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 1.12 2007-11-07 10:19:12 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
  18 #if HAVE_ICU
  19 #include <yaz/icu_I18N.h>
  20
  21 #include <yaz/log.h>
  22
  23 #include <string.h>
  24 #include <stdlib.h>
  25 #include <stdio.h>
  26
  27 #include <unicode/ustring.h>  /* some more string fcns*/
  28 #include <unicode/uchar.h>    /* char names           */
  29
  30
  31 #include <unicode/ucol.h>
  32
  33
  34 int icu_check_status (UErrorCode status)
  35 {
  36     if(U_FAILURE(status)){
  37         yaz_log(YLOG_WARN,
  38                 "ICU: %d %s\n", status, u_errorName(status));
  39         return 0;
  40     }
  41     return 1;
  42
  43 }
  44
  45
  46
  47 struct icu_buf_utf16 * icu_buf_utf16_create(size_t capacity)
  48 {
  49     struct icu_buf_utf16 * buf16
  50         = (struct icu_buf_utf16 *) malloc(sizeof(struct icu_buf_utf16));
  51
  52     buf16->utf16 = 0;
  53     buf16->utf16_len = 0;
  54     buf16->utf16_cap = 0;
  55
  56     if (capacity > 0){
  57         buf16->utf16 = (UChar *) malloc(sizeof(UChar) * capacity);
  58         buf16->utf16[0] = (UChar) 0;
  59         buf16->utf16_cap = capacity;
  60     }
  61     return buf16;
  62 }
  63
  64 struct icu_buf_utf16 * icu_buf_utf16_clear(struct icu_buf_utf16 * buf16)
  65 {
  66     if (buf16){
  67         if (buf16->utf16)
  68             buf16->utf16[0] = (UChar) 0;
  69         buf16->utf16_len = 0;
  70     }
  71     return buf16;
  72 }
  73
  74 struct icu_buf_utf16 * icu_buf_utf16_resize(struct icu_buf_utf16 * buf16,
  75                                             size_t capacity)
  76 {
  77     if (!buf16)
  78         return 0;
  79
  80     if (capacity >  0){
  81         if (0 == buf16->utf16)
  82             buf16->utf16 = (UChar *) malloc(sizeof(UChar) * capacity);
  83         else
  84             buf16->utf16
  85                 = (UChar *) realloc(buf16->utf16, sizeof(UChar) * capacity);
  86
  87         icu_buf_utf16_clear(buf16);
  88         buf16->utf16_cap = capacity;
  89     }
  90     else {
  91         if (buf16->utf16)
  92             free(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         if (buf16->utf16)
 123             free(buf16->utf16);
 124         free(buf16);
 125     }
 126 }
 127
 128
 129
 130
 131
 132
 133 struct icu_buf_utf8 * icu_buf_utf8_create(size_t capacity)
 134 {
 135     struct icu_buf_utf8 * buf8
 136         = (struct icu_buf_utf8 *) malloc(sizeof(struct icu_buf_utf8));
 137
 138     buf8->utf8 = 0;
 139     buf8->utf8_len = 0;
 140     buf8->utf8_cap = 0;
 141
 142     if (capacity > 0){
 143         buf8->utf8 = (uint8_t *) malloc(sizeof(uint8_t) * capacity);
 144         buf8->utf8[0] = (uint8_t) 0;
 145         buf8->utf8_cap = capacity;
 146     }
 147     return buf8;
 148 }
 149
 150
 151 struct icu_buf_utf8 * icu_buf_utf8_clear(struct icu_buf_utf8 * buf8)
 152 {
 153     if (buf8){
 154         if (buf8->utf8)
 155             buf8->utf8[0] = (uint8_t) 0;
 156         buf8->utf8_len = 0;
 157     }
 158     return buf8;
 159 }
 160
 161
 162 struct icu_buf_utf8 * icu_buf_utf8_resize(struct icu_buf_utf8 * buf8,
 163                                           size_t capacity)
 164 {
 165     if (!buf8)
 166         return 0;
 167
 168     if (capacity >  0){
 169         if (0 == buf8->utf8)
 170             buf8->utf8 = (uint8_t *) malloc(sizeof(uint8_t) * capacity);
 171         else
 172             buf8->utf8
 173                 = (uint8_t *) realloc(buf8->utf8, sizeof(uint8_t) * capacity);
 174
 175         icu_buf_utf8_clear(buf8);
 176         buf8->utf8_cap = capacity;
 177     }
 178     else {
 179         if (buf8->utf8)
 180             free(buf8->utf8);
 181         buf8->utf8 = 0;
 182         buf8->utf8_len = 0;
 183         buf8->utf8_cap = 0;
 184     }
 185
 186     return buf8;
 187 }
 188
 189
 190 struct icu_buf_utf8 * icu_buf_utf8_copy(struct icu_buf_utf8 * dest8,
 191                                           struct icu_buf_utf8 * src8)
 192 {
 193     if(!dest8 || !src8
 194        || dest8 == src8)
 195         return 0;
 196
 197
 198     if (dest8->utf8_cap < src8->utf8_len)
 199         icu_buf_utf8_resize(dest8, src8->utf8_len * 2);
 200
 201     strncpy((char*) dest8->utf8, (char*) src8->utf8, src8->utf8_len);
 202
 203     return dest8;
 204 }
 205
 206
 207 const char *icu_buf_utf8_to_cstr(struct icu_buf_utf8 *src8)
 208 {
 209     if (!src8 || src8->utf8_len == 0)
 210         return "";
 211
 212     if (src8->utf8_len == src8->utf8_cap)
 213         src8 = icu_buf_utf8_resize(src8, src8->utf8_len * 2 + 1);
 214
 215     src8->utf8[src8->utf8_len] = '\0';
 216
 217     return (const char *) src8->utf8;
 218 }
 219
 220
 221 void icu_buf_utf8_destroy(struct icu_buf_utf8 * buf8)
 222 {
 223     if (buf8){
 224         if (buf8->utf8)
 225             free(buf8->utf8);
 226         free(buf8);
 227     }
 228 }
 229
 230
 231
 232 UErrorCode icu_utf16_from_utf8(struct icu_buf_utf16 * dest16,
 233                                struct icu_buf_utf8 * src8,
 234                                UErrorCode * status)
 235 {
 236     int32_t utf16_len = 0;
 237
 238     u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 239                   &utf16_len,
 240                   (const char *) src8->utf8, src8->utf8_len, status);
 241
 242     /* check for buffer overflow, resize and retry */
 243     if (*status == U_BUFFER_OVERFLOW_ERROR)
 244     {
 245         icu_buf_utf16_resize(dest16, utf16_len * 2);
 246         *status = U_ZERO_ERROR;
 247         u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 248                       &utf16_len,
 249                       (const char *) src8->utf8, src8->utf8_len, status);
 250     }
 251
 252     if (U_SUCCESS(*status)
 253         && utf16_len <= dest16->utf16_cap)
 254         dest16->utf16_len = utf16_len;
 255     else
 256         icu_buf_utf16_clear(dest16);
 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     *status = U_ZERO_ERROR;
 271     src8cstr_len = strlen(src8cstr);
 272
 273     u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 274                   &utf16_len,
 275                   src8cstr, src8cstr_len, status);
 276
 277     /* check for buffer overflow, resize and retry */
 278     if (*status == U_BUFFER_OVERFLOW_ERROR)
 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 (U_SUCCESS(*status)
 288         && utf16_len <= dest16->utf16_cap)
 289         dest16->utf16_len = utf16_len;
 290     else
 291         icu_buf_utf16_clear(dest16);
 292
 293     return *status;
 294 }
 295
 296
 297
 298
 299 UErrorCode icu_utf16_to_utf8(struct icu_buf_utf8 * dest8,
 300                              struct icu_buf_utf16 * src16,
 301                              UErrorCode * status)
 302 {
 303     int32_t utf8_len = 0;
 304
 305     u_strToUTF8((char *) dest8->utf8, dest8->utf8_cap,
 306                 &utf8_len,
 307                 src16->utf16, src16->utf16_len, status);
 308
 309     /* check for buffer overflow, resize and retry */
 310     if (*status == U_BUFFER_OVERFLOW_ERROR)
 311     {
 312         icu_buf_utf8_resize(dest8, utf8_len * 2);
 313         *status = U_ZERO_ERROR;
 314         u_strToUTF8((char *) dest8->utf8, dest8->utf8_cap,
 315                     &utf8_len,
 316                     src16->utf16, src16->utf16_len, status);
 317
 318     }
 319
 320     if (U_SUCCESS(*status)
 321         && utf8_len <= dest8->utf8_cap)
 322         dest8->utf8_len = utf8_len;
 323     else
 324         icu_buf_utf8_clear(dest8);
 325
 326     return *status;
 327 }
 328
 329
 330
 331 struct icu_casemap * icu_casemap_create(const char *locale, char action,
 332                                         UErrorCode *status)
 333 {
 334     struct icu_casemap * casemap
 335         = (struct icu_casemap *) malloc(sizeof(struct icu_casemap));
 336     strcpy(casemap->locale, locale);
 337     casemap->action = action;
 338
 339     switch(casemap->action) {
 340     case 'l':
 341     case 'L':
 342     case 'u':
 343     case 'U':
 344     case 't':
 345     case 'T':
 346     case 'f':
 347     case 'F':
 348         break;
 349     default:
 350         icu_casemap_destroy(casemap);
 351         return 0;
 352     }
 353
 354     return casemap;
 355 }
 356
 357 void icu_casemap_destroy(struct icu_casemap * casemap)
 358 {
 359     if (casemap)
 360         free(casemap);
 361 }
 362
 363
 364 int icu_casemap_casemap(struct icu_casemap * casemap,
 365                         struct icu_buf_utf16 * dest16,
 366                         struct icu_buf_utf16 * src16,
 367                         UErrorCode *status)
 368 {
 369     if(!casemap)
 370         return 0;
 371
 372     return icu_utf16_casemap(dest16, src16,
 373                              casemap->locale, casemap->action, status);
 374 }
 375
 376
 377 int icu_utf16_casemap(struct icu_buf_utf16 * dest16,
 378                       struct icu_buf_utf16 * src16,
 379                       const char *locale, char action,
 380                       UErrorCode *status)
 381 {
 382     int32_t dest16_len = 0;
 383
 384
 385     if (!src16->utf16_len){           //guarding for empty source string
 386         if (dest16->utf16)
 387             dest16->utf16[0] = (UChar) 0;
 388         dest16->utf16_len = 0;
 389         return U_ZERO_ERROR;
 390     }
 391
 392
 393     switch(action) {
 394     case 'l':
 395     case 'L':
 396         dest16_len = u_strToLower(dest16->utf16, dest16->utf16_cap,
 397                                   src16->utf16, src16->utf16_len,
 398                                   locale, status);
 399         break;
 400     case 'u':
 401     case 'U':
 402         dest16_len = u_strToUpper(dest16->utf16, dest16->utf16_cap,
 403                                   src16->utf16, src16->utf16_len,
 404                                   locale, status);
 405         break;
 406     case 't':
 407     case 'T':
 408         dest16_len = u_strToTitle(dest16->utf16, dest16->utf16_cap,
 409                                   src16->utf16, src16->utf16_len,
 410                                   0, locale, status);
 411         break;
 412     case 'f':
 413     case 'F':
 414         dest16_len = u_strFoldCase(dest16->utf16, dest16->utf16_cap,
 415                                    src16->utf16, src16->utf16_len,
 416                                    U_FOLD_CASE_DEFAULT, status);
 417         break;
 418
 419     default:
 420         return U_UNSUPPORTED_ERROR;
 421         break;
 422     }
 423
 424     /* check for buffer overflow, resize and retry */
 425     if (*status == U_BUFFER_OVERFLOW_ERROR
 426         && dest16 != src16        /* do not resize if in-place conversion */
 427         ){
 428         icu_buf_utf16_resize(dest16, dest16_len * 2);
 429         *status = U_ZERO_ERROR;
 430
 431
 432         switch(action) {
 433         case 'l':
 434         case 'L':
 435             dest16_len = u_strToLower(dest16->utf16, dest16->utf16_cap,
 436                                       src16->utf16, src16->utf16_len,
 437                                       locale, status);
 438             break;
 439         case 'u':
 440         case 'U':
 441             dest16_len = u_strToUpper(dest16->utf16, dest16->utf16_cap,
 442                                       src16->utf16, src16->utf16_len,
 443                                       locale, status);
 444             break;
 445         case 't':
 446         case 'T':
 447             dest16_len = u_strToTitle(dest16->utf16, dest16->utf16_cap,
 448                                       src16->utf16, src16->utf16_len,
 449                                       0, locale, status);
 450             break;
 451         case 'f':
 452         case 'F':
 453             dest16_len = u_strFoldCase(dest16->utf16, dest16->utf16_cap,
 454                                        src16->utf16, src16->utf16_len,
 455                                        U_FOLD_CASE_DEFAULT, status);
 456             break;
 457
 458         default:
 459             return U_UNSUPPORTED_ERROR;
 460             break;
 461         }
 462     }
 463
 464     if (U_SUCCESS(*status)
 465         && dest16_len <= dest16->utf16_cap)
 466         dest16->utf16_len = dest16_len;
 467     else {
 468         if (dest16->utf16)
 469             dest16->utf16[0] = (UChar) 0;
 470         dest16->utf16_len = 0;
 471     }
 472
 473     return *status;
 474 }
 475
 476
 477
 478 UErrorCode icu_sortkey8_from_utf16(UCollator *coll,
 479                                    struct icu_buf_utf8 * dest8,
 480                                    struct icu_buf_utf16 * src16,
 481                                    UErrorCode * status)
 482 {
 483
 484     int32_t sortkey_len = 0;
 485
 486     sortkey_len = ucol_getSortKey(coll, src16->utf16, src16->utf16_len,
 487                                   dest8->utf8, dest8->utf8_cap);
 488
 489     /* check for buffer overflow, resize and retry */
 490     if (sortkey_len > dest8->utf8_cap) {
 491         icu_buf_utf8_resize(dest8, sortkey_len * 2);
 492         sortkey_len = ucol_getSortKey(coll, src16->utf16, src16->utf16_len,
 493                                       dest8->utf8, dest8->utf8_cap);
 494     }
 495
 496     if (U_SUCCESS(*status)
 497         && sortkey_len > 0)
 498         dest8->utf8_len = sortkey_len;
 499     else
 500         icu_buf_utf8_clear(dest8);
 501
 502     return sortkey_len;
 503 }
 504
 505
 506
 507 struct icu_tokenizer * icu_tokenizer_create(const char *locale, char action,
 508                                             UErrorCode *status)
 509 {
 510     struct icu_tokenizer * tokenizer
 511         = (struct icu_tokenizer *) malloc(sizeof(struct icu_tokenizer));
 512
 513     strcpy(tokenizer->locale, locale);
 514     tokenizer->action = action;
 515     tokenizer->bi = 0;
 516     tokenizer->buf16 = 0;
 517     tokenizer->token_count = 0;
 518     tokenizer->token_id = 0;
 519     tokenizer->token_start = 0;
 520     tokenizer->token_end = 0;
 521
 522
 523     switch(tokenizer->action) {
 524     case 'l':
 525     case 'L':
 526         tokenizer->bi
 527             = ubrk_open(UBRK_LINE, tokenizer->locale,
 528                         0, 0, status);
 529         break;
 530     case 's':
 531     case 'S':
 532         tokenizer->bi
 533             = ubrk_open(UBRK_SENTENCE, tokenizer->locale,
 534                         0, 0, status);
 535         break;
 536     case 'w':
 537     case 'W':
 538         tokenizer->bi
 539             = ubrk_open(UBRK_WORD, tokenizer->locale,
 540                         0, 0, status);
 541         break;
 542     case 'c':
 543     case 'C':
 544         tokenizer->bi
 545             = ubrk_open(UBRK_CHARACTER, tokenizer->locale,
 546                         0, 0, status);
 547         break;
 548     case 't':
 549     case 'T':
 550         tokenizer->bi
 551             = ubrk_open(UBRK_TITLE, tokenizer->locale,
 552                         0, 0, status);
 553         break;
 554     default:
 555         *status = U_UNSUPPORTED_ERROR;
 556         return 0;
 557         break;
 558     }
 559
 560     /* ICU error stuff is a very  funny business */
 561     if (U_SUCCESS(*status))
 562         return tokenizer;
 563
 564     /* freeing if failed */
 565     icu_tokenizer_destroy(tokenizer);
 566     return 0;
 567 }
 568
 569 void icu_tokenizer_destroy(struct icu_tokenizer * tokenizer)
 570 {
 571     if (tokenizer) {
 572         if (tokenizer->bi)
 573             ubrk_close(tokenizer->bi);
 574         free(tokenizer);
 575     }
 576 }
 577
 578 int icu_tokenizer_attach(struct icu_tokenizer * tokenizer,
 579                          struct icu_buf_utf16 * src16,
 580                          UErrorCode *status)
 581 {
 582     if (!tokenizer || !tokenizer->bi || !src16)
 583         return 0;
 584
 585
 586     tokenizer->buf16 = src16;
 587     tokenizer->token_count = 0;
 588     tokenizer->token_id = 0;
 589     tokenizer->token_start = 0;
 590     tokenizer->token_end = 0;
 591
 592     ubrk_setText(tokenizer->bi, src16->utf16, src16->utf16_len, status);
 593
 594
 595     if (U_FAILURE(*status))
 596         return 0;
 597
 598     return 1;
 599 };
 600
 601 int32_t icu_tokenizer_next_token(struct icu_tokenizer * tokenizer,
 602                          struct icu_buf_utf16 * tkn16,
 603                          UErrorCode *status)
 604 {
 605     int32_t tkn_start = 0;
 606     int32_t tkn_end = 0;
 607     int32_t tkn_len = 0;
 608
 609
 610     if (!tokenizer || !tokenizer->bi
 611         || !tokenizer->buf16 || !tokenizer->buf16->utf16_len)
 612         return 0;
 613
 614     /*
 615     never change tokenizer->buf16 and keep always invariant
 616     0 <= tokenizer->token_start
 617        <= tokenizer->token_end
 618        <= tokenizer->buf16->utf16_len
 619     returns length of token
 620     */
 621
 622     if (0 == tokenizer->token_end) /* first call */
 623         tkn_start = ubrk_first(tokenizer->bi);
 624     else /* successive calls */
 625         tkn_start = tokenizer->token_end;
 626
 627     /* get next position */
 628     tkn_end = ubrk_next(tokenizer->bi);
 629
 630     /* repairing invariant at end of ubrk, which is UBRK_DONE = -1 */
 631     if (UBRK_DONE == tkn_end)
 632         tkn_end = tokenizer->buf16->utf16_len;
 633
 634     /* copy out if everything is well */
 635     if(U_FAILURE(*status))
 636         return 0;
 637
 638     /* everything OK, now update internal state */
 639     tkn_len = tkn_end - tkn_start;
 640
 641     if (0 < tkn_len){
 642         tokenizer->token_count++;
 643         tokenizer->token_id++;
 644     } else {
 645         tokenizer->token_id = 0;
 646     }
 647     tokenizer->token_start = tkn_start;
 648     tokenizer->token_end = tkn_end;
 649
 650
 651     /* copying into token buffer if it exists */
 652     if (tkn16){
 653         if (tkn16->utf16_cap < tkn_len)
 654             icu_buf_utf16_resize(tkn16, (size_t) tkn_len * 2);
 655
 656         u_strncpy(tkn16->utf16, &(tokenizer->buf16->utf16)[tkn_start],
 657                   tkn_len);
 658
 659         tkn16->utf16_len = tkn_len;
 660     }
 661
 662     return tkn_len;
 663 }
 664
 665
 666 int32_t icu_tokenizer_token_id(struct icu_tokenizer * tokenizer)
 667 {
 668     return tokenizer->token_id;
 669 }
 670
 671 int32_t icu_tokenizer_token_start(struct icu_tokenizer * tokenizer)
 672 {
 673     return tokenizer->token_start;
 674 }
 675
 676 int32_t icu_tokenizer_token_end(struct icu_tokenizer * tokenizer)
 677 {
 678     return tokenizer->token_end;
 679 }
 680
 681 int32_t icu_tokenizer_token_length(struct icu_tokenizer * tokenizer)
 682 {
 683     return (tokenizer->token_end - tokenizer->token_start);
 684 }
 685
 686 int32_t icu_tokenizer_token_count(struct icu_tokenizer * tokenizer)
 687 {
 688     return tokenizer->token_count;
 689 }
 690
 691
 692
 693 struct icu_normalizer * icu_normalizer_create(const char *rules, char action,
 694                                               UErrorCode *status)
 695 {
 696
 697     struct icu_normalizer * normalizer
 698         = (struct icu_normalizer *) malloc(sizeof(struct icu_normalizer));
 699
 700     normalizer->action = action;
 701     normalizer->trans = 0;
 702     normalizer->rules16 =  icu_buf_utf16_create(0);
 703     icu_utf16_from_utf8_cstr(normalizer->rules16, rules, status);
 704
 705     switch(normalizer->action) {
 706     case 'f':
 707     case 'F':
 708         normalizer->trans
 709             = utrans_openU(normalizer->rules16->utf16,
 710                            normalizer->rules16->utf16_len,
 711                            UTRANS_FORWARD,
 712                            0, 0,
 713                            normalizer->parse_error, status);
 714         break;
 715     case 'r':
 716     case 'R':
 717         normalizer->trans
 718             = utrans_openU(normalizer->rules16->utf16,
 719                            normalizer->rules16->utf16_len,
 720                            UTRANS_REVERSE ,
 721                            0, 0,
 722                            normalizer->parse_error, status);
 723         break;
 724     default:
 725         *status = U_UNSUPPORTED_ERROR;
 726         return 0;
 727         break;
 728     }
 729
 730     if (U_SUCCESS(*status))
 731         return normalizer;
 732
 733     /* freeing if failed */
 734     icu_normalizer_destroy(normalizer);
 735     return 0;
 736 }
 737
 738
 739 void icu_normalizer_destroy(struct icu_normalizer * normalizer){
 740     if (normalizer) {
 741         if (normalizer->rules16)
 742             icu_buf_utf16_destroy(normalizer->rules16);
 743         if (normalizer->trans)
 744             utrans_close(normalizer->trans);
 745         free(normalizer);
 746     }
 747 }
 748
 749
 750
 751 int icu_normalizer_normalize(struct icu_normalizer * normalizer,
 752                              struct icu_buf_utf16 * dest16,
 753                              struct icu_buf_utf16 * src16,
 754                              UErrorCode *status)
 755 {
 756     if (!normalizer || !normalizer->trans
 757         || !src16
 758         || !dest16)
 759         return 0;
 760
 761     if (!src16->utf16_len){           //guarding for empty source string
 762         icu_buf_utf16_clear(dest16);
 763         return 0;
 764     }
 765
 766     if (!icu_buf_utf16_copy(dest16, src16))
 767         return 0;
 768
 769
 770     utrans_transUChars (normalizer->trans,
 771                         dest16->utf16, &(dest16->utf16_len),
 772                         dest16->utf16_cap,
 773                         0, &(src16->utf16_len), status);
 774
 775     if (U_FAILURE(*status))
 776         icu_buf_utf16_clear(dest16);
 777
 778     return dest16->utf16_len;
 779 }
 780
 781
 782
 783
 784 struct icu_chain_step * icu_chain_step_create(struct icu_chain * chain,
 785                                               enum icu_chain_step_type type,
 786                                               const uint8_t * rule,
 787                                               struct icu_buf_utf16 * buf16,
 788                                               UErrorCode *status)
 789 {
 790     struct icu_chain_step * step = 0;
 791
 792     if(!chain || !type || !rule)
 793         return 0;
 794
 795     step = (struct icu_chain_step *) malloc(sizeof(struct icu_chain_step));
 796
 797     step->type = type;
 798
 799     step->buf16 = buf16;
 800
 801     /* create auxilary objects */
 802     switch(step->type) {
 803     case ICU_chain_step_type_display:
 804         break;
 805     case ICU_chain_step_type_casemap:
 806         step->u.casemap = icu_casemap_create((char *) chain->locale,
 807                                              (char) rule[0], status);
 808         break;
 809     case ICU_chain_step_type_normalize:
 810         step->u.normalizer = icu_normalizer_create((char *) rule, 'f', status);
 811         break;
 812     case ICU_chain_step_type_tokenize:
 813         step->u.tokenizer = icu_tokenizer_create((char *) chain->locale,
 814                                                  (char) rule[0], status);
 815         break;
 816     default:
 817         break;
 818     }
 819
 820     return step;
 821 }
 822
 823
 824 void icu_chain_step_destroy(struct icu_chain_step * step){
 825
 826     if (!step)
 827         return;
 828
 829     icu_chain_step_destroy(step->previous);
 830
 831     switch(step->type) {
 832     case ICU_chain_step_type_display:
 833         break;
 834     case ICU_chain_step_type_casemap:
 835         icu_casemap_destroy(step->u.casemap);
 836         icu_buf_utf16_destroy(step->buf16);
 837         break;
 838     case ICU_chain_step_type_normalize:
 839         icu_normalizer_destroy(step->u.normalizer);
 840         icu_buf_utf16_destroy(step->buf16);
 841         break;
 842     case ICU_chain_step_type_tokenize:
 843         icu_tokenizer_destroy(step->u.tokenizer);
 844         icu_buf_utf16_destroy(step->buf16);
 845         break;
 846     default:
 847         break;
 848     }
 849     free(step);
 850 }
 851
 852
 853
 854 struct icu_chain * icu_chain_create(const char *locale,
 855                                     int sort,
 856                                     UErrorCode * status)
 857 {
 858
 859     struct icu_chain * chain
 860         = (struct icu_chain *) malloc(sizeof(struct icu_chain));
 861
 862     *status = U_ZERO_ERROR;
 863
 864     strncpy((char *) chain->locale, (const char *) locale, 16);
 865     chain->locale[16 - 1] = '\0';
 866
 867     chain->sort = sort;
 868
 869     chain->coll = ucol_open((const char *) chain->locale, status);
 870
 871     if (U_FAILURE(*status))
 872         return 0;
 873
 874
 875     chain->token_count = 0;
 876
 877     chain->src8cstr = 0;
 878
 879     chain->display8 = icu_buf_utf8_create(0);
 880     chain->norm8 = icu_buf_utf8_create(0);
 881     chain->sort8 = icu_buf_utf8_create(0);
 882
 883     chain->src16 = icu_buf_utf16_create(0);
 884
 885     chain->steps = 0;
 886
 887     return chain;
 888 }
 889
 890
 891 void icu_chain_destroy(struct icu_chain * chain)
 892 {
 893     if (chain){
 894
 895         if (chain->coll)
 896             ucol_close(chain->coll);
 897
 898         icu_buf_utf8_destroy(chain->display8);
 899         icu_buf_utf8_destroy(chain->norm8);
 900         icu_buf_utf8_destroy(chain->sort8);
 901
 902         icu_buf_utf16_destroy(chain->src16);
 903
 904         icu_chain_step_destroy(chain->steps);
 905         free(chain);
 906     }
 907 }
 908
 909
 910
 911 struct icu_chain * icu_chain_xml_config(const xmlNode *xml_node,
 912                                         const char *locale,
 913                                         int sort,
 914                                         UErrorCode * status){
 915
 916     xmlNode *node = 0;
 917     struct icu_chain * chain = 0;
 918
 919     *status = U_ZERO_ERROR;
 920
 921     if (!xml_node
 922         ||xml_node->type != XML_ELEMENT_NODE
 923         // || strcmp((const char *) xml_node->name, "icu_chain")
 924         )
 925
 926         return 0;
 927
 928         chain = icu_chain_create(locale, sort, status);
 929
 930     if (!chain)
 931         return 0;
 932
 933
 934     for (node = xml_node->children; node; node = node->next)
 935     {
 936         xmlChar *xml_rule;
 937         struct icu_chain_step * step = 0;
 938
 939         if (node->type != XML_ELEMENT_NODE)
 940             continue;
 941
 942         xml_rule = xmlGetProp(node, (xmlChar *) "rule");
 943
 944         if (!strcmp((const char *) node->name,
 945                     (const char *) "casemap")){
 946             step = icu_chain_insert_step(chain, ICU_chain_step_type_casemap,
 947                                          (const uint8_t *) xml_rule, status);
 948         }
 949         else if (!strcmp((const char *) node->name,
 950                          (const char *) "normalize")){
 951             step = icu_chain_insert_step(chain, ICU_chain_step_type_normalize,
 952                                          (const uint8_t *) xml_rule, status);
 953         }
 954         else if (!strcmp((const char *) node->name,
 955                          (const char *) "tokenize")){
 956             step = icu_chain_insert_step(chain, ICU_chain_step_type_tokenize,
 957                                          (const uint8_t *) xml_rule, status);
 958         }
 959         else if (!strcmp((const char *) node->name,
 960                          (const char *) "display")){
 961             step = icu_chain_insert_step(chain, ICU_chain_step_type_display,
 962                                          (const uint8_t *) "", status);
 963         }
 964         xmlFree(xml_rule);
 965         if (!step || U_FAILURE(*status)){
 966             icu_chain_destroy(chain);
 967             return 0;
 968         }
 969
 970
 971     }
 972
 973     return chain;
 974 }
 975
 976
 977
 978 struct icu_chain_step * icu_chain_insert_step(struct icu_chain * chain,
 979                                               enum icu_chain_step_type type,
 980                                               const uint8_t * rule,
 981                                               UErrorCode *status)
 982 {
 983     struct icu_chain_step * step = 0;
 984     struct icu_buf_utf16 * src16 = 0;
 985     struct icu_buf_utf16 * buf16 = 0;
 986
 987     if (!chain || !type || !rule)
 988         return 0;
 989
 990     /* assign utf16 src buffers as needed */
 991     if (chain->steps && chain->steps->buf16)
 992         src16 = chain->steps->buf16;
 993     else if (chain->src16)
 994         src16 = chain->src16;
 995     else
 996         return 0;
 997
 998
 999     /* create utf16 destination buffers as needed, or */
1000     switch(type) {
1001     case ICU_chain_step_type_display:
1002         buf16 = src16;
1003         break;
1004     case ICU_chain_step_type_casemap:
1005         buf16 = icu_buf_utf16_create(0);
1006         break;
1007     case ICU_chain_step_type_normalize:
1008         buf16 = icu_buf_utf16_create(0);
1009         break;
1010     case ICU_chain_step_type_tokenize:
1011         buf16 = icu_buf_utf16_create(0);
1012         break;
1013     default:
1014         break;
1015     }
1016
1017     /* create actual chain step with this buffer */
1018     step = icu_chain_step_create(chain, type, rule, buf16, status);
1019
1020     step->previous = chain->steps;
1021     chain->steps = step;
1022
1023     return step;
1024 }
1025
1026
1027 int icu_chain_step_next_token(struct icu_chain * chain,
1028                               struct icu_chain_step * step,
1029                               UErrorCode *status)
1030 {
1031     struct icu_buf_utf16 * src16 = 0;
1032     int got_new_token = 0;
1033
1034     if (!chain || !chain->src16 || !step || !step->more_tokens)
1035         return 0;
1036
1037     /* assign utf16 src buffers as neeed, advance in previous steps
1038        tokens until non-zero token met, and setting stop condition */
1039
1040     if (step->previous){
1041         src16 = step->previous->buf16;
1042         /* tokens might be killed in previous steps, therefore looping */
1043
1044         while (step->need_new_token
1045                && step->previous->more_tokens
1046                && !got_new_token)
1047             got_new_token
1048                 = icu_chain_step_next_token(chain, step->previous, status);
1049     }
1050     else { /* first step can only work once on chain->src16 input buffer */
1051         src16 = chain->src16;
1052         step->more_tokens = 0;
1053         got_new_token = 1;
1054     }
1055
1056     if (!src16)
1057         return 0;
1058
1059     /* stop if nothing to process */
1060     if (step->need_new_token && !got_new_token){
1061         step->more_tokens = 0;
1062         return 0;
1063     }
1064
1065     /* either an old token not finished yet, or a new token, thus
1066        perform the work, eventually put this steps output in
1067        step->buf16 or the chains UTF8 output buffers  */
1068
1069     switch(step->type) {
1070     case ICU_chain_step_type_display:
1071         icu_utf16_to_utf8(chain->display8, src16, status);
1072         break;
1073     case ICU_chain_step_type_casemap:
1074         icu_casemap_casemap(step->u.casemap,
1075                             step->buf16, src16, status);
1076         break;
1077     case ICU_chain_step_type_normalize:
1078         icu_normalizer_normalize(step->u.normalizer,
1079                                  step->buf16, src16, status);
1080         break;
1081     case ICU_chain_step_type_tokenize:
1082         /* attach to new src16 token only first time during splitting */
1083         if (step->need_new_token){
1084             icu_tokenizer_attach(step->u.tokenizer, src16, status);
1085             step->need_new_token = 0;
1086         }
1087
1088
1089         /* splitting one src16 token into multiple buf16 tokens */
1090         step->more_tokens
1091             = icu_tokenizer_next_token(step->u.tokenizer,
1092                                        step->buf16, status);
1093
1094         /* make sure to get new previous token if this one had been used up
1095            by recursive call to _same_ step */
1096
1097         if (!step->more_tokens){
1098             step->more_tokens = icu_chain_step_next_token(chain, step, status);
1099             return step->more_tokens;  // avoid one token count too much!
1100         }
1101
1102         break;
1103     default:
1104         return 0;
1105         break;
1106     }
1107
1108     if (U_FAILURE(*status))
1109         return 0;
1110
1111     /* if token disappered into thin air, tell caller */
1112     /* if (!step->buf16->utf16_len && !step->more_tokens) */
1113     /*    return 0; */
1114
1115     return 1;
1116 }
1117
1118
1119 int icu_chain_assign_cstr(struct icu_chain * chain,
1120                           const char * src8cstr,
1121                           UErrorCode *status)
1122 {
1123     struct icu_chain_step * stp = 0;
1124
1125     if (!chain || !src8cstr)
1126         return 0;
1127
1128     chain->src8cstr = src8cstr;
1129
1130     stp = chain->steps;
1131
1132     /* clear token count */
1133     chain->token_count = 0;
1134
1135     /* clear all steps stop states */
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 if needed */
1143     if (chain->steps || chain->sort)
1144         icu_utf16_from_utf8_cstr(chain->src16, chain->src8cstr, status);
1145
1146     if (U_FAILURE(*status))
1147         return 0;
1148
1149     return 1;
1150 }
1151
1152
1153
1154 int icu_chain_next_token(struct icu_chain * chain,
1155                          UErrorCode *status)
1156 {
1157     int got_token = 0;
1158
1159     *status = U_ZERO_ERROR;
1160
1161     if (!chain)
1162         return 0;
1163
1164     /* special case with no steps - same as index type binary */
1165     if (!chain->steps){
1166         if (chain->token_count)
1167             return 0;
1168         else {
1169             chain->token_count++;
1170
1171             if (chain->sort)
1172                 icu_sortkey8_from_utf16(chain->coll,
1173                                         chain->sort8, chain->steps->buf16,
1174                                         status);
1175             return chain->token_count;
1176         }
1177     }
1178     /* usual case, one or more icu chain steps existing */
1179     else {
1180
1181         while(!got_token && chain->steps && chain->steps->more_tokens)
1182             got_token = icu_chain_step_next_token(chain, chain->steps, status);
1183
1184         if (got_token){
1185             chain->token_count++;
1186
1187             icu_utf16_to_utf8(chain->norm8, chain->steps->buf16, status);
1188
1189             if (chain->sort)
1190                 icu_sortkey8_from_utf16(chain->coll,
1191                                         chain->sort8, chain->steps->buf16,
1192                                         status);
1193
1194             return chain->token_count;
1195         }
1196     }
1197
1198     return 0;
1199 }
1200
1201 int icu_chain_token_number(struct icu_chain * chain)
1202 {
1203     if (!chain)
1204         return 0;
1205
1206     return chain->token_count;
1207 }
1208
1209
1210 const char * icu_chain_token_display(struct icu_chain * chain)
1211 {
1212     if (chain->display8)
1213         return icu_buf_utf8_to_cstr(chain->display8);
1214
1215     return 0;
1216 }
1217
1218 const char * icu_chain_token_norm(struct icu_chain * chain)
1219 {
1220     if (!chain->steps)
1221         return chain->src8cstr;
1222
1223     if (chain->norm8)
1224         return icu_buf_utf8_to_cstr(chain->norm8);
1225
1226     return 0;
1227 }
1228
1229 const char * icu_chain_token_sortkey(struct icu_chain * chain)
1230 {
1231     if (chain->sort8)
1232         return icu_buf_utf8_to_cstr(chain->sort8);
1233
1234     return 0;
1235 }
1236
1237 const UCollator * icu_chain_get_coll(struct icu_chain * chain)
1238 {
1239     return chain->coll;
1240 }
1241
1242
1243
1244
1245 #endif /* HAVE_ICU */
1246
1247
1248
1249
1250 /*
1251  * Local variables:
1252  * c-basic-offset: 4
1253  * indent-tabs-mode: nil
1254  * End:
1255  * vim: shiftwidth=4 tabstop=8 expandtab
1256  */