { 1293 1294 p = ((mid - 1) * NJ_INDEX_SIZE) + ptr; 1295 1296 for (cmp = mid - 1; cmp >= 0; cmp--) { 1297 que_id = GET_UINT16(p); 1298 str = get_string(iwnn, handle, que_id, &slen); 1299 1300 #ifdef IWNN_ERR_CHECK 1301 if (iwnn->err_check_flg == 4) { 1302 str = NULL; 1303 } 1304 #endif 1305 if (str == NULL) { 1306 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_DIC_BROKEN); 1307 } 1308 1309 if (op != NJ_CUR_OP_FORE) { 1310 ret = nj_strncmp(yomi, str, len); 1311 if (ret == 0) { 1312 if (len > (NJ_UINT16)slen) { 1313 ret = 1; 1314 } else if (len < (NJ_UINT16)slen) { 1315 ret = -1; 1316 } 1317 } 1318 if (ret > 0) { 1319 1320 break; 1321 } 1322 } else { 1323 1324 if (nj_strncmp(yomi, str, len) != 0) { 1325 break; 1326 } 1327 } 1328 p -= NJ_INDEX_SIZE; 1329 } 1330 if (cmp < 0) { 1331 *from = 0; 1332 } else { 1333 *from = (NJ_UINT16)cmp + 1; 1334 } 1335 } 1336 1337 #ifdef LEARN_DEBUG 1338 printf(" >> from:(%d)\n", *from); 1339 #endif 1340 1341 #ifdef IWNN_ERR_CHECK 1342 if (iwnn->err_check_flg == 5) { 1343 mid = max - 2; 1344 } 1345 #endif 1346 if ((mid + 1) >= max) { 1347 *to = mid; 1348 } else { 1349 1350 p = ((mid + 1) * NJ_INDEX_SIZE) + ptr; 1351 1352 for (right = mid + 1; right < max; right++) { 1353 que_id = GET_UINT16(p); 1354 str = get_string(iwnn, handle, que_id, &slen); 1355 1356 #ifdef IWNN_ERR_CHECK 1357 if (iwnn->err_check_flg == 5) { 1358 str = NULL; 1359 } 1360 #endif 1361 if (str == NULL) { 1362 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_DIC_BROKEN); 1363 } 1364 1365 if (op != NJ_CUR_OP_FORE) { 1366 ret = nj_strncmp(yomi, str, len); 1367 if (ret == 0) { 1368 if (len > (NJ_UINT16)slen) { 1369 ret = 1; 1370 } else if (len < (NJ_UINT16)slen) { 1371 ret = -1; 1372 } 1373 } 1374 if (ret < 0) { 1375 1376 break; 1377 } 1378 } else { 1379 1380 if (nj_strncmp(yomi, str, len) != 0) { 1381 break; 1382 } 1383 } 1384 p += NJ_INDEX_SIZE; 1385 } 1386 *to = right - 1; 1387 } 1388 1389 #ifdef LEARN_DEBUG 1390 printf(" >> to:(%d)\n", *to); 1391 #endif 1392 return 1; 1393 } 1394 1395 static NJ_INT16 search_range_by_yomi_multi(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, 1396 NJ_CHAR *yomi, NJ_UINT16 len, NJ_UINT16 *from, NJ_UINT16 *to) { 1397 NJ_UINT16 right, mid = 0, left, max = 0; 1398 NJ_UINT16 que_id; 1399 NJ_UINT8 *ptr, *p; 1400 NJ_INT16 ret = 0; 1401 NJ_UINT16 comp_len; 1402 NJ_UINT16 i, char_len; 1403 NJ_INT32 found = 0; 1404 NJ_INT32 cmp; 1405 NJ_CHAR comp_yomi[NJ_MAX_LEN + NJ_TERM_LEN]; 1406 NJ_CHAR *pYomi; 1407 1408 1409 1410 1411 ptr = LEARN_INDEX_TOP_ADDR(handle); 1412 1413 1414 max = GET_LEARN_WORD_COUNT(handle); 1415 1416 #ifdef LEARN_DEBUG 1417 printf("src:[%s]\n", yomi); 1418 #endif 1419 1420 comp_len = 0; 1421 pYomi = &yomi[0]; 1422 while (comp_len < len) { 1423 1424 1425 char_len = NJ_CHAR_LEN(pYomi); 1426 for (i = 0; i < char_len; i++) { 1427 *(comp_yomi + comp_len) = *pYomi; 1428 comp_len++; 1429 pYomi++; 1430 } 1431 *(comp_yomi + comp_len) = NJ_CHAR_NUL; 1432 1433 right = max - 1; 1434 left = 0; 1435 while (left <= right) { 1436 mid = left + ((right - left) / 2); 1437 p = ptr + (mid * NJ_INDEX_SIZE); 1438 que_id = GET_UINT16(p); 1439 1440 1441 ret = str_que_cmp(iwnn, handle, comp_yomi, comp_len, que_id, 1); 1442 if (ret < 0) { 1443 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI_MULTI, NJ_ERR_DIC_BROKEN); 1444 } 1445 1446 #ifdef LEARN_DEBUG 1447 printf(" [%d][%d][%d]COMPARE:[%s] = %d\n", left, mid, right, str, ret); 1448 #endif 1449 if (ret == 1) { 1450 1451 found = 1; 1452 break; 1453 } else if (ret == 0) { 1454 1455 right = mid - 1; 1456 if (mid == 0) { 1457 break; 1458 } 1459 } else { 1460 1461 left = mid + 1; 1462 } 1463 } 1464 1465 if (found) { 1466 break; 1467 } 1468 } 1469 1470 if (!found) { 1471 1472 return 0; 1473 } 1474 1475 1476 if (mid == 0) { 1477 *from = mid; 1478 } else { 1479 1480 p = ((mid - 1) * NJ_INDEX_SIZE) + ptr; 1481 1482 for (cmp = mid - 1; cmp >= 0; cmp--) { 1483 que_id = GET_UINT16(p); 1484 ret = str_que_cmp(iwnn, handle, comp_yomi, comp_len, que_id, 1); 1485 if (ret < 0) { 1486 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI_MULTI, NJ_ERR_DIC_BROKEN); 1487 } 1488 if (ret == 2) { 1489 break; 1490 } 1491 p -= NJ_INDEX_SIZE; 1492 } 1493 if (cmp < 0) { 1494 *from = 0; 1495 } else { 1496 *from = (NJ_UINT16)cmp + 1; 1497 } 1498 } 1499 1500 #ifdef LEARN_DEBUG 1501 printf(" >> from:(%d)\n", *from); 1502 #endif 1503 1504 1505 if ((mid + 1) >= max) { 1506 *to = mid; 1507 } else { 1508 1509 p = ((mid + 1) * NJ_INDEX_SIZE) + ptr; 1510 1511 for (right = mid + 1; right < max; right++) { 1512 que_id = GET_UINT16(p); 1513 ret = str_que_cmp(iwnn, handle, yomi, len, que_id, 1); 1514 if (ret < 0) { 1515 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI_MULTI, NJ_ERR_DIC_BROKEN); 1516 } 1517 if (ret == 0) { 1518 break; 1519 } 1520 p += NJ_INDEX_SIZE; 1521 } 1522 *to = right - 1; 1523 } 1524 1525 #ifdef LEARN_DEBUG 1526 printf(" >> to:(%d)\n", *to); 1527 #endif 1528 return 1; 1529 } 1530 1531 static NJ_INT16 str_que_cmp(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_CHAR *yomi, 1532 NJ_UINT16 yomiLen, NJ_UINT16 que_id, NJ_UINT8 mode) { 1533 NJ_UINT8 *queYomi; 1534 NJ_UINT8 *yomiPtr; 1535 NJ_UINT16 yomiByte; 1536 NJ_UINT16 yomiPos; 1537 NJ_UINT8 queYomiByte, queKouhoByte; 1538 NJ_UINT8 queYomiPos, queYomiSearchArea; 1539 NJ_INT16 complete; 1540 NJ_UINT8 *top_addr; 1541 NJ_UINT8 *bottom_addr; 1542 NJ_UINT16 que_size; 1543 1544 1545 #ifdef IWNN_ERR_CHECK 1546 if (iwnn->err_check_flg == 6) { 1547 que_id = GET_LEARN_MAX_WORD_COUNT(handle); 1548 } 1549 #endif 1550 if (que_id >= GET_LEARN_MAX_WORD_COUNT(handle)) { 1551 1552 return NJ_SET_ERR_VAL(NJ_FUNC_STR_QUE_CMP, NJ_ERR_DIC_BROKEN); 1553 } 1554 1555 queYomi = POS_TO_ADDRESS(handle, que_id); 1556 #ifdef IWNN_ERR_CHECK 1557 if (iwnn->err_check_flg == 7) { 1558 *queYomi = 0x03; 1559 } 1560 #endif 1561 switch (GET_TYPE_FROM_DATA(queYomi)) { 1562 case QUE_TYPE_EMPTY: 1563 case QUE_TYPE_JIRI: 1564 case QUE_TYPE_FZK: 1565 break; 1566 default: 1567 1568 return NJ_SET_ERR_VAL(NJ_FUNC_STR_QUE_CMP, NJ_ERR_DIC_BROKEN); 1569 } 1570 1571 1572 if ((mode == 2) && (yomiLen == 0)) { 1573 return 1; 1574 } 1575 1576 1577 queYomiByte = GET_YSIZE_FROM_DATA(queYomi); 1578 queKouhoByte= GET_KSIZE_FROM_DATA(queYomi); 1579 1580 top_addr = LEARN_DATA_TOP_ADDR(handle); 1581 que_size = QUE_SIZE(handle); 1582 1583 1584 queYomi += LEARN_QUE_STRING_OFFSET; 1585 queYomiSearchArea = (NJ_UINT8)(QUE_SIZE(handle) - LEARN_QUE_STRING_OFFSET); 1586 1587 complete = 0; 1588 yomiPos = 0; queYomiPos = 0; 1589 yomiPtr = (NJ_UINT8*)yomi; 1590 yomiByte = yomiLen * sizeof(NJ_CHAR); 1591 1592 1593 while ((complete = (*yomiPtr - *queYomi)) == 0) { 1594 yomiPos++; queYomiPos++; 1595 1596 if (queYomiPos >= queYomiByte) { 1597 if (queYomiByte == yomiByte) { 1598 1599 return 1; 1600 } else if (mode == 2) { 1601 1602 return 2; 1603 } else { 1604 1605 return (mode + 1); 1606 } 1607 } 1608 if (yomiPos >= yomiByte) { 1609 1610 break; 1611 } else { 1612 yomiPtr++; queYomi++; 1613 #ifdef IWNN_ERR_CHECK 1614 if (iwnn->err_check_flg == 8) { 1615 queYomiPos = queYomiSearchArea; 1616 } 1617 #endif 1618 if (queYomiPos >= queYomiSearchArea) { 1619 1620 bottom_addr = top_addr; 1621 bottom_addr += que_size * GET_LEARN_MAX_WORD_COUNT(handle) - 1; 1622 if (queYomi >= bottom_addr) { 1623 queYomi = top_addr; 1624 } 1625 1626 1627 if (*queYomi++ != QUE_TYPE_NEXT) { 1628 1629 return NJ_SET_ERR_VAL(NJ_FUNC_STR_QUE_CMP, NJ_ERR_DIC_BROKEN); 1630 } 1631 queYomiSearchArea += (NJ_UINT8)(que_size - 1); 1632 } 1633 } 1634 } 1635 if (complete == 0) { 1636 if (yomiByte < queYomiByte) { 1637 1638 if (mode == 2) { 1639 return 1; 1640 } 1641 1642 return 0; 1643 } else { 1644 1645 return 2; 1646 } 1647 } else if (complete < 0) { 1648 1649 return 0; 1650 } else { 1651 1652 return 2; 1653 } 1654 } 1655 1656 static NJ_HINDO calculate_hindo(NJ_DIC_HANDLE handle, NJ_INT32 freq, NJ_DIC_FREQ *dic_freq, NJ_INT16 freq_max, NJ_INT16 freq_min) { 1657 NJ_UINT16 max; 1658 NJ_HINDO hindo; 1659 1660 1661 max = GET_LEARN_MAX_WORD_COUNT(handle); 1662 1663 1664 1665 1666 if (NJ_GET_DIC_TYPE(handle) == NJ_DIC_TYPE_USER) { 1667 1668 hindo = (NJ_INT16)dic_freq->base; 1669 } else { 1670 1671 if (max > 1) { 1672 1673 hindo = CALCULATE_HINDO(freq, dic_freq->base, dic_freq->high, (max-1)); 1674 } else { 1675 1676 hindo = (NJ_INT16)dic_freq->high; 1677 } 1678 } 1679 return NORMALIZE_HINDO(hindo, freq_max, freq_min); 1680 } 1681 1682 static NJ_HINDO get_hindo(NJ_CLASS *iwnn, NJ_SEARCH_LOCATION_SET *loctset, 1683 NJ_UINT8 search_pattern) { 1684 NJ_WQUE *que; 1685 NJ_UINT16 que_id, oldest; 1686 NJ_UINT8 offset; 1687 NJ_INT32 dic_freq; 1688 NJ_UINT16 max; 1689 NJ_UINT8 *learn_index_top_addr; 1690 1691 1692 1693 learn_index_top_addr = get_search_index_address(loctset->loct.handle, search_pattern); 1694 1695 que_id = (NJ_UINT16)GET_UINT16(learn_index_top_addr + 1696 ((loctset->loct.current & 0xffffU) * NJ_INDEX_SIZE)); 1697 oldest = GET_LEARN_NEXT_WORD_POS(loctset->loct.handle); 1698 1699 offset = (loctset->loct.current_info & 0x0f); 1700 while (offset--) { 1701 que_id = search_next_que(loctset->loct.handle, que_id); 1702 } 1703 1704 que = get_que(iwnn, loctset->loct.handle, que_id); 1705 if (que == NULL) { 1706 return INIT_HINDO; 1707 } 1708 1709 max = GET_LEARN_MAX_WORD_COUNT(loctset->loct.handle); 1710 if (que_id >= oldest) { 1711 dic_freq = que_id - oldest; 1712 } else { 1713 dic_freq = que_id - oldest + max; 1714 } 1715 1716 1717 return calculate_hindo(loctset->loct.handle, dic_freq, &(loctset->dic_freq), 1000, 0); 1718 } 1719 1720 NJ_INT16 njd_l_get_word(NJ_CLASS *iwnn, NJ_SEARCH_LOCATION_SET *loctset, NJ_WORD *word) { 1721 NJ_WQUE *que; 1722 NJ_UINT16 que_id; 1723 NJ_UINT8 offset; 1724 NJ_UINT8 *learn_index_top_addr; 1725 1726 1727 1728 learn_index_top_addr = get_search_index_address(loctset->loct.handle, GET_LOCATION_OPERATION(loctset->loct.status)); 1729 1730 que_id = (NJ_UINT16)GET_UINT16(learn_index_top_addr + 1731 ((loctset->loct.current & 0xffff) * NJ_INDEX_SIZE)); 1732 1733 offset = (loctset->loct.current_info & 0x0f); 1734 while (offset--) { 1735 que_id = search_next_que(loctset->loct.handle, que_id); 1736 } 1737 1738 que = get_que(iwnn, loctset->loct.handle, que_id); 1739 if (que == NULL) { 1740 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_GET_WORD, NJ_ERR_CANNOT_GET_QUE); 1741 } 1742 1743 word->stem.loc = loctset->loct; 1744 1745 word->stem.loc.current &= 0x0000ffff; 1746 word->stem.loc.current |= ((NJ_UINT32)que_id << 16); 1747 1748 1749 word->stem.hindo = loctset->cache_freq; 1750 1751 NJ_SET_FPOS_TO_STEM(word, que->mae_hinsi); 1752 NJ_SET_YLEN_TO_STEM(word, que->yomi_len); 1753 if (que->hyouki_len > 0) { 1754 NJ_SET_KLEN_TO_STEM(word, que->hyouki_len); 1755 } else { 1756 1757 NJ_SET_KLEN_TO_STEM(word, que->yomi_len); 1758 } 1759 NJ_SET_BPOS_TO_STEM(word, que->ato_hinsi); 1760 1761 1762 word->stem.type = 0; 1763 1764 return 1; 1765 } 1766 1767 NJ_INT16 njd_l_get_stroke(NJ_CLASS *iwnn, NJ_WORD *word, NJ_CHAR *stroke, NJ_UINT16 size) { 1768 NJ_UINT16 que_id; 1769 NJ_CHAR *str; 1770 NJ_UINT8 slen; 1771 NJ_UINT8 ylen; 1772 1773 1774 que_id = (NJ_UINT16)(word->stem.loc.current >> 16); 1775 1776 1777 ylen = (NJ_UINT8)NJ_GET_YLEN_FROM_STEM(word); 1778 1779 if ((NJ_UINT16)((ylen+ NJ_TERM_LEN)*sizeof(NJ_CHAR)) > size) { 1780 1781 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_GET_STROKE, NJ_ERR_BUFFER_NOT_ENOUGH); 1782 } 1783 if (ylen == 0) { 1784 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_GET_STROKE, NJ_ERR_INVALID_RESULT); 1785 } 1786 str = get_string(iwnn, word->stem.loc.handle, que_id, &slen); 1787 1788 #ifdef IWNN_ERR_CHECK 1789 if (iwnn->err_check_flg == 9) { 1790 str = NULL; 1791 } 1792 #endif 1793 1794 if (str == NULL) { 1795 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_GET_STROKE, NJ_ERR_DIC_BROKEN); 1796 } 1797 1798 1799 nj_strcpy(stroke, str); 1800 1801 return slen; 1802 } 1803 1804 NJ_INT16 njd_l_get_candidate(NJ_CLASS *iwnn, NJ_WORD *word, 1805 NJ_CHAR *candidate, NJ_UINT16 size) { 1806 NJ_UINT16 que_id; 1807 NJ_CHAR *str; 1808 NJ_UINT16 klen; 1809 NJ_UINT8 slen; 1810 1811 1812 que_id = (NJ_UINT16)(word->stem.loc.current >> 16); 1813 1814 1815 klen = NJ_GET_KLEN_FROM_STEM(word); 1816 1817 if (size < ((klen+NJ_TERM_LEN)*sizeof(NJ_CHAR))) { 1818 1819 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_GET_CANDIDATE, NJ_ERR_BUFFER_NOT_ENOUGH); 1820 } 1821 str = get_hyouki(iwnn, word->stem.loc.handle, que_id, &slen); 1822 #ifdef IWNN_ERR_CHECK 1823 if (iwnn->err_check_flg == 10) { 1824 str = NULL; 1825 } 1826 #endif 1827 if (str == NULL) { 1828 1829 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_GET_CANDIDATE, NJ_ERR_DIC_BROKEN); 1830 } 1831 1832 1833 nj_strcpy(candidate, str); 1834 1835 return klen; 1836 } 1837 1838 NJ_INT16 njd_l_check_dic(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle) { 1839 NJ_UINT16 flg; 1840 NJ_UINT16 word_cnt, max; 1841 NJ_UINT8 *ptr; 1842 NJ_UINT16 target_id; 1843 NJ_UINT16 i; 1844 NJ_UINT16 id1 = 0; 1845 NJ_UINT8 slen; 1846 1847 1848 1849 if ((NJ_GET_DIC_TYPE(handle) != NJ_DIC_TYPE_USER)) { 1850 1851 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_CHECK_DIC, NJ_ERR_DIC_TYPE_INVALID); 1852 } 1853 1854 1855 word_cnt = GET_LEARN_WORD_COUNT(handle); 1856 max = GET_LEARN_MAX_WORD_COUNT(handle); 1857 if (word_cnt > max) { 1858 1859 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_CHECK_DIC, 1860 NJ_ERR_DIC_BROKEN); 1861 } 1862 1863 ptr = LEARN_INDEX_TOP_ADDR(handle); 1864 for (i = 0; i < word_cnt; i++) { 1865 id1 = GET_UINT16(ptr); 1866 1867 if (id1 >= max) { 1868 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_CHECK_DIC, 1869 NJ_ERR_DIC_BROKEN); 1870 } 1871 ptr += NJ_INDEX_SIZE; 1872 } 1873 1874 1875 ptr = LEARN_INDEX_TOP_ADDR2(handle); 1876 for (i = 0; i < word_cnt; i++) { 1877 id1 = GET_UINT16(ptr); 1878 1879 if (id1 >= max) { 1880 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_CHECK_DIC, 1881 NJ_ERR_DIC_BROKEN); 1882 } 1883 ptr += NJ_INDEX_SIZE; 1884 } 1885 1886 1887 flg = GET_UINT16(handle + POS_WRITE_FLG); 1888 1889 target_id = GET_UINT16(handle + POS_WRITE_FLG + 2); 1890 1891 1892 1893 if (((flg != word_cnt) && (flg != (word_cnt + 1)) && (flg != (word_cnt - 1))) || 1894 (target_id >= max)) { 1895 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_CHECK_DIC, 1896 NJ_ERR_DIC_BROKEN); 1897 } 1898 1899 1900 if (flg == (word_cnt + 1)) { 1901 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_CHECK_DIC, NJ_ERR_DIC_BROKEN); 1902 } else if (flg == (word_cnt - 1)) { 1903 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_CHECK_DIC, NJ_ERR_DIC_BROKEN); 1904 } 1905 1906 word_cnt = GET_LEARN_WORD_COUNT(handle); 1907 1908 ptr = LEARN_INDEX_TOP_ADDR(handle); 1909 for (i = 0; i < word_cnt; i++) { 1910 id1 = GET_UINT16(ptr); 1911 if (get_hyouki(iwnn, handle, id1, &slen) == NULL) { 1912 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_CHECK_DIC, 1913 NJ_ERR_DIC_BROKEN); 1914 } 1915 ptr += NJ_INDEX_SIZE; 1916 } 1917 1918 ptr = LEARN_INDEX_TOP_ADDR2(handle); 1919 for (i = 0; i < word_cnt; i++) { 1920 id1 = GET_UINT16(ptr); 1921 1922 if (id1 >= max) { 1923 return NJ_SET_ERR_VAL(NJ_FUNC_NJD_L_CHECK_DIC, 1924 NJ_ERR_DIC_BROKEN); 1925 } 1926 ptr += NJ_INDEX_SIZE; 1927 } 1928 1929 return 0; 1930 } 1931 1932 static NJ_INT16 get_cand_by_evaluate2(NJ_CLASS *iwnn, NJ_SEARCH_CONDITION *cond, 1933 NJ_SEARCH_LOCATION_SET *loctset, 1934 NJ_UINT8 search_pattern, 1935 NJ_UINT16 idx) { 1936 NJ_UINT16 from, to, i; 1937 NJ_UINT16 que_id, oldest; 1938 NJ_UINT32 max_value, eval, current; 1939 NJ_UINT8 *ptr, *p; 1940 NJ_WQUE *que; 1941 NJ_INT16 ret = 0; 1942 NJ_INT32 found = 0; 1943 NJ_UINT8 forward_flag = 0; 1944 1945 1946 NJ_UINT16 abIdx; 1947 NJ_UINT16 abIdx_old; 1948 NJ_UINT16 tmp_len; 1949 NJ_UINT16 yomi_clen; 1950 NJ_UINT16 j,l,m; 1951 NJ_UINT8 cmpflg; 1952 NJ_UINT8 endflg = 0; 1953 NJ_CHAR *str; 1954 NJ_CHAR *key; 1955 NJ_CHAR char_tmp[NJ_MAX_LEN + NJ_TERM_LEN]; 1956 NJ_CHAR *pchar_tmp; 1957 NJ_SEARCH_CACHE *psrhCache = cond->ds->dic[idx].srhCache; 1958 NJ_UINT16 endIdx; 1959 NJ_UINT8 slen; 1960 NJ_UINT16 addcnt = 0; 1961 NJ_CHAR *yomi; 1962 NJ_UINT8 aimai_flg = 0x01; 1963 NJ_CHARSET *pCharset = cond->charset; 1964 1965 1966 if (NJ_GET_CACHEOVER_FROM_SCACHE(psrhCache)) { 1967 aimai_flg = 0x00; 1968 } 1969 1970 1971 ptr = get_search_index_address(loctset->loct.handle, cond->operation); 1972 1973 1974 oldest = GET_LEARN_NEXT_WORD_POS(loctset->loct.handle); 1975 max_value = oldest; 1976 1977 1978 current = 0; 1979 if (GET_LOCATION_STATUS(loctset->loct.status) == NJ_ST_SEARCH_NO_INIT) { 1980 1981 1982 key = cond->ds->keyword; 1983 yomi = cond->yomi; 1984 yomi_clen = cond->yclen; 1985 1986 1987 endflg = 0x00; 1988 1989 if (psrhCache->keyPtr[0] == 0xFFFF) { 1990 cmpflg = 0x01; 1991 psrhCache->keyPtr[0] = 0x0000; 1992 } else { 1993 cmpflg = 0x00; 1994 } 1995 1996 for (i = 0; i < yomi_clen; i++) { 1997 j = i; 1998 1999 2000 if (!cmpflg) { 2001 2002 if (((j != 0) && (psrhCache->keyPtr[j] == 0)) || (psrhCache->keyPtr[j+1] == 0)) { 2003 2004 cmpflg = 0x01; 2005 } else { 2006 2007 } 2008 } 2009 2010 if (cmpflg) { 2011 2012 if (!j) { 2013 2014 abIdx = 0; 2015 addcnt = 0; 2016 nj_charncpy(char_tmp, yomi, 1); 2017 tmp_len = nj_strlen(char_tmp); 2018 ret = search_range_by_yomi(iwnn, loctset->loct.handle, search_pattern, 2019 char_tmp, tmp_len, &from, 2020 &to, &forward_flag); 2021 if (ret < 0) { 2022 2023 2024 psrhCache->keyPtr[j+1] = abIdx; 2025 loctset->loct.status = NJ_ST_SEARCH_END; 2026 return ret; 2027 } else if (ret > 0) { 2028 2029 psrhCache->storebuff[abIdx].top = from; 2030 psrhCache->storebuff[abIdx].bottom = to; 2031 psrhCache->storebuff[abIdx].idx_no = (NJ_INT8)tmp_len; 2032 addcnt++; 2033 abIdx++; 2034 psrhCache->keyPtr[j+1] = abIdx; 2035 } else { 2036 psrhCache->keyPtr[j+1] = abIdx; 2037 } 2038 2039 if ((!endflg) && (pCharset != NULL) && aimai_flg) { 2040 2041 for (l = 0; l < pCharset->charset_count; l++) { 2042 2043 if (nj_charncmp(yomi, pCharset->from[l], 1) == 0) { 2044 2045 nj_strcpy(char_tmp, pCharset->to[l]); 2046 tmp_len = nj_strlen(char_tmp); 2047 ret = search_range_by_yomi(iwnn, loctset->loct.handle, search_pattern, 2048 char_tmp, tmp_len, &from, &to, &forward_flag); 2049 if (ret < 0) { 2050 2051 2052 psrhCache->keyPtr[j+1] = abIdx; 2053 loctset->loct.status = NJ_ST_SEARCH_END; 2054 return ret; 2055 } else if (ret > 0) { 2056 2057 2058 if (abIdx >= NJ_SEARCH_CACHE_SIZE) { 2059 psrhCache->keyPtr[j+1] = 0; 2060 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_CACHE_NOT_ENOUGH); 2061 } 2062 psrhCache->storebuff[abIdx].top = from; 2063 psrhCache->storebuff[abIdx].bottom = to; 2064 psrhCache->storebuff[abIdx].idx_no = (NJ_INT8)tmp_len; 2065 if (addcnt == 0) { 2066 psrhCache->keyPtr[j] = abIdx; 2067 } 2068 abIdx++; 2069 addcnt++; 2070 psrhCache->keyPtr[j+1] = abIdx; 2071 } else { 2072 psrhCache->keyPtr[j+1] = abIdx; 2073 } 2074 } 2075 } 2076 } 2077 } else { 2078 2079 if (psrhCache->keyPtr[j] == psrhCache->keyPtr[j-1]) { 2080 2081 psrhCache->keyPtr[j+1] = psrhCache->keyPtr[j-1]; 2082 endflg = 0x01; 2083 } else { 2084 2085 endIdx = psrhCache->keyPtr[j]; 2086 abIdx_old = psrhCache->keyPtr[j-1]; 2087 2088 if (NJ_GET_CACHEOVER_FROM_SCACHE(psrhCache)) { 2089 abIdx = psrhCache->keyPtr[j - 1]; 2090 psrhCache->keyPtr[j] = abIdx; 2091 } else { 2092 abIdx = psrhCache->keyPtr[j]; 2093 } 2094 addcnt = 0; 2095 2096 if ((abIdx > NJ_SEARCH_CACHE_SIZE) || (abIdx_old >= NJ_SEARCH_CACHE_SIZE) || 2097 (endIdx > NJ_SEARCH_CACHE_SIZE)) { 2098 2099 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_CACHE_BROKEN); 2100 } 2101 for (m = abIdx_old; m < endIdx; m++) { 2102 2103 p = ptr + (psrhCache->storebuff[m].top * NJ_INDEX_SIZE); 2104 que_id = GET_UINT16(p); 2105 2106 2107 str = get_string(iwnn, loctset->loct.handle, que_id, &slen); 2108 2109 if (str == NULL) { 2110 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_DIC_BROKEN); 2111 } 2112 2113 2114 nj_strncpy(char_tmp, str, psrhCache->storebuff[m].idx_no); 2115 char_tmp[psrhCache->storebuff[m].idx_no] = NJ_CHAR_NUL; 2116 2117 pchar_tmp = &char_tmp[psrhCache->storebuff[m].idx_no]; 2118 nj_charncpy(pchar_tmp, yomi, 1); 2119 tmp_len = nj_strlen(char_tmp); 2120 2121 2122 ret = search_range_by_yomi2(iwnn, loctset->loct.handle, search_pattern, 2123 char_tmp, tmp_len, 2124 (NJ_UINT16)(psrhCache->storebuff[m].top), 2125 (NJ_UINT16)(psrhCache->storebuff[m].bottom), 2126 &from, &to, &forward_flag); 2127 if (ret < 0) { 2128 2129 2130 psrhCache->keyPtr[j+1] = abIdx; 2131 loctset->loct.status = NJ_ST_SEARCH_END; 2132 return ret; 2133 } else if (ret > 0) { 2134 2135 2136 if (abIdx >= NJ_SEARCH_CACHE_SIZE) { 2137 psrhCache->keyPtr[j+1] = 0; 2138 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_CACHE_NOT_ENOUGH); 2139 } 2140 psrhCache->storebuff[abIdx].top = from; 2141 psrhCache->storebuff[abIdx].bottom = to; 2142 psrhCache->storebuff[abIdx].idx_no = (NJ_INT8)tmp_len; 2143 if (addcnt == 0) { 2144 psrhCache->keyPtr[j] = abIdx; 2145 } 2146 abIdx++; 2147 addcnt++; 2148 psrhCache->keyPtr[j+1] = abIdx; 2149 } else { 2150 psrhCache->keyPtr[j+1] = abIdx; 2151 } 2152 2153 if ((!endflg) && (pCharset != NULL) && aimai_flg) { 2154 2155 for (l = 0; l < pCharset->charset_count; l++) { 2156 2157 if (nj_charncmp(yomi, pCharset->from[l], 1) == 0) { 2158 2159 tmp_len = nj_strlen(pCharset->to[l]); 2160 2161 nj_strncpy(pchar_tmp, pCharset->to[l], tmp_len); 2162 *(pchar_tmp + tmp_len) = NJ_CHAR_NUL; 2163 tmp_len = nj_strlen(char_tmp); 2164 ret = search_range_by_yomi2(iwnn, loctset->loct.handle, search_pattern, 2165 char_tmp, tmp_len, 2166 (NJ_UINT16)(psrhCache->storebuff[m].top), 2167 (NJ_UINT16)(psrhCache->storebuff[m].bottom), 2168 &from, &to, &forward_flag); 2169 if (ret < 0) { 2170 2171 2172 psrhCache->keyPtr[j+1] = abIdx; 2173 loctset->loct.status = NJ_ST_SEARCH_END; 2174 return ret; 2175 } else if (ret > 0) { 2176 2177 2178 if (abIdx >= NJ_SEARCH_CACHE_SIZE) { 2179 psrhCache->keyPtr[j+1] = 0; 2180 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_CACHE_NOT_ENOUGH); 2181 } 2182 psrhCache->storebuff[abIdx].top = from; 2183 psrhCache->storebuff[abIdx].bottom = to; 2184 psrhCache->storebuff[abIdx].idx_no = (NJ_INT8)tmp_len; 2185 abIdx++; 2186 addcnt++; 2187 psrhCache->keyPtr[j+1] = abIdx; 2188 } else { 2189 psrhCache->keyPtr[j+1] = abIdx; 2190 } 2191 } 2192 } 2193 } 2194 } 2195 } 2196 } 2197 } 2198 yomi += UTL_CHAR(yomi); 2199 key += UTL_CHAR(key); 2200 } 2201 2202 2203 if ((addcnt == 0) && (psrhCache->keyPtr[yomi_clen - 1] == psrhCache->keyPtr[yomi_clen])) { 2204 endflg = 0x01; 2205 } 2206 2207 if (endflg) { 2208 loctset->loct.status = NJ_ST_SEARCH_END; 2209 return 0; 2210 } 2211 } else if (GET_LOCATION_STATUS(loctset->loct.status) == NJ_ST_SEARCH_READY) { 2212 current = GET_UINT16(ptr + (loctset->loct.current * NJ_INDEX_SIZE)); 2213 if (current < oldest) { 2214 current += GET_LEARN_MAX_WORD_COUNT(loctset->loct.handle); 2215 } 2216 } else { 2217 loctset->loct.status = NJ_ST_SEARCH_END; 2218 return 0; 2219 } 2220 2221 2222 j = cond->yclen - 1; 2223 2224 abIdx = psrhCache->keyPtr[j]; 2225 abIdx_old = psrhCache->keyPtr[j+1]; 2226 2227 endIdx = abIdx_old; 2228 if ((abIdx >= NJ_SEARCH_CACHE_SIZE) || (abIdx_old > NJ_SEARCH_CACHE_SIZE)) { 2229 2230 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_CACHE_BROKEN); 2231 } 2232 p = ptr + (psrhCache->storebuff[abIdx].top * NJ_INDEX_SIZE); 2233 que_id = GET_UINT16(p); 2234 eval = current; 2235 2236 2237 2238 if (psrhCache->keyPtr[j] < psrhCache->keyPtr[j + 1]) { 2239 if (GET_LOCATION_STATUS(loctset->loct.status) == NJ_ST_SEARCH_NO_INIT) { 2240 endIdx = abIdx + 1; 2241 NJ_SET_AIMAI_TO_SCACHE(psrhCache); 2242 } 2243 2244 for (m = abIdx; m < endIdx; m++) { 2245 p = ptr + (psrhCache->storebuff[m].top * NJ_INDEX_SIZE); 2246 que_id = GET_UINT16(p); 2247 eval = current; 2248 2249 for (i = (NJ_UINT16)psrhCache->storebuff[m].top; i <= (NJ_UINT16)psrhCache->storebuff[m].bottom; i++) { 2250 que_id = GET_UINT16(p); 2251 if (que_id < oldest) { 2252 eval = que_id + GET_LEARN_MAX_WORD_COUNT(loctset->loct.handle); 2253 } else { 2254 eval = que_id; 2255 } 2256 #ifdef LEARN_DEBUG 2257 printf("que(%d) : eval = %d : %d\n", que_id, eval, i); 2258 #endif 2259 if (eval >= max_value) { 2260 if ((GET_LOCATION_STATUS(loctset->loct.status) == NJ_ST_SEARCH_NO_INIT) 2261 || ((GET_LOCATION_STATUS(loctset->loct.status) == NJ_ST_SEARCH_READY) 2262 && (NJ_GET_AIMAI_FROM_SCACHE(psrhCache))) 2263 || (eval < current)) { 2264 2265 2266 2267 str = get_string(iwnn, loctset->loct.handle, que_id, &slen); 2268 if (str == NULL) { 2269 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_DIC_BROKEN); 2270 } 2271 2272 2273 2274 que = get_que_allHinsi(iwnn, loctset->loct.handle, que_id); 2275 if (njd_connect_test(cond, que->mae_hinsi, que->ato_hinsi)) { 2276 2277 loctset->loct.current_info = (NJ_UINT8)0x10; 2278 loctset->loct.current = i; 2279 max_value = eval; 2280 found = 1; 2281 #ifdef LEARN_DEBUG 2282 printf("---keep."); 2283 #endif 2284 } 2285 } 2286 } 2287 p += NJ_INDEX_SIZE; 2288 } 2289 } 2290 } 2291 2292 if (GET_LOCATION_STATUS(loctset->loct.status) != NJ_ST_SEARCH_NO_INIT) { 2293 NJ_UNSET_AIMAI_TO_SCACHE(psrhCache); 2294 } 2295 2296 2297 if (found == 0) { 2298 loctset->loct.status = NJ_ST_SEARCH_END; 2299 return 0; 2300 } else { 2301 loctset->loct.status = NJ_ST_SEARCH_READY; 2302 loctset->cache_freq = get_hindo(iwnn, loctset, search_pattern); 2303 return 1; 2304 } 2305 } 2306 2307 static NJ_INT16 search_range_by_yomi2(NJ_CLASS *iwnn, NJ_DIC_HANDLE handle, NJ_UINT8 op, 2308 NJ_CHAR *yomi, NJ_UINT16 len, 2309 NJ_UINT16 sfrom, NJ_UINT16 sto, 2310 NJ_UINT16 *from, NJ_UINT16 *to, 2311 NJ_UINT8 *forward_flag) { 2312 NJ_UINT16 right, mid = 0, left, max; 2313 NJ_UINT16 que_id; 2314 NJ_UINT8 *ptr, *p; 2315 NJ_CHAR *str; 2316 NJ_INT16 ret = 0; 2317 NJ_INT32 found = 0; 2318 NJ_UINT8 slen; 2319 NJ_INT32 cmp; 2320 2321 2322 2323 ptr = get_search_index_address(handle, op); 2324 2325 max = GET_LEARN_WORD_COUNT(handle); 2326 2327 right = sto; 2328 left = sfrom; 2329 2330 #ifdef LEARN_DEBUG 2331 printf("src:[%s]\n", yomi); 2332 #endif 2333 2334 *forward_flag = 0; 2335 2336 while (left <= right) { 2337 mid = left + ((right - left) / 2); 2338 p = ptr + (mid * NJ_INDEX_SIZE); 2339 que_id = GET_UINT16(p); 2340 str = get_string(iwnn, handle, que_id, &slen); 2341 if (str == NULL) { 2342 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_DIC_BROKEN); 2343 } 2344 2345 2346 ret = nj_strncmp(yomi, str, len); 2347 2348 #ifdef LEARN_DEBUG 2349 printf(" [%d][%d][%d]COMPARE:[%s] = %d\n", left, mid, right, str, ret); 2350 #endif 2351 if (ret == 0) { 2352 2353 found = 1; 2354 break; 2355 } else if (ret < 0) { 2356 2357 right = mid - 1; 2358 if (mid == 0) { 2359 break; 2360 } 2361 } else { 2362 2363 left = mid + 1; 2364 } 2365 } 2366 2367 if (!found) { 2368 return 0; 2369 } 2370 2371 if (mid == 0) { 2372 *from = mid; 2373 } else { 2374 2375 p = ((mid - 1) * NJ_INDEX_SIZE) + ptr; 2376 2377 for (cmp = mid - 1; cmp >= 0; cmp--) { 2378 que_id = GET_UINT16(p); 2379 str = get_string(iwnn, handle, que_id, &slen); 2380 if (str == NULL) { 2381 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_DIC_BROKEN); 2382 } 2383 2384 2385 if (nj_strncmp(yomi, str, len) != 0) { 2386 break; 2387 } 2388 p -= NJ_INDEX_SIZE; 2389 } 2390 if (cmp < 0) { 2391 *from = 0; 2392 } else { 2393 *from = (NJ_UINT16)cmp + 1; 2394 } 2395 } 2396 2397 #ifdef LEARN_DEBUG 2398 printf(" >> from:(%d)\n", *from); 2399 #endif 2400 2401 if ((mid + 1) >= max) { 2402 *to = mid; 2403 } else { 2404 2405 p = ((mid + 1) * NJ_INDEX_SIZE) + ptr; 2406 2407 for (right = mid + 1; right < max; right++) { 2408 que_id = GET_UINT16(p); 2409 str = get_string(iwnn, handle, que_id, &slen); 2410 if (str == NULL) { 2411 return NJ_SET_ERR_VAL(NJ_FUNC_SEARCH_RANGE_BY_YOMI, NJ_ERR_DIC_BROKEN); 2412 } 2413 2414 2415 if (nj_strncmp(yomi, str, len) != 0) { 2416 break; 2417 } 2418 p += NJ_INDEX_SIZE; 2419 } 2420 *to = right - 1; 2421 } 2422 2423 #ifdef LEARN_DEBUG 2424 printf(" >> to:(%d)\n", *to); 2425 #endif 2426 return 1; 2427 } 2428 2429
