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 1.16 2007-11-08 17:15:13 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     tokenizer->action = action;
 513     tokenizer->bi = 0;
 514     tokenizer->buf16 = 0;
 515     tokenizer->token_count = 0;
 516     tokenizer->token_id = 0;
 517     tokenizer->token_start = 0;
 518     tokenizer->token_end = 0;
 519
 520
 521     switch(tokenizer->action) {
 522     case 'l':
 523     case 'L':
 524         tokenizer->bi = ubrk_open(UBRK_LINE, locale, 0, 0, status);
 525         break;
 526     case 's':
 527     case 'S':
 528         tokenizer->bi = ubrk_open(UBRK_SENTENCE, locale, 0, 0, status);
 529         break;
 530     case 'w':
 531     case 'W':
 532         tokenizer->bi = ubrk_open(UBRK_WORD, locale, 0, 0, status);
 533         break;
 534     case 'c':
 535     case 'C':
 536         tokenizer->bi = ubrk_open(UBRK_CHARACTER, locale, 0, 0, status);
 537         break;
 538     case 't':
 539     case 'T':
 540         tokenizer->bi = ubrk_open(UBRK_TITLE, locale, 0, 0, status);
 541         break;
 542     default:
 543         *status = U_UNSUPPORTED_ERROR;
 544         return 0;
 545         break;
 546     }
 547
 548     /* ICU error stuff is a very  funny business */
 549     if (U_SUCCESS(*status))
 550         return tokenizer;
 551
 552     /* freeing if failed */
 553     icu_tokenizer_destroy(tokenizer);
 554     return 0;
 555 }
 556
 557 void icu_tokenizer_destroy(struct icu_tokenizer * tokenizer)
 558 {
 559     if (tokenizer) {
 560         if (tokenizer->bi)
 561             ubrk_close(tokenizer->bi);
 562         free(tokenizer);
 563     }
 564 }
 565
 566 int icu_tokenizer_attach(struct icu_tokenizer * tokenizer,
 567                          struct icu_buf_utf16 * src16,
 568                          UErrorCode *status)
 569 {
 570     if (!tokenizer || !tokenizer->bi || !src16)
 571         return 0;
 572
 573
 574     tokenizer->buf16 = src16;
 575     tokenizer->token_count = 0;
 576     tokenizer->token_id = 0;
 577     tokenizer->token_start = 0;
 578     tokenizer->token_end = 0;
 579
 580     ubrk_setText(tokenizer->bi, src16->utf16, src16->utf16_len, status);
 581
 582
 583     if (U_FAILURE(*status))
 584         return 0;
 585
 586     return 1;
 587 };
 588
 589 int32_t icu_tokenizer_next_token(struct icu_tokenizer * tokenizer,
 590                          struct icu_buf_utf16 * tkn16,
 591                          UErrorCode *status)
 592 {
 593     int32_t tkn_start = 0;
 594     int32_t tkn_end = 0;
 595     int32_t tkn_len = 0;
 596
 597
 598     if (!tokenizer || !tokenizer->bi
 599         || !tokenizer->buf16 || !tokenizer->buf16->utf16_len)
 600         return 0;
 601
 602     /*
 603     never change tokenizer->buf16 and keep always invariant
 604     0 <= tokenizer->token_start
 605        <= tokenizer->token_end
 606        <= tokenizer->buf16->utf16_len
 607     returns length of token
 608     */
 609
 610     if (0 == tokenizer->token_end) /* first call */
 611         tkn_start = ubrk_first(tokenizer->bi);
 612     else /* successive calls */
 613         tkn_start = tokenizer->token_end;
 614
 615     /* get next position */
 616     tkn_end = ubrk_next(tokenizer->bi);
 617
 618     /* repairing invariant at end of ubrk, which is UBRK_DONE = -1 */
 619     if (UBRK_DONE == tkn_end)
 620         tkn_end = tokenizer->buf16->utf16_len;
 621
 622     /* copy out if everything is well */
 623     if(U_FAILURE(*status))
 624         return 0;
 625
 626     /* everything OK, now update internal state */
 627     tkn_len = tkn_end - tkn_start;
 628
 629     if (0 < tkn_len){
 630         tokenizer->token_count++;
 631         tokenizer->token_id++;
 632     } else {
 633         tokenizer->token_id = 0;
 634     }
 635     tokenizer->token_start = tkn_start;
 636     tokenizer->token_end = tkn_end;
 637
 638
 639     /* copying into token buffer if it exists */
 640     if (tkn16){
 641         if (tkn16->utf16_cap < tkn_len)
 642             icu_buf_utf16_resize(tkn16, (size_t) tkn_len * 2);
 643
 644         u_strncpy(tkn16->utf16, &(tokenizer->buf16->utf16)[tkn_start],
 645                   tkn_len);
 646
 647         tkn16->utf16_len = tkn_len;
 648     }
 649
 650     return tkn_len;
 651 }
 652
 653
 654 int32_t icu_tokenizer_token_id(struct icu_tokenizer * tokenizer)
 655 {
 656     return tokenizer->token_id;
 657 }
 658
 659 int32_t icu_tokenizer_token_start(struct icu_tokenizer * tokenizer)
 660 {
 661     return tokenizer->token_start;
 662 }
 663
 664 int32_t icu_tokenizer_token_end(struct icu_tokenizer * tokenizer)
 665 {
 666     return tokenizer->token_end;
 667 }
 668
 669 int32_t icu_tokenizer_token_length(struct icu_tokenizer * tokenizer)
 670 {
 671     return (tokenizer->token_end - tokenizer->token_start);
 672 }
 673
 674 int32_t icu_tokenizer_token_count(struct icu_tokenizer * tokenizer)
 675 {
 676     return tokenizer->token_count;
 677 }
 678
 679
 680
 681 struct icu_normalizer * icu_normalizer_create(const char *rules, char action,
 682                                               UErrorCode *status)
 683 {
 684
 685     struct icu_normalizer * normalizer
 686         = (struct icu_normalizer *) malloc(sizeof(struct icu_normalizer));
 687
 688     normalizer->action = action;
 689     normalizer->trans = 0;
 690     normalizer->rules16 =  icu_buf_utf16_create(0);
 691     icu_utf16_from_utf8_cstr(normalizer->rules16, rules, status);
 692
 693     switch(normalizer->action) {
 694     case 'f':
 695     case 'F':
 696         normalizer->trans
 697             = utrans_openU(normalizer->rules16->utf16,
 698                            normalizer->rules16->utf16_len,
 699                            UTRANS_FORWARD,
 700                            0, 0,
 701                            normalizer->parse_error, status);
 702         break;
 703     case 'r':
 704     case 'R':
 705         normalizer->trans
 706             = utrans_openU(normalizer->rules16->utf16,
 707                            normalizer->rules16->utf16_len,
 708                            UTRANS_REVERSE ,
 709                            0, 0,
 710                            normalizer->parse_error, status);
 711         break;
 712     default:
 713         *status = U_UNSUPPORTED_ERROR;
 714         return 0;
 715         break;
 716     }
 717
 718     if (U_SUCCESS(*status))
 719         return normalizer;
 720
 721     /* freeing if failed */
 722     icu_normalizer_destroy(normalizer);
 723     return 0;
 724 }
 725
 726
 727 void icu_normalizer_destroy(struct icu_normalizer * normalizer){
 728     if (normalizer) {
 729         if (normalizer->rules16)
 730             icu_buf_utf16_destroy(normalizer->rules16);
 731         if (normalizer->trans)
 732             utrans_close(normalizer->trans);
 733         free(normalizer);
 734     }
 735 }
 736
 737
 738
 739 int icu_normalizer_normalize(struct icu_normalizer * normalizer,
 740                              struct icu_buf_utf16 * dest16,
 741                              struct icu_buf_utf16 * src16,
 742                              UErrorCode *status)
 743 {
 744     if (!normalizer || !normalizer->trans
 745         || !src16
 746         || !dest16)
 747         return 0;
 748
 749     if (!src16->utf16_len){           //guarding for empty source string
 750         icu_buf_utf16_clear(dest16);
 751         return 0;
 752     }
 753
 754     if (!icu_buf_utf16_copy(dest16, src16))
 755         return 0;
 756
 757
 758     utrans_transUChars (normalizer->trans,
 759                         dest16->utf16, &(dest16->utf16_len),
 760                         dest16->utf16_cap,
 761                         0, &(src16->utf16_len), status);
 762
 763     if (U_FAILURE(*status))
 764         icu_buf_utf16_clear(dest16);
 765
 766     return dest16->utf16_len;
 767 }
 768
 769
 770
 771
 772 struct icu_chain_step * icu_chain_step_create(struct icu_chain * chain,
 773                                               enum icu_chain_step_type type,
 774                                               const uint8_t * rule,
 775                                               struct icu_buf_utf16 * buf16,
 776                                               UErrorCode *status)
 777 {
 778     struct icu_chain_step * step = 0;
 779
 780     if(!chain || !type || !rule)
 781         return 0;
 782
 783     step = (struct icu_chain_step *) malloc(sizeof(struct icu_chain_step));
 784
 785     step->type = type;
 786
 787     step->buf16 = buf16;
 788
 789     /* create auxilary objects */
 790     switch(step->type) {
 791     case ICU_chain_step_type_display:
 792         break;
 793     case ICU_chain_step_type_casemap:
 794         step->u.casemap = icu_casemap_create(rule[0], status);
 795         break;
 796     case ICU_chain_step_type_normalize:
 797         step->u.normalizer = icu_normalizer_create((char *) rule, 'f', status);
 798         break;
 799     case ICU_chain_step_type_tokenize:
 800         step->u.tokenizer = icu_tokenizer_create((char *) chain->locale,
 801                                                  (char) rule[0], status);
 802         break;
 803     default:
 804         break;
 805     }
 806
 807     return step;
 808 }
 809
 810
 811 void icu_chain_step_destroy(struct icu_chain_step * step){
 812
 813     if (!step)
 814         return;
 815
 816     icu_chain_step_destroy(step->previous);
 817
 818     switch(step->type) {
 819     case ICU_chain_step_type_display:
 820         break;
 821     case ICU_chain_step_type_casemap:
 822         icu_casemap_destroy(step->u.casemap);
 823         icu_buf_utf16_destroy(step->buf16);
 824         break;
 825     case ICU_chain_step_type_normalize:
 826         icu_normalizer_destroy(step->u.normalizer);
 827         icu_buf_utf16_destroy(step->buf16);
 828         break;
 829     case ICU_chain_step_type_tokenize:
 830         icu_tokenizer_destroy(step->u.tokenizer);
 831         icu_buf_utf16_destroy(step->buf16);
 832         break;
 833     default:
 834         break;
 835     }
 836     free(step);
 837 }
 838
 839
 840
 841 struct icu_chain * icu_chain_create(const char *locale,
 842                                     int sort,
 843                                     UErrorCode * status)
 844 {
 845
 846     struct icu_chain * chain
 847         = (struct icu_chain *) malloc(sizeof(struct icu_chain));
 848
 849     *status = U_ZERO_ERROR;
 850
 851     strncpy((char *) chain->locale, (const char *) locale, 16);
 852     chain->locale[16 - 1] = '\0';
 853
 854     chain->sort = sort;
 855
 856     chain->coll = ucol_open((const char *) chain->locale, status);
 857
 858     if (U_FAILURE(*status))
 859         return 0;
 860
 861
 862     chain->token_count = 0;
 863
 864     chain->src8cstr = 0;
 865
 866     chain->display8 = icu_buf_utf8_create(0);
 867     chain->norm8 = icu_buf_utf8_create(0);
 868     chain->sort8 = icu_buf_utf8_create(0);
 869
 870     chain->src16 = icu_buf_utf16_create(0);
 871
 872     chain->steps = 0;
 873
 874     return chain;
 875 }
 876
 877
 878 void icu_chain_destroy(struct icu_chain * chain)
 879 {
 880     if (chain){
 881
 882         if (chain->coll)
 883             ucol_close(chain->coll);
 884
 885         icu_buf_utf8_destroy(chain->display8);
 886         icu_buf_utf8_destroy(chain->norm8);
 887         icu_buf_utf8_destroy(chain->sort8);
 888
 889         icu_buf_utf16_destroy(chain->src16);
 890
 891         icu_chain_step_destroy(chain->steps);
 892         free(chain);
 893     }
 894 }
 895
 896
 897
 898 struct icu_chain * icu_chain_xml_config(const xmlNode *xml_node,
 899                                         int sort,
 900                                         UErrorCode * status)
 901 {
 902     xmlNode *node = 0;
 903     struct icu_chain * chain = 0;
 904
 905     *status = U_ZERO_ERROR;
 906
 907     if (!xml_node ||xml_node->type != XML_ELEMENT_NODE)
 908         return 0;
 909
 910     {
 911         xmlChar * xml_locale = xmlGetProp((xmlNode *) xml_node,
 912                                           (xmlChar *) "locale");
 913
 914         if (xml_locale)
 915         {
 916             chain = icu_chain_create((const char *) xml_locale, sort, status);
 917             xmlFree(xml_locale);
 918         }
 919
 920     }
 921     if (!chain)
 922         return 0;
 923
 924     for (node = xml_node->children; node; node = node->next)
 925     {
 926         xmlChar *xml_rule;
 927         struct icu_chain_step * step = 0;
 928
 929         if (node->type != XML_ELEMENT_NODE)
 930             continue;
 931
 932         xml_rule = xmlGetProp(node, (xmlChar *) "rule");
 933
 934         if (!strcmp((const char *) node->name,
 935                     (const char *) "casemap")){
 936             step = icu_chain_insert_step(chain, ICU_chain_step_type_casemap,
 937                                          (const uint8_t *) xml_rule, status);
 938         }
 939         else if (!strcmp((const char *) node->name,
 940                          (const char *) "normalize")){
 941             step = icu_chain_insert_step(chain, ICU_chain_step_type_normalize,
 942                                          (const uint8_t *) xml_rule, status);
 943         }
 944         else if (!strcmp((const char *) node->name,
 945                          (const char *) "tokenize")){
 946             step = icu_chain_insert_step(chain, ICU_chain_step_type_tokenize,
 947                                          (const uint8_t *) xml_rule, status);
 948         }
 949         else if (!strcmp((const char *) node->name,
 950                          (const char *) "display")){
 951             step = icu_chain_insert_step(chain, ICU_chain_step_type_display,
 952                                          (const uint8_t *) "", status);
 953         }
 954         xmlFree(xml_rule);
 955         if (!step || U_FAILURE(*status)){
 956             icu_chain_destroy(chain);
 957             return 0;
 958         }
 959
 960
 961     }
 962
 963     return chain;
 964 }
 965
 966
 967
 968 struct icu_chain_step * icu_chain_insert_step(struct icu_chain * chain,
 969                                               enum icu_chain_step_type type,
 970                                               const uint8_t * rule,
 971                                               UErrorCode *status)
 972 {
 973     struct icu_chain_step * step = 0;
 974     struct icu_buf_utf16 * src16 = 0;
 975     struct icu_buf_utf16 * buf16 = 0;
 976
 977     if (!chain || !type || !rule)
 978         return 0;
 979
 980     /* assign utf16 src buffers as needed */
 981     if (chain->steps && chain->steps->buf16)
 982         src16 = chain->steps->buf16;
 983     else if (chain->src16)
 984         src16 = chain->src16;
 985     else
 986         return 0;
 987
 988
 989     /* create utf16 destination buffers as needed, or */
 990     switch(type) {
 991     case ICU_chain_step_type_display:
 992         buf16 = src16;
 993         break;
 994     case ICU_chain_step_type_casemap:
 995         buf16 = icu_buf_utf16_create(0);
 996         break;
 997     case ICU_chain_step_type_normalize:
 998         buf16 = icu_buf_utf16_create(0);
 999         break;
1000     case ICU_chain_step_type_tokenize:
1001         buf16 = icu_buf_utf16_create(0);
1002         break;
1003     default:
1004         break;
1005     }
1006
1007     /* create actual chain step with this buffer */
1008     step = icu_chain_step_create(chain, type, rule, buf16, status);
1009
1010     step->previous = chain->steps;
1011     chain->steps = step;
1012
1013     return step;
1014 }
1015
1016
1017 int icu_chain_step_next_token(struct icu_chain * chain,
1018                               struct icu_chain_step * step,
1019                               UErrorCode *status)
1020 {
1021     struct icu_buf_utf16 * src16 = 0;
1022     int got_new_token = 0;
1023
1024     if (!chain || !chain->src16 || !step || !step->more_tokens)
1025         return 0;
1026
1027     /* assign utf16 src buffers as neeed, advance in previous steps
1028        tokens until non-zero token met, and setting stop condition */
1029
1030     if (step->previous){
1031         src16 = step->previous->buf16;
1032         /* tokens might be killed in previous steps, therefore looping */
1033
1034         while (step->need_new_token
1035                && step->previous->more_tokens
1036                && !got_new_token)
1037             got_new_token
1038                 = icu_chain_step_next_token(chain, step->previous, status);
1039     }
1040     else { /* first step can only work once on chain->src16 input buffer */
1041         src16 = chain->src16;
1042         step->more_tokens = 0;
1043         got_new_token = 1;
1044     }
1045
1046     if (!src16)
1047         return 0;
1048
1049     /* stop if nothing to process */
1050     if (step->need_new_token && !got_new_token){
1051         step->more_tokens = 0;
1052         return 0;
1053     }
1054
1055     /* either an old token not finished yet, or a new token, thus
1056        perform the work, eventually put this steps output in
1057        step->buf16 or the chains UTF8 output buffers  */
1058
1059     switch(step->type) {
1060     case ICU_chain_step_type_display:
1061         icu_utf16_to_utf8(chain->display8, src16, status);
1062         break;
1063     case ICU_chain_step_type_casemap:
1064         icu_casemap_casemap(step->u.casemap,
1065                             step->buf16, src16, status,
1066                             chain->locale);
1067         break;
1068     case ICU_chain_step_type_normalize:
1069         icu_normalizer_normalize(step->u.normalizer,
1070                                  step->buf16, src16, status);
1071         break;
1072     case ICU_chain_step_type_tokenize:
1073         /* attach to new src16 token only first time during splitting */
1074         if (step->need_new_token){
1075             icu_tokenizer_attach(step->u.tokenizer, src16, status);
1076             step->need_new_token = 0;
1077         }
1078
1079
1080         /* splitting one src16 token into multiple buf16 tokens */
1081         step->more_tokens
1082             = icu_tokenizer_next_token(step->u.tokenizer,
1083                                        step->buf16, status);
1084
1085         /* make sure to get new previous token if this one had been used up
1086            by recursive call to _same_ step */
1087
1088         if (!step->more_tokens){
1089             step->more_tokens = icu_chain_step_next_token(chain, step, status);
1090             return step->more_tokens;  // avoid one token count too much!
1091         }
1092
1093         break;
1094     default:
1095         return 0;
1096         break;
1097     }
1098
1099     if (U_FAILURE(*status))
1100         return 0;
1101
1102     /* if token disappered into thin air, tell caller */
1103     /* if (!step->buf16->utf16_len && !step->more_tokens) */
1104     /*    return 0; */
1105
1106     return 1;
1107 }
1108
1109
1110 int icu_chain_assign_cstr(struct icu_chain * chain,
1111                           const char * src8cstr,
1112                           UErrorCode *status)
1113 {
1114     struct icu_chain_step * stp = 0;
1115
1116     if (!chain || !src8cstr)
1117         return 0;
1118
1119     chain->src8cstr = src8cstr;
1120
1121     stp = chain->steps;
1122
1123     /* clear token count */
1124     chain->token_count = 0;
1125
1126     /* clear all steps stop states */
1127     while (stp){
1128         stp->more_tokens = 1;
1129         stp->need_new_token = 1;
1130         stp = stp->previous;
1131     }
1132
1133     /* finally convert UTF8 to UTF16 string if needed */
1134     if (chain->steps || chain->sort)
1135         icu_utf16_from_utf8_cstr(chain->src16, chain->src8cstr, status);
1136
1137     if (U_FAILURE(*status))
1138         return 0;
1139
1140     return 1;
1141 }
1142
1143
1144
1145 int icu_chain_next_token(struct icu_chain * chain,
1146                          UErrorCode *status)
1147 {
1148     int got_token = 0;
1149
1150     *status = U_ZERO_ERROR;
1151
1152     if (!chain)
1153         return 0;
1154
1155     /* special case with no steps - same as index type binary */
1156     if (!chain->steps){
1157         if (chain->token_count)
1158             return 0;
1159         else {
1160             chain->token_count++;
1161
1162             if (chain->sort)
1163                 icu_sortkey8_from_utf16(chain->coll,
1164                                         chain->sort8, chain->steps->buf16,
1165                                         status);
1166             return chain->token_count;
1167         }
1168     }
1169     /* usual case, one or more icu chain steps existing */
1170     else {
1171
1172         while(!got_token && chain->steps && chain->steps->more_tokens)
1173             got_token = icu_chain_step_next_token(chain, chain->steps, status);
1174
1175         if (got_token){
1176             chain->token_count++;
1177
1178             icu_utf16_to_utf8(chain->norm8, chain->steps->buf16, status);
1179
1180             if (chain->sort)
1181                 icu_sortkey8_from_utf16(chain->coll,
1182                                         chain->sort8, chain->steps->buf16,
1183                                         status);
1184
1185             return chain->token_count;
1186         }
1187     }
1188
1189     return 0;
1190 }
1191
1192 int icu_chain_token_number(struct icu_chain * chain)
1193 {
1194     if (!chain)
1195         return 0;
1196
1197     return chain->token_count;
1198 }
1199
1200
1201 const char * icu_chain_token_display(struct icu_chain * chain)
1202 {
1203     if (chain->display8)
1204         return icu_buf_utf8_to_cstr(chain->display8);
1205
1206     return 0;
1207 }
1208
1209 const char * icu_chain_token_norm(struct icu_chain * chain)
1210 {
1211     if (!chain->steps)
1212         return chain->src8cstr;
1213
1214     if (chain->norm8)
1215         return icu_buf_utf8_to_cstr(chain->norm8);
1216
1217     return 0;
1218 }
1219
1220 const char * icu_chain_token_sortkey(struct icu_chain * chain)
1221 {
1222     if (chain->sort8)
1223         return icu_buf_utf8_to_cstr(chain->sort8);
1224
1225     return 0;
1226 }
1227
1228 const UCollator * icu_chain_get_coll(struct icu_chain * chain)
1229 {
1230     return chain->coll;
1231 }
1232
1233
1234 #endif /* YAZ_HAVE_ICU */
1235
1236
1237
1238
1239 /*
1240  * Local variables:
1241  * c-basic-offset: 4
1242  * indent-tabs-mode: nil
1243  * End:
1244  * vim: shiftwidth=4 tabstop=8 expandtab
1245  */