intl/unicharutil/ucdata.c

   1 /*
   2  * Copyright 1996, 1997, 1998 Computing Research Labs,
   3  * New Mexico State University
   4  *
   5  * Permission is hereby granted, free of charge, to any person obtaining a
   6  * copy of this software and associated documentation files (the "Software"),
   7  * to deal in the Software without restriction, including without limitation
   8  * the rights to use, copy, modify, merge, publish, distribute, sublicense,
   9  * and/or sell copies of the Software, and to permit persons to whom the
  10  * Software is furnished to do so, subject to the following conditions:
  11  *
  12  * The above copyright notice and this permission notice shall be included in
  13  * all copies or substantial portions of the Software.
  14  *
  15  * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
  16  * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
  17  * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
  18  * THE COMPUTING RESEARCH LAB OR NEW MEXICO STATE UNIVERSITY BE LIABLE FOR ANY
  19  * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT
  20  * OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR
  21  * THE USE OR OTHER DEALINGS IN THE SOFTWARE.
  22  */
  23 #ifndef lint
  24 #ifdef __GNUC__
  25 static char rcsid[] __attribute__ ((unused)) = "$Id: ucdata.c,v 1.1 1999/01/08 00:19:11 ftang%netscape.com Exp $";
  26 #else
  27 static char rcsid[] = "$Id: ucdata.c,v 1.1 1999/01/08 00:19:11 ftang%netscape.com Exp $";
  28 #endif
  29 #endif
  30
  31 #include <stdio.h>
  32 #include <stdlib.h>
  33 #include <string.h>
  34 #ifndef WIN32
  35 #include <unistd.h>
  36 #endif
  37
  38 #include "ucdata.h"
  39
  40 /**************************************************************************
  41  *
  42  * Miscellaneous types, data, and support functions.
  43  *
  44  **************************************************************************/
  45
  46 typedef struct {
  47     unsigned short bom;
  48     unsigned short cnt;
  49     union {
  50         unsigned long bytes;
  51         unsigned short len[2];
  52     } size;
  53 } _ucheader_t;
  54
  55 /*
  56  * A simple array of 32-bit masks for lookup.
  57  */
  58 static unsigned long masks32[32] = {
  59     0x00000001, 0x00000002, 0x00000004, 0x00000008, 0x00000010, 0x00000020,
  60     0x00000040, 0x00000080, 0x00000100, 0x00000200, 0x00000400, 0x00000800,
  61     0x00001000, 0x00002000, 0x00004000, 0x00008000, 0x00010000, 0x00020000,
  62     0x00040000, 0x00080000, 0x00100000, 0x00200000, 0x00400000, 0x00800000,
  63     0x01000000, 0x02000000, 0x04000000, 0x08000000, 0x10000000, 0x20000000,
  64     0x40000000, 0x80000000
  65 };
  66
  67 #define endian_short(cc) (((cc) >> 8) | (((cc) & 0xff) << 8))
  68 #define endian_long(cc) ((((cc) & 0xff) << 24)|((((cc) >> 8) & 0xff) << 16)|\
  69                         ((((cc) >> 16) & 0xff) << 8)|((cc) >> 24))
  70
  71 static FILE *
  72 #ifdef __STDC__
  73 _ucopenfile(char *paths, char *filename, char *mode)
  74 #else
  75 _ucopenfile(paths, filename, mode)
  76 char *paths, *filename, *mode;
  77 #endif
  78 {
  79     FILE *f;
  80     char *fp, *dp, *pp, path[BUFSIZ];
  81
  82     if (filename == 0 || *filename == 0)
  83       return 0;
  84
  85     dp = paths;
  86     while (dp && *dp) {
  87         pp = path;
  88         while (*dp && *dp != ':')
  89           *pp++ = *dp++;
  90         *pp++ = '/';
  91
  92         fp = filename;
  93         while (*fp)
  94           *pp++ = *fp++;
  95         *pp = 0;
  96
  97         if ((f = fopen(path, mode)) != 0)
  98           return f;
  99
 100         if (*dp == ':')
 101           dp++;
 102     }
 103
 104     return 0;
 105 }
 106
 107 /**************************************************************************
 108  *
 109  * Support for the character properties.
 110  *
 111  **************************************************************************/
 112
 113 static unsigned long  _ucprop_size;
 114 static unsigned short *_ucprop_offsets;
 115 static unsigned long  *_ucprop_ranges;
 116
 117 static void
 118 #ifdef __STDC__
 119 _ucprop_load(char *paths, int reload)
 120 #else
 121 _ucprop_load(paths, reload)
 122 char *paths;
 123 int reload;
 124 #endif
 125 {
 126     FILE *in;
 127     unsigned long size, i;
 128     _ucheader_t hdr;
 129
 130     if (_ucprop_size > 0) {
 131         if (!reload)
 132           /*
 133            * The character properties have already been loaded.
 134            */
 135           return;
 136
 137         /*
 138          * Unload the current character property data in preparation for
 139          * loading a new copy.  Only the first array has to be deallocated
 140          * because all the memory for the arrays is allocated as a single
 141          * block.
 142          */
 143         free((char *) _ucprop_offsets);
 144         _ucprop_size = 0;
 145     }
 146
 147     if ((in = _ucopenfile(paths, "ctype.dat", "rb")) == 0)
 148       return;
 149
 150     /*
 151      * Load the header.
 152      */
 153     fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
 154
 155     if (hdr.bom == 0xfffe) {
 156         hdr.cnt = endian_short(hdr.cnt);
 157         hdr.size.bytes = endian_long(hdr.size.bytes);
 158     }
 159
 160     if ((_ucprop_size = hdr.cnt) == 0) {
 161         fclose(in);
 162         return;
 163     }
 164
 165     /*
 166      * Allocate all the storage needed for the lookup table.
 167      */
 168     _ucprop_offsets = (unsigned short *) malloc(hdr.size.bytes);
 169
 170     /*
 171      * Calculate the offset into the storage for the ranges.  The offsets
 172      * array is on a 4-byte boundary and one larger than the value provided in
 173      * the header count field.  This means the offset to the ranges must be
 174      * calculated after aligning the count to a 4-byte boundary.
 175      */
 176     if ((size = ((hdr.cnt + 1) * sizeof(unsigned short))) & 3)
 177       size += 4 - (size & 3);
 178     size >>= 1;
 179     _ucprop_ranges = (unsigned long *) (_ucprop_offsets + size);
 180
 181     /*
 182      * Load the offset array.
 183      */
 184     fread((char *) _ucprop_offsets, sizeof(unsigned short), size, in);
 185
 186     /*
 187      * Do an endian swap if necessary.  Don't forget there is an extra node on
 188      * the end with the final index.
 189      */
 190     if (hdr.bom == 0xfffe) {
 191         for (i = 0; i <= _ucprop_size; i++)
 192           _ucprop_offsets[i] = endian_short(_ucprop_offsets[i]);
 193     }
 194
 195     /*
 196      * Load the ranges.  The number of elements is in the last array position
 197      * of the offsets.
 198      */
 199     fread((char *) _ucprop_ranges, sizeof(unsigned long),
 200           _ucprop_offsets[_ucprop_size], in);
 201
 202     fclose(in);
 203
 204     /*
 205      * Do an endian swap if necessary.
 206      */
 207     if (hdr.bom == 0xfffe) {
 208         for (i = 0; i < _ucprop_offsets[_ucprop_size]; i++)
 209           _ucprop_ranges[i] = endian_long(_ucprop_ranges[i]);
 210     }
 211 }
 212
 213 static void
 214 #ifdef __STDC__
 215 _ucprop_unload(void)
 216 #else
 217 _ucprop_unload()
 218 #endif
 219 {
 220     if (_ucprop_size == 0)
 221       return;
 222
 223     /*
 224      * Only need to free the offsets because the memory is allocated as a
 225      * single block.
 226      */
 227     free((char *) _ucprop_offsets);
 228     _ucprop_size = 0;
 229 }
 230
 231 static int
 232 #ifdef __STDC__
 233 _ucprop_lookup(unsigned long code, unsigned long n)
 234 #else
 235 _ucprop_lookup(code, n)
 236 unsigned long code, n;
 237 #endif
 238 {
 239     long l, r, m;
 240
 241     /*
 242      * There is an extra node on the end of the offsets to allow this routine
 243      * to work right.  If the index is 0xffff, then there are no nodes for the
 244      * property.
 245      */
 246     if ((l = _ucprop_offsets[n]) == 0xffff)
 247       return 0;
 248
 249     /*
 250      * Locate the next offset that is not 0xffff.  The sentinel at the end of
 251      * the array is the max index value.
 252      */
 253     for (m = 1;
 254          n + m < _ucprop_size && _ucprop_offsets[n + m] == 0xffff; m++) ;
 255
 256     r = _ucprop_offsets[n + m] - 1;
 257
 258     while (l <= r) {
 259         /*
 260          * Determine a "mid" point and adjust to make sure the mid point is at
 261          * the beginning of a range pair.
 262          */
 263         m = (l + r) >> 1;
 264         m -= (m & 1);
 265         if (code > _ucprop_ranges[m + 1])
 266           l = m + 2;
 267         else if (code < _ucprop_ranges[m])
 268           r = m - 2;
 269         else if (code >= _ucprop_ranges[m] && code <= _ucprop_ranges[m + 1])
 270           return 1;
 271     }
 272     return 0;
 273 }
 274
 275 int
 276 #ifdef __STDC__
 277 ucisprop(unsigned long code, unsigned long mask1, unsigned long mask2)
 278 #else
 279 ucisprop(code, mask1, mask2)
 280 unsigned long code, mask1, mask2;
 281 #endif
 282 {
 283     unsigned long i;
 284
 285     if (mask1 == 0 && mask2 == 0)
 286       return 0;
 287
 288     for (i = 0; mask1 && i < 32; i++) {
 289         if ((mask1 & masks32[i]) && _ucprop_lookup(code, i))
 290           return 1;
 291     }
 292
 293     for (i = 32; mask2 && i < _ucprop_size; i++) {
 294         if ((mask2 & masks32[i & 31]) && _ucprop_lookup(code, i))
 295           return 1;
 296     }
 297
 298     return 0;
 299 }
 300
 301 /**************************************************************************
 302  *
 303  * Support for case mapping.
 304  *
 305  **************************************************************************/
 306
 307 static unsigned long _uccase_size;
 308 static unsigned short _uccase_len[2];
 309 static unsigned long *_uccase_map;
 310
 311 static void
 312 #ifdef __STDC__
 313 _uccase_load(char *paths, int reload)
 314 #else
 315 _uccase_load(paths, reload)
 316 char *paths;
 317 int reload;
 318 #endif
 319 {
 320     FILE *in;
 321     unsigned long i;
 322     _ucheader_t hdr;
 323
 324     if (_uccase_size > 0) {
 325         if (!reload)
 326           /*
 327            * The case mappings have already been loaded.
 328            */
 329           return;
 330
 331         free((char *) _uccase_map);
 332         _uccase_size = 0;
 333     }
 334
 335     if ((in = _ucopenfile(paths, "case.dat", "rb")) == 0)
 336       return;
 337
 338     /*
 339      * Load the header.
 340      */
 341     fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
 342
 343     if (hdr.bom == 0xfffe) {
 344         hdr.cnt = endian_short(hdr.cnt);
 345         hdr.size.len[0] = endian_short(hdr.size.len[0]);
 346         hdr.size.len[1] = endian_short(hdr.size.len[1]);
 347     }
 348
 349     /*
 350      * Set the node count and lengths of the upper and lower case mapping
 351      * tables.
 352      */
 353     _uccase_size = hdr.cnt * 3;
 354     _uccase_len[0] = hdr.size.len[0] * 3;
 355     _uccase_len[1] = hdr.size.len[1] * 3;
 356
 357     _uccase_map = (unsigned long *)
 358         malloc(_uccase_size * sizeof(unsigned long));
 359
 360     /*
 361      * Load the case mapping table.
 362      */
 363     fread((char *) _uccase_map, sizeof(unsigned long), _uccase_size, in);
 364
 365     /*
 366      * Do an endian swap if necessary.
 367      */
 368     if (hdr.bom == 0xfffe) {
 369         for (i = 0; i < _uccase_size; i++)
 370           _uccase_map[i] = endian_long(_uccase_map[i]);
 371     }
 372 }
 373
 374 static void
 375 #ifdef __STDC__
 376 _uccase_unload(void)
 377 #else
 378 _uccase_unload()
 379 #endif
 380 {
 381     if (_uccase_size == 0)
 382       return;
 383
 384     free((char *) _uccase_map);
 385     _uccase_size = 0;
 386 }
 387
 388 static unsigned long
 389 #ifdef __STDC__
 390 _uccase_lookup(unsigned long code, long l, long r, int field)
 391 #else
 392 _uccase_lookup(code, l, r, field)
 393 unsigned long code;
 394 long l, r;
 395 int field;
 396 #endif
 397 {
 398     long m;
 399
 400     /*
 401      * Do the binary search.
 402      */
 403     while (l <= r) {
 404         /*
 405          * Determine a "mid" point and adjust to make sure the mid point is at
 406          * the beginning of a case mapping triple.
 407          */
 408         m = (l + r) >> 1;
 409         m -= (m % 3);
 410         if (code > _uccase_map[m])
 411           l = m + 3;
 412         else if (code < _uccase_map[m])
 413           r = m - 3;
 414         else if (code == _uccase_map[m])
 415           return _uccase_map[m + field];
 416     }
 417
 418     return code;
 419 }
 420
 421 unsigned long
 422 #ifdef __STDC__
 423 uctoupper(unsigned long code)
 424 #else
 425 uctoupper(code)
 426 unsigned long code;
 427 #endif
 428 {
 429     int field;
 430     long l, r;
 431
 432     if (ucisupper(code))
 433       return code;
 434
 435     if (ucislower(code)) {
 436         /*
 437          * The character is lower case.
 438          */
 439         field = 1;
 440         l = _uccase_len[0];
 441         r = (l + _uccase_len[1]) - 1;
 442     } else {
 443         /*
 444          * The character is title case.
 445          */
 446         field = 2;
 447         l = _uccase_len[0] + _uccase_len[1];
 448         r = _uccase_size - 1;
 449     }
 450     return _uccase_lookup(code, l, r, field);
 451 }
 452
 453 unsigned long
 454 #ifdef __STDC__
 455 uctolower(unsigned long code)
 456 #else
 457 uctolower(code)
 458 unsigned long code;
 459 #endif
 460 {
 461     int field;
 462     long l, r;
 463
 464     if (ucislower(code))
 465       return code;
 466
 467     if (ucisupper(code)) {
 468         /*
 469          * The character is upper case.
 470          */
 471         field = 1;
 472         l = 0;
 473         r = _uccase_len[0] - 1;
 474     } else {
 475         /*
 476          * The character is title case.
 477          */
 478         field = 2;
 479         l = _uccase_len[0] + _uccase_len[1];
 480         r = _uccase_size - 1;
 481     }
 482     return _uccase_lookup(code, l, r, field);
 483 }
 484
 485 unsigned long
 486 #ifdef __STDC__
 487 uctotitle(unsigned long code)
 488 #else
 489 uctotitle(code)
 490 unsigned long code;
 491 #endif
 492 {
 493     int field;
 494     long l, r;
 495
 496     if (ucistitle(code))
 497       return code;
 498
 499     /*
 500      * The offset will always be the same for converting to title case.
 501      */
 502     field = 2;
 503
 504     if (ucisupper(code)) {
 505         /*
 506          * The character is upper case.
 507          */
 508         l = 0;
 509         r = _uccase_len[0] - 1;
 510     } else {
 511         /*
 512          * The character is lower case.
 513          */
 514         l = _uccase_len[0];
 515         r = (l + _uccase_len[1]) - 1;
 516     }
 517     return _uccase_lookup(code, l, r, field);
 518 }
 519
 520 /**************************************************************************
 521  *
 522  * Support for decompositions.
 523  *
 524  **************************************************************************/
 525
 526 static unsigned long  _ucdcmp_size;
 527 static unsigned long *_ucdcmp_nodes;
 528 static unsigned long *_ucdcmp_decomp;
 529
 530 static void
 531 #ifdef __STDC__
 532 _ucdcmp_load(char *paths, int reload)
 533 #else
 534 _ucdcmp_load(paths, reload)
 535 char *paths;
 536 int reload;
 537 #endif
 538 {
 539     FILE *in;
 540     unsigned long size, i;
 541     _ucheader_t hdr;
 542
 543     if (_ucdcmp_size > 0) {
 544         if (!reload)
 545           /*
 546            * The decompositions have already been loaded.
 547            */
 548           return;
 549
 550         free((char *) _ucdcmp_nodes);
 551         _ucdcmp_size = 0;
 552     }
 553
 554     if ((in = _ucopenfile(paths, "decomp.dat", "rb")) == 0)
 555       return;
 556
 557     /*
 558      * Load the header.
 559      */
 560     fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
 561
 562     if (hdr.bom == 0xfffe) {
 563         hdr.cnt = endian_short(hdr.cnt);
 564         hdr.size.bytes = endian_long(hdr.size.bytes);
 565     }
 566
 567     _ucdcmp_size = hdr.cnt << 1;
 568     _ucdcmp_nodes = (unsigned long *) malloc(hdr.size.bytes);
 569     _ucdcmp_decomp = _ucdcmp_nodes + (_ucdcmp_size + 1);
 570
 571     /*
 572      * Read the decomposition data in.
 573      */
 574     size = hdr.size.bytes / sizeof(unsigned long);
 575     fread((char *) _ucdcmp_nodes, sizeof(unsigned long), size, in);
 576
 577     /*
 578      * Do an endian swap if necessary.
 579      */
 580     if (hdr.bom == 0xfffe) {
 581         for (i = 0; i < size; i++)
 582           _ucdcmp_nodes[i] = endian_long(_ucdcmp_nodes[i]);
 583     }
 584 }
 585
 586 static void
 587 #ifdef __STDC__
 588 _ucdcmp_unload(void)
 589 #else
 590 _ucdcmp_unload()
 591 #endif
 592 {
 593     if (_ucdcmp_size == 0)
 594       return;
 595
 596     /*
 597      * Only need to free the offsets because the memory is allocated as a
 598      * single block.
 599      */
 600     free((char *) _ucdcmp_nodes);
 601     _ucdcmp_size = 0;
 602 }
 603
 604 int
 605 #ifdef __STDC__
 606 ucdecomp(unsigned long code, unsigned long *num, unsigned long **decomp)
 607 #else
 608 ucdecomp(code, num, decomp)
 609 unsigned long code, *num, **decomp;
 610 #endif
 611 {
 612     long l, r, m;
 613
 614     l = 0;
 615     r = _ucdcmp_nodes[_ucdcmp_size] - 1;
 616
 617     while (l <= r) {
 618         /*
 619          * Determine a "mid" point and adjust to make sure the mid point is at
 620          * the beginning of a code+offset pair.
 621          */
 622         m = (l + r) >> 1;
 623         m -= (m & 1);
 624         if (code > _ucdcmp_nodes[m])
 625           l = m + 2;
 626         else if (code < _ucdcmp_nodes[m])
 627           r = m - 2;
 628         else if (code == _ucdcmp_nodes[m]) {
 629             *num = _ucdcmp_nodes[m + 3] - _ucdcmp_nodes[m + 1];
 630             *decomp = &_ucdcmp_decomp[_ucdcmp_nodes[m + 1]];
 631             return 1;
 632         }
 633     }
 634     return 0;
 635 }
 636
 637 int
 638 #ifdef __STDC__
 639 ucdecomp_hangul(unsigned long code, unsigned long *num, unsigned long decomp[])
 640 #else
 641 ucdecomp_hangul(code, num, decomp)
 642 unsigned long code, *num, decomp[];
 643 #endif
 644 {
 645     if (!ucishangul(code))
 646       return 0;
 647
 648     code -= 0xac00;
 649     decomp[0] = 0x1100 + (unsigned long) (code / 588);
 650     decomp[1] = 0x1161 + (unsigned long) ((code % 588) / 28);
 651     decomp[2] = 0x11a7 + (unsigned long) (code % 28);
 652     *num = (decomp[2] != 0x11a7) ? 3 : 2;
 653
 654     return 1;
 655 }
 656
 657 /**************************************************************************
 658  *
 659  * Support for combining classes.
 660  *
 661  **************************************************************************/
 662
 663 static unsigned long  _uccmcl_size;
 664 static unsigned long *_uccmcl_nodes;
 665
 666 static void
 667 #ifdef __STDC__
 668 _uccmcl_load(char *paths, int reload)
 669 #else
 670 _uccmcl_load(paths, reload)
 671 char *paths;
 672 int reload;
 673 #endif
 674 {
 675     FILE *in;
 676     unsigned long i;
 677     _ucheader_t hdr;
 678
 679     if (_uccmcl_size > 0) {
 680         if (!reload)
 681           /*
 682            * The combining classes have already been loaded.
 683            */
 684           return;
 685
 686         free((char *) _uccmcl_nodes);
 687         _uccmcl_size = 0;
 688     }
 689
 690     if ((in = _ucopenfile(paths, "cmbcl.dat", "rb")) == 0)
 691       return;
 692
 693     /*
 694      * Load the header.
 695      */
 696     fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
 697
 698     if (hdr.bom == 0xfffe) {
 699         hdr.cnt = endian_short(hdr.cnt);
 700         hdr.size.bytes = endian_long(hdr.size.bytes);
 701     }
 702
 703     _uccmcl_size = hdr.cnt * 3;
 704     _uccmcl_nodes = (unsigned long *) malloc(hdr.size.bytes);
 705
 706     /*
 707      * Read the combining classes in.
 708      */
 709     fread((char *) _uccmcl_nodes, sizeof(unsigned long), _uccmcl_size, in);
 710
 711     /*
 712      * Do an endian swap if necessary.
 713      */
 714     if (hdr.bom == 0xfffe) {
 715         for (i = 0; i < _uccmcl_size; i++)
 716           _uccmcl_nodes[i] = endian_long(_uccmcl_nodes[i]);
 717     }
 718 }
 719
 720 static void
 721 #ifdef __STDC__
 722 _uccmcl_unload(void)
 723 #else
 724 _uccmcl_unload()
 725 #endif
 726 {
 727     if (_uccmcl_size == 0)
 728       return;
 729
 730     free((char *) _uccmcl_nodes);
 731     _uccmcl_size = 0;
 732 }
 733
 734 unsigned long
 735 #ifdef __STDC__
 736 uccombining_class(unsigned long code)
 737 #else
 738 uccombining_class(code)
 739 unsigned long code;
 740 #endif
 741 {
 742     long l, r, m;
 743
 744     l = 0;
 745     r = _uccmcl_size - 1;
 746
 747     while (l <= r) {
 748         m = (l + r) >> 1;
 749         m -= (m % 3);
 750         if (code > _uccmcl_nodes[m + 1])
 751           l = m + 3;
 752         else if (code < _uccmcl_nodes[m])
 753           r = m - 3;
 754         else if (code >= _uccmcl_nodes[m] && code <= _uccmcl_nodes[m + 1])
 755           return _uccmcl_nodes[m + 2];
 756     }
 757     return 0;
 758 }
 759
 760 /**************************************************************************
 761  *
 762  * Support for numeric values.
 763  *
 764  **************************************************************************/
 765
 766 static unsigned long *_ucnum_nodes;
 767 static unsigned long _ucnum_size;
 768 static short *_ucnum_vals;
 769
 770 static void
 771 #ifdef __STDC__
 772 _ucnumb_load(char *paths, int reload)
 773 #else
 774 _ucnumb_load(paths, reload)
 775 char *paths;
 776 int reload;
 777 #endif
 778 {
 779     FILE *in;
 780     unsigned long size, i;
 781     _ucheader_t hdr;
 782
 783     if (_ucnum_size > 0) {
 784         if (!reload)
 785           /*
 786            * The numbers have already been loaded.
 787            */
 788           return;
 789
 790         free((char *) _ucnum_nodes);
 791         _ucnum_size = 0;
 792     }
 793
 794     if ((in = _ucopenfile(paths, "num.dat", "rb")) == 0)
 795       return;
 796
 797     /*
 798      * Load the header.
 799      */
 800     fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
 801
 802     if (hdr.bom == 0xfffe) {
 803         hdr.cnt = endian_short(hdr.cnt);
 804         hdr.size.bytes = endian_long(hdr.size.bytes);
 805     }
 806
 807     _ucnum_size = hdr.cnt;
 808     _ucnum_nodes = (unsigned long *) malloc(hdr.size.bytes);
 809     _ucnum_vals = (short *) (_ucnum_nodes + _ucnum_size);
 810
 811     /*
 812      * Read the combining classes in.
 813      */
 814     fread((char *) _ucnum_nodes, sizeof(unsigned char), hdr.size.bytes, in);
 815
 816     /*
 817      * Do an endian swap if necessary.
 818      */
 819     if (hdr.bom == 0xfffe) {
 820         for (i = 0; i < _ucnum_size; i++)
 821           _ucnum_nodes[i] = endian_long(_ucnum_nodes[i]);
 822
 823         /*
 824          * Determine the number of values that have to be adjusted.
 825          */
 826         size = (hdr.size.bytes -
 827                 (_ucnum_size * (sizeof(unsigned long) << 1))) /
 828             sizeof(short);
 829
 830         for (i = 0; i < size; i++)
 831           _ucnum_vals[i] = endian_short(_ucnum_vals[i]);
 832     }
 833 }
 834
 835 static void
 836 #ifdef __STDC__
 837 _ucnumb_unload(void)
 838 #else
 839 _ucnumb_unload()
 840 #endif
 841 {
 842     if (_ucnum_size == 0)
 843       return;
 844
 845     free((char *) _ucnum_nodes);
 846     _ucnum_size = 0;
 847 }
 848
 849 int
 850 #ifdef __STDC__
 851 ucnumber_lookup(unsigned long code, struct ucnumber *num)
 852 #else
 853 ucnumber_lookup(code, num)
 854 unsigned long code;
 855 struct ucnumber *num;
 856 #endif
 857 {
 858     long l, r, m;
 859     short *vp;
 860
 861     l = 0;
 862     r = _ucnum_size - 1;
 863     while (l <= r) {
 864         /*
 865          * Determine a "mid" point and adjust to make sure the mid point is at
 866          * the beginning of a code+offset pair.
 867          */
 868         m = (l + r) >> 1;
 869         m -= (m & 1);
 870         if (code > _ucnum_nodes[m])
 871           l = m + 2;
 872         else if (code < _ucnum_nodes[m])
 873           r = m - 2;
 874         else {
 875             vp = _ucnum_vals + _ucnum_nodes[m + 1];
 876             num->numerator = (int) *vp++;
 877             num->denominator = (int) *vp;
 878             return 1;
 879         }
 880     }
 881     return 0;
 882 }
 883
 884 int
 885 #ifdef __STDC__
 886 ucdigit_lookup(unsigned long code, int *digit)
 887 #else
 888 ucdigit_lookup(code, digit)
 889 unsigned long code;
 890 int *digit;
 891 #endif
 892 {
 893     long l, r, m;
 894     short *vp;
 895
 896     l = 0;
 897     r = _ucnum_size - 1;
 898     while (l <= r) {
 899         /*
 900          * Determine a "mid" point and adjust to make sure the mid point is at
 901          * the beginning of a code+offset pair.
 902          */
 903         m = (l + r) >> 1;
 904         m -= (m & 1);
 905         if (code > _ucnum_nodes[m])
 906           l = m + 2;
 907         else if (code < _ucnum_nodes[m])
 908           r = m - 2;
 909         else {
 910             vp = _ucnum_vals + _ucnum_nodes[m + 1];
 911             if (*vp == *(vp + 1)) {
 912               *digit = *vp;
 913               return 1;
 914             }
 915             return 0;
 916         }
 917     }
 918     return 0;
 919 }
 920
 921 struct ucnumber
 922 #ifdef __STDC__
 923 ucgetnumber(unsigned long code)
 924 #else
 925 ucgetnumber(code)
 926 unsigned long code;
 927 #endif
 928 {
 929     struct ucnumber num;
 930
 931     /*
 932      * Initialize with some arbitrary value, because the caller simply cannot
 933      * tell for sure if the code is a number without calling the ucisnumber()
 934      * macro before calling this function.
 935      */
 936     num.numerator = num.denominator = -111;
 937
 938     (void) ucnumber_lookup(code, &num);
 939
 940     return num;
 941 }
 942
 943 int
 944 #ifdef __STDC__
 945 ucgetdigit(unsigned long code)
 946 #else
 947 ucgetdigit(code)
 948 unsigned long code;
 949 #endif
 950 {
 951     int dig;
 952
 953     /*
 954      * Initialize with some arbitrary value, because the caller simply cannot
 955      * tell for sure if the code is a number without calling the ucisdigit()
 956      * macro before calling this function.
 957      */
 958     dig = -111;
 959
 960     (void) ucdigit_lookup(code, &dig);
 961
 962     return dig;
 963 }
 964
 965 /**************************************************************************
 966  *
 967  * Setup and cleanup routines.
 968  *
 969  **************************************************************************/
 970
 971 void
 972 #ifdef __STDC__
 973 ucdata_load(char *paths, int masks)
 974 #else
 975 ucdata_load(paths, masks)
 976 char *paths;
 977 int masks;
 978 #endif
 979 {
 980     if (masks & UCDATA_CTYPE)
 981       _ucprop_load(paths, 0);
 982     if (masks & UCDATA_CASE)
 983       _uccase_load(paths, 0);
 984     if (masks & UCDATA_DECOMP)
 985       _ucdcmp_load(paths, 0);
 986     if (masks & UCDATA_CMBCL)
 987       _uccmcl_load(paths, 0);
 988     if (masks & UCDATA_NUM)
 989       _ucnumb_load(paths, 0);
 990 }
 991
 992 void
 993 #ifdef __STDC__
 994 ucdata_unload(int masks)
 995 #else
 996 ucdata_unload(masks)
 997 int masks;
 998 #endif
 999 {
1000     if (masks & UCDATA_CTYPE)
1001       _ucprop_unload();
1002     if (masks & UCDATA_CASE)
1003       _uccase_unload();
1004     if (masks & UCDATA_DECOMP)
1005       _ucdcmp_unload();
1006     if (masks & UCDATA_CMBCL)
1007       _uccmcl_unload();
1008     if (masks & UCDATA_NUM)
1009       _ucnumb_unload();
1010 }
1011
1012 void
1013 #ifdef __STDC__
1014 ucdata_reload(char *paths, int masks)
1015 #else
1016 ucdata_reload(paths, masks)
1017 char *paths;
1018 int masks;
1019 #endif
1020 {
1021     if (masks & UCDATA_CTYPE)
1022       _ucprop_load(paths, 1);
1023     if (masks & UCDATA_CASE)
1024       _uccase_load(paths, 1);
1025     if (masks & UCDATA_DECOMP)
1026       _ucdcmp_load(paths, 1);
1027     if (masks & UCDATA_CMBCL)
1028       _uccmcl_load(paths, 1);
1029     if (masks & UCDATA_NUM)
1030       _ucnumb_load(paths, 1);
1031 }
1032
1033 #ifdef TEST
1034
1035 void
1036 #ifdef __STDC__
1037 main(void)
1038 #else
1039 main()
1040 #endif
1041 {
1042     int dig;
1043     unsigned long i, lo, *dec;
1044     struct ucnumber num;
1045
1046     ucdata_setup(".");
1047
1048     if (ucisweak(0x30))
1049       printf("WEAK\n");
1050     else
1051       printf("NOT WEAK\n");
1052
1053     printf("LOWER 0x%04lX\n", uctolower(0xff3a));
1054     printf("UPPER 0x%04lX\n", uctoupper(0xff5a));
1055
1056     if (ucisalpha(0x1d5))
1057       printf("ALPHA\n");
1058     else
1059       printf("NOT ALPHA\n");
1060
1061     if (ucisupper(0x1d5)) {
1062         printf("UPPER\n");
1063         lo = uctolower(0x1d5);
1064         printf("0x%04lx\n", lo);
1065         lo = uctotitle(0x1d5);
1066         printf("0x%04lx\n", lo);
1067     } else
1068       printf("NOT UPPER\n");
1069
1070     if (ucistitle(0x1d5))
1071       printf("TITLE\n");
1072     else
1073       printf("NOT TITLE\n");
1074
1075     if (uciscomposite(0x1d5))
1076       printf("COMPOSITE\n");
1077     else
1078       printf("NOT COMPOSITE\n");
1079
1080     if (ucdecomp(0x1d5, &lo, &dec)) {
1081         for (i = 0; i < lo; i++)
1082           printf("0x%04lx ", dec[i]);
1083         putchar('\n');
1084     }
1085
1086     if ((lo = uccombining_class(0x41)) != 0)
1087       printf("0x41 CCL %ld\n", lo);
1088
1089     if (ucisxdigit(0xfeff))
1090       printf("0xFEFF HEX DIGIT\n");
1091     else
1092       printf("0xFEFF NOT HEX DIGIT\n");
1093
1094     if (ucisdefined(0x10000))
1095       printf("0x10000 DEFINED\n");
1096     else
1097       printf("0x10000 NOT DEFINED\n");
1098
1099     if (ucnumber_lookup(0x30, &num)) {
1100         if (num.numerator != num.denominator)
1101           printf("UCNUMBER: 0x30 = %d/%d\n", num.numerator, num.denominator);
1102         else
1103           printf("UCNUMBER: 0x30 = %d\n", num.numerator);
1104     } else
1105       printf("UCNUMBER: 0x30 NOT A NUMBER\n");
1106
1107     if (ucnumber_lookup(0xbc, &num)) {
1108         if (num.numerator != num.denominator)
1109           printf("UCNUMBER: 0xbc = %d/%d\n", num.numerator, num.denominator);
1110         else
1111           printf("UCNUMBER: 0xbc = %d\n", num.numerator);
1112     } else
1113       printf("UCNUMBER: 0xbc NOT A NUMBER\n");
1114
1115
1116     if (ucnumber_lookup(0xff19, &num)) {
1117         if (num.numerator != num.denominator)
1118           printf("UCNUMBER: 0xff19 = %d/%d\n", num.numerator, num.denominator);
1119         else
1120           printf("UCNUMBER: 0xff19 = %d\n", num.numerator);
1121     } else
1122       printf("UCNUMBER: 0xff19 NOT A NUMBER\n");
1123
1124     if (ucnumber_lookup(0x4e00, &num)) {
1125         if (num.numerator != num.denominator)
1126           printf("UCNUMBER: 0x4e00 = %d/%d\n", num.numerator, num.denominator);
1127         else
1128           printf("UCNUMBER: 0x4e00 = %d\n", num.numerator);
1129     } else
1130       printf("UCNUMBER: 0x4e00 NOT A NUMBER\n");
1131
1132     if (ucdigit_lookup(0x06f9, &dig))
1133       printf("UCDIGIT: 0x6f9 = %d\n", dig);
1134     else
1135       printf("UCDIGIT: 0x6f9 NOT A NUMBER\n");
1136
1137     dig = ucgetdigit(0x0969);
1138     printf("UCGETDIGIT: 0x969 = %d\n", dig);
1139
1140     num = ucgetnumber(0x30);
1141     if (num.numerator != num.denominator)
1142       printf("UCGETNUMBER: 0x30 = %d/%d\n", num.numerator, num.denominator);
1143     else
1144       printf("UCGETNUMBER: 0x30 = %d\n", num.numerator);
1145
1146     num = ucgetnumber(0xbc);
1147     if (num.numerator != num.denominator)
1148       printf("UCGETNUMBER: 0xbc = %d/%d\n", num.numerator, num.denominator);
1149     else
1150       printf("UCGETNUMBER: 0xbc = %d\n", num.numerator);
1151
1152     num = ucgetnumber(0xff19);
1153     if (num.numerator != num.denominator)
1154       printf("UCGETNUMBER: 0xff19 = %d/%d\n", num.numerator, num.denominator);
1155     else
1156       printf("UCGETNUMBER: 0xff19 = %d\n", num.numerator);
1157
1158     ucdata_cleanup();
1159     exit(0);
1160 }
1161
1162 #endif /* TEST */