src/icu_I18N.c

   1 /* $Id: icu_I18N.c,v 1.18 2007-05-21 10:14:08 marc Exp $
   2    Copyright (c) 2006-2007, Index Data.
   3
   4    This file is part of Pazpar2.
   5
   6    Pazpar2 is free software; you can redistribute it and/or modify it under
   7    the terms of the GNU General Public License as published by the Free
   8    Software Foundation; either version 2, or (at your option) any later
   9    version.
  10
  11    Pazpar2 is distributed in the hope that it will be useful, but WITHOUT ANY
  12    WARRANTY; without even the implied warranty of MERCHANTABILITY or
  13    FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License
  14    for more details.
  15
  16    You should have received a copy of the GNU General Public License
  17    along with Pazpar2; see the file LICENSE.  If not, write to the
  18    Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
  19    02111-1307, USA.
  20 */
  21
  22 #if HAVE_CONFIG_H
  23 #include "cconfig.h"
  24 #endif
  25
  26 #define USE_TIMING 0
  27 #if USE_TIMING
  28 #include <yaz/timing.h>
  29 #endif
  30
  31
  32 #ifdef HAVE_ICU
  33 #include "icu_I18N.h"
  34
  35 #include <yaz/log.h>
  36
  37 #include <string.h>
  38 #include <stdlib.h>
  39 #include <stdio.h>
  40
  41 #include <unicode/ustring.h>  /* some more string fcns*/
  42 #include <unicode/uchar.h>    /* char names           */
  43
  44
  45 //#include <unicode/ustdio.h>
  46 //#include <unicode/utypes.h>   /* Basic ICU data types */
  47 #include <unicode/ucol.h>
  48 //#include <unicode/ucnv.h>     /* C   Converter API    */
  49 //#include <unicode/uloc.h>
  50 //#include <unicode/ubrk.h>
  51 /* #include <unicode/unistr.h> */
  52
  53
  54
  55
  56 int icu_check_status (UErrorCode status)
  57 {
  58     if(U_FAILURE(status)){
  59         yaz_log(YLOG_WARN,
  60                 "ICU: %d %s\n", status, u_errorName(status));
  61         return 0;
  62     }
  63     return 1;
  64
  65 }
  66
  67
  68
  69 struct icu_buf_utf16 * icu_buf_utf16_create(size_t capacity)
  70 {
  71     struct icu_buf_utf16 * buf16
  72         = (struct icu_buf_utf16 *) malloc(sizeof(struct icu_buf_utf16));
  73
  74     buf16->utf16 = 0;
  75     buf16->utf16_len = 0;
  76     buf16->utf16_cap = 0;
  77
  78     if (capacity > 0){
  79         buf16->utf16 = (UChar *) malloc(sizeof(UChar) * capacity);
  80         buf16->utf16[0] = (UChar) 0;
  81         buf16->utf16_cap = capacity;
  82     }
  83     return buf16;
  84 };
  85
  86
  87 struct icu_buf_utf16 * icu_buf_utf16_resize(struct icu_buf_utf16 * buf16,
  88                                             size_t capacity)
  89 {
  90     if (buf16){
  91         if (capacity >  0){
  92             if (0 == buf16->utf16)
  93                 buf16->utf16 = (UChar *) malloc(sizeof(UChar) * capacity);
  94             else
  95                 buf16->utf16
  96                     = (UChar *) realloc(buf16->utf16, sizeof(UChar) * capacity);
  97             buf16->utf16[0] = (UChar) 0;
  98             buf16->utf16_len = 0;
  99             buf16->utf16_cap = capacity;
 100         }
 101         else {
 102             if (buf16->utf16)
 103                 free(buf16->utf16);
 104             buf16->utf16 = 0;
 105             buf16->utf16_len = 0;
 106             buf16->utf16_cap = 0;
 107         }
 108     }
 109
 110     return buf16;
 111 };
 112
 113
 114 struct icu_buf_utf16 * icu_buf_utf16_copy(struct icu_buf_utf16 * dest16,
 115                                           struct icu_buf_utf16 * src16)
 116 {
 117     if(!dest16 || !src16
 118        || dest16 == src16)
 119         return 0;
 120
 121     if (dest16->utf16_cap < src16->utf16_len)
 122         icu_buf_utf16_resize(dest16, src16->utf16_len * 2);
 123
 124     u_strncpy(dest16->utf16, src16->utf16, src16->utf16_len);
 125     dest16->utf16_len = src16->utf16_len;
 126
 127     return dest16;
 128 };
 129
 130
 131 void icu_buf_utf16_destroy(struct icu_buf_utf16 * buf16)
 132 {
 133     if (buf16){
 134         if (buf16->utf16)
 135             free(buf16->utf16);
 136         free(buf16);
 137     }
 138 };
 139
 140
 141
 142
 143
 144
 145 struct icu_buf_utf8 * icu_buf_utf8_create(size_t capacity)
 146 {
 147     struct icu_buf_utf8 * buf8
 148         = (struct icu_buf_utf8 *) malloc(sizeof(struct icu_buf_utf8));
 149
 150     buf8->utf8 = 0;
 151     buf8->utf8_len = 0;
 152     buf8->utf8_cap = 0;
 153
 154     if (capacity > 0){
 155         buf8->utf8 = (uint8_t *) malloc(sizeof(uint8_t) * capacity);
 156         buf8->utf8[0] = (uint8_t) 0;
 157         buf8->utf8_cap = capacity;
 158     }
 159     return buf8;
 160 };
 161
 162
 163
 164 struct icu_buf_utf8 * icu_buf_utf8_resize(struct icu_buf_utf8 * buf8,
 165                                           size_t capacity)
 166 {
 167     if (buf8){
 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,
 174                                           sizeof(uint8_t) * capacity);
 175             buf8->utf8[0] = (uint8_t) 0;
 176             buf8->utf8_len = 0;
 177             buf8->utf8_cap = capacity;
 178         }
 179         else {
 180             if (buf8->utf8)
 181                 free(buf8->utf8);
 182             buf8->utf8 = 0;
 183             buf8->utf8_len = 0;
 184             buf8->utf8_cap = 0;
 185         }
 186     }
 187
 188     return buf8;
 189 };
 190
 191
 192 struct icu_buf_utf8 * icu_buf_utf8_copy(struct icu_buf_utf8 * dest8,
 193                                           struct icu_buf_utf8 * src8)
 194 {
 195     if(!dest8 || !src8
 196        || dest8 == src8)
 197         return 0;
 198
 199
 200     if (dest8->utf8_cap < src8->utf8_len)
 201         icu_buf_utf8_resize(dest8, src8->utf8_len * 2);
 202
 203     strncpy((char*) dest8->utf8, (char*) src8->utf8, src8->utf8_len);
 204
 205     return dest8;
 206 };
 207
 208
 209
 210 void icu_buf_utf8_destroy(struct icu_buf_utf8 * buf8)
 211 {
 212     if (buf8){
 213         if (buf8->utf8)
 214             free(buf8->utf8);
 215         free(buf8);
 216     }
 217 };
 218
 219
 220
 221 UErrorCode icu_utf16_from_utf8(struct icu_buf_utf16 * dest16,
 222                                struct icu_buf_utf8 * src8,
 223                                UErrorCode * status)
 224 {
 225     int32_t utf16_len = 0;
 226
 227     u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 228                   &utf16_len,
 229                   (const char *) src8->utf8, src8->utf8_len, status);
 230
 231     // check for buffer overflow, resize and retry
 232     if (*status == U_BUFFER_OVERFLOW_ERROR
 233         //|| dest16->utf16_len > dest16->utf16_cap
 234         ){
 235         icu_buf_utf16_resize(dest16, utf16_len * 2);
 236         *status = U_ZERO_ERROR;
 237         u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 238                       &utf16_len,
 239                       (const char *) src8->utf8, src8->utf8_len, status);
 240     }
 241
 242     //if (*status != U_BUFFER_OVERFLOW_ERROR
 243     if (U_SUCCESS(*status)
 244         && utf16_len < dest16->utf16_cap)
 245         dest16->utf16_len = utf16_len;
 246     else {
 247         dest16->utf16[0] = (UChar) 0;
 248         dest16->utf16_len = 0;
 249     }
 250
 251     return *status;
 252 };
 253
 254
 255
 256 UErrorCode icu_utf16_from_utf8_cstr(struct icu_buf_utf16 * dest16,
 257                                     const char * src8cstr,
 258                                     UErrorCode * status)
 259 {
 260     size_t src8cstr_len = 0;
 261     int32_t utf16_len = 0;
 262
 263     src8cstr_len = strlen(src8cstr);
 264
 265     u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 266                   &utf16_len,
 267                   src8cstr, src8cstr_len, status);
 268
 269     // check for buffer overflow, resize and retry
 270     if (*status == U_BUFFER_OVERFLOW_ERROR
 271         //|| dest16->utf16_len > dest16->utf16_cap
 272         ){
 273         icu_buf_utf16_resize(dest16, utf16_len * 2);
 274         *status = U_ZERO_ERROR;
 275         u_strFromUTF8(dest16->utf16, dest16->utf16_cap,
 276                       &utf16_len,
 277                       src8cstr, src8cstr_len, status);
 278     }
 279
 280     //  if (*status != U_BUFFER_OVERFLOW_ERROR
 281     if (U_SUCCESS(*status)
 282         && utf16_len < dest16->utf16_cap)
 283         dest16->utf16_len = utf16_len;
 284     else {
 285         dest16->utf16[0] = (UChar) 0;
 286         dest16->utf16_len = 0;
 287     }
 288
 289     return *status;
 290 };
 291
 292
 293
 294
 295 UErrorCode icu_utf16_to_utf8(struct icu_buf_utf8 * dest8,
 296                              struct icu_buf_utf16 * src16,
 297                              UErrorCode * status)
 298 {
 299     int32_t utf8_len = 0;
 300
 301     u_strToUTF8((char *) dest8->utf8, dest8->utf8_cap,
 302                 &utf8_len,
 303                 src16->utf16, src16->utf16_len, status);
 304
 305     // check for buffer overflow, resize and retry
 306     if (*status == U_BUFFER_OVERFLOW_ERROR
 307         //|| dest8->utf8_len > dest8->utf8_cap
 308         ){
 309         icu_buf_utf8_resize(dest8, utf8_len * 2);
 310         *status = U_ZERO_ERROR;
 311         u_strToUTF8((char *) dest8->utf8, dest8->utf8_cap,
 312                     &utf8_len,
 313                     src16->utf16, src16->utf16_len, status);
 314
 315     }
 316
 317     //if (*status != U_BUFFER_OVERFLOW_ERROR
 318     if (U_SUCCESS(*status)
 319         && utf8_len < dest8->utf8_cap)
 320         dest8->utf8_len = utf8_len;
 321     else {
 322         dest8->utf8[0] = (uint8_t) 0;
 323         dest8->utf8_len = 0;
 324     }
 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         break;
 342     case 'u':
 343         break;
 344     case 't':
 345         break;
 346     case 'f':
 347         break;
 348     default:
 349         icu_casemap_destroy(casemap);
 350         return 0;
 351     }
 352
 353     return casemap;
 354 };
 355
 356 void icu_casemap_destroy(struct icu_casemap * casemap)
 357 {
 358     if (casemap)
 359         free(casemap);
 360 };
 361
 362
 363 int icu_casemap_casemap(struct icu_casemap * casemap,
 364                         struct icu_buf_utf16 * dest16,
 365                         struct icu_buf_utf16 * src16,
 366                         UErrorCode *status)
 367 {
 368     if(!casemap)
 369         return 0;
 370
 371     return icu_utf16_casemap(dest16, src16,
 372                              casemap->locale, 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     switch(action) {
 384     case 'l':
 385         dest16_len = u_strToLower(dest16->utf16, dest16->utf16_cap,
 386                                   src16->utf16, src16->utf16_len,
 387                                   locale, status);
 388         break;
 389     case 'u':
 390         dest16_len = u_strToUpper(dest16->utf16, dest16->utf16_cap,
 391                                   src16->utf16, src16->utf16_len,
 392                                   locale, status);
 393         break;
 394     case 't':
 395         dest16_len = u_strToTitle(dest16->utf16, dest16->utf16_cap,
 396                                   src16->utf16, src16->utf16_len,
 397                                   0, locale, status);
 398         break;
 399     case 'f':
 400         dest16_len = u_strFoldCase(dest16->utf16, dest16->utf16_cap,
 401                                    src16->utf16, src16->utf16_len,
 402                                    U_FOLD_CASE_DEFAULT, status);
 403         break;
 404
 405     default:
 406         return U_UNSUPPORTED_ERROR;
 407         break;
 408     }
 409
 410     // check for buffer overflow, resize and retry
 411     if (*status == U_BUFFER_OVERFLOW_ERROR
 412         && dest16 != src16        // do not resize if in-place conversion
 413         //|| dest16_len > dest16->utf16_cap
 414         ){
 415         icu_buf_utf16_resize(dest16, dest16_len * 2);
 416         *status = U_ZERO_ERROR;
 417
 418
 419         switch(action) {
 420         case 'l':
 421             dest16_len = u_strToLower(dest16->utf16, dest16->utf16_cap,
 422                                       src16->utf16, src16->utf16_len,
 423                                       locale, status);
 424             break;
 425         case 'u':
 426             dest16_len = u_strToUpper(dest16->utf16, dest16->utf16_cap,
 427                                       src16->utf16, src16->utf16_len,
 428                                       locale, status);
 429             break;
 430         case 't':
 431             dest16_len = u_strToTitle(dest16->utf16, dest16->utf16_cap,
 432                                       src16->utf16, src16->utf16_len,
 433                                       0, locale, status);
 434             break;
 435         case 'f':
 436             dest16_len = u_strFoldCase(dest16->utf16, dest16->utf16_cap,
 437                                        src16->utf16, src16->utf16_len,
 438                                        U_FOLD_CASE_DEFAULT, status);
 439             break;
 440
 441         default:
 442             return U_UNSUPPORTED_ERROR;
 443             break;
 444         }
 445     }
 446
 447     if (U_SUCCESS(*status)
 448         && dest16_len < dest16->utf16_cap)
 449         dest16->utf16_len = dest16_len;
 450     else {
 451         dest16->utf16[0] = (UChar) 0;
 452         dest16->utf16_len = 0;
 453     }
 454
 455     return *status;
 456 };
 457
 458
 459
 460 UErrorCode icu_sortkey8_from_utf16(UCollator *coll,
 461                                    struct icu_buf_utf8 * dest8,
 462                                    struct icu_buf_utf16 * src16,
 463                                    UErrorCode * status)
 464 {
 465
 466     int32_t sortkey_len = 0;
 467
 468     sortkey_len = ucol_getSortKey(coll, src16->utf16, src16->utf16_len,
 469                                   dest8->utf8, dest8->utf8_cap);
 470
 471     // check for buffer overflow, resize and retry
 472     if (sortkey_len > dest8->utf8_cap) {
 473         icu_buf_utf8_resize(dest8, sortkey_len * 2);
 474         sortkey_len = ucol_getSortKey(coll, src16->utf16, src16->utf16_len,
 475                                       dest8->utf8, dest8->utf8_cap);
 476     }
 477
 478     if (U_SUCCESS(*status)
 479         && sortkey_len > 0)
 480         dest8->utf8_len = sortkey_len;
 481     else {
 482         dest8->utf8[0] = (UChar) 0;
 483         dest8->utf8_len = 0;
 484     }
 485
 486     return sortkey_len;
 487 };
 488
 489
 490
 491 struct icu_tokenizer * icu_tokenizer_create(const char *locale, char action,
 492                                             UErrorCode *status)
 493 {
 494     struct icu_tokenizer * tokenizer
 495         = (struct icu_tokenizer *) malloc(sizeof(struct icu_tokenizer));
 496
 497     strcpy(tokenizer->locale, locale);
 498     tokenizer->action = action;
 499     tokenizer->bi = 0;
 500     tokenizer->buf16 = 0;
 501     tokenizer->token_count = 0;
 502     tokenizer->token_id = 0;
 503     tokenizer->token_start = 0;
 504     tokenizer->token_end = 0;
 505
 506
 507     switch(tokenizer->action) {
 508     case 'l':
 509         tokenizer->bi
 510             = ubrk_open(UBRK_LINE, tokenizer->locale,
 511                         0, 0, status);
 512         break;
 513     case 's':
 514         tokenizer->bi
 515             = ubrk_open(UBRK_SENTENCE, tokenizer->locale,
 516                         0, 0, status);
 517         break;
 518     case 'w':
 519         tokenizer->bi
 520             = ubrk_open(UBRK_WORD, tokenizer->locale,
 521                         0, 0, status);
 522         break;
 523     case 'c':
 524         tokenizer->bi
 525             = ubrk_open(UBRK_CHARACTER, tokenizer->locale,
 526                         0, 0, status);
 527         break;
 528     case 't':
 529         tokenizer->bi
 530             = ubrk_open(UBRK_TITLE, tokenizer->locale,
 531                         0, 0, status);
 532         break;
 533     default:
 534         *status = U_UNSUPPORTED_ERROR;
 535         return 0;
 536         break;
 537     }
 538
 539     // ICU error stuff is a very  funny business
 540     if (U_SUCCESS(*status))
 541         return tokenizer;
 542
 543     // freeing if failed
 544     icu_tokenizer_destroy(tokenizer);
 545     return 0;
 546 };
 547
 548 void icu_tokenizer_destroy(struct icu_tokenizer * tokenizer)
 549 {
 550     if (tokenizer) {
 551         if (tokenizer->bi)
 552             ubrk_close(tokenizer->bi);
 553         free(tokenizer);
 554     }
 555 };
 556
 557 int icu_tokenizer_attach(struct icu_tokenizer * tokenizer,
 558                          struct icu_buf_utf16 * src16,
 559                          UErrorCode *status)
 560 {
 561     if (!tokenizer || !tokenizer->bi || !src16)
 562         return 0;
 563
 564
 565     tokenizer->buf16 = src16;
 566     tokenizer->token_count = 0;
 567     tokenizer->token_id = 0;
 568     tokenizer->token_start = 0;
 569     tokenizer->token_end = 0;
 570
 571     ubrk_setText(tokenizer->bi, src16->utf16, src16->utf16_len, status);
 572
 573
 574     if (U_FAILURE(*status))
 575         return 0;
 576
 577     return 1;
 578 };
 579
 580 int32_t icu_tokenizer_next_token(struct icu_tokenizer * tokenizer,
 581                          struct icu_buf_utf16 * tkn16,
 582                          UErrorCode *status)
 583 {
 584     int32_t tkn_start = 0;
 585     int32_t tkn_end = 0;
 586     int32_t tkn_len = 0;
 587
 588
 589     if (!tokenizer || !tokenizer->bi
 590         || !tokenizer->buf16 || !tokenizer->buf16->utf16_len)
 591         return 0;
 592
 593     // never change tokenizer->buf16 and keep always invariant
 594     // 0 <= tokenizer->token_start
 595     //   <= tokenizer->token_end
 596     //   <= tokenizer->buf16->utf16_len
 597     // returns length of token
 598
 599     if (0 == tokenizer->token_end) // first call
 600         tkn_start = ubrk_first(tokenizer->bi);
 601     else //successive calls
 602         tkn_start = tokenizer->token_end;
 603
 604     // get next position
 605     tkn_end = ubrk_next(tokenizer->bi);
 606
 607     // repairing invariant at end of ubrk, which is UBRK_DONE = -1
 608     if (UBRK_DONE == tkn_end)
 609         tkn_end = tokenizer->buf16->utf16_len;
 610
 611     // copy out if everything is well
 612     if(U_FAILURE(*status))
 613         return 0;
 614
 615     // everything OK, now update internal state
 616     tkn_len = tkn_end - tkn_start;
 617
 618     if (0 < tkn_len){
 619         tokenizer->token_count++;
 620         tokenizer->token_id++;
 621     } else {
 622         tokenizer->token_id = 0;
 623     }
 624     tokenizer->token_start = tkn_start;
 625     tokenizer->token_end = tkn_end;
 626
 627
 628     // copying into token buffer if it exists
 629     if (tkn16){
 630         if (tkn16->utf16_cap < tkn_len)
 631             icu_buf_utf16_resize(tkn16, (size_t) tkn_len * 2);
 632
 633         u_strncpy(tkn16->utf16, &(tokenizer->buf16->utf16)[tkn_start],
 634                   tkn_len);
 635
 636         tkn16->utf16_len = tkn_len;
 637     }
 638
 639     return tkn_len;
 640 }
 641
 642
 643 int32_t icu_tokenizer_token_id(struct icu_tokenizer * tokenizer)
 644 {
 645     return tokenizer->token_id;
 646 };
 647
 648 int32_t icu_tokenizer_token_start(struct icu_tokenizer * tokenizer)
 649 {
 650     return tokenizer->token_start;
 651 };
 652
 653 int32_t icu_tokenizer_token_end(struct icu_tokenizer * tokenizer)
 654 {
 655     return tokenizer->token_end;
 656 };
 657
 658 int32_t icu_tokenizer_token_length(struct icu_tokenizer * tokenizer)
 659 {
 660     return (tokenizer->token_end - tokenizer->token_start);
 661 };
 662
 663 int32_t icu_tokenizer_token_count(struct icu_tokenizer * tokenizer)
 664 {
 665     return tokenizer->token_count;
 666 };
 667
 668
 669
 670 struct icu_normalizer * icu_normalizer_create(const char *rules, char action,
 671                                               UErrorCode *status)
 672 {
 673
 674     struct icu_normalizer * normalizer
 675         = (struct icu_normalizer *) malloc(sizeof(struct icu_normalizer));
 676
 677     normalizer->action = action;
 678     normalizer->trans = 0;
 679     normalizer->rules16 =  icu_buf_utf16_create(0);
 680     icu_utf16_from_utf8_cstr(normalizer->rules16, rules, status);
 681
 682     switch(normalizer->action) {
 683     case 'f':
 684         normalizer->trans
 685             = utrans_openU(normalizer->rules16->utf16,
 686                            normalizer->rules16->utf16_len,
 687                            UTRANS_FORWARD,
 688                            0, 0,
 689                            normalizer->parse_error, status);
 690         break;
 691     case 'r':
 692         normalizer->trans
 693             = utrans_openU(normalizer->rules16->utf16,
 694                            normalizer->rules16->utf16_len,
 695                            UTRANS_REVERSE ,
 696                            0, 0,
 697                            normalizer->parse_error, status);
 698         break;
 699     default:
 700         *status = U_UNSUPPORTED_ERROR;
 701         return 0;
 702         break;
 703     }
 704
 705     if (U_SUCCESS(*status))
 706         return normalizer;
 707
 708     // freeing if failed
 709     icu_normalizer_destroy(normalizer);
 710     return 0;
 711 };
 712
 713
 714 void icu_normalizer_destroy(struct icu_normalizer * normalizer){
 715     if (normalizer) {
 716         if (normalizer->rules16)
 717             icu_buf_utf16_destroy(normalizer->rules16);
 718         if (normalizer->trans)
 719             utrans_close(normalizer->trans);
 720         free(normalizer);
 721     }
 722 };
 723
 724
 725
 726 int icu_normalizer_normalize(struct icu_normalizer * normalizer,
 727                              struct icu_buf_utf16 * dest16,
 728                              struct icu_buf_utf16 * src16,
 729                              UErrorCode *status)
 730 {
 731     if (!normalizer || !normalizer->trans || !src16 || !dest16)
 732         return 0;
 733
 734     if (!icu_buf_utf16_copy(dest16, src16))
 735         return 0;
 736
 737     utrans_transUChars (normalizer->trans,
 738                         dest16->utf16, &(dest16->utf16_len),
 739                         dest16->utf16_cap,
 740                         0, &(src16->utf16_len), status);
 741
 742     if (U_FAILURE(*status)){
 743         dest16->utf16[0] = (UChar) 0;
 744         dest16->utf16_len = 0;
 745     }
 746
 747     return dest16->utf16_len;
 748 }
 749
 750
 751
 752
 753 struct icu_chain_step * icu_chain_step_create(struct icu_chain * chain,
 754                                               enum icu_chain_step_type type,
 755                                               const uint8_t * rule,
 756                                               struct icu_buf_utf16 * buf16,
 757                                               UErrorCode *status)
 758 {
 759     struct icu_chain_step * step = 0;
 760
 761     if(!chain || !type || !rule)
 762         return 0;
 763
 764     step = (struct icu_chain_step *) malloc(sizeof(struct icu_chain_step));
 765
 766     step->type = type;
 767     step->more_tokens = 0;
 768     step->need_new_token = 1;
 769
 770     if (buf16)
 771         step->buf16 = buf16;
 772     else
 773         step->buf16 = 0;
 774
 775     // create auxilary objects
 776     switch(step->type) {
 777     case ICU_chain_step_type_display:
 778         break;
 779     case ICU_chain_step_type_norm:
 780         break;
 781     case ICU_chain_step_type_sort:
 782         break;
 783     case ICU_chain_step_type_casemap:
 784         step->u.casemap = icu_casemap_create((char *) chain->locale,
 785                                              (char) rule[0], status);
 786         break;
 787     case ICU_chain_step_type_normalize:
 788         step->u.normalizer = icu_normalizer_create((char *) rule, 'f', status);
 789         break;
 790     case ICU_chain_step_type_tokenize:
 791         step->u.tokenizer = icu_tokenizer_create((char *) chain->locale,
 792                                                  (char) rule[0], status);
 793         break;
 794     default:
 795         break;
 796     }
 797
 798     return step;
 799 };
 800
 801
 802 void icu_chain_step_destroy(struct icu_chain_step * step){
 803
 804     if (!step)
 805         return;
 806
 807     icu_chain_step_destroy(step->previous);
 808
 809     switch(step->type) {
 810     case ICU_chain_step_type_display:
 811         break;
 812     case ICU_chain_step_type_norm:
 813         break;
 814     case ICU_chain_step_type_sort:
 815         break;
 816     case ICU_chain_step_type_casemap:
 817         icu_casemap_destroy(step->u.casemap);
 818         icu_buf_utf16_destroy(step->buf16);
 819         break;
 820     case ICU_chain_step_type_normalize:
 821         icu_normalizer_destroy(step->u.normalizer);
 822         icu_buf_utf16_destroy(step->buf16);
 823         break;
 824     case ICU_chain_step_type_tokenize:
 825         icu_tokenizer_destroy(step->u.tokenizer);
 826         icu_buf_utf16_destroy(step->buf16);
 827         break;
 828     default:
 829         break;
 830     }
 831
 832
 833 };
 834
 835
 836
 837 struct icu_chain * icu_chain_create(const uint8_t * identifier,
 838                                     const uint8_t * locale)
 839 {
 840
 841     struct icu_chain * chain
 842         = (struct icu_chain *) malloc(sizeof(struct icu_chain));
 843
 844     strncpy((char *) chain->identifier, (const char *) identifier, 128);
 845     chain->identifier[128 - 1] = '\0';
 846     strncpy((char *) chain->locale, (const char *) locale, 16);
 847     chain->locale[16 - 1] = '\0';
 848
 849     chain->token_count = 0;
 850
 851     chain->display8 = icu_buf_utf8_create(0);
 852     chain->norm8 = icu_buf_utf8_create(0);
 853     chain->sort8 = icu_buf_utf8_create(0);
 854
 855     chain->src16 = icu_buf_utf16_create(0);
 856
 857     chain->steps = 0;
 858
 859     return chain;
 860 };
 861
 862
 863 void icu_chain_destroy(struct icu_chain * chain)
 864 {
 865     if (chain){
 866         icu_buf_utf8_destroy(chain->display8);
 867         icu_buf_utf8_destroy(chain->norm8);
 868         icu_buf_utf8_destroy(chain->sort8);
 869
 870         icu_buf_utf16_destroy(chain->src16);
 871
 872         icu_chain_step_destroy(chain->steps);
 873     }
 874 };
 875
 876
 877
 878 struct icu_chain * icu_chain_xml_config(xmlNode *xml_node,
 879                                         UErrorCode * status){
 880
 881     xmlNode *node = 0;
 882     struct icu_chain * chain = 0;
 883
 884     if (!xml_node
 885         ||xml_node->type != XML_ELEMENT_NODE
 886         || strcmp((const char *) xml_node->name, "icu_chain"))
 887
 888         return 0;
 889
 890     xmlChar *xml_id = xmlGetProp(xml_node, (xmlChar *) "id");
 891     xmlChar *xml_locale = xmlGetProp(xml_node, (xmlChar *) "locale");
 892
 893     if (!xml_id || !strlen((const char *) xml_id)
 894         || !xml_locale || !strlen((const char *) xml_locale))
 895         return 0;
 896
 897     chain = icu_chain_create((const uint8_t *) xml_id,
 898                              (const uint8_t *) xml_locale);
 899
 900     if (!chain)
 901         return 0;
 902
 903     for (node = xml_node->children; node; node = node->next)
 904     {
 905         if (node->type != XML_ELEMENT_NODE)
 906             continue;
 907
 908         xmlChar *xml_rule = xmlGetProp(node, (xmlChar *) "rule");
 909         struct icu_chain_step * step = 0;
 910
 911         if (!strcmp((const char *) node->name,
 912                     (const char *) "casemap")){
 913             step = icu_chain_insert_step(chain, ICU_chain_step_type_casemap,
 914                                          (const uint8_t *) xml_rule, status);
 915         }
 916         else if (!strcmp((const char *) node->name,
 917                          (const char *) "normalize")){
 918             step = icu_chain_insert_step(chain, ICU_chain_step_type_normalize,
 919                                          (const uint8_t *) xml_rule, status);
 920         }
 921         else if (!strcmp((const char *) node->name,
 922                          (const char *) "tokenize")){
 923             step = icu_chain_insert_step(chain, ICU_chain_step_type_tokenize,
 924                                          (const uint8_t *) xml_rule, status);
 925         }
 926         else if (!strcmp((const char *) node->name,
 927                          (const char *) "display")){
 928             step = icu_chain_insert_step(chain, ICU_chain_step_type_display,
 929                                          (const uint8_t *) "", status);
 930         }
 931         else if (!strcmp((const char *) node->name,
 932                          (const char *) "normal")){
 933             step = icu_chain_insert_step(chain, ICU_chain_step_type_norm,
 934                                          (const uint8_t *) "", status);
 935         }
 936         else if (!strcmp((const char *) node->name,
 937                          (const char *) "sort")){
 938             step = icu_chain_insert_step(chain, ICU_chain_step_type_sort,
 939                                          (const uint8_t *) "", status);
 940         }
 941
 942         if (!step || U_FAILURE(*status)){
 943             icu_chain_destroy(chain);
 944             return 0;
 945         }
 946
 947
 948     }
 949
 950     return chain;
 951 };
 952
 953
 954
 955 struct icu_chain_step * icu_chain_insert_step(struct icu_chain * chain,
 956                                               enum icu_chain_step_type type,
 957                                               const uint8_t * rule,
 958                                               UErrorCode *status)
 959 {
 960     struct icu_chain_step * step = 0;
 961     struct icu_buf_utf16 * src16 = 0;
 962     struct icu_buf_utf16 * buf16 = 0;
 963
 964     if (!chain || !type || !rule)
 965         return 0;
 966
 967     // assign utf16 src buffers as needed
 968     if (chain->steps && chain->steps->buf16)
 969         src16 = chain->steps->buf16;
 970     else if (chain->src16)
 971         src16 = chain->src16;
 972     else
 973         return 0;
 974
 975
 976     // create utf16 destination buffers as needed, or
 977     switch(type) {
 978     case ICU_chain_step_type_display:
 979         buf16 = src16;
 980         break;
 981     case ICU_chain_step_type_norm:
 982         buf16 = src16;
 983         break;
 984     case ICU_chain_step_type_sort:
 985         buf16 = src16;
 986         break;
 987     case ICU_chain_step_type_casemap:
 988         buf16 = icu_buf_utf16_create(0);
 989         break;
 990     case ICU_chain_step_type_normalize:
 991         buf16 = icu_buf_utf16_create(0);
 992         break;
 993     case ICU_chain_step_type_tokenize:
 994         buf16 = icu_buf_utf16_create(0);
 995         break;
 996     default:
 997         break;
 998     }
 999
1000     // create actual chain step with this buffer
1001     step = icu_chain_step_create(chain, type, rule, buf16, status);
1002
1003     step->previous = chain->steps;
1004     chain->steps = step;
1005
1006     return step;
1007 };
1008
1009
1010 int icu_chain_step_next_token(struct icu_chain * chain,
1011                               struct icu_chain_step * step,
1012                               UErrorCode *status)
1013 {
1014     struct icu_buf_utf16 * src16 = 0;
1015
1016     //printf("icu_chain_step_next_token %d\n", (int) step);
1017
1018     if (!chain || !chain->src16 || !step || !step->more_tokens)
1019         return 0;
1020
1021     // assign utf16 src buffers as neeed, advance in previous steps
1022     // tokens until non-zero token met, and setting stop condition
1023     if (step->previous){
1024         src16 = step->previous->buf16;
1025         if (step->need_new_token)
1026             //while (step->more_tokens &&  !src16->utf16_len)
1027                 step->more_tokens
1028                     = icu_chain_step_next_token(chain, step->previous, status);
1029     }
1030     else { // first step can only work once on chain->src16 input buffer
1031         src16 = chain->src16;
1032         step->more_tokens = 1;
1033     }
1034
1035     // stop if nothing to process
1036     // i.e new token source was not properly assigned
1037     if (!step->more_tokens || !src16) // || !src16->utf16_len
1038         return 0;
1039
1040     //printf("icu_chain_step_next_token %d working\n", (int) step);
1041
1042
1043     // perform the work, eventually put this steps output in
1044     // step->buf16 or the chains UTF8 output buffers
1045     switch(step->type) {
1046     case ICU_chain_step_type_display:
1047         icu_utf16_to_utf8(chain->display8, src16, status);
1048         break;
1049     case ICU_chain_step_type_norm:
1050         icu_utf16_to_utf8(chain->norm8, src16, status);
1051         break;
1052     case ICU_chain_step_type_sort:
1053         icu_utf16_to_utf8(chain->sort8, src16, status);
1054         break;
1055     case ICU_chain_step_type_casemap:
1056         icu_casemap_casemap(step->u.casemap,
1057                             step->buf16, src16, status);
1058         break;
1059     case ICU_chain_step_type_normalize:
1060         icu_normalizer_normalize(step->u.normalizer,
1061                                  step->buf16, src16, status);
1062         break;
1063     case ICU_chain_step_type_tokenize:
1064         // attach to new src16 token only first time during splitting
1065         if (step->need_new_token){
1066             icu_tokenizer_attach(step->u.tokenizer, src16, status);
1067             step->need_new_token = 0;
1068         }
1069         // splitting one src16 token into multiple buf16 tokens
1070         step->more_tokens
1071             = icu_tokenizer_next_token(step->u.tokenizer,
1072                                        step->buf16, status);
1073         // make sure to get new previous token if this one had been used up
1074         if (step->previous && !step->more_tokens){
1075             if (icu_chain_step_next_token(chain, step->previous, status)){
1076                 icu_tokenizer_attach(step->u.tokenizer, src16, status);
1077                 step->need_new_token = 0;
1078                 step->more_tokens
1079                     = icu_tokenizer_next_token(step->u.tokenizer,
1080                                                step->buf16, status);
1081             }
1082         }
1083         if (0 == step->more_tokens)
1084             return 0;
1085         break;
1086     default:
1087         return 0;
1088         break;
1089     }
1090
1091
1092
1093     // stop further token processing if last step and
1094     // new tokens are needed from previous (non-existing) step
1095     if (!step->previous && step->need_new_token)
1096         step->more_tokens = 0;
1097
1098     //printf("%d %d %d\n",
1099     //       step->more_tokens, src16->utf16_len, step->buf16->utf16_len);
1100
1101
1102     if (U_FAILURE(*status))
1103         return 0;
1104
1105     return 1;
1106 };
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     stp = chain->steps;
1120
1121     // clear token count
1122     chain->token_count = 0;
1123
1124     // clear all steps stop states
1125
1126     while (stp){
1127         stp->more_tokens = 1;
1128         stp = stp->previous;
1129     }
1130
1131     // finally convert UTF8 to UTF16 string
1132     icu_utf16_from_utf8_cstr(chain->src16, src8cstr, status);
1133
1134     if (U_FAILURE(*status))
1135         return 0;
1136
1137     return 1;
1138 };
1139
1140
1141
1142 int icu_chain_next_token(struct icu_chain * chain,
1143                          UErrorCode *status)
1144 {
1145     int success = 0;
1146
1147     if (!chain || !chain->steps)
1148         return 0;
1149
1150     success = icu_chain_step_next_token(chain, chain->steps, status);
1151
1152     if (success){
1153         chain->token_count++;
1154         return chain->token_count;
1155     }
1156
1157     return 0;
1158 };
1159
1160 int icu_chain_get_token_count(struct icu_chain * chain)
1161 {
1162     if (!chain)
1163         return 0;
1164
1165     return chain->token_count;
1166 };
1167
1168
1169
1170 const char * icu_chain_get_display(struct icu_chain * chain)
1171 {
1172     if (chain->display8)
1173         return (const char *) chain->display8->utf8;
1174
1175     return 0;
1176 };
1177
1178 const char * icu_chain_get_norm(struct icu_chain * chain)
1179 {
1180     if (chain->norm8)
1181         return (const char *) chain->norm8->utf8;
1182
1183     return 0;
1184 };
1185
1186 const char * icu_chain_get_sort(struct icu_chain * chain)
1187 {
1188     if (chain->sort8)
1189         return (const char *) chain->sort8->utf8;
1190
1191     return 0;
1192 };
1193
1194
1195
1196
1197 #endif // HAVE_ICU
1198
1199
1200
1201
1202 /*
1203  * Local variables:
1204  * c-basic-offset: 4
1205  * indent-tabs-mode: nil
1206  * End:
1207  * vim: shiftwidth=4 tabstop=8 expandtab
1208  */