src/icu_I18N.c

   1 /*
   2  * Copyright (C) 1995-2007, Index Data ApS
   3  * See the file LICENSE for details.
   4  *
   5  * $Id: icu_I18N.c,v 1.21 2007-11-30 11:43:47 adam Exp $
   6  */
   7
   8 #if HAVE_CONFIG_H
   9 #include "config.h"
  10 #endif
  11
  12 #define USE_TIMING 0
  13 #if USE_TIMING
  14 #include <yaz/timing.h>
  15 #endif
  16
  17 #if YAZ_HAVE_ICU
  18 #include <yaz/xmalloc.h>
  19
  20 #include <yaz/icu_I18N.h>
  21
  22 #include <yaz/log.h>
  23
  24 #include <string.h>
  25 #include <stdlib.h>
  26 #include <stdio.h>
  27
  28 #include <unicode/ustring.h>  /* some more string fcns*/
  29 #include <unicode/uchar.h>    /* char names           */
  30
  31
  32 #include <unicode/ucol.h>
  33
  34
  35 int icu_check_status (UErrorCode status)
  36 {
  37     if (U_FAILURE(status))
  38     {
  39         yaz_log(YLOG_WARN, "ICU: %d %s\n", status, u_errorName(status));
  40         return 0;
  41     }
  42     return 1;
  43
  44 }
  45
  46
  47
  48 struct icu_buf_utf16 * icu_buf_utf16_create(size_t capacity)
  49 {
  50     struct icu_buf_utf16 * buf16
  51         = (struct icu_buf_utf16 *) xmalloc(sizeof(struct icu_buf_utf16));
  52
  53     buf16->utf16 = 0;
  54     buf16->utf16_len = 0;
  55     buf16->utf16_cap = 0;
  56
  57     if (capacity > 0){
  58         buf16->utf16 = (UChar *) xmalloc(sizeof(UChar) * capacity);
  59         buf16->utf16[0] = (UChar) 0;
  60         buf16->utf16_cap = capacity;
  61     }
  62     return buf16;
  63 }
  64
  65 struct icu_buf_utf16 * icu_buf_utf16_clear(struct icu_buf_utf16 * buf16)
  66 {
  67     if (buf16){
  68         if (buf16->utf16)
  69             buf16->utf16[0] = (UChar) 0;
  70         buf16->utf16_len = 0;
  71     }
  72     return buf16;
  73 }
  74
  75 struct icu_buf_utf16 * icu_buf_utf16_resize(struct icu_buf_utf16 * buf16,
  76                                             size_t capacity)
  77 {
  78     if (!buf16)
  79         return 0;
  80
  81     if (capacity >  0){
  82         if (0 == buf16->utf16)
  83             buf16->utf16 = (UChar *) xmalloc(sizeof(UChar) * capacity);
  84         else
  85             buf16->utf16
  86                 = (UChar *) xrealloc(buf16->utf16, sizeof(UChar) * capacity);
  87
  88         icu_buf_utf16_clear(buf16);
  89         buf16->utf16_cap = capacity;
  90     }
  91     else {
  92         xfree(buf16->utf16);
  93         buf16->utf16 = 0;
  94         buf16->utf16_len = 0;
  95         buf16->utf16_cap = 0;
  96     }
  97
  98     return buf16;
  99 }
 100
 101
 102 struct icu_buf_utf16 * icu_buf_utf16_copy(struct icu_buf_utf16 * dest16,
 103                                           struct icu_buf_utf16 * src16)
 104 {
 105     if(!dest16 || !src16
 106        || dest16 == src16)
 107         return 0;
 108
 109     if (dest16->utf16_cap < src16->utf16_len)
 110         icu_buf_utf16_resize(dest16, src16->utf16_len * 2);
 111
 112     u_strncpy(dest16->utf16, src16->utf16, src16->utf16_len);
 113     dest16->utf16_len = src16->utf16_len;
 114
 115     return dest16;
 116 }
 117
 118
 119 void icu_buf_utf16_destroy(struct icu_buf_utf16 * buf16)
 120 {
 121     if (buf16)
 122         xfree(buf16->utf16);
 123     xfree(buf16);
 124 }
 125
 126
 127
 128 struct icu_buf_utf8 * icu_buf_utf8_create(size_t capacity)
 129 {
 130     struct icu_buf_utf8 * buf8
 131         = (struct icu_buf_utf8 *) xmalloc(sizeof(struct icu_buf_utf8));
 132
 133     buf8->utf8 = 0;
 134     buf8->utf8_len = 0;
 135     buf8->utf8_cap = 0;
 136
 137     if (capacity > 0){
 138         buf8->utf8 = (uint8_t *) xmalloc(sizeof(uint8_t) * capacity);
 139         buf8->utf8[0] = (uint8_t) 0;
 140         buf8->utf8_cap = capacity;
 141     }
 142     return buf8;
 143 }
 144
 145
 146 struct icu_buf_utf8 * icu_buf_utf8_clear(struct icu_buf_utf8 * buf8)
 147 {
 148     if (buf8){
 149         if (buf8->utf8)
 150             buf8->utf8[0] = (uint8_t) 0;
 151         buf8->utf8_len = 0;
 152     }
 153     return buf8;
 154 }
 155
 156
 157 struct icu_buf_utf8 * icu_buf_utf8_resize(struct icu_buf_utf8 * buf8,
 158                                           size_t capacity)
 159 {
 160     if (!buf8)
 161         return 0;
 162
 163     if (capacity >  0){
 164         if (0 == buf8->utf8)
 165             buf8->utf8 = (uint8_t *) xmalloc(sizeof(uint8_t) * capacity);
 166         else
 167             buf8->utf8
 168                 = (uint8_t *) xrealloc(buf8->utf8, sizeof(uint8_t) * capacity);
 169
 170         icu_buf_utf8_clear(buf8);
 171         buf8->utf8_cap = capacity;
 172     }
 173     else {
 174         xfree(buf8->utf8);
 175         buf8->utf8 = 0;
 176         buf8->utf8_len = 0;
 177         buf8->utf8_cap = 0;
 178     }
 179
 180     return buf8;
 181 }
 182
 183
 184 struct icu_buf_utf8 * icu_buf_utf8_copy(struct icu_buf_utf8 * dest8,
 185                                           struct icu_buf_utf8 * src8)
 186 {
 187     if(!dest8 || !src8
 188        || dest8 == src8)
 189         return 0;
 190
 191
 192     if (dest8->utf8_cap < src8->utf8_len)
 193         icu_buf_utf8_resize(dest8, src8->utf8_len * 2);
 194
 195     strncpy((char*) dest8->utf8, (char*) src8->utf8, src8->utf8_len);
 196
 197     return dest8;
 198 }
 199
 200
 201 const char *icu_buf_utf8_to_cstr(struct icu_buf_utf8 *src8)
 202 {
 203     if (!src8 || src8->utf8_len == 0)
 204         return "";
 205
 206     if (src8->utf8_len == src8->utf8_cap)
 207         src8 = icu_buf_utf8_resize(src8, src8->utf8_len * 2 + 1);
 208
 209     src8->utf8[src8->utf8_len] = '\0';
 210
 211     return (const char *) src8->utf8;
 212 }
 213
 214
 215 void icu_buf_utf8_destroy(struct icu_buf_utf8 * buf8)
 216 {
 217     if (buf8)
 218         xfree(buf8->utf8);
 219     xfree(buf8);
 220 }
 221
 222
 223
 224 UErrorCode icu_utf16_from_utf8(struct icu_buf_utf16 * dest16,
 225                                struct icu_buf_utf8 * src8,
 226                                UErrorCode * status)
 227 {
 228     int32_t utf16_len = 0;
 229
 230     u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 231                   &utf16_len,
 232                   (const char *) src8->utf8, src8->utf8_len, status);
 233
 234     /* check for buffer overflow, resize and retry */
 235     if (*status == U_BUFFER_OVERFLOW_ERROR)
 236     {
 237         icu_buf_utf16_resize(dest16, utf16_len * 2);
 238         *status = U_ZERO_ERROR;
 239         u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 240                       &utf16_len,
 241                       (const char *) src8->utf8, src8->utf8_len, status);
 242     }
 243
 244     if (U_SUCCESS(*status)
 245         && utf16_len <= dest16->utf16_cap)
 246         dest16->utf16_len = utf16_len;
 247     else
 248         icu_buf_utf16_clear(dest16);
 249
 250     return *status;
 251 }
 252
 253
 254
 255 UErrorCode icu_utf16_from_utf8_cstr(struct icu_buf_utf16 * dest16,
 256                                     const char * src8cstr,
 257                                     UErrorCode * status)
 258 {
 259     size_t src8cstr_len = 0;
 260     int32_t utf16_len = 0;
 261
 262     *status = U_ZERO_ERROR;
 263     src8cstr_len = strlen(src8cstr);
 264
 265     u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 266                   &utf16_len,
 267                   src8cstr, src8cstr_len, status);
 268
 269     /* check for buffer overflow, resize and retry */
 270     if (*status == U_BUFFER_OVERFLOW_ERROR)
 271     {
 272         icu_buf_utf16_resize(dest16, utf16_len * 2);
 273         *status = U_ZERO_ERROR;
 274         u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 275                       &utf16_len,
 276                       src8cstr, src8cstr_len, status);
 277     }
 278
 279     if (U_SUCCESS(*status)
 280         && utf16_len <= dest16->utf16_cap)
 281         dest16->utf16_len = utf16_len;
 282     else
 283         icu_buf_utf16_clear(dest16);
 284
 285     return *status;
 286 }
 287
 288
 289
 290
 291 UErrorCode icu_utf16_to_utf8(struct icu_buf_utf8 * dest8,
 292                              struct icu_buf_utf16 * src16,
 293                              UErrorCode * status)
 294 {
 295     int32_t utf8_len = 0;
 296
 297     u_strToUTF8((char *) dest8->utf8, dest8->utf8_cap,
 298                 &utf8_len,
 299                 src16->utf16, src16->utf16_len, status);
 300
 301     /* check for buffer overflow, resize and retry */
 302     if (*status == U_BUFFER_OVERFLOW_ERROR)
 303     {
 304         icu_buf_utf8_resize(dest8, utf8_len * 2);
 305         *status = U_ZERO_ERROR;
 306         u_strToUTF8((char *) dest8->utf8, dest8->utf8_cap,
 307                     &utf8_len,
 308                     src16->utf16, src16->utf16_len, status);
 309
 310     }
 311
 312     if (U_SUCCESS(*status)
 313         && utf8_len <= dest8->utf8_cap)
 314         dest8->utf8_len = utf8_len;
 315     else
 316         icu_buf_utf8_clear(dest8);
 317
 318     return *status;
 319 }
 320
 321
 322
 323 struct icu_casemap * icu_casemap_create(char action, UErrorCode *status)
 324 {
 325     struct icu_casemap * casemap
 326         = (struct icu_casemap *) xmalloc(sizeof(struct icu_casemap));
 327     casemap->action = action;
 328
 329     switch(casemap->action) {
 330     case 'l':
 331     case 'L':
 332     case 'u':
 333     case 'U':
 334     case 't':
 335     case 'T':
 336     case 'f':
 337     case 'F':
 338         break;
 339     default:
 340         icu_casemap_destroy(casemap);
 341         return 0;
 342     }
 343
 344     return casemap;
 345 }
 346
 347 void icu_casemap_destroy(struct icu_casemap * casemap)
 348 {
 349     xfree(casemap);
 350 }
 351
 352
 353 int icu_casemap_casemap(struct icu_casemap * casemap,
 354                         struct icu_buf_utf16 * dest16,
 355                         struct icu_buf_utf16 * src16,
 356                         UErrorCode *status,
 357                         const char *locale)
 358 {
 359     if(!casemap)
 360         return 0;
 361
 362     return icu_utf16_casemap(dest16, src16, locale,
 363                              casemap->action, status);
 364 }
 365
 366
 367 int icu_utf16_casemap(struct icu_buf_utf16 * dest16,
 368                       struct icu_buf_utf16 * src16,
 369                       const char *locale, char action,
 370                       UErrorCode *status)
 371 {
 372     int32_t dest16_len = 0;
 373
 374
 375     if (!src16->utf16_len){           /* guarding for empty source string */
 376         if (dest16->utf16)
 377             dest16->utf16[0] = (UChar) 0;
 378         dest16->utf16_len = 0;
 379         return U_ZERO_ERROR;
 380     }
 381
 382
 383     switch(action) {
 384     case 'l':
 385     case 'L':
 386         dest16_len = u_strToLower(dest16->utf16, dest16->utf16_cap,
 387                                   src16->utf16, src16->utf16_len,
 388                                   locale, status);
 389         break;
 390     case 'u':
 391     case 'U':
 392         dest16_len = u_strToUpper(dest16->utf16, dest16->utf16_cap,
 393                                   src16->utf16, src16->utf16_len,
 394                                   locale, status);
 395         break;
 396     case 't':
 397     case 'T':
 398         dest16_len = u_strToTitle(dest16->utf16, dest16->utf16_cap,
 399                                   src16->utf16, src16->utf16_len,
 400                                   0, locale, status);
 401         break;
 402     case 'f':
 403     case 'F':
 404         dest16_len = u_strFoldCase(dest16->utf16, dest16->utf16_cap,
 405                                    src16->utf16, src16->utf16_len,
 406                                    U_FOLD_CASE_DEFAULT, status);
 407         break;
 408
 409     default:
 410         return U_UNSUPPORTED_ERROR;
 411         break;
 412     }
 413
 414     /* check for buffer overflow, resize and retry */
 415     if (*status == U_BUFFER_OVERFLOW_ERROR
 416         && dest16 != src16        /* do not resize if in-place conversion */
 417         ){
 418         icu_buf_utf16_resize(dest16, dest16_len * 2);
 419         *status = U_ZERO_ERROR;
 420
 421
 422         switch(action) {
 423         case 'l':
 424         case 'L':
 425             dest16_len = u_strToLower(dest16->utf16, dest16->utf16_cap,
 426                                       src16->utf16, src16->utf16_len,
 427                                       locale, status);
 428             break;
 429         case 'u':
 430         case 'U':
 431             dest16_len = u_strToUpper(dest16->utf16, dest16->utf16_cap,
 432                                       src16->utf16, src16->utf16_len,
 433                                       locale, status);
 434             break;
 435         case 't':
 436         case 'T':
 437             dest16_len = u_strToTitle(dest16->utf16, dest16->utf16_cap,
 438                                       src16->utf16, src16->utf16_len,
 439                                       0, locale, status);
 440             break;
 441         case 'f':
 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         if (dest16->utf16)
 459             dest16->utf16[0] = (UChar) 0;
 460         dest16->utf16_len = 0;
 461     }
 462
 463     return *status;
 464 }
 465
 466
 467
 468 void icu_sortkey8_from_utf16(UCollator *coll,
 469                              struct icu_buf_utf8 * dest8,
 470                              struct icu_buf_utf16 * src16,
 471                              UErrorCode * status)
 472 {
 473
 474     int32_t sortkey_len = 0;
 475
 476     sortkey_len = ucol_getSortKey(coll, src16->utf16, src16->utf16_len,
 477                                   dest8->utf8, dest8->utf8_cap);
 478
 479     /* check for buffer overflow, resize and retry */
 480     if (sortkey_len > dest8->utf8_cap) {
 481         icu_buf_utf8_resize(dest8, sortkey_len * 2);
 482         sortkey_len = ucol_getSortKey(coll, src16->utf16, src16->utf16_len,
 483                                       dest8->utf8, dest8->utf8_cap);
 484     }
 485
 486     if (U_SUCCESS(*status)
 487         && sortkey_len > 0)
 488         dest8->utf8_len = sortkey_len;
 489     else
 490         icu_buf_utf8_clear(dest8);
 491 }
 492
 493
 494
 495 struct icu_tokenizer * icu_tokenizer_create(const char *locale, char action,
 496                                             UErrorCode *status)
 497 {
 498     struct icu_tokenizer * tokenizer
 499         = (struct icu_tokenizer *) xmalloc(sizeof(struct icu_tokenizer));
 500
 501     tokenizer->action = action;
 502     tokenizer->bi = 0;
 503     tokenizer->buf16 = 0;
 504     tokenizer->token_count = 0;
 505     tokenizer->token_id = 0;
 506     tokenizer->token_start = 0;
 507     tokenizer->token_end = 0;
 508
 509
 510     switch(tokenizer->action) {
 511     case 'l':
 512     case 'L':
 513         tokenizer->bi = ubrk_open(UBRK_LINE, locale, 0, 0, status);
 514         break;
 515     case 's':
 516     case 'S':
 517         tokenizer->bi = ubrk_open(UBRK_SENTENCE, locale, 0, 0, status);
 518         break;
 519     case 'w':
 520     case 'W':
 521         tokenizer->bi = ubrk_open(UBRK_WORD, locale, 0, 0, status);
 522         break;
 523     case 'c':
 524     case 'C':
 525         tokenizer->bi = ubrk_open(UBRK_CHARACTER, locale, 0, 0, status);
 526         break;
 527     case 't':
 528     case 'T':
 529         tokenizer->bi = ubrk_open(UBRK_TITLE, locale, 0, 0, status);
 530         break;
 531     default:
 532         *status = U_UNSUPPORTED_ERROR;
 533         return 0;
 534         break;
 535     }
 536
 537     /* ICU error stuff is a very  funny business */
 538     if (U_SUCCESS(*status))
 539         return tokenizer;
 540
 541     /* freeing if failed */
 542     icu_tokenizer_destroy(tokenizer);
 543     return 0;
 544 }
 545
 546 void icu_tokenizer_destroy(struct icu_tokenizer * tokenizer)
 547 {
 548     if (tokenizer) {
 549         if (tokenizer->bi)
 550             ubrk_close(tokenizer->bi);
 551         xfree(tokenizer);
 552     }
 553 }
 554
 555 int icu_tokenizer_attach(struct icu_tokenizer * tokenizer,
 556                          struct icu_buf_utf16 * src16,
 557                          UErrorCode *status)
 558 {
 559     if (!tokenizer || !tokenizer->bi || !src16)
 560         return 0;
 561
 562
 563     tokenizer->buf16 = src16;
 564     tokenizer->token_count = 0;
 565     tokenizer->token_id = 0;
 566     tokenizer->token_start = 0;
 567     tokenizer->token_end = 0;
 568
 569     ubrk_setText(tokenizer->bi, src16->utf16, src16->utf16_len, status);
 570
 571
 572     if (U_FAILURE(*status))
 573         return 0;
 574
 575     return 1;
 576 };
 577
 578 int32_t icu_tokenizer_next_token(struct icu_tokenizer * tokenizer,
 579                          struct icu_buf_utf16 * tkn16,
 580                          UErrorCode *status)
 581 {
 582     int32_t tkn_start = 0;
 583     int32_t tkn_end = 0;
 584     int32_t tkn_len = 0;
 585
 586
 587     if (!tokenizer || !tokenizer->bi
 588         || !tokenizer->buf16 || !tokenizer->buf16->utf16_len)
 589         return 0;
 590
 591     /*
 592     never change tokenizer->buf16 and keep always invariant
 593     0 <= tokenizer->token_start
 594        <= tokenizer->token_end
 595        <= tokenizer->buf16->utf16_len
 596     returns length of token
 597     */
 598
 599     if (0 == tokenizer->token_end) /* first call */
 600         tkn_start = ubrk_first(tokenizer->bi);
 601     else /* successive calls */
 602         tkn_start = tokenizer->token_end;
 603
 604     /* get next position */
 605     tkn_end = ubrk_next(tokenizer->bi);
 606
 607     /* repairing invariant at end of ubrk, which is UBRK_DONE = -1 */
 608     if (UBRK_DONE == tkn_end)
 609         tkn_end = tokenizer->buf16->utf16_len;
 610
 611     /* copy out if everything is well */
 612     if(U_FAILURE(*status))
 613         return 0;
 614
 615     /* everything OK, now update internal state */
 616     tkn_len = tkn_end - tkn_start;
 617
 618     if (0 < tkn_len){
 619         tokenizer->token_count++;
 620         tokenizer->token_id++;
 621     } else {
 622         tokenizer->token_id = 0;
 623     }
 624     tokenizer->token_start = tkn_start;
 625     tokenizer->token_end = tkn_end;
 626
 627
 628     /* copying into token buffer if it exists */
 629     if (tkn16){
 630         if (tkn16->utf16_cap < tkn_len)
 631             icu_buf_utf16_resize(tkn16, (size_t) tkn_len * 2);
 632
 633         u_strncpy(tkn16->utf16, &(tokenizer->buf16->utf16)[tkn_start],
 634                   tkn_len);
 635
 636         tkn16->utf16_len = tkn_len;
 637     }
 638
 639     return tkn_len;
 640 }
 641
 642
 643 int32_t icu_tokenizer_token_id(struct icu_tokenizer * tokenizer)
 644 {
 645     return tokenizer->token_id;
 646 }
 647
 648 int32_t icu_tokenizer_token_start(struct icu_tokenizer * tokenizer)
 649 {
 650     return tokenizer->token_start;
 651 }
 652
 653 int32_t icu_tokenizer_token_end(struct icu_tokenizer * tokenizer)
 654 {
 655     return tokenizer->token_end;
 656 }
 657
 658 int32_t icu_tokenizer_token_length(struct icu_tokenizer * tokenizer)
 659 {
 660     return (tokenizer->token_end - tokenizer->token_start);
 661 }
 662
 663 int32_t icu_tokenizer_token_count(struct icu_tokenizer * tokenizer)
 664 {
 665     return tokenizer->token_count;
 666 }
 667
 668
 669
 670 struct icu_normalizer * icu_normalizer_create(const char *rules, char action,
 671                                               UErrorCode *status)
 672 {
 673
 674     struct icu_normalizer * normalizer
 675         = (struct icu_normalizer *) xmalloc(sizeof(struct icu_normalizer));
 676
 677     normalizer->action = action;
 678     normalizer->trans = 0;
 679     normalizer->rules16 =  icu_buf_utf16_create(0);
 680     icu_utf16_from_utf8_cstr(normalizer->rules16, rules, status);
 681
 682     switch(normalizer->action) {
 683     case 'f':
 684     case 'F':
 685         normalizer->trans
 686             = utrans_openU(normalizer->rules16->utf16,
 687                            normalizer->rules16->utf16_len,
 688                            UTRANS_FORWARD,
 689                            0, 0,
 690                            normalizer->parse_error, status);
 691         break;
 692     case 'r':
 693     case 'R':
 694         normalizer->trans
 695             = utrans_openU(normalizer->rules16->utf16,
 696                            normalizer->rules16->utf16_len,
 697                            UTRANS_REVERSE ,
 698                            0, 0,
 699                            normalizer->parse_error, status);
 700         break;
 701     default:
 702         *status = U_UNSUPPORTED_ERROR;
 703         return 0;
 704         break;
 705     }
 706
 707     if (U_SUCCESS(*status))
 708         return normalizer;
 709
 710     /* freeing if failed */
 711     icu_normalizer_destroy(normalizer);
 712     return 0;
 713 }
 714
 715
 716 void icu_normalizer_destroy(struct icu_normalizer * normalizer){
 717     if (normalizer) {
 718         if (normalizer->rules16)
 719             icu_buf_utf16_destroy(normalizer->rules16);
 720         if (normalizer->trans)
 721             utrans_close(normalizer->trans);
 722         xfree(normalizer);
 723     }
 724 }
 725
 726
 727
 728 int icu_normalizer_normalize(struct icu_normalizer * normalizer,
 729                              struct icu_buf_utf16 * dest16,
 730                              struct icu_buf_utf16 * src16,
 731                              UErrorCode *status)
 732 {
 733     if (!normalizer || !normalizer->trans
 734         || !src16
 735         || !dest16)
 736         return 0;
 737
 738     if (!src16->utf16_len){           /* guarding for empty source string */
 739         icu_buf_utf16_clear(dest16);
 740         return 0;
 741     }
 742
 743     if (!icu_buf_utf16_copy(dest16, src16))
 744         return 0;
 745
 746
 747     utrans_transUChars (normalizer->trans,
 748                         dest16->utf16, &(dest16->utf16_len),
 749                         dest16->utf16_cap,
 750                         0, &(src16->utf16_len), status);
 751
 752     if (U_FAILURE(*status))
 753         icu_buf_utf16_clear(dest16);
 754
 755     return dest16->utf16_len;
 756 }
 757
 758
 759
 760
 761 struct icu_chain_step * icu_chain_step_create(struct icu_chain * chain,
 762                                               enum icu_chain_step_type type,
 763                                               const uint8_t * rule,
 764                                               struct icu_buf_utf16 * buf16,
 765                                               UErrorCode *status)
 766 {
 767     struct icu_chain_step * step = 0;
 768
 769     if(!chain || !type || !rule)
 770         return 0;
 771
 772     step = (struct icu_chain_step *) xmalloc(sizeof(struct icu_chain_step));
 773
 774     step->type = type;
 775
 776     step->buf16 = buf16;
 777
 778     /* create auxilary objects */
 779     switch(step->type) {
 780     case ICU_chain_step_type_display:
 781         break;
 782     case ICU_chain_step_type_casemap:
 783         step->u.casemap = icu_casemap_create(rule[0], status);
 784         break;
 785     case ICU_chain_step_type_normalize:
 786         step->u.normalizer = icu_normalizer_create((char *) rule, 'f', status);
 787         break;
 788     case ICU_chain_step_type_tokenize:
 789         step->u.tokenizer = icu_tokenizer_create((char *) chain->locale,
 790                                                  (char) rule[0], status);
 791         break;
 792     default:
 793         break;
 794     }
 795
 796     return step;
 797 }
 798
 799
 800 void icu_chain_step_destroy(struct icu_chain_step * step){
 801
 802     if (!step)
 803         return;
 804
 805     icu_chain_step_destroy(step->previous);
 806
 807     switch(step->type) {
 808     case ICU_chain_step_type_display:
 809         break;
 810     case ICU_chain_step_type_casemap:
 811         icu_casemap_destroy(step->u.casemap);
 812         icu_buf_utf16_destroy(step->buf16);
 813         break;
 814     case ICU_chain_step_type_normalize:
 815         icu_normalizer_destroy(step->u.normalizer);
 816         icu_buf_utf16_destroy(step->buf16);
 817         break;
 818     case ICU_chain_step_type_tokenize:
 819         icu_tokenizer_destroy(step->u.tokenizer);
 820         icu_buf_utf16_destroy(step->buf16);
 821         break;
 822     default:
 823         break;
 824     }
 825     xfree(step);
 826 }
 827
 828
 829
 830 struct icu_chain * icu_chain_create(const char *locale,  int sort,
 831                                     UErrorCode * status)
 832 {
 833     struct icu_chain * chain
 834         = (struct icu_chain *) xmalloc(sizeof(struct icu_chain));
 835
 836     *status = U_ZERO_ERROR;
 837
 838     chain->locale = xstrdup(locale);
 839
 840     chain->sort = sort;
 841
 842     chain->coll = ucol_open((const char *) chain->locale, status);
 843
 844     if (U_FAILURE(*status))
 845         return 0;
 846
 847     chain->token_count = 0;
 848
 849     chain->src8cstr = 0;
 850
 851     chain->display8 = icu_buf_utf8_create(0);
 852     chain->norm8 = icu_buf_utf8_create(0);
 853     chain->sort8 = icu_buf_utf8_create(0);
 854
 855     chain->src16 = icu_buf_utf16_create(0);
 856
 857     chain->steps = 0;
 858
 859     return chain;
 860 }
 861
 862
 863 void icu_chain_destroy(struct icu_chain * chain)
 864 {
 865     if (chain)
 866     {
 867         if (chain->coll)
 868             ucol_close(chain->coll);
 869
 870         icu_buf_utf8_destroy(chain->display8);
 871         icu_buf_utf8_destroy(chain->norm8);
 872         icu_buf_utf8_destroy(chain->sort8);
 873
 874         icu_buf_utf16_destroy(chain->src16);
 875
 876         icu_chain_step_destroy(chain->steps);
 877         xfree(chain->locale);
 878         xfree(chain);
 879     }
 880 }
 881
 882
 883
 884 struct icu_chain * icu_chain_xml_config(const xmlNode *xml_node,
 885                                         int sort,
 886                                         UErrorCode * status)
 887 {
 888     xmlNode *node = 0;
 889     struct icu_chain * chain = 0;
 890
 891     *status = U_ZERO_ERROR;
 892
 893     if (!xml_node ||xml_node->type != XML_ELEMENT_NODE)
 894         return 0;
 895
 896     {
 897         xmlChar * xml_locale = xmlGetProp((xmlNode *) xml_node,
 898                                           (xmlChar *) "locale");
 899
 900         if (xml_locale)
 901         {
 902             chain = icu_chain_create((const char *) xml_locale, sort, status);
 903             xmlFree(xml_locale);
 904         }
 905
 906     }
 907     if (!chain)
 908         return 0;
 909
 910     for (node = xml_node->children; node; node = node->next)
 911     {
 912         xmlChar *xml_rule;
 913         struct icu_chain_step * step = 0;
 914
 915         if (node->type != XML_ELEMENT_NODE)
 916             continue;
 917
 918         xml_rule = xmlGetProp(node, (xmlChar *) "rule");
 919
 920         if (!strcmp((const char *) node->name, "casemap"))
 921             step = icu_chain_insert_step(chain, ICU_chain_step_type_casemap,
 922                                          (const uint8_t *) xml_rule, status);
 923         else if (!strcmp((const char *) node->name, "transform"))
 924             step = icu_chain_insert_step(chain, ICU_chain_step_type_normalize,
 925                                          (const uint8_t *) xml_rule, status);
 926         else if (!strcmp((const char *) node->name, "tokenize"))
 927             step = icu_chain_insert_step(chain, ICU_chain_step_type_tokenize,
 928                                          (const uint8_t *) xml_rule, status);
 929         else if (!strcmp((const char *) node->name, "display"))
 930             step = icu_chain_insert_step(chain, ICU_chain_step_type_display,
 931                                          (const uint8_t *) "", status);
 932         xmlFree(xml_rule);
 933         if (!step || U_FAILURE(*status))
 934         {
 935             icu_chain_destroy(chain);
 936             return 0;
 937         }
 938
 939
 940     }
 941     return chain;
 942 }
 943
 944
 945
 946 struct icu_chain_step * icu_chain_insert_step(struct icu_chain * chain,
 947                                               enum icu_chain_step_type type,
 948                                               const uint8_t * rule,
 949                                               UErrorCode *status)
 950 {
 951     struct icu_chain_step * step = 0;
 952     struct icu_buf_utf16 * src16 = 0;
 953     struct icu_buf_utf16 * buf16 = 0;
 954
 955     if (!chain || !type || !rule)
 956         return 0;
 957
 958     /* assign utf16 src buffers as needed */
 959     if (chain->steps && chain->steps->buf16)
 960         src16 = chain->steps->buf16;
 961     else if (chain->src16)
 962         src16 = chain->src16;
 963     else
 964         return 0;
 965
 966
 967     /* create utf16 destination buffers as needed, or */
 968     switch(type)
 969     {
 970     case ICU_chain_step_type_display:
 971         buf16 = src16;
 972         break;
 973     case ICU_chain_step_type_casemap:
 974         buf16 = icu_buf_utf16_create(0);
 975         break;
 976     case ICU_chain_step_type_normalize:
 977         buf16 = icu_buf_utf16_create(0);
 978         break;
 979     case ICU_chain_step_type_tokenize:
 980         buf16 = icu_buf_utf16_create(0);
 981         break;
 982     default:
 983         break;
 984     }
 985
 986     /* create actual chain step with this buffer */
 987     step = icu_chain_step_create(chain, type, rule, buf16, status);
 988
 989     step->previous = chain->steps;
 990     chain->steps = step;
 991
 992     return step;
 993 }
 994
 995
 996 int icu_chain_step_next_token(struct icu_chain * chain,
 997                               struct icu_chain_step * step,
 998                               UErrorCode *status)
 999 {
1000     struct icu_buf_utf16 * src16 = 0;
1001     int got_new_token = 0;
1002
1003     if (!chain || !chain->src16 || !step || !step->more_tokens)
1004         return 0;
1005
1006     /* assign utf16 src buffers as neeed, advance in previous steps
1007        tokens until non-zero token met, and setting stop condition */
1008
1009     if (step->previous)
1010     {
1011         src16 = step->previous->buf16;
1012         /* tokens might be killed in previous steps, therefore looping */
1013
1014         while (step->need_new_token
1015                && step->previous->more_tokens
1016                && !got_new_token)
1017             got_new_token
1018                 = icu_chain_step_next_token(chain, step->previous, status);
1019     }
1020     else
1021     { /* first step can only work once on chain->src16 input buffer */
1022         src16 = chain->src16;
1023         step->more_tokens = 0;
1024         got_new_token = 1;
1025     }
1026
1027     if (!src16)
1028         return 0;
1029
1030     /* stop if nothing to process */
1031     if (step->need_new_token && !got_new_token)
1032     {
1033         step->more_tokens = 0;
1034         return 0;
1035     }
1036
1037     /* either an old token not finished yet, or a new token, thus
1038        perform the work, eventually put this steps output in
1039        step->buf16 or the chains UTF8 output buffers  */
1040
1041     switch(step->type)
1042     {
1043     case ICU_chain_step_type_display:
1044         icu_utf16_to_utf8(chain->display8, src16, status);
1045         break;
1046     case ICU_chain_step_type_casemap:
1047         icu_casemap_casemap(step->u.casemap,
1048                             step->buf16, src16, status,
1049                             chain->locale);
1050         break;
1051     case ICU_chain_step_type_normalize:
1052         icu_normalizer_normalize(step->u.normalizer,
1053                                  step->buf16, src16, status);
1054         break;
1055     case ICU_chain_step_type_tokenize:
1056         /* attach to new src16 token only first time during splitting */
1057         if (step->need_new_token)
1058         {
1059             icu_tokenizer_attach(step->u.tokenizer, src16, status);
1060             step->need_new_token = 0;
1061         }
1062
1063         /* splitting one src16 token into multiple buf16 tokens */
1064         step->more_tokens
1065             = icu_tokenizer_next_token(step->u.tokenizer,
1066                                        step->buf16, status);
1067
1068         /* make sure to get new previous token if this one had been used up
1069            by recursive call to _same_ step */
1070
1071         if (!step->more_tokens)
1072         {
1073             step->more_tokens = icu_chain_step_next_token(chain, step, status);
1074             return step->more_tokens;  /* avoid one token count too much! */
1075         }
1076         break;
1077     default:
1078         return 0;
1079         break;
1080     }
1081
1082     if (U_FAILURE(*status))
1083         return 0;
1084
1085     /* if token disappered into thin air, tell caller */
1086     /* if (!step->buf16->utf16_len && !step->more_tokens) */
1087     /*    return 0; */
1088
1089     return 1;
1090 }
1091
1092
1093 int icu_chain_assign_cstr(struct icu_chain * chain,
1094                           const char * src8cstr,
1095                           UErrorCode *status)
1096 {
1097     struct icu_chain_step * stp = 0;
1098
1099     if (!chain || !src8cstr)
1100         return 0;
1101
1102     chain->src8cstr = src8cstr;
1103
1104     stp = chain->steps;
1105
1106     /* clear token count */
1107     chain->token_count = 0;
1108
1109     /* clear all steps stop states */
1110     while (stp)
1111     {
1112         stp->more_tokens = 1;
1113         stp->need_new_token = 1;
1114         stp = stp->previous;
1115     }
1116
1117     /* finally convert UTF8 to UTF16 string if needed */
1118     if (chain->steps || chain->sort)
1119         icu_utf16_from_utf8_cstr(chain->src16, chain->src8cstr, status);
1120
1121     if (U_FAILURE(*status))
1122         return 0;
1123
1124     return 1;
1125 }
1126
1127
1128
1129 int icu_chain_next_token(struct icu_chain * chain,
1130                          UErrorCode *status)
1131 {
1132     int got_token = 0;
1133
1134     *status = U_ZERO_ERROR;
1135
1136     if (!chain)
1137         return 0;
1138
1139     /* special case with no steps - same as index type binary */
1140     if (!chain->steps)
1141     {
1142         if (chain->token_count)
1143             return 0;
1144         else
1145         {
1146             chain->token_count++;
1147
1148             if (chain->sort)
1149                 icu_sortkey8_from_utf16(chain->coll,
1150                                         chain->sort8, chain->steps->buf16,
1151                                         status);
1152             return chain->token_count;
1153         }
1154     }
1155     /* usual case, one or more icu chain steps existing */
1156     else
1157     {
1158         while(!got_token && chain->steps && chain->steps->more_tokens)
1159             got_token = icu_chain_step_next_token(chain, chain->steps, status);
1160
1161         if (got_token)
1162         {
1163             chain->token_count++;
1164
1165             icu_utf16_to_utf8(chain->norm8, chain->steps->buf16, status);
1166
1167             if (chain->sort)
1168                 icu_sortkey8_from_utf16(chain->coll,
1169                                         chain->sort8, chain->steps->buf16,
1170                                         status);
1171
1172             return chain->token_count;
1173         }
1174     }
1175
1176     return 0;
1177 }
1178
1179 int icu_chain_token_number(struct icu_chain * chain)
1180 {
1181     if (!chain)
1182         return 0;
1183
1184     return chain->token_count;
1185 }
1186
1187
1188 const char * icu_chain_token_display(struct icu_chain * chain)
1189 {
1190     if (chain->display8)
1191         return icu_buf_utf8_to_cstr(chain->display8);
1192
1193     return 0;
1194 }
1195
1196 const char * icu_chain_token_norm(struct icu_chain * chain)
1197 {
1198     if (!chain->steps)
1199         return chain->src8cstr;
1200
1201     if (chain->norm8)
1202         return icu_buf_utf8_to_cstr(chain->norm8);
1203
1204     return 0;
1205 }
1206
1207 const char * icu_chain_token_sortkey(struct icu_chain * chain)
1208 {
1209     if (chain->sort8)
1210         return icu_buf_utf8_to_cstr(chain->sort8);
1211
1212     return 0;
1213 }
1214
1215 const UCollator * icu_chain_get_coll(struct icu_chain * chain)
1216 {
1217     return chain->coll;
1218 }
1219
1220 #endif /* YAZ_HAVE_ICU */
1221
1222 /*
1223  * Local variables:
1224  * c-basic-offset: 4
1225  * indent-tabs-mode: nil
1226  * End:
1227  * vim: shiftwidth=4 tabstop=8 expandtab
1228  */