src/icu_I18N.c

   1 /* $Id: icu_I18N.c,v 1.1 2007-10-22 12:21:39 adam Exp $
   2    Copyright (c) 2006-2007, Index Data.
   3
   4    This file is part of Pazpar2.
   5
   6    Pazpar2 is free software; you can redistribute it and/or modify it under
   7    the terms of the GNU General Public License as published by the Free
   8    Software Foundation; either version 2, or (at your option) any later
   9    version.
  10
  11    Pazpar2 is distributed in the hope that it will be useful, but WITHOUT ANY
  12    WARRANTY; without even the implied warranty of MERCHANTABILITY or
  13    FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  14    for more details.
  15
  16    You should have received a copy of the GNU General Public License
  17    along with Pazpar2; see the file LICENSE.  If not, write to the
  18    Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
  19    02111-1307, USA.
  20 */
  21
  22 #if HAVE_CONFIG_H
  23 #include "cconfig.h"
  24 #endif
  25
  26 #define USE_TIMING 0
  27 #if USE_TIMING
  28 #include <yaz/timing.h>
  29 #endif
  30
  31
  32 #ifdef HAVE_ICU
  33 #include <yaz/icu_I18N.h>
  34
  35 #include <yaz/log.h>
  36
  37 #include <string.h>
  38 #include <stdlib.h>
  39 #include <stdio.h>
  40
  41 #include <unicode/ustring.h>  /* some more string fcns*/
  42 #include <unicode/uchar.h>    /* char names           */
  43
  44
  45 //#include <unicode/ustdio.h>
  46 //#include <unicode/utypes.h>   /* Basic ICU data types */
  47 #include <unicode/ucol.h>
  48 //#include <unicode/ucnv.h>     /* C   Converter API    */
  49 //#include <unicode/uloc.h>
  50 //#include <unicode/ubrk.h>
  51 /* #include <unicode/unistr.h> */
  52
  53
  54
  55
  56 int icu_check_status (UErrorCode status)
  57 {
  58     if(U_FAILURE(status)){
  59         yaz_log(YLOG_WARN,
  60                 "ICU: %d %s\n", status, u_errorName(status));
  61         return 0;
  62     }
  63     return 1;
  64
  65 }
  66
  67
  68
  69 struct icu_buf_utf16 * icu_buf_utf16_create(size_t capacity)
  70 {
  71     struct icu_buf_utf16 * buf16
  72         = (struct icu_buf_utf16 *) malloc(sizeof(struct icu_buf_utf16));
  73
  74     buf16->utf16 = 0;
  75     buf16->utf16_len = 0;
  76     buf16->utf16_cap = 0;
  77
  78     if (capacity > 0){
  79         buf16->utf16 = (UChar *) malloc(sizeof(UChar) * capacity);
  80         buf16->utf16[0] = (UChar) 0;
  81         buf16->utf16_cap = capacity;
  82     }
  83     return buf16;
  84 };
  85
  86 struct icu_buf_utf16 * icu_buf_utf16_resize(struct icu_buf_utf16 * buf16,
  87                                             size_t capacity)
  88 {
  89     if (buf16){
  90         if (capacity >  0){
  91             if (0 == buf16->utf16)
  92                 buf16->utf16 = (UChar *) malloc(sizeof(UChar) * capacity);
  93             else
  94                 buf16->utf16
  95                     = (UChar *) realloc(buf16->utf16, sizeof(UChar) * capacity);
  96             buf16->utf16[0] = (UChar) 0;
  97             buf16->utf16_len = 0;
  98             buf16->utf16_cap = capacity;
  99         }
 100         else {
 101             if (buf16->utf16)
 102                 free(buf16->utf16);
 103             buf16->utf16 = 0;
 104             buf16->utf16_len = 0;
 105             buf16->utf16_cap = 0;
 106         }
 107     }
 108
 109     return buf16;
 110 };
 111
 112
 113 struct icu_buf_utf16 * icu_buf_utf16_copy(struct icu_buf_utf16 * dest16,
 114                                           struct icu_buf_utf16 * src16)
 115 {
 116     if(!dest16 || !src16
 117        || dest16 == src16)
 118         return 0;
 119
 120     if (dest16->utf16_cap < src16->utf16_len)
 121         icu_buf_utf16_resize(dest16, src16->utf16_len * 2);
 122
 123     u_strncpy(dest16->utf16, src16->utf16, src16->utf16_len);
 124     dest16->utf16_len = src16->utf16_len;
 125
 126     return dest16;
 127 };
 128
 129
 130 void icu_buf_utf16_destroy(struct icu_buf_utf16 * buf16)
 131 {
 132     if (buf16){
 133         if (buf16->utf16)
 134             free(buf16->utf16);
 135         free(buf16);
 136     }
 137 };
 138
 139
 140
 141
 142
 143
 144 struct icu_buf_utf8 * icu_buf_utf8_create(size_t capacity)
 145 {
 146     struct icu_buf_utf8 * buf8
 147         = (struct icu_buf_utf8 *) malloc(sizeof(struct icu_buf_utf8));
 148
 149     buf8->utf8 = 0;
 150     buf8->utf8_len = 0;
 151     buf8->utf8_cap = 0;
 152
 153     if (capacity > 0){
 154         buf8->utf8 = (uint8_t *) malloc(sizeof(uint8_t) * capacity);
 155         buf8->utf8[0] = (uint8_t) 0;
 156         buf8->utf8_cap = capacity;
 157     }
 158     return buf8;
 159 };
 160
 161
 162
 163 struct icu_buf_utf8 * icu_buf_utf8_resize(struct icu_buf_utf8 * buf8,
 164                                           size_t capacity)
 165 {
 166     if (buf8){
 167         if (capacity >  0){
 168             if (0 == buf8->utf8)
 169                 buf8->utf8 = (uint8_t *) malloc(sizeof(uint8_t) * capacity);
 170             else
 171                 buf8->utf8
 172                     = (uint8_t *) realloc(buf8->utf8,
 173                                           sizeof(uint8_t) * capacity);
 174             buf8->utf8_cap = capacity;
 175         }
 176         else {
 177             if (buf8->utf8)
 178                 free(buf8->utf8);
 179             buf8->utf8 = 0;
 180             buf8->utf8_len = 0;
 181             buf8->utf8_cap = 0;
 182         }
 183     }
 184
 185     return buf8;
 186 };
 187
 188
 189 struct icu_buf_utf8 * icu_buf_utf8_copy(struct icu_buf_utf8 * dest8,
 190                                           struct icu_buf_utf8 * src8)
 191 {
 192     if(!dest8 || !src8
 193        || dest8 == src8)
 194         return 0;
 195
 196
 197     if (dest8->utf8_cap < src8->utf8_len)
 198         icu_buf_utf8_resize(dest8, src8->utf8_len * 2);
 199
 200     strncpy((char*) dest8->utf8, (char*) src8->utf8, src8->utf8_len);
 201
 202     return dest8;
 203 };
 204
 205
 206 const char *icu_buf_utf8_to_cstr(struct icu_buf_utf8 *src8)
 207 {
 208     if (!src8 || src8->utf8_len == 0)
 209         return "";
 210     if (src8->utf8_len == src8->utf8_cap)
 211         src8 = icu_buf_utf8_resize(src8, src8->utf8_len * 2 + 1);
 212     src8->utf8[src8->utf8_len] = '\0';
 213     return (const char *) src8->utf8;
 214 }
 215
 216
 217 void icu_buf_utf8_destroy(struct icu_buf_utf8 * buf8)
 218 {
 219     if (buf8){
 220         if (buf8->utf8)
 221             free(buf8->utf8);
 222         free(buf8);
 223     }
 224 };
 225
 226
 227
 228 UErrorCode icu_utf16_from_utf8(struct icu_buf_utf16 * dest16,
 229                                struct icu_buf_utf8 * src8,
 230                                UErrorCode * status)
 231 {
 232     int32_t utf16_len = 0;
 233
 234     u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 235                   &utf16_len,
 236                   (const char *) src8->utf8, src8->utf8_len, status);
 237
 238     // check for buffer overflow, resize and retry
 239     if (*status == U_BUFFER_OVERFLOW_ERROR
 240         //|| dest16->utf16_len > dest16->utf16_cap
 241         ){
 242         icu_buf_utf16_resize(dest16, utf16_len * 2);
 243         *status = U_ZERO_ERROR;
 244         u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 245                       &utf16_len,
 246                       (const char *) src8->utf8, src8->utf8_len, status);
 247     }
 248
 249     //if (*status != U_BUFFER_OVERFLOW_ERROR
 250     if (U_SUCCESS(*status)
 251         && utf16_len <= dest16->utf16_cap)
 252         dest16->utf16_len = utf16_len;
 253     else {
 254         dest16->utf16[0] = (UChar) 0;
 255         dest16->utf16_len = 0;
 256     }
 257
 258     return *status;
 259 };
 260
 261
 262
 263 UErrorCode icu_utf16_from_utf8_cstr(struct icu_buf_utf16 * dest16,
 264                                     const char * src8cstr,
 265                                     UErrorCode * status)
 266 {
 267     size_t src8cstr_len = 0;
 268     int32_t utf16_len = 0;
 269
 270     src8cstr_len = strlen(src8cstr);
 271
 272     u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 273                   &utf16_len,
 274                   src8cstr, src8cstr_len, status);
 275
 276     // check for buffer overflow, resize and retry
 277     if (*status == U_BUFFER_OVERFLOW_ERROR
 278         //|| dest16->utf16_len > dest16->utf16_cap
 279         ){
 280         icu_buf_utf16_resize(dest16, utf16_len * 2);
 281         *status = U_ZERO_ERROR;
 282         u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 283                       &utf16_len,
 284                       src8cstr, src8cstr_len, status);
 285     }
 286
 287     //  if (*status != U_BUFFER_OVERFLOW_ERROR
 288     if (U_SUCCESS(*status)
 289         && utf16_len <= dest16->utf16_cap)
 290         dest16->utf16_len = utf16_len;
 291     else {
 292         dest16->utf16[0] = (UChar) 0;
 293         dest16->utf16_len = 0;
 294     }
 295
 296     return *status;
 297 };
 298
 299
 300
 301
 302 UErrorCode icu_utf16_to_utf8(struct icu_buf_utf8 * dest8,
 303                              struct icu_buf_utf16 * src16,
 304                              UErrorCode * status)
 305 {
 306     int32_t utf8_len = 0;
 307
 308     u_strToUTF8((char *) dest8->utf8, dest8->utf8_cap,
 309                 &utf8_len,
 310                 src16->utf16, src16->utf16_len, status);
 311
 312     // check for buffer overflow, resize and retry
 313     if (*status == U_BUFFER_OVERFLOW_ERROR
 314         //|| dest8->utf8_len > dest8->utf8_cap
 315         ){
 316         icu_buf_utf8_resize(dest8, utf8_len * 2);
 317         *status = U_ZERO_ERROR;
 318         u_strToUTF8((char *) dest8->utf8, dest8->utf8_cap,
 319                     &utf8_len,
 320                     src16->utf16, src16->utf16_len, status);
 321
 322     }
 323
 324     //if (*status != U_BUFFER_OVERFLOW_ERROR
 325     if (U_SUCCESS(*status)
 326         && utf8_len <= dest8->utf8_cap)
 327         dest8->utf8_len = utf8_len;
 328     else {
 329         dest8->utf8[0] = (uint8_t) 0;
 330         dest8->utf8_len = 0;
 331     }
 332
 333     return *status;
 334 };
 335
 336
 337
 338 struct icu_casemap * icu_casemap_create(const char *locale, char action,
 339                                         UErrorCode *status)
 340 {
 341     struct icu_casemap * casemap
 342         = (struct icu_casemap *) malloc(sizeof(struct icu_casemap));
 343     strcpy(casemap->locale, locale);
 344     casemap->action = action;
 345
 346     switch(casemap->action) {
 347     case 'l':
 348         break;
 349     case 'u':
 350         break;
 351     case 't':
 352         break;
 353     case 'f':
 354         break;
 355     default:
 356         icu_casemap_destroy(casemap);
 357         return 0;
 358     }
 359
 360     return casemap;
 361 };
 362
 363 void icu_casemap_destroy(struct icu_casemap * casemap)
 364 {
 365     if (casemap)
 366         free(casemap);
 367 };
 368
 369
 370 int icu_casemap_casemap(struct icu_casemap * casemap,
 371                         struct icu_buf_utf16 * dest16,
 372                         struct icu_buf_utf16 * src16,
 373                         UErrorCode *status)
 374 {
 375     if(!casemap)
 376         return 0;
 377
 378     return icu_utf16_casemap(dest16, src16,
 379                              casemap->locale, casemap->action, status);
 380 };
 381
 382
 383 int icu_utf16_casemap(struct icu_buf_utf16 * dest16,
 384                       struct icu_buf_utf16 * src16,
 385                       const char *locale, char action,
 386                       UErrorCode *status)
 387 {
 388     int32_t dest16_len = 0;
 389
 390     switch(action) {
 391     case 'l':
 392         dest16_len = u_strToLower(dest16->utf16, dest16->utf16_cap,
 393                                   src16->utf16, src16->utf16_len,
 394                                   locale, status);
 395         break;
 396     case 'u':
 397         dest16_len = u_strToUpper(dest16->utf16, dest16->utf16_cap,
 398                                   src16->utf16, src16->utf16_len,
 399                                   locale, status);
 400         break;
 401     case 't':
 402         dest16_len = u_strToTitle(dest16->utf16, dest16->utf16_cap,
 403                                   src16->utf16, src16->utf16_len,
 404                                   0, locale, status);
 405         break;
 406     case 'f':
 407         dest16_len = u_strFoldCase(dest16->utf16, dest16->utf16_cap,
 408                                    src16->utf16, src16->utf16_len,
 409                                    U_FOLD_CASE_DEFAULT, status);
 410         break;
 411
 412     default:
 413         return U_UNSUPPORTED_ERROR;
 414         break;
 415     }
 416
 417     // check for buffer overflow, resize and retry
 418     if (*status == U_BUFFER_OVERFLOW_ERROR
 419         && dest16 != src16        // do not resize if in-place conversion
 420         //|| dest16_len > dest16->utf16_cap
 421         ){
 422         icu_buf_utf16_resize(dest16, dest16_len * 2);
 423         *status = U_ZERO_ERROR;
 424
 425
 426         switch(action) {
 427         case 'l':
 428             dest16_len = u_strToLower(dest16->utf16, dest16->utf16_cap,
 429                                       src16->utf16, src16->utf16_len,
 430                                       locale, status);
 431             break;
 432         case 'u':
 433             dest16_len = u_strToUpper(dest16->utf16, dest16->utf16_cap,
 434                                       src16->utf16, src16->utf16_len,
 435                                       locale, status);
 436             break;
 437         case 't':
 438             dest16_len = u_strToTitle(dest16->utf16, dest16->utf16_cap,
 439                                       src16->utf16, src16->utf16_len,
 440                                       0, locale, status);
 441             break;
 442         case 'f':
 443             dest16_len = u_strFoldCase(dest16->utf16, dest16->utf16_cap,
 444                                        src16->utf16, src16->utf16_len,
 445                                        U_FOLD_CASE_DEFAULT, status);
 446             break;
 447
 448         default:
 449             return U_UNSUPPORTED_ERROR;
 450             break;
 451         }
 452     }
 453
 454     if (U_SUCCESS(*status)
 455         && dest16_len <= dest16->utf16_cap)
 456         dest16->utf16_len = dest16_len;
 457     else {
 458         dest16->utf16[0] = (UChar) 0;
 459         dest16->utf16_len = 0;
 460     }
 461
 462     return *status;
 463 };
 464
 465
 466
 467 UErrorCode icu_sortkey8_from_utf16(UCollator *coll,
 468                                    struct icu_buf_utf8 * dest8,
 469                                    struct icu_buf_utf16 * src16,
 470                                    UErrorCode * status)
 471 {
 472
 473     int32_t sortkey_len = 0;
 474
 475     sortkey_len = ucol_getSortKey(coll, src16->utf16, src16->utf16_len,
 476                                   dest8->utf8, dest8->utf8_cap);
 477
 478     // check for buffer overflow, resize and retry
 479     if (sortkey_len > dest8->utf8_cap) {
 480         icu_buf_utf8_resize(dest8, sortkey_len * 2);
 481         sortkey_len = ucol_getSortKey(coll, src16->utf16, src16->utf16_len,
 482                                       dest8->utf8, dest8->utf8_cap);
 483     }
 484
 485     if (U_SUCCESS(*status)
 486         && sortkey_len > 0)
 487         dest8->utf8_len = sortkey_len;
 488     else {
 489         dest8->utf8[0] = (UChar) 0;
 490         dest8->utf8_len = 0;
 491     }
 492
 493     return sortkey_len;
 494 };
 495
 496
 497
 498 struct icu_tokenizer * icu_tokenizer_create(const char *locale, char action,
 499                                             UErrorCode *status)
 500 {
 501     struct icu_tokenizer * tokenizer
 502         = (struct icu_tokenizer *) malloc(sizeof(struct icu_tokenizer));
 503
 504     strcpy(tokenizer->locale, locale);
 505     tokenizer->action = action;
 506     tokenizer->bi = 0;
 507     tokenizer->buf16 = 0;
 508     tokenizer->token_count = 0;
 509     tokenizer->token_id = 0;
 510     tokenizer->token_start = 0;
 511     tokenizer->token_end = 0;
 512
 513
 514     switch(tokenizer->action) {
 515     case 'l':
 516         tokenizer->bi
 517             = ubrk_open(UBRK_LINE, tokenizer->locale,
 518                         0, 0, status);
 519         break;
 520     case 's':
 521         tokenizer->bi
 522             = ubrk_open(UBRK_SENTENCE, tokenizer->locale,
 523                         0, 0, status);
 524         break;
 525     case 'w':
 526         tokenizer->bi
 527             = ubrk_open(UBRK_WORD, tokenizer->locale,
 528                         0, 0, status);
 529         break;
 530     case 'c':
 531         tokenizer->bi
 532             = ubrk_open(UBRK_CHARACTER, tokenizer->locale,
 533                         0, 0, status);
 534         break;
 535     case 't':
 536         tokenizer->bi
 537             = ubrk_open(UBRK_TITLE, tokenizer->locale,
 538                         0, 0, status);
 539         break;
 540     default:
 541         *status = U_UNSUPPORTED_ERROR;
 542         return 0;
 543         break;
 544     }
 545
 546     // ICU error stuff is a very  funny business
 547     if (U_SUCCESS(*status))
 548         return tokenizer;
 549
 550     // freeing if failed
 551     icu_tokenizer_destroy(tokenizer);
 552     return 0;
 553 };
 554
 555 void icu_tokenizer_destroy(struct icu_tokenizer * tokenizer)
 556 {
 557     if (tokenizer) {
 558         if (tokenizer->bi)
 559             ubrk_close(tokenizer->bi);
 560         free(tokenizer);
 561     }
 562 };
 563
 564 int icu_tokenizer_attach(struct icu_tokenizer * tokenizer,
 565                          struct icu_buf_utf16 * src16,
 566                          UErrorCode *status)
 567 {
 568     if (!tokenizer || !tokenizer->bi || !src16)
 569         return 0;
 570
 571
 572     tokenizer->buf16 = src16;
 573     tokenizer->token_count = 0;
 574     tokenizer->token_id = 0;
 575     tokenizer->token_start = 0;
 576     tokenizer->token_end = 0;
 577
 578     ubrk_setText(tokenizer->bi, src16->utf16, src16->utf16_len, status);
 579
 580
 581     if (U_FAILURE(*status))
 582         return 0;
 583
 584     return 1;
 585 };
 586
 587 int32_t icu_tokenizer_next_token(struct icu_tokenizer * tokenizer,
 588                          struct icu_buf_utf16 * tkn16,
 589                          UErrorCode *status)
 590 {
 591     int32_t tkn_start = 0;
 592     int32_t tkn_end = 0;
 593     int32_t tkn_len = 0;
 594
 595
 596     if (!tokenizer || !tokenizer->bi
 597         || !tokenizer->buf16 || !tokenizer->buf16->utf16_len)
 598         return 0;
 599
 600     // never change tokenizer->buf16 and keep always invariant
 601     // 0 <= tokenizer->token_start
 602     //   <= tokenizer->token_end
 603     //   <= tokenizer->buf16->utf16_len
 604     // returns length of token
 605
 606     if (0 == tokenizer->token_end) // first call
 607         tkn_start = ubrk_first(tokenizer->bi);
 608     else //successive calls
 609         tkn_start = tokenizer->token_end;
 610
 611     // get next position
 612     tkn_end = ubrk_next(tokenizer->bi);
 613
 614     // repairing invariant at end of ubrk, which is UBRK_DONE = -1
 615     if (UBRK_DONE == tkn_end)
 616         tkn_end = tokenizer->buf16->utf16_len;
 617
 618     // copy out if everything is well
 619     if(U_FAILURE(*status))
 620         return 0;
 621
 622     // everything OK, now update internal state
 623     tkn_len = tkn_end - tkn_start;
 624
 625     if (0 < tkn_len){
 626         tokenizer->token_count++;
 627         tokenizer->token_id++;
 628     } else {
 629         tokenizer->token_id = 0;
 630     }
 631     tokenizer->token_start = tkn_start;
 632     tokenizer->token_end = tkn_end;
 633
 634
 635     // copying into token buffer if it exists
 636     if (tkn16){
 637         if (tkn16->utf16_cap < tkn_len)
 638             icu_buf_utf16_resize(tkn16, (size_t) tkn_len * 2);
 639
 640         u_strncpy(tkn16->utf16, &(tokenizer->buf16->utf16)[tkn_start],
 641                   tkn_len);
 642
 643         tkn16->utf16_len = tkn_len;
 644     }
 645
 646     return tkn_len;
 647 }
 648
 649
 650 int32_t icu_tokenizer_token_id(struct icu_tokenizer * tokenizer)
 651 {
 652     return tokenizer->token_id;
 653 };
 654
 655 int32_t icu_tokenizer_token_start(struct icu_tokenizer * tokenizer)
 656 {
 657     return tokenizer->token_start;
 658 };
 659
 660 int32_t icu_tokenizer_token_end(struct icu_tokenizer * tokenizer)
 661 {
 662     return tokenizer->token_end;
 663 };
 664
 665 int32_t icu_tokenizer_token_length(struct icu_tokenizer * tokenizer)
 666 {
 667     return (tokenizer->token_end - tokenizer->token_start);
 668 };
 669
 670 int32_t icu_tokenizer_token_count(struct icu_tokenizer * tokenizer)
 671 {
 672     return tokenizer->token_count;
 673 };
 674
 675
 676
 677 struct icu_normalizer * icu_normalizer_create(const char *rules, char action,
 678                                               UErrorCode *status)
 679 {
 680
 681     struct icu_normalizer * normalizer
 682         = (struct icu_normalizer *) malloc(sizeof(struct icu_normalizer));
 683
 684     normalizer->action = action;
 685     normalizer->trans = 0;
 686     normalizer->rules16 =  icu_buf_utf16_create(0);
 687     icu_utf16_from_utf8_cstr(normalizer->rules16, rules, status);
 688
 689     switch(normalizer->action) {
 690     case 'f':
 691         normalizer->trans
 692             = utrans_openU(normalizer->rules16->utf16,
 693                            normalizer->rules16->utf16_len,
 694                            UTRANS_FORWARD,
 695                            0, 0,
 696                            normalizer->parse_error, status);
 697         // yaz_log(YLOG_LOG, "utrans_open %p", normalizer->trans);
 698         break;
 699     case 'r':
 700         normalizer->trans
 701             = utrans_openU(normalizer->rules16->utf16,
 702                            normalizer->rules16->utf16_len,
 703                            UTRANS_REVERSE ,
 704                            0, 0,
 705                            normalizer->parse_error, status);
 706         // yaz_log(YLOG_LOG, "utrans_open %p", normalizer->trans);
 707         break;
 708     default:
 709         *status = U_UNSUPPORTED_ERROR;
 710         return 0;
 711         break;
 712     }
 713
 714     if (U_SUCCESS(*status))
 715         return normalizer;
 716
 717     // freeing if failed
 718     icu_normalizer_destroy(normalizer);
 719     return 0;
 720 };
 721
 722
 723 void icu_normalizer_destroy(struct icu_normalizer * normalizer){
 724     if (normalizer) {
 725         if (normalizer->rules16)
 726             icu_buf_utf16_destroy(normalizer->rules16);
 727         if (normalizer->trans)
 728         {
 729             // yaz_log(YLOG_LOG, "utrans_close %p", normalizer->trans);
 730             utrans_close(normalizer->trans);
 731         }
 732         free(normalizer);
 733     }
 734 };
 735
 736
 737
 738 int icu_normalizer_normalize(struct icu_normalizer * normalizer,
 739                              struct icu_buf_utf16 * dest16,
 740                              struct icu_buf_utf16 * src16,
 741                              UErrorCode *status)
 742 {
 743     if (!normalizer || !normalizer->trans || !src16 || !dest16)
 744         return 0;
 745
 746     if (!icu_buf_utf16_copy(dest16, src16))
 747         return 0;
 748
 749     utrans_transUChars (normalizer->trans,
 750                         dest16->utf16, &(dest16->utf16_len),
 751                         dest16->utf16_cap,
 752                         0, &(src16->utf16_len), status);
 753
 754     if (U_FAILURE(*status)){
 755         dest16->utf16[0] = (UChar) 0;
 756         dest16->utf16_len = 0;
 757     }
 758
 759     return dest16->utf16_len;
 760 }
 761
 762
 763
 764
 765 struct icu_chain_step * icu_chain_step_create(struct icu_chain * chain,
 766                                               enum icu_chain_step_type type,
 767                                               const uint8_t * rule,
 768                                               struct icu_buf_utf16 * buf16,
 769                                               UErrorCode *status)
 770 {
 771     struct icu_chain_step * step = 0;
 772
 773     if(!chain || !type || !rule)
 774         return 0;
 775
 776     step = (struct icu_chain_step *) malloc(sizeof(struct icu_chain_step));
 777
 778     step->type = type;
 779
 780     step->buf16 = buf16;
 781
 782     // create auxilary objects
 783     switch(step->type) {
 784     case ICU_chain_step_type_display:
 785         break;
 786     case ICU_chain_step_type_index:
 787         break;
 788     case ICU_chain_step_type_sortkey:
 789         break;
 790     case ICU_chain_step_type_casemap:
 791         step->u.casemap = icu_casemap_create((char *) chain->locale,
 792                                              (char) rule[0], status);
 793         break;
 794     case ICU_chain_step_type_normalize:
 795         step->u.normalizer = icu_normalizer_create((char *) rule, 'f', status);
 796         break;
 797     case ICU_chain_step_type_tokenize:
 798         step->u.tokenizer = icu_tokenizer_create((char *) chain->locale,
 799                                                  (char) rule[0], status);
 800         break;
 801     default:
 802         break;
 803     }
 804
 805     return step;
 806 };
 807
 808
 809 void icu_chain_step_destroy(struct icu_chain_step * step){
 810
 811     if (!step)
 812         return;
 813
 814     icu_chain_step_destroy(step->previous);
 815
 816     switch(step->type) {
 817     case ICU_chain_step_type_display:
 818         break;
 819     case ICU_chain_step_type_index:
 820         break;
 821     case ICU_chain_step_type_sortkey:
 822         break;
 823     case ICU_chain_step_type_casemap:
 824         icu_casemap_destroy(step->u.casemap);
 825         icu_buf_utf16_destroy(step->buf16);
 826         break;
 827     case ICU_chain_step_type_normalize:
 828         icu_normalizer_destroy(step->u.normalizer);
 829         icu_buf_utf16_destroy(step->buf16);
 830         break;
 831     case ICU_chain_step_type_tokenize:
 832         icu_tokenizer_destroy(step->u.tokenizer);
 833         icu_buf_utf16_destroy(step->buf16);
 834         break;
 835     default:
 836         break;
 837     }
 838     free(step);
 839 };
 840
 841
 842
 843 struct icu_chain * icu_chain_create(const uint8_t * identifier,
 844                                     const uint8_t * locale)
 845 {
 846
 847     struct icu_chain * chain
 848         = (struct icu_chain *) malloc(sizeof(struct icu_chain));
 849
 850     strncpy((char *) chain->identifier, (const char *) identifier, 128);
 851     chain->identifier[128 - 1] = '\0';
 852     strncpy((char *) chain->locale, (const char *) locale, 16);
 853     chain->locale[16 - 1] = '\0';
 854
 855     chain->token_count = 0;
 856
 857     chain->display8 = icu_buf_utf8_create(0);
 858     chain->norm8 = icu_buf_utf8_create(0);
 859     chain->sort8 = icu_buf_utf8_create(0);
 860
 861     chain->src16 = icu_buf_utf16_create(0);
 862
 863     chain->steps = 0;
 864
 865     return chain;
 866 };
 867
 868
 869 void icu_chain_destroy(struct icu_chain * chain)
 870 {
 871     if (chain){
 872         icu_buf_utf8_destroy(chain->display8);
 873         icu_buf_utf8_destroy(chain->norm8);
 874         icu_buf_utf8_destroy(chain->sort8);
 875
 876         icu_buf_utf16_destroy(chain->src16);
 877
 878         icu_chain_step_destroy(chain->steps);
 879         free(chain);
 880     }
 881 };
 882
 883
 884
 885 struct icu_chain * icu_chain_xml_config(xmlNode *xml_node,
 886                                         UErrorCode * status){
 887
 888     xmlNode *node = 0;
 889     struct icu_chain * chain = 0;
 890
 891     if (!xml_node
 892         ||xml_node->type != XML_ELEMENT_NODE
 893         || strcmp((const char *) xml_node->name, "icu_chain"))
 894
 895         return 0;
 896
 897     xmlChar *xml_id = xmlGetProp(xml_node, (xmlChar *) "id");
 898     xmlChar *xml_locale = xmlGetProp(xml_node, (xmlChar *) "locale");
 899
 900     if (!xml_id || !strlen((const char *) xml_id)
 901         || !xml_locale || !strlen((const char *) xml_locale))
 902         return 0;
 903
 904     chain = icu_chain_create((const uint8_t *) xml_id,
 905                              (const uint8_t *) xml_locale);
 906
 907     xmlFree(xml_id);
 908     xmlFree(xml_locale);
 909     if (!chain)
 910         return 0;
 911
 912     for (node = xml_node->children; node; node = node->next)
 913     {
 914         if (node->type != XML_ELEMENT_NODE)
 915             continue;
 916
 917         xmlChar *xml_rule = xmlGetProp(node, (xmlChar *) "rule");
 918         struct icu_chain_step * step = 0;
 919
 920         if (!strcmp((const char *) node->name,
 921                     (const char *) "casemap")){
 922             step = icu_chain_insert_step(chain, ICU_chain_step_type_casemap,
 923                                          (const uint8_t *) xml_rule, status);
 924         }
 925         else if (!strcmp((const char *) node->name,
 926                          (const char *) "normalize")){
 927             step = icu_chain_insert_step(chain, ICU_chain_step_type_normalize,
 928                                          (const uint8_t *) xml_rule, status);
 929         }
 930         else if (!strcmp((const char *) node->name,
 931                          (const char *) "tokenize")){
 932             step = icu_chain_insert_step(chain, ICU_chain_step_type_tokenize,
 933                                          (const uint8_t *) xml_rule, status);
 934         }
 935         else if (!strcmp((const char *) node->name,
 936                          (const char *) "display")){
 937             step = icu_chain_insert_step(chain, ICU_chain_step_type_display,
 938                                          (const uint8_t *) "", status);
 939         }
 940         else if (!strcmp((const char *) node->name,
 941                          (const char *) "index")){
 942             step = icu_chain_insert_step(chain, ICU_chain_step_type_index,
 943                                          (const uint8_t *) "", status);
 944         }
 945         else if (!strcmp((const char *) node->name,
 946                          (const char *) "sortkey")){
 947             step = icu_chain_insert_step(chain, ICU_chain_step_type_sortkey,
 948                                          (const uint8_t *) "", status);
 949         }
 950
 951         xmlFree(xml_rule);
 952         if (!step || U_FAILURE(*status)){
 953             icu_chain_destroy(chain);
 954             return 0;
 955         }
 956
 957
 958     }
 959
 960     return chain;
 961 };
 962
 963
 964
 965 struct icu_chain_step * icu_chain_insert_step(struct icu_chain * chain,
 966                                               enum icu_chain_step_type type,
 967                                               const uint8_t * rule,
 968                                               UErrorCode *status)
 969 {
 970     struct icu_chain_step * step = 0;
 971     struct icu_buf_utf16 * src16 = 0;
 972     struct icu_buf_utf16 * buf16 = 0;
 973
 974     if (!chain || !type || !rule)
 975         return 0;
 976
 977     // assign utf16 src buffers as needed
 978     if (chain->steps && chain->steps->buf16)
 979         src16 = chain->steps->buf16;
 980     else if (chain->src16)
 981         src16 = chain->src16;
 982     else
 983         return 0;
 984
 985
 986     // create utf16 destination buffers as needed, or
 987     switch(type) {
 988     case ICU_chain_step_type_display:
 989         buf16 = src16;
 990         break;
 991     case ICU_chain_step_type_index:
 992         buf16 = src16;
 993         break;
 994     case ICU_chain_step_type_sortkey:
 995         buf16 = src16;
 996         break;
 997     case ICU_chain_step_type_casemap:
 998         buf16 = icu_buf_utf16_create(0);
 999         break;
1000     case ICU_chain_step_type_normalize:
1001         buf16 = icu_buf_utf16_create(0);
1002         break;
1003     case ICU_chain_step_type_tokenize:
1004         buf16 = icu_buf_utf16_create(0);
1005         break;
1006     default:
1007         break;
1008     }
1009
1010     // create actual chain step with this buffer
1011     step = icu_chain_step_create(chain, type, rule, buf16, status);
1012
1013     step->previous = chain->steps;
1014     chain->steps = step;
1015
1016     return step;
1017 };
1018
1019
1020 int icu_chain_step_next_token(struct icu_chain * chain,
1021                               struct icu_chain_step * step,
1022                               UErrorCode *status)
1023 {
1024     struct icu_buf_utf16 * src16 = 0;
1025
1026     //printf("icu_chain_step_next_token %d\n", (int) step);
1027
1028     if (!chain || !chain->src16 || !step || !step->more_tokens)
1029         return 0;
1030
1031     // assign utf16 src buffers as neeed, advance in previous steps
1032     // tokens until non-zero token met, and setting stop condition
1033     if (step->previous){
1034         src16 = step->previous->buf16;
1035         if (step->need_new_token)
1036             //while (step->more_tokens &&  !src16->utf16_len)
1037                 step->more_tokens
1038                     = icu_chain_step_next_token(chain, step->previous, status);
1039     }
1040     else { // first step can only work once on chain->src16 input buffer
1041         src16 = chain->src16;
1042         step->more_tokens = 1;
1043     }
1044
1045     // stop if nothing to process
1046     // i.e new token source was not properly assigned
1047     if (!step->more_tokens || !src16) // || !src16->utf16_len
1048         return 0;
1049
1050     //printf("icu_chain_step_next_token %d working\n", (int) step);
1051
1052
1053     // perform the work, eventually put this steps output in
1054     // step->buf16 or the chains UTF8 output buffers
1055     switch(step->type) {
1056     case ICU_chain_step_type_display:
1057         icu_utf16_to_utf8(chain->display8, src16, status);
1058         break;
1059     case ICU_chain_step_type_index:
1060         icu_utf16_to_utf8(chain->norm8, src16, status);
1061         break;
1062     case ICU_chain_step_type_sortkey:
1063         icu_utf16_to_utf8(chain->sort8, src16, status);
1064         break;
1065     case ICU_chain_step_type_casemap:
1066         icu_casemap_casemap(step->u.casemap,
1067                             step->buf16, src16, status);
1068         break;
1069     case ICU_chain_step_type_normalize:
1070         icu_normalizer_normalize(step->u.normalizer,
1071                                  step->buf16, src16, status);
1072         break;
1073     case ICU_chain_step_type_tokenize:
1074         // attach to new src16 token only first time during splitting
1075         if (step->need_new_token){
1076             icu_tokenizer_attach(step->u.tokenizer, src16, status);
1077             step->need_new_token = 0;
1078         }
1079         // splitting one src16 token into multiple buf16 tokens
1080         step->more_tokens
1081             = icu_tokenizer_next_token(step->u.tokenizer,
1082                                        step->buf16, status);
1083         // make sure to get new previous token if this one had been used up
1084         if (step->previous && !step->more_tokens){
1085             if (icu_chain_step_next_token(chain, step->previous, status)){
1086                 icu_tokenizer_attach(step->u.tokenizer, src16, status);
1087                 step->need_new_token = 0;
1088                 step->more_tokens
1089                     = icu_tokenizer_next_token(step->u.tokenizer,
1090                                                step->buf16, status);
1091             }
1092         }
1093         if (0 == step->more_tokens)
1094             return 0;
1095         break;
1096     default:
1097         return 0;
1098         break;
1099     }
1100
1101
1102
1103     // stop further token processing if last step and
1104     // new tokens are needed from previous (non-existing) step
1105     if (!step->previous && step->need_new_token)
1106         step->more_tokens = 0;
1107
1108     //printf("%d %d %d\n",
1109     //       step->more_tokens, src16->utf16_len, step->buf16->utf16_len);
1110
1111
1112     if (U_FAILURE(*status))
1113         return 0;
1114
1115     return 1;
1116 };
1117
1118
1119
1120 int icu_chain_assign_cstr(struct icu_chain * chain,
1121                           const char * src8cstr,
1122                           UErrorCode *status)
1123 {
1124     struct icu_chain_step * stp = 0;
1125
1126     if (!chain || !src8cstr)
1127         return 0;
1128
1129     stp = chain->steps;
1130
1131     // clear token count
1132     chain->token_count = 0;
1133
1134     // clear all steps stop states
1135
1136     while (stp){
1137         stp->more_tokens = 1;
1138         stp->need_new_token = 1;
1139         stp = stp->previous;
1140     }
1141
1142     // finally convert UTF8 to UTF16 string
1143     icu_utf16_from_utf8_cstr(chain->src16, src8cstr, status);
1144
1145     if (U_FAILURE(*status))
1146         return 0;
1147
1148     return 1;
1149 };
1150
1151
1152
1153 int icu_chain_next_token(struct icu_chain * chain,
1154                          UErrorCode *status)
1155 {
1156     int success = 0;
1157
1158     if (!chain || !chain->steps)
1159         return 0;
1160
1161     success = icu_chain_step_next_token(chain, chain->steps, status);
1162
1163     if (success){
1164         chain->token_count++;
1165         return chain->token_count;
1166     }
1167
1168     return 0;
1169 };
1170
1171 int icu_chain_get_token_count(struct icu_chain * chain)
1172 {
1173     if (!chain)
1174         return 0;
1175
1176     return chain->token_count;
1177 };
1178
1179
1180
1181 const char * icu_chain_get_display(struct icu_chain * chain)
1182 {
1183     if (chain->display8)
1184         return icu_buf_utf8_to_cstr(chain->display8);
1185
1186     return 0;
1187 };
1188
1189 const char * icu_chain_get_norm(struct icu_chain * chain)
1190 {
1191     if (chain->norm8)
1192         return icu_buf_utf8_to_cstr(chain->norm8);
1193
1194     return 0;
1195 };
1196
1197 const char * icu_chain_get_sort(struct icu_chain * chain)
1198 {
1199     if (chain->sort8)
1200         return icu_buf_utf8_to_cstr(chain->sort8);
1201
1202     return 0;
1203 };
1204
1205
1206
1207
1208 #endif // HAVE_ICU
1209
1210
1211
1212
1213 /*
1214  * Local variables:
1215  * c-basic-offset: 4
1216  * indent-tabs-mode: nil
1217  * End:
1218  * vim: shiftwidth=4 tabstop=8 expandtab
1219  */