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 1.15 2007-11-08 17:08:11 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 YAZ_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(char action, UErrorCode *status)
 332 {
 333     struct icu_casemap * casemap
 334         = (struct icu_casemap *) malloc(sizeof(struct icu_casemap));
 335     casemap->action = action;
 336
 337     switch(casemap->action) {
 338     case 'l':
 339     case 'L':
 340     case 'u':
 341     case 'U':
 342     case 't':
 343     case 'T':
 344     case 'f':
 345     case 'F':
 346         break;
 347     default:
 348         icu_casemap_destroy(casemap);
 349         return 0;
 350     }
 351
 352     return casemap;
 353 }
 354
 355 void icu_casemap_destroy(struct icu_casemap * casemap)
 356 {
 357     if (casemap)
 358         free(casemap);
 359 }
 360
 361
 362 int icu_casemap_casemap(struct icu_casemap * casemap,
 363                         struct icu_buf_utf16 * dest16,
 364                         struct icu_buf_utf16 * src16,
 365                         UErrorCode *status,
 366                         const char *locale)
 367 {
 368     if(!casemap)
 369         return 0;
 370
 371     return icu_utf16_casemap(dest16, src16, locale,
 372                              casemap->action, status);
 373 }
 374
 375
 376 int icu_utf16_casemap(struct icu_buf_utf16 * dest16,
 377                       struct icu_buf_utf16 * src16,
 378                       const char *locale, char action,
 379                       UErrorCode *status)
 380 {
 381     int32_t dest16_len = 0;
 382
 383
 384     if (!src16->utf16_len){           //guarding for empty source string
 385         if (dest16->utf16)
 386             dest16->utf16[0] = (UChar) 0;
 387         dest16->utf16_len = 0;
 388         return U_ZERO_ERROR;
 389     }
 390
 391
 392     switch(action) {
 393     case 'l':
 394     case 'L':
 395         dest16_len = u_strToLower(dest16->utf16, dest16->utf16_cap,
 396                                   src16->utf16, src16->utf16_len,
 397                                   locale, status);
 398         break;
 399     case 'u':
 400     case 'U':
 401         dest16_len = u_strToUpper(dest16->utf16, dest16->utf16_cap,
 402                                   src16->utf16, src16->utf16_len,
 403                                   locale, status);
 404         break;
 405     case 't':
 406     case 'T':
 407         dest16_len = u_strToTitle(dest16->utf16, dest16->utf16_cap,
 408                                   src16->utf16, src16->utf16_len,
 409                                   0, locale, status);
 410         break;
 411     case 'f':
 412     case 'F':
 413         dest16_len = u_strFoldCase(dest16->utf16, dest16->utf16_cap,
 414                                    src16->utf16, src16->utf16_len,
 415                                    U_FOLD_CASE_DEFAULT, status);
 416         break;
 417
 418     default:
 419         return U_UNSUPPORTED_ERROR;
 420         break;
 421     }
 422
 423     /* check for buffer overflow, resize and retry */
 424     if (*status == U_BUFFER_OVERFLOW_ERROR
 425         && dest16 != src16        /* do not resize if in-place conversion */
 426         ){
 427         icu_buf_utf16_resize(dest16, dest16_len * 2);
 428         *status = U_ZERO_ERROR;
 429
 430
 431         switch(action) {
 432         case 'l':
 433         case 'L':
 434             dest16_len = u_strToLower(dest16->utf16, dest16->utf16_cap,
 435                                       src16->utf16, src16->utf16_len,
 436                                       locale, status);
 437             break;
 438         case 'u':
 439         case 'U':
 440             dest16_len = u_strToUpper(dest16->utf16, dest16->utf16_cap,
 441                                       src16->utf16, src16->utf16_len,
 442                                       locale, status);
 443             break;
 444         case 't':
 445         case 'T':
 446             dest16_len = u_strToTitle(dest16->utf16, dest16->utf16_cap,
 447                                       src16->utf16, src16->utf16_len,
 448                                       0, locale, status);
 449             break;
 450         case 'f':
 451         case 'F':
 452             dest16_len = u_strFoldCase(dest16->utf16, dest16->utf16_cap,
 453                                        src16->utf16, src16->utf16_len,
 454                                        U_FOLD_CASE_DEFAULT, status);
 455             break;
 456
 457         default:
 458             return U_UNSUPPORTED_ERROR;
 459             break;
 460         }
 461     }
 462
 463     if (U_SUCCESS(*status)
 464         && dest16_len <= dest16->utf16_cap)
 465         dest16->utf16_len = dest16_len;
 466     else {
 467         if (dest16->utf16)
 468             dest16->utf16[0] = (UChar) 0;
 469         dest16->utf16_len = 0;
 470     }
 471
 472     return *status;
 473 }
 474
 475
 476
 477 UErrorCode icu_sortkey8_from_utf16(UCollator *coll,
 478                                    struct icu_buf_utf8 * dest8,
 479                                    struct icu_buf_utf16 * src16,
 480                                    UErrorCode * status)
 481 {
 482
 483     int32_t sortkey_len = 0;
 484
 485     sortkey_len = ucol_getSortKey(coll, src16->utf16, src16->utf16_len,
 486                                   dest8->utf8, dest8->utf8_cap);
 487
 488     /* check for buffer overflow, resize and retry */
 489     if (sortkey_len > dest8->utf8_cap) {
 490         icu_buf_utf8_resize(dest8, sortkey_len * 2);
 491         sortkey_len = ucol_getSortKey(coll, src16->utf16, src16->utf16_len,
 492                                       dest8->utf8, dest8->utf8_cap);
 493     }
 494
 495     if (U_SUCCESS(*status)
 496         && sortkey_len > 0)
 497         dest8->utf8_len = sortkey_len;
 498     else
 499         icu_buf_utf8_clear(dest8);
 500
 501     return sortkey_len;
 502 }
 503
 504
 505
 506 struct icu_tokenizer * icu_tokenizer_create(const char *locale, char action,
 507                                             UErrorCode *status)
 508 {
 509     struct icu_tokenizer * tokenizer
 510         = (struct icu_tokenizer *) malloc(sizeof(struct icu_tokenizer));
 511
 512     strcpy(tokenizer->locale, locale);
 513     tokenizer->action = action;
 514     tokenizer->bi = 0;
 515     tokenizer->buf16 = 0;
 516     tokenizer->token_count = 0;
 517     tokenizer->token_id = 0;
 518     tokenizer->token_start = 0;
 519     tokenizer->token_end = 0;
 520
 521
 522     switch(tokenizer->action) {
 523     case 'l':
 524     case 'L':
 525         tokenizer->bi
 526             = ubrk_open(UBRK_LINE, tokenizer->locale,
 527                         0, 0, status);
 528         break;
 529     case 's':
 530     case 'S':
 531         tokenizer->bi
 532             = ubrk_open(UBRK_SENTENCE, tokenizer->locale,
 533                         0, 0, status);
 534         break;
 535     case 'w':
 536     case 'W':
 537         tokenizer->bi
 538             = ubrk_open(UBRK_WORD, tokenizer->locale,
 539                         0, 0, status);
 540         break;
 541     case 'c':
 542     case 'C':
 543         tokenizer->bi
 544             = ubrk_open(UBRK_CHARACTER, tokenizer->locale,
 545                         0, 0, status);
 546         break;
 547     case 't':
 548     case 'T':
 549         tokenizer->bi
 550             = ubrk_open(UBRK_TITLE, tokenizer->locale,
 551                         0, 0, status);
 552         break;
 553     default:
 554         *status = U_UNSUPPORTED_ERROR;
 555         return 0;
 556         break;
 557     }
 558
 559     /* ICU error stuff is a very  funny business */
 560     if (U_SUCCESS(*status))
 561         return tokenizer;
 562
 563     /* freeing if failed */
 564     icu_tokenizer_destroy(tokenizer);
 565     return 0;
 566 }
 567
 568 void icu_tokenizer_destroy(struct icu_tokenizer * tokenizer)
 569 {
 570     if (tokenizer) {
 571         if (tokenizer->bi)
 572             ubrk_close(tokenizer->bi);
 573         free(tokenizer);
 574     }
 575 }
 576
 577 int icu_tokenizer_attach(struct icu_tokenizer * tokenizer,
 578                          struct icu_buf_utf16 * src16,
 579                          UErrorCode *status)
 580 {
 581     if (!tokenizer || !tokenizer->bi || !src16)
 582         return 0;
 583
 584
 585     tokenizer->buf16 = src16;
 586     tokenizer->token_count = 0;
 587     tokenizer->token_id = 0;
 588     tokenizer->token_start = 0;
 589     tokenizer->token_end = 0;
 590
 591     ubrk_setText(tokenizer->bi, src16->utf16, src16->utf16_len, status);
 592
 593
 594     if (U_FAILURE(*status))
 595         return 0;
 596
 597     return 1;
 598 };
 599
 600 int32_t icu_tokenizer_next_token(struct icu_tokenizer * tokenizer,
 601                          struct icu_buf_utf16 * tkn16,
 602                          UErrorCode *status)
 603 {
 604     int32_t tkn_start = 0;
 605     int32_t tkn_end = 0;
 606     int32_t tkn_len = 0;
 607
 608
 609     if (!tokenizer || !tokenizer->bi
 610         || !tokenizer->buf16 || !tokenizer->buf16->utf16_len)
 611         return 0;
 612
 613     /*
 614     never change tokenizer->buf16 and keep always invariant
 615     0 <= tokenizer->token_start
 616        <= tokenizer->token_end
 617        <= tokenizer->buf16->utf16_len
 618     returns length of token
 619     */
 620
 621     if (0 == tokenizer->token_end) /* first call */
 622         tkn_start = ubrk_first(tokenizer->bi);
 623     else /* successive calls */
 624         tkn_start = tokenizer->token_end;
 625
 626     /* get next position */
 627     tkn_end = ubrk_next(tokenizer->bi);
 628
 629     /* repairing invariant at end of ubrk, which is UBRK_DONE = -1 */
 630     if (UBRK_DONE == tkn_end)
 631         tkn_end = tokenizer->buf16->utf16_len;
 632
 633     /* copy out if everything is well */
 634     if(U_FAILURE(*status))
 635         return 0;
 636
 637     /* everything OK, now update internal state */
 638     tkn_len = tkn_end - tkn_start;
 639
 640     if (0 < tkn_len){
 641         tokenizer->token_count++;
 642         tokenizer->token_id++;
 643     } else {
 644         tokenizer->token_id = 0;
 645     }
 646     tokenizer->token_start = tkn_start;
 647     tokenizer->token_end = tkn_end;
 648
 649
 650     /* copying into token buffer if it exists */
 651     if (tkn16){
 652         if (tkn16->utf16_cap < tkn_len)
 653             icu_buf_utf16_resize(tkn16, (size_t) tkn_len * 2);
 654
 655         u_strncpy(tkn16->utf16, &(tokenizer->buf16->utf16)[tkn_start],
 656                   tkn_len);
 657
 658         tkn16->utf16_len = tkn_len;
 659     }
 660
 661     return tkn_len;
 662 }
 663
 664
 665 int32_t icu_tokenizer_token_id(struct icu_tokenizer * tokenizer)
 666 {
 667     return tokenizer->token_id;
 668 }
 669
 670 int32_t icu_tokenizer_token_start(struct icu_tokenizer * tokenizer)
 671 {
 672     return tokenizer->token_start;
 673 }
 674
 675 int32_t icu_tokenizer_token_end(struct icu_tokenizer * tokenizer)
 676 {
 677     return tokenizer->token_end;
 678 }
 679
 680 int32_t icu_tokenizer_token_length(struct icu_tokenizer * tokenizer)
 681 {
 682     return (tokenizer->token_end - tokenizer->token_start);
 683 }
 684
 685 int32_t icu_tokenizer_token_count(struct icu_tokenizer * tokenizer)
 686 {
 687     return tokenizer->token_count;
 688 }
 689
 690
 691
 692 struct icu_normalizer * icu_normalizer_create(const char *rules, char action,
 693                                               UErrorCode *status)
 694 {
 695
 696     struct icu_normalizer * normalizer
 697         = (struct icu_normalizer *) malloc(sizeof(struct icu_normalizer));
 698
 699     normalizer->action = action;
 700     normalizer->trans = 0;
 701     normalizer->rules16 =  icu_buf_utf16_create(0);
 702     icu_utf16_from_utf8_cstr(normalizer->rules16, rules, status);
 703
 704     switch(normalizer->action) {
 705     case 'f':
 706     case 'F':
 707         normalizer->trans
 708             = utrans_openU(normalizer->rules16->utf16,
 709                            normalizer->rules16->utf16_len,
 710                            UTRANS_FORWARD,
 711                            0, 0,
 712                            normalizer->parse_error, status);
 713         break;
 714     case 'r':
 715     case 'R':
 716         normalizer->trans
 717             = utrans_openU(normalizer->rules16->utf16,
 718                            normalizer->rules16->utf16_len,
 719                            UTRANS_REVERSE ,
 720                            0, 0,
 721                            normalizer->parse_error, status);
 722         break;
 723     default:
 724         *status = U_UNSUPPORTED_ERROR;
 725         return 0;
 726         break;
 727     }
 728
 729     if (U_SUCCESS(*status))
 730         return normalizer;
 731
 732     /* freeing if failed */
 733     icu_normalizer_destroy(normalizer);
 734     return 0;
 735 }
 736
 737
 738 void icu_normalizer_destroy(struct icu_normalizer * normalizer){
 739     if (normalizer) {
 740         if (normalizer->rules16)
 741             icu_buf_utf16_destroy(normalizer->rules16);
 742         if (normalizer->trans)
 743             utrans_close(normalizer->trans);
 744         free(normalizer);
 745     }
 746 }
 747
 748
 749
 750 int icu_normalizer_normalize(struct icu_normalizer * normalizer,
 751                              struct icu_buf_utf16 * dest16,
 752                              struct icu_buf_utf16 * src16,
 753                              UErrorCode *status)
 754 {
 755     if (!normalizer || !normalizer->trans
 756         || !src16
 757         || !dest16)
 758         return 0;
 759
 760     if (!src16->utf16_len){           //guarding for empty source string
 761         icu_buf_utf16_clear(dest16);
 762         return 0;
 763     }
 764
 765     if (!icu_buf_utf16_copy(dest16, src16))
 766         return 0;
 767
 768
 769     utrans_transUChars (normalizer->trans,
 770                         dest16->utf16, &(dest16->utf16_len),
 771                         dest16->utf16_cap,
 772                         0, &(src16->utf16_len), status);
 773
 774     if (U_FAILURE(*status))
 775         icu_buf_utf16_clear(dest16);
 776
 777     return dest16->utf16_len;
 778 }
 779
 780
 781
 782
 783 struct icu_chain_step * icu_chain_step_create(struct icu_chain * chain,
 784                                               enum icu_chain_step_type type,
 785                                               const uint8_t * rule,
 786                                               struct icu_buf_utf16 * buf16,
 787                                               UErrorCode *status)
 788 {
 789     struct icu_chain_step * step = 0;
 790
 791     if(!chain || !type || !rule)
 792         return 0;
 793
 794     step = (struct icu_chain_step *) malloc(sizeof(struct icu_chain_step));
 795
 796     step->type = type;
 797
 798     step->buf16 = buf16;
 799
 800     /* create auxilary objects */
 801     switch(step->type) {
 802     case ICU_chain_step_type_display:
 803         break;
 804     case ICU_chain_step_type_casemap:
 805         step->u.casemap = icu_casemap_create(rule[0], status);
 806         break;
 807     case ICU_chain_step_type_normalize:
 808         step->u.normalizer = icu_normalizer_create((char *) rule, 'f', status);
 809         break;
 810     case ICU_chain_step_type_tokenize:
 811         step->u.tokenizer = icu_tokenizer_create((char *) chain->locale,
 812                                                  (char) rule[0], status);
 813         break;
 814     default:
 815         break;
 816     }
 817
 818     return step;
 819 }
 820
 821
 822 void icu_chain_step_destroy(struct icu_chain_step * step){
 823
 824     if (!step)
 825         return;
 826
 827     icu_chain_step_destroy(step->previous);
 828
 829     switch(step->type) {
 830     case ICU_chain_step_type_display:
 831         break;
 832     case ICU_chain_step_type_casemap:
 833         icu_casemap_destroy(step->u.casemap);
 834         icu_buf_utf16_destroy(step->buf16);
 835         break;
 836     case ICU_chain_step_type_normalize:
 837         icu_normalizer_destroy(step->u.normalizer);
 838         icu_buf_utf16_destroy(step->buf16);
 839         break;
 840     case ICU_chain_step_type_tokenize:
 841         icu_tokenizer_destroy(step->u.tokenizer);
 842         icu_buf_utf16_destroy(step->buf16);
 843         break;
 844     default:
 845         break;
 846     }
 847     free(step);
 848 }
 849
 850
 851
 852 struct icu_chain * icu_chain_create(const char *locale,
 853                                     int sort,
 854                                     UErrorCode * status)
 855 {
 856
 857     struct icu_chain * chain
 858         = (struct icu_chain *) malloc(sizeof(struct icu_chain));
 859
 860     *status = U_ZERO_ERROR;
 861
 862     strncpy((char *) chain->locale, (const char *) locale, 16);
 863     chain->locale[16 - 1] = '\0';
 864
 865     chain->sort = sort;
 866
 867     chain->coll = ucol_open((const char *) chain->locale, status);
 868
 869     if (U_FAILURE(*status))
 870         return 0;
 871
 872
 873     chain->token_count = 0;
 874
 875     chain->src8cstr = 0;
 876
 877     chain->display8 = icu_buf_utf8_create(0);
 878     chain->norm8 = icu_buf_utf8_create(0);
 879     chain->sort8 = icu_buf_utf8_create(0);
 880
 881     chain->src16 = icu_buf_utf16_create(0);
 882
 883     chain->steps = 0;
 884
 885     return chain;
 886 }
 887
 888
 889 void icu_chain_destroy(struct icu_chain * chain)
 890 {
 891     if (chain){
 892
 893         if (chain->coll)
 894             ucol_close(chain->coll);
 895
 896         icu_buf_utf8_destroy(chain->display8);
 897         icu_buf_utf8_destroy(chain->norm8);
 898         icu_buf_utf8_destroy(chain->sort8);
 899
 900         icu_buf_utf16_destroy(chain->src16);
 901
 902         icu_chain_step_destroy(chain->steps);
 903         free(chain);
 904     }
 905 }
 906
 907
 908
 909 struct icu_chain * icu_chain_xml_config(const xmlNode *xml_node,
 910                                         int sort,
 911                                         UErrorCode * status)
 912 {
 913     xmlNode *node = 0;
 914     struct icu_chain * chain = 0;
 915
 916     *status = U_ZERO_ERROR;
 917
 918     if (!xml_node ||xml_node->type != XML_ELEMENT_NODE)
 919         return 0;
 920
 921     {
 922         xmlChar * xml_locale = xmlGetProp((xmlNode *) xml_node,
 923                                           (xmlChar *) "locale");
 924
 925         if (xml_locale)
 926         {
 927             chain = icu_chain_create((const char *) xml_locale, sort, status);
 928             xmlFree(xml_locale);
 929         }
 930
 931     }
 932     if (!chain)
 933         return 0;
 934
 935     for (node = xml_node->children; node; node = node->next)
 936     {
 937         xmlChar *xml_rule;
 938         struct icu_chain_step * step = 0;
 939
 940         if (node->type != XML_ELEMENT_NODE)
 941             continue;
 942
 943         xml_rule = xmlGetProp(node, (xmlChar *) "rule");
 944
 945         if (!strcmp((const char *) node->name,
 946                     (const char *) "casemap")){
 947             step = icu_chain_insert_step(chain, ICU_chain_step_type_casemap,
 948                                          (const uint8_t *) xml_rule, status);
 949         }
 950         else if (!strcmp((const char *) node->name,
 951                          (const char *) "normalize")){
 952             step = icu_chain_insert_step(chain, ICU_chain_step_type_normalize,
 953                                          (const uint8_t *) xml_rule, status);
 954         }
 955         else if (!strcmp((const char *) node->name,
 956                          (const char *) "tokenize")){
 957             step = icu_chain_insert_step(chain, ICU_chain_step_type_tokenize,
 958                                          (const uint8_t *) xml_rule, status);
 959         }
 960         else if (!strcmp((const char *) node->name,
 961                          (const char *) "display")){
 962             step = icu_chain_insert_step(chain, ICU_chain_step_type_display,
 963                                          (const uint8_t *) "", status);
 964         }
 965         xmlFree(xml_rule);
 966         if (!step || U_FAILURE(*status)){
 967             icu_chain_destroy(chain);
 968             return 0;
 969         }
 970
 971
 972     }
 973
 974     return chain;
 975 }
 976
 977
 978
 979 struct icu_chain_step * icu_chain_insert_step(struct icu_chain * chain,
 980                                               enum icu_chain_step_type type,
 981                                               const uint8_t * rule,
 982                                               UErrorCode *status)
 983 {
 984     struct icu_chain_step * step = 0;
 985     struct icu_buf_utf16 * src16 = 0;
 986     struct icu_buf_utf16 * buf16 = 0;
 987
 988     if (!chain || !type || !rule)
 989         return 0;
 990
 991     /* assign utf16 src buffers as needed */
 992     if (chain->steps && chain->steps->buf16)
 993         src16 = chain->steps->buf16;
 994     else if (chain->src16)
 995         src16 = chain->src16;
 996     else
 997         return 0;
 998
 999
1000     /* create utf16 destination buffers as needed, or */
1001     switch(type) {
1002     case ICU_chain_step_type_display:
1003         buf16 = src16;
1004         break;
1005     case ICU_chain_step_type_casemap:
1006         buf16 = icu_buf_utf16_create(0);
1007         break;
1008     case ICU_chain_step_type_normalize:
1009         buf16 = icu_buf_utf16_create(0);
1010         break;
1011     case ICU_chain_step_type_tokenize:
1012         buf16 = icu_buf_utf16_create(0);
1013         break;
1014     default:
1015         break;
1016     }
1017
1018     /* create actual chain step with this buffer */
1019     step = icu_chain_step_create(chain, type, rule, buf16, status);
1020
1021     step->previous = chain->steps;
1022     chain->steps = step;
1023
1024     return step;
1025 }
1026
1027
1028 int icu_chain_step_next_token(struct icu_chain * chain,
1029                               struct icu_chain_step * step,
1030                               UErrorCode *status)
1031 {
1032     struct icu_buf_utf16 * src16 = 0;
1033     int got_new_token = 0;
1034
1035     if (!chain || !chain->src16 || !step || !step->more_tokens)
1036         return 0;
1037
1038     /* assign utf16 src buffers as neeed, advance in previous steps
1039        tokens until non-zero token met, and setting stop condition */
1040
1041     if (step->previous){
1042         src16 = step->previous->buf16;
1043         /* tokens might be killed in previous steps, therefore looping */
1044
1045         while (step->need_new_token
1046                && step->previous->more_tokens
1047                && !got_new_token)
1048             got_new_token
1049                 = icu_chain_step_next_token(chain, step->previous, status);
1050     }
1051     else { /* first step can only work once on chain->src16 input buffer */
1052         src16 = chain->src16;
1053         step->more_tokens = 0;
1054         got_new_token = 1;
1055     }
1056
1057     if (!src16)
1058         return 0;
1059
1060     /* stop if nothing to process */
1061     if (step->need_new_token && !got_new_token){
1062         step->more_tokens = 0;
1063         return 0;
1064     }
1065
1066     /* either an old token not finished yet, or a new token, thus
1067        perform the work, eventually put this steps output in
1068        step->buf16 or the chains UTF8 output buffers  */
1069
1070     switch(step->type) {
1071     case ICU_chain_step_type_display:
1072         icu_utf16_to_utf8(chain->display8, src16, status);
1073         break;
1074     case ICU_chain_step_type_casemap:
1075         icu_casemap_casemap(step->u.casemap,
1076                             step->buf16, src16, status,
1077                             chain->locale);
1078         break;
1079     case ICU_chain_step_type_normalize:
1080         icu_normalizer_normalize(step->u.normalizer,
1081                                  step->buf16, src16, status);
1082         break;
1083     case ICU_chain_step_type_tokenize:
1084         /* attach to new src16 token only first time during splitting */
1085         if (step->need_new_token){
1086             icu_tokenizer_attach(step->u.tokenizer, src16, status);
1087             step->need_new_token = 0;
1088         }
1089
1090
1091         /* splitting one src16 token into multiple buf16 tokens */
1092         step->more_tokens
1093             = icu_tokenizer_next_token(step->u.tokenizer,
1094                                        step->buf16, status);
1095
1096         /* make sure to get new previous token if this one had been used up
1097            by recursive call to _same_ step */
1098
1099         if (!step->more_tokens){
1100             step->more_tokens = icu_chain_step_next_token(chain, step, status);
1101             return step->more_tokens;  // avoid one token count too much!
1102         }
1103
1104         break;
1105     default:
1106         return 0;
1107         break;
1108     }
1109
1110     if (U_FAILURE(*status))
1111         return 0;
1112
1113     /* if token disappered into thin air, tell caller */
1114     /* if (!step->buf16->utf16_len && !step->more_tokens) */
1115     /*    return 0; */
1116
1117     return 1;
1118 }
1119
1120
1121 int icu_chain_assign_cstr(struct icu_chain * chain,
1122                           const char * src8cstr,
1123                           UErrorCode *status)
1124 {
1125     struct icu_chain_step * stp = 0;
1126
1127     if (!chain || !src8cstr)
1128         return 0;
1129
1130     chain->src8cstr = src8cstr;
1131
1132     stp = chain->steps;
1133
1134     /* clear token count */
1135     chain->token_count = 0;
1136
1137     /* clear all steps stop states */
1138     while (stp){
1139         stp->more_tokens = 1;
1140         stp->need_new_token = 1;
1141         stp = stp->previous;
1142     }
1143
1144     /* finally convert UTF8 to UTF16 string if needed */
1145     if (chain->steps || chain->sort)
1146         icu_utf16_from_utf8_cstr(chain->src16, chain->src8cstr, status);
1147
1148     if (U_FAILURE(*status))
1149         return 0;
1150
1151     return 1;
1152 }
1153
1154
1155
1156 int icu_chain_next_token(struct icu_chain * chain,
1157                          UErrorCode *status)
1158 {
1159     int got_token = 0;
1160
1161     *status = U_ZERO_ERROR;
1162
1163     if (!chain)
1164         return 0;
1165
1166     /* special case with no steps - same as index type binary */
1167     if (!chain->steps){
1168         if (chain->token_count)
1169             return 0;
1170         else {
1171             chain->token_count++;
1172
1173             if (chain->sort)
1174                 icu_sortkey8_from_utf16(chain->coll,
1175                                         chain->sort8, chain->steps->buf16,
1176                                         status);
1177             return chain->token_count;
1178         }
1179     }
1180     /* usual case, one or more icu chain steps existing */
1181     else {
1182
1183         while(!got_token && chain->steps && chain->steps->more_tokens)
1184             got_token = icu_chain_step_next_token(chain, chain->steps, status);
1185
1186         if (got_token){
1187             chain->token_count++;
1188
1189             icu_utf16_to_utf8(chain->norm8, chain->steps->buf16, status);
1190
1191             if (chain->sort)
1192                 icu_sortkey8_from_utf16(chain->coll,
1193                                         chain->sort8, chain->steps->buf16,
1194                                         status);
1195
1196             return chain->token_count;
1197         }
1198     }
1199
1200     return 0;
1201 }
1202
1203 int icu_chain_token_number(struct icu_chain * chain)
1204 {
1205     if (!chain)
1206         return 0;
1207
1208     return chain->token_count;
1209 }
1210
1211
1212 const char * icu_chain_token_display(struct icu_chain * chain)
1213 {
1214     if (chain->display8)
1215         return icu_buf_utf8_to_cstr(chain->display8);
1216
1217     return 0;
1218 }
1219
1220 const char * icu_chain_token_norm(struct icu_chain * chain)
1221 {
1222     if (!chain->steps)
1223         return chain->src8cstr;
1224
1225     if (chain->norm8)
1226         return icu_buf_utf8_to_cstr(chain->norm8);
1227
1228     return 0;
1229 }
1230
1231 const char * icu_chain_token_sortkey(struct icu_chain * chain)
1232 {
1233     if (chain->sort8)
1234         return icu_buf_utf8_to_cstr(chain->sort8);
1235
1236     return 0;
1237 }
1238
1239 const UCollator * icu_chain_get_coll(struct icu_chain * chain)
1240 {
1241     return chain->coll;
1242 }
1243
1244
1245 #endif /* YAZ_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  */