1.1 --- /dev/null Thu Jan 01 00:00:00 1970 +0000
1.2 +++ b/intl/unicharutil/src/ucdata.c Wed Dec 31 06:09:35 2014 +0100
1.3 @@ -0,0 +1,1162 @@
1.4 +/*
1.5 + * Copyright 1996, 1997, 1998 Computing Research Labs,
1.6 + * New Mexico State University
1.7 + *
1.8 + * Permission is hereby granted, free of charge, to any person obtaining a
1.9 + * copy of this software and associated documentation files (the "Software"),
1.10 + * to deal in the Software without restriction, including without limitation
1.11 + * the rights to use, copy, modify, merge, publish, distribute, sublicense,
1.12 + * and/or sell copies of the Software, and to permit persons to whom the
1.13 + * Software is furnished to do so, subject to the following conditions:
1.14 + *
1.15 + * The above copyright notice and this permission notice shall be included in
1.16 + * all copies or substantial portions of the Software.
1.17 + *
1.18 + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
1.19 + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
1.20 + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
1.21 + * THE COMPUTING RESEARCH LAB OR NEW MEXICO STATE UNIVERSITY BE LIABLE FOR ANY
1.22 + * CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT
1.23 + * OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR
1.24 + * THE USE OR OTHER DEALINGS IN THE SOFTWARE.
1.25 + */
1.26 +#ifndef lint
1.27 +#ifdef __GNUC__
1.28 +static char rcsid[] __attribute__ ((unused)) = "$Id: ucdata.c,v 1.1 1999/01/08 00:19:11 ftang%netscape.com Exp $";
1.29 +#else
1.30 +static char rcsid[] = "$Id: ucdata.c,v 1.1 1999/01/08 00:19:11 ftang%netscape.com Exp $";
1.31 +#endif
1.32 +#endif
1.33 +
1.34 +#include <stdio.h>
1.35 +#include <stdlib.h>
1.36 +#include <string.h>
1.37 +#ifndef WIN32
1.38 +#include <unistd.h>
1.39 +#endif
1.40 +
1.41 +#include "ucdata.h"
1.42 +
1.43 +/**************************************************************************
1.44 + *
1.45 + * Miscellaneous types, data, and support functions.
1.46 + *
1.47 + **************************************************************************/
1.48 +
1.49 +typedef struct {
1.50 + unsigned short bom;
1.51 + unsigned short cnt;
1.52 + union {
1.53 + unsigned long bytes;
1.54 + unsigned short len[2];
1.55 + } size;
1.56 +} _ucheader_t;
1.57 +
1.58 +/*
1.59 + * A simple array of 32-bit masks for lookup.
1.60 + */
1.61 +static unsigned long masks32[32] = {
1.62 + 0x00000001, 0x00000002, 0x00000004, 0x00000008, 0x00000010, 0x00000020,
1.63 + 0x00000040, 0x00000080, 0x00000100, 0x00000200, 0x00000400, 0x00000800,
1.64 + 0x00001000, 0x00002000, 0x00004000, 0x00008000, 0x00010000, 0x00020000,
1.65 + 0x00040000, 0x00080000, 0x00100000, 0x00200000, 0x00400000, 0x00800000,
1.66 + 0x01000000, 0x02000000, 0x04000000, 0x08000000, 0x10000000, 0x20000000,
1.67 + 0x40000000, 0x80000000
1.68 +};
1.69 +
1.70 +#define endian_short(cc) (((cc) >> 8) | (((cc) & 0xff) << 8))
1.71 +#define endian_long(cc) ((((cc) & 0xff) << 24)|((((cc) >> 8) & 0xff) << 16)|\
1.72 + ((((cc) >> 16) & 0xff) << 8)|((cc) >> 24))
1.73 +
1.74 +static FILE *
1.75 +#ifdef __STDC__
1.76 +_ucopenfile(char *paths, char *filename, char *mode)
1.77 +#else
1.78 +_ucopenfile(paths, filename, mode)
1.79 +char *paths, *filename, *mode;
1.80 +#endif
1.81 +{
1.82 + FILE *f;
1.83 + char *fp, *dp, *pp, path[BUFSIZ];
1.84 +
1.85 + if (filename == 0 || *filename == 0)
1.86 + return 0;
1.87 +
1.88 + dp = paths;
1.89 + while (dp && *dp) {
1.90 + pp = path;
1.91 + while (*dp && *dp != ':')
1.92 + *pp++ = *dp++;
1.93 + *pp++ = '/';
1.94 +
1.95 + fp = filename;
1.96 + while (*fp)
1.97 + *pp++ = *fp++;
1.98 + *pp = 0;
1.99 +
1.100 + if ((f = fopen(path, mode)) != 0)
1.101 + return f;
1.102 +
1.103 + if (*dp == ':')
1.104 + dp++;
1.105 + }
1.106 +
1.107 + return 0;
1.108 +}
1.109 +
1.110 +/**************************************************************************
1.111 + *
1.112 + * Support for the character properties.
1.113 + *
1.114 + **************************************************************************/
1.115 +
1.116 +static unsigned long _ucprop_size;
1.117 +static unsigned short *_ucprop_offsets;
1.118 +static unsigned long *_ucprop_ranges;
1.119 +
1.120 +static void
1.121 +#ifdef __STDC__
1.122 +_ucprop_load(char *paths, int reload)
1.123 +#else
1.124 +_ucprop_load(paths, reload)
1.125 +char *paths;
1.126 +int reload;
1.127 +#endif
1.128 +{
1.129 + FILE *in;
1.130 + unsigned long size, i;
1.131 + _ucheader_t hdr;
1.132 +
1.133 + if (_ucprop_size > 0) {
1.134 + if (!reload)
1.135 + /*
1.136 + * The character properties have already been loaded.
1.137 + */
1.138 + return;
1.139 +
1.140 + /*
1.141 + * Unload the current character property data in preparation for
1.142 + * loading a new copy. Only the first array has to be deallocated
1.143 + * because all the memory for the arrays is allocated as a single
1.144 + * block.
1.145 + */
1.146 + free((char *) _ucprop_offsets);
1.147 + _ucprop_size = 0;
1.148 + }
1.149 +
1.150 + if ((in = _ucopenfile(paths, "ctype.dat", "rb")) == 0)
1.151 + return;
1.152 +
1.153 + /*
1.154 + * Load the header.
1.155 + */
1.156 + fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
1.157 +
1.158 + if (hdr.bom == 0xfffe) {
1.159 + hdr.cnt = endian_short(hdr.cnt);
1.160 + hdr.size.bytes = endian_long(hdr.size.bytes);
1.161 + }
1.162 +
1.163 + if ((_ucprop_size = hdr.cnt) == 0) {
1.164 + fclose(in);
1.165 + return;
1.166 + }
1.167 +
1.168 + /*
1.169 + * Allocate all the storage needed for the lookup table.
1.170 + */
1.171 + _ucprop_offsets = (unsigned short *) malloc(hdr.size.bytes);
1.172 +
1.173 + /*
1.174 + * Calculate the offset into the storage for the ranges. The offsets
1.175 + * array is on a 4-byte boundary and one larger than the value provided in
1.176 + * the header count field. This means the offset to the ranges must be
1.177 + * calculated after aligning the count to a 4-byte boundary.
1.178 + */
1.179 + if ((size = ((hdr.cnt + 1) * sizeof(unsigned short))) & 3)
1.180 + size += 4 - (size & 3);
1.181 + size >>= 1;
1.182 + _ucprop_ranges = (unsigned long *) (_ucprop_offsets + size);
1.183 +
1.184 + /*
1.185 + * Load the offset array.
1.186 + */
1.187 + fread((char *) _ucprop_offsets, sizeof(unsigned short), size, in);
1.188 +
1.189 + /*
1.190 + * Do an endian swap if necessary. Don't forget there is an extra node on
1.191 + * the end with the final index.
1.192 + */
1.193 + if (hdr.bom == 0xfffe) {
1.194 + for (i = 0; i <= _ucprop_size; i++)
1.195 + _ucprop_offsets[i] = endian_short(_ucprop_offsets[i]);
1.196 + }
1.197 +
1.198 + /*
1.199 + * Load the ranges. The number of elements is in the last array position
1.200 + * of the offsets.
1.201 + */
1.202 + fread((char *) _ucprop_ranges, sizeof(unsigned long),
1.203 + _ucprop_offsets[_ucprop_size], in);
1.204 +
1.205 + fclose(in);
1.206 +
1.207 + /*
1.208 + * Do an endian swap if necessary.
1.209 + */
1.210 + if (hdr.bom == 0xfffe) {
1.211 + for (i = 0; i < _ucprop_offsets[_ucprop_size]; i++)
1.212 + _ucprop_ranges[i] = endian_long(_ucprop_ranges[i]);
1.213 + }
1.214 +}
1.215 +
1.216 +static void
1.217 +#ifdef __STDC__
1.218 +_ucprop_unload(void)
1.219 +#else
1.220 +_ucprop_unload()
1.221 +#endif
1.222 +{
1.223 + if (_ucprop_size == 0)
1.224 + return;
1.225 +
1.226 + /*
1.227 + * Only need to free the offsets because the memory is allocated as a
1.228 + * single block.
1.229 + */
1.230 + free((char *) _ucprop_offsets);
1.231 + _ucprop_size = 0;
1.232 +}
1.233 +
1.234 +static int
1.235 +#ifdef __STDC__
1.236 +_ucprop_lookup(unsigned long code, unsigned long n)
1.237 +#else
1.238 +_ucprop_lookup(code, n)
1.239 +unsigned long code, n;
1.240 +#endif
1.241 +{
1.242 + long l, r, m;
1.243 +
1.244 + /*
1.245 + * There is an extra node on the end of the offsets to allow this routine
1.246 + * to work right. If the index is 0xffff, then there are no nodes for the
1.247 + * property.
1.248 + */
1.249 + if ((l = _ucprop_offsets[n]) == 0xffff)
1.250 + return 0;
1.251 +
1.252 + /*
1.253 + * Locate the next offset that is not 0xffff. The sentinel at the end of
1.254 + * the array is the max index value.
1.255 + */
1.256 + for (m = 1;
1.257 + n + m < _ucprop_size && _ucprop_offsets[n + m] == 0xffff; m++) ;
1.258 +
1.259 + r = _ucprop_offsets[n + m] - 1;
1.260 +
1.261 + while (l <= r) {
1.262 + /*
1.263 + * Determine a "mid" point and adjust to make sure the mid point is at
1.264 + * the beginning of a range pair.
1.265 + */
1.266 + m = (l + r) >> 1;
1.267 + m -= (m & 1);
1.268 + if (code > _ucprop_ranges[m + 1])
1.269 + l = m + 2;
1.270 + else if (code < _ucprop_ranges[m])
1.271 + r = m - 2;
1.272 + else if (code >= _ucprop_ranges[m] && code <= _ucprop_ranges[m + 1])
1.273 + return 1;
1.274 + }
1.275 + return 0;
1.276 +}
1.277 +
1.278 +int
1.279 +#ifdef __STDC__
1.280 +ucisprop(unsigned long code, unsigned long mask1, unsigned long mask2)
1.281 +#else
1.282 +ucisprop(code, mask1, mask2)
1.283 +unsigned long code, mask1, mask2;
1.284 +#endif
1.285 +{
1.286 + unsigned long i;
1.287 +
1.288 + if (mask1 == 0 && mask2 == 0)
1.289 + return 0;
1.290 +
1.291 + for (i = 0; mask1 && i < 32; i++) {
1.292 + if ((mask1 & masks32[i]) && _ucprop_lookup(code, i))
1.293 + return 1;
1.294 + }
1.295 +
1.296 + for (i = 32; mask2 && i < _ucprop_size; i++) {
1.297 + if ((mask2 & masks32[i & 31]) && _ucprop_lookup(code, i))
1.298 + return 1;
1.299 + }
1.300 +
1.301 + return 0;
1.302 +}
1.303 +
1.304 +/**************************************************************************
1.305 + *
1.306 + * Support for case mapping.
1.307 + *
1.308 + **************************************************************************/
1.309 +
1.310 +static unsigned long _uccase_size;
1.311 +static unsigned short _uccase_len[2];
1.312 +static unsigned long *_uccase_map;
1.313 +
1.314 +static void
1.315 +#ifdef __STDC__
1.316 +_uccase_load(char *paths, int reload)
1.317 +#else
1.318 +_uccase_load(paths, reload)
1.319 +char *paths;
1.320 +int reload;
1.321 +#endif
1.322 +{
1.323 + FILE *in;
1.324 + unsigned long i;
1.325 + _ucheader_t hdr;
1.326 +
1.327 + if (_uccase_size > 0) {
1.328 + if (!reload)
1.329 + /*
1.330 + * The case mappings have already been loaded.
1.331 + */
1.332 + return;
1.333 +
1.334 + free((char *) _uccase_map);
1.335 + _uccase_size = 0;
1.336 + }
1.337 +
1.338 + if ((in = _ucopenfile(paths, "case.dat", "rb")) == 0)
1.339 + return;
1.340 +
1.341 + /*
1.342 + * Load the header.
1.343 + */
1.344 + fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
1.345 +
1.346 + if (hdr.bom == 0xfffe) {
1.347 + hdr.cnt = endian_short(hdr.cnt);
1.348 + hdr.size.len[0] = endian_short(hdr.size.len[0]);
1.349 + hdr.size.len[1] = endian_short(hdr.size.len[1]);
1.350 + }
1.351 +
1.352 + /*
1.353 + * Set the node count and lengths of the upper and lower case mapping
1.354 + * tables.
1.355 + */
1.356 + _uccase_size = hdr.cnt * 3;
1.357 + _uccase_len[0] = hdr.size.len[0] * 3;
1.358 + _uccase_len[1] = hdr.size.len[1] * 3;
1.359 +
1.360 + _uccase_map = (unsigned long *)
1.361 + malloc(_uccase_size * sizeof(unsigned long));
1.362 +
1.363 + /*
1.364 + * Load the case mapping table.
1.365 + */
1.366 + fread((char *) _uccase_map, sizeof(unsigned long), _uccase_size, in);
1.367 +
1.368 + /*
1.369 + * Do an endian swap if necessary.
1.370 + */
1.371 + if (hdr.bom == 0xfffe) {
1.372 + for (i = 0; i < _uccase_size; i++)
1.373 + _uccase_map[i] = endian_long(_uccase_map[i]);
1.374 + }
1.375 +}
1.376 +
1.377 +static void
1.378 +#ifdef __STDC__
1.379 +_uccase_unload(void)
1.380 +#else
1.381 +_uccase_unload()
1.382 +#endif
1.383 +{
1.384 + if (_uccase_size == 0)
1.385 + return;
1.386 +
1.387 + free((char *) _uccase_map);
1.388 + _uccase_size = 0;
1.389 +}
1.390 +
1.391 +static unsigned long
1.392 +#ifdef __STDC__
1.393 +_uccase_lookup(unsigned long code, long l, long r, int field)
1.394 +#else
1.395 +_uccase_lookup(code, l, r, field)
1.396 +unsigned long code;
1.397 +long l, r;
1.398 +int field;
1.399 +#endif
1.400 +{
1.401 + long m;
1.402 +
1.403 + /*
1.404 + * Do the binary search.
1.405 + */
1.406 + while (l <= r) {
1.407 + /*
1.408 + * Determine a "mid" point and adjust to make sure the mid point is at
1.409 + * the beginning of a case mapping triple.
1.410 + */
1.411 + m = (l + r) >> 1;
1.412 + m -= (m % 3);
1.413 + if (code > _uccase_map[m])
1.414 + l = m + 3;
1.415 + else if (code < _uccase_map[m])
1.416 + r = m - 3;
1.417 + else if (code == _uccase_map[m])
1.418 + return _uccase_map[m + field];
1.419 + }
1.420 +
1.421 + return code;
1.422 +}
1.423 +
1.424 +unsigned long
1.425 +#ifdef __STDC__
1.426 +uctoupper(unsigned long code)
1.427 +#else
1.428 +uctoupper(code)
1.429 +unsigned long code;
1.430 +#endif
1.431 +{
1.432 + int field;
1.433 + long l, r;
1.434 +
1.435 + if (ucisupper(code))
1.436 + return code;
1.437 +
1.438 + if (ucislower(code)) {
1.439 + /*
1.440 + * The character is lower case.
1.441 + */
1.442 + field = 1;
1.443 + l = _uccase_len[0];
1.444 + r = (l + _uccase_len[1]) - 1;
1.445 + } else {
1.446 + /*
1.447 + * The character is title case.
1.448 + */
1.449 + field = 2;
1.450 + l = _uccase_len[0] + _uccase_len[1];
1.451 + r = _uccase_size - 1;
1.452 + }
1.453 + return _uccase_lookup(code, l, r, field);
1.454 +}
1.455 +
1.456 +unsigned long
1.457 +#ifdef __STDC__
1.458 +uctolower(unsigned long code)
1.459 +#else
1.460 +uctolower(code)
1.461 +unsigned long code;
1.462 +#endif
1.463 +{
1.464 + int field;
1.465 + long l, r;
1.466 +
1.467 + if (ucislower(code))
1.468 + return code;
1.469 +
1.470 + if (ucisupper(code)) {
1.471 + /*
1.472 + * The character is upper case.
1.473 + */
1.474 + field = 1;
1.475 + l = 0;
1.476 + r = _uccase_len[0] - 1;
1.477 + } else {
1.478 + /*
1.479 + * The character is title case.
1.480 + */
1.481 + field = 2;
1.482 + l = _uccase_len[0] + _uccase_len[1];
1.483 + r = _uccase_size - 1;
1.484 + }
1.485 + return _uccase_lookup(code, l, r, field);
1.486 +}
1.487 +
1.488 +unsigned long
1.489 +#ifdef __STDC__
1.490 +uctotitle(unsigned long code)
1.491 +#else
1.492 +uctotitle(code)
1.493 +unsigned long code;
1.494 +#endif
1.495 +{
1.496 + int field;
1.497 + long l, r;
1.498 +
1.499 + if (ucistitle(code))
1.500 + return code;
1.501 +
1.502 + /*
1.503 + * The offset will always be the same for converting to title case.
1.504 + */
1.505 + field = 2;
1.506 +
1.507 + if (ucisupper(code)) {
1.508 + /*
1.509 + * The character is upper case.
1.510 + */
1.511 + l = 0;
1.512 + r = _uccase_len[0] - 1;
1.513 + } else {
1.514 + /*
1.515 + * The character is lower case.
1.516 + */
1.517 + l = _uccase_len[0];
1.518 + r = (l + _uccase_len[1]) - 1;
1.519 + }
1.520 + return _uccase_lookup(code, l, r, field);
1.521 +}
1.522 +
1.523 +/**************************************************************************
1.524 + *
1.525 + * Support for decompositions.
1.526 + *
1.527 + **************************************************************************/
1.528 +
1.529 +static unsigned long _ucdcmp_size;
1.530 +static unsigned long *_ucdcmp_nodes;
1.531 +static unsigned long *_ucdcmp_decomp;
1.532 +
1.533 +static void
1.534 +#ifdef __STDC__
1.535 +_ucdcmp_load(char *paths, int reload)
1.536 +#else
1.537 +_ucdcmp_load(paths, reload)
1.538 +char *paths;
1.539 +int reload;
1.540 +#endif
1.541 +{
1.542 + FILE *in;
1.543 + unsigned long size, i;
1.544 + _ucheader_t hdr;
1.545 +
1.546 + if (_ucdcmp_size > 0) {
1.547 + if (!reload)
1.548 + /*
1.549 + * The decompositions have already been loaded.
1.550 + */
1.551 + return;
1.552 +
1.553 + free((char *) _ucdcmp_nodes);
1.554 + _ucdcmp_size = 0;
1.555 + }
1.556 +
1.557 + if ((in = _ucopenfile(paths, "decomp.dat", "rb")) == 0)
1.558 + return;
1.559 +
1.560 + /*
1.561 + * Load the header.
1.562 + */
1.563 + fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
1.564 +
1.565 + if (hdr.bom == 0xfffe) {
1.566 + hdr.cnt = endian_short(hdr.cnt);
1.567 + hdr.size.bytes = endian_long(hdr.size.bytes);
1.568 + }
1.569 +
1.570 + _ucdcmp_size = hdr.cnt << 1;
1.571 + _ucdcmp_nodes = (unsigned long *) malloc(hdr.size.bytes);
1.572 + _ucdcmp_decomp = _ucdcmp_nodes + (_ucdcmp_size + 1);
1.573 +
1.574 + /*
1.575 + * Read the decomposition data in.
1.576 + */
1.577 + size = hdr.size.bytes / sizeof(unsigned long);
1.578 + fread((char *) _ucdcmp_nodes, sizeof(unsigned long), size, in);
1.579 +
1.580 + /*
1.581 + * Do an endian swap if necessary.
1.582 + */
1.583 + if (hdr.bom == 0xfffe) {
1.584 + for (i = 0; i < size; i++)
1.585 + _ucdcmp_nodes[i] = endian_long(_ucdcmp_nodes[i]);
1.586 + }
1.587 +}
1.588 +
1.589 +static void
1.590 +#ifdef __STDC__
1.591 +_ucdcmp_unload(void)
1.592 +#else
1.593 +_ucdcmp_unload()
1.594 +#endif
1.595 +{
1.596 + if (_ucdcmp_size == 0)
1.597 + return;
1.598 +
1.599 + /*
1.600 + * Only need to free the offsets because the memory is allocated as a
1.601 + * single block.
1.602 + */
1.603 + free((char *) _ucdcmp_nodes);
1.604 + _ucdcmp_size = 0;
1.605 +}
1.606 +
1.607 +int
1.608 +#ifdef __STDC__
1.609 +ucdecomp(unsigned long code, unsigned long *num, unsigned long **decomp)
1.610 +#else
1.611 +ucdecomp(code, num, decomp)
1.612 +unsigned long code, *num, **decomp;
1.613 +#endif
1.614 +{
1.615 + long l, r, m;
1.616 +
1.617 + l = 0;
1.618 + r = _ucdcmp_nodes[_ucdcmp_size] - 1;
1.619 +
1.620 + while (l <= r) {
1.621 + /*
1.622 + * Determine a "mid" point and adjust to make sure the mid point is at
1.623 + * the beginning of a code+offset pair.
1.624 + */
1.625 + m = (l + r) >> 1;
1.626 + m -= (m & 1);
1.627 + if (code > _ucdcmp_nodes[m])
1.628 + l = m + 2;
1.629 + else if (code < _ucdcmp_nodes[m])
1.630 + r = m - 2;
1.631 + else if (code == _ucdcmp_nodes[m]) {
1.632 + *num = _ucdcmp_nodes[m + 3] - _ucdcmp_nodes[m + 1];
1.633 + *decomp = &_ucdcmp_decomp[_ucdcmp_nodes[m + 1]];
1.634 + return 1;
1.635 + }
1.636 + }
1.637 + return 0;
1.638 +}
1.639 +
1.640 +int
1.641 +#ifdef __STDC__
1.642 +ucdecomp_hangul(unsigned long code, unsigned long *num, unsigned long decomp[])
1.643 +#else
1.644 +ucdecomp_hangul(code, num, decomp)
1.645 +unsigned long code, *num, decomp[];
1.646 +#endif
1.647 +{
1.648 + if (!ucishangul(code))
1.649 + return 0;
1.650 +
1.651 + code -= 0xac00;
1.652 + decomp[0] = 0x1100 + (unsigned long) (code / 588);
1.653 + decomp[1] = 0x1161 + (unsigned long) ((code % 588) / 28);
1.654 + decomp[2] = 0x11a7 + (unsigned long) (code % 28);
1.655 + *num = (decomp[2] != 0x11a7) ? 3 : 2;
1.656 +
1.657 + return 1;
1.658 +}
1.659 +
1.660 +/**************************************************************************
1.661 + *
1.662 + * Support for combining classes.
1.663 + *
1.664 + **************************************************************************/
1.665 +
1.666 +static unsigned long _uccmcl_size;
1.667 +static unsigned long *_uccmcl_nodes;
1.668 +
1.669 +static void
1.670 +#ifdef __STDC__
1.671 +_uccmcl_load(char *paths, int reload)
1.672 +#else
1.673 +_uccmcl_load(paths, reload)
1.674 +char *paths;
1.675 +int reload;
1.676 +#endif
1.677 +{
1.678 + FILE *in;
1.679 + unsigned long i;
1.680 + _ucheader_t hdr;
1.681 +
1.682 + if (_uccmcl_size > 0) {
1.683 + if (!reload)
1.684 + /*
1.685 + * The combining classes have already been loaded.
1.686 + */
1.687 + return;
1.688 +
1.689 + free((char *) _uccmcl_nodes);
1.690 + _uccmcl_size = 0;
1.691 + }
1.692 +
1.693 + if ((in = _ucopenfile(paths, "cmbcl.dat", "rb")) == 0)
1.694 + return;
1.695 +
1.696 + /*
1.697 + * Load the header.
1.698 + */
1.699 + fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
1.700 +
1.701 + if (hdr.bom == 0xfffe) {
1.702 + hdr.cnt = endian_short(hdr.cnt);
1.703 + hdr.size.bytes = endian_long(hdr.size.bytes);
1.704 + }
1.705 +
1.706 + _uccmcl_size = hdr.cnt * 3;
1.707 + _uccmcl_nodes = (unsigned long *) malloc(hdr.size.bytes);
1.708 +
1.709 + /*
1.710 + * Read the combining classes in.
1.711 + */
1.712 + fread((char *) _uccmcl_nodes, sizeof(unsigned long), _uccmcl_size, in);
1.713 +
1.714 + /*
1.715 + * Do an endian swap if necessary.
1.716 + */
1.717 + if (hdr.bom == 0xfffe) {
1.718 + for (i = 0; i < _uccmcl_size; i++)
1.719 + _uccmcl_nodes[i] = endian_long(_uccmcl_nodes[i]);
1.720 + }
1.721 +}
1.722 +
1.723 +static void
1.724 +#ifdef __STDC__
1.725 +_uccmcl_unload(void)
1.726 +#else
1.727 +_uccmcl_unload()
1.728 +#endif
1.729 +{
1.730 + if (_uccmcl_size == 0)
1.731 + return;
1.732 +
1.733 + free((char *) _uccmcl_nodes);
1.734 + _uccmcl_size = 0;
1.735 +}
1.736 +
1.737 +unsigned long
1.738 +#ifdef __STDC__
1.739 +uccombining_class(unsigned long code)
1.740 +#else
1.741 +uccombining_class(code)
1.742 +unsigned long code;
1.743 +#endif
1.744 +{
1.745 + long l, r, m;
1.746 +
1.747 + l = 0;
1.748 + r = _uccmcl_size - 1;
1.749 +
1.750 + while (l <= r) {
1.751 + m = (l + r) >> 1;
1.752 + m -= (m % 3);
1.753 + if (code > _uccmcl_nodes[m + 1])
1.754 + l = m + 3;
1.755 + else if (code < _uccmcl_nodes[m])
1.756 + r = m - 3;
1.757 + else if (code >= _uccmcl_nodes[m] && code <= _uccmcl_nodes[m + 1])
1.758 + return _uccmcl_nodes[m + 2];
1.759 + }
1.760 + return 0;
1.761 +}
1.762 +
1.763 +/**************************************************************************
1.764 + *
1.765 + * Support for numeric values.
1.766 + *
1.767 + **************************************************************************/
1.768 +
1.769 +static unsigned long *_ucnum_nodes;
1.770 +static unsigned long _ucnum_size;
1.771 +static short *_ucnum_vals;
1.772 +
1.773 +static void
1.774 +#ifdef __STDC__
1.775 +_ucnumb_load(char *paths, int reload)
1.776 +#else
1.777 +_ucnumb_load(paths, reload)
1.778 +char *paths;
1.779 +int reload;
1.780 +#endif
1.781 +{
1.782 + FILE *in;
1.783 + unsigned long size, i;
1.784 + _ucheader_t hdr;
1.785 +
1.786 + if (_ucnum_size > 0) {
1.787 + if (!reload)
1.788 + /*
1.789 + * The numbers have already been loaded.
1.790 + */
1.791 + return;
1.792 +
1.793 + free((char *) _ucnum_nodes);
1.794 + _ucnum_size = 0;
1.795 + }
1.796 +
1.797 + if ((in = _ucopenfile(paths, "num.dat", "rb")) == 0)
1.798 + return;
1.799 +
1.800 + /*
1.801 + * Load the header.
1.802 + */
1.803 + fread((char *) &hdr, sizeof(_ucheader_t), 1, in);
1.804 +
1.805 + if (hdr.bom == 0xfffe) {
1.806 + hdr.cnt = endian_short(hdr.cnt);
1.807 + hdr.size.bytes = endian_long(hdr.size.bytes);
1.808 + }
1.809 +
1.810 + _ucnum_size = hdr.cnt;
1.811 + _ucnum_nodes = (unsigned long *) malloc(hdr.size.bytes);
1.812 + _ucnum_vals = (short *) (_ucnum_nodes + _ucnum_size);
1.813 +
1.814 + /*
1.815 + * Read the combining classes in.
1.816 + */
1.817 + fread((char *) _ucnum_nodes, sizeof(unsigned char), hdr.size.bytes, in);
1.818 +
1.819 + /*
1.820 + * Do an endian swap if necessary.
1.821 + */
1.822 + if (hdr.bom == 0xfffe) {
1.823 + for (i = 0; i < _ucnum_size; i++)
1.824 + _ucnum_nodes[i] = endian_long(_ucnum_nodes[i]);
1.825 +
1.826 + /*
1.827 + * Determine the number of values that have to be adjusted.
1.828 + */
1.829 + size = (hdr.size.bytes -
1.830 + (_ucnum_size * (sizeof(unsigned long) << 1))) /
1.831 + sizeof(short);
1.832 +
1.833 + for (i = 0; i < size; i++)
1.834 + _ucnum_vals[i] = endian_short(_ucnum_vals[i]);
1.835 + }
1.836 +}
1.837 +
1.838 +static void
1.839 +#ifdef __STDC__
1.840 +_ucnumb_unload(void)
1.841 +#else
1.842 +_ucnumb_unload()
1.843 +#endif
1.844 +{
1.845 + if (_ucnum_size == 0)
1.846 + return;
1.847 +
1.848 + free((char *) _ucnum_nodes);
1.849 + _ucnum_size = 0;
1.850 +}
1.851 +
1.852 +int
1.853 +#ifdef __STDC__
1.854 +ucnumber_lookup(unsigned long code, struct ucnumber *num)
1.855 +#else
1.856 +ucnumber_lookup(code, num)
1.857 +unsigned long code;
1.858 +struct ucnumber *num;
1.859 +#endif
1.860 +{
1.861 + long l, r, m;
1.862 + short *vp;
1.863 +
1.864 + l = 0;
1.865 + r = _ucnum_size - 1;
1.866 + while (l <= r) {
1.867 + /*
1.868 + * Determine a "mid" point and adjust to make sure the mid point is at
1.869 + * the beginning of a code+offset pair.
1.870 + */
1.871 + m = (l + r) >> 1;
1.872 + m -= (m & 1);
1.873 + if (code > _ucnum_nodes[m])
1.874 + l = m + 2;
1.875 + else if (code < _ucnum_nodes[m])
1.876 + r = m - 2;
1.877 + else {
1.878 + vp = _ucnum_vals + _ucnum_nodes[m + 1];
1.879 + num->numerator = (int) *vp++;
1.880 + num->denominator = (int) *vp;
1.881 + return 1;
1.882 + }
1.883 + }
1.884 + return 0;
1.885 +}
1.886 +
1.887 +int
1.888 +#ifdef __STDC__
1.889 +ucdigit_lookup(unsigned long code, int *digit)
1.890 +#else
1.891 +ucdigit_lookup(code, digit)
1.892 +unsigned long code;
1.893 +int *digit;
1.894 +#endif
1.895 +{
1.896 + long l, r, m;
1.897 + short *vp;
1.898 +
1.899 + l = 0;
1.900 + r = _ucnum_size - 1;
1.901 + while (l <= r) {
1.902 + /*
1.903 + * Determine a "mid" point and adjust to make sure the mid point is at
1.904 + * the beginning of a code+offset pair.
1.905 + */
1.906 + m = (l + r) >> 1;
1.907 + m -= (m & 1);
1.908 + if (code > _ucnum_nodes[m])
1.909 + l = m + 2;
1.910 + else if (code < _ucnum_nodes[m])
1.911 + r = m - 2;
1.912 + else {
1.913 + vp = _ucnum_vals + _ucnum_nodes[m + 1];
1.914 + if (*vp == *(vp + 1)) {
1.915 + *digit = *vp;
1.916 + return 1;
1.917 + }
1.918 + return 0;
1.919 + }
1.920 + }
1.921 + return 0;
1.922 +}
1.923 +
1.924 +struct ucnumber
1.925 +#ifdef __STDC__
1.926 +ucgetnumber(unsigned long code)
1.927 +#else
1.928 +ucgetnumber(code)
1.929 +unsigned long code;
1.930 +#endif
1.931 +{
1.932 + struct ucnumber num;
1.933 +
1.934 + /*
1.935 + * Initialize with some arbitrary value, because the caller simply cannot
1.936 + * tell for sure if the code is a number without calling the ucisnumber()
1.937 + * macro before calling this function.
1.938 + */
1.939 + num.numerator = num.denominator = -111;
1.940 +
1.941 + (void) ucnumber_lookup(code, &num);
1.942 +
1.943 + return num;
1.944 +}
1.945 +
1.946 +int
1.947 +#ifdef __STDC__
1.948 +ucgetdigit(unsigned long code)
1.949 +#else
1.950 +ucgetdigit(code)
1.951 +unsigned long code;
1.952 +#endif
1.953 +{
1.954 + int dig;
1.955 +
1.956 + /*
1.957 + * Initialize with some arbitrary value, because the caller simply cannot
1.958 + * tell for sure if the code is a number without calling the ucisdigit()
1.959 + * macro before calling this function.
1.960 + */
1.961 + dig = -111;
1.962 +
1.963 + (void) ucdigit_lookup(code, &dig);
1.964 +
1.965 + return dig;
1.966 +}
1.967 +
1.968 +/**************************************************************************
1.969 + *
1.970 + * Setup and cleanup routines.
1.971 + *
1.972 + **************************************************************************/
1.973 +
1.974 +void
1.975 +#ifdef __STDC__
1.976 +ucdata_load(char *paths, int masks)
1.977 +#else
1.978 +ucdata_load(paths, masks)
1.979 +char *paths;
1.980 +int masks;
1.981 +#endif
1.982 +{
1.983 + if (masks & UCDATA_CTYPE)
1.984 + _ucprop_load(paths, 0);
1.985 + if (masks & UCDATA_CASE)
1.986 + _uccase_load(paths, 0);
1.987 + if (masks & UCDATA_DECOMP)
1.988 + _ucdcmp_load(paths, 0);
1.989 + if (masks & UCDATA_CMBCL)
1.990 + _uccmcl_load(paths, 0);
1.991 + if (masks & UCDATA_NUM)
1.992 + _ucnumb_load(paths, 0);
1.993 +}
1.994 +
1.995 +void
1.996 +#ifdef __STDC__
1.997 +ucdata_unload(int masks)
1.998 +#else
1.999 +ucdata_unload(masks)
1.1000 +int masks;
1.1001 +#endif
1.1002 +{
1.1003 + if (masks & UCDATA_CTYPE)
1.1004 + _ucprop_unload();
1.1005 + if (masks & UCDATA_CASE)
1.1006 + _uccase_unload();
1.1007 + if (masks & UCDATA_DECOMP)
1.1008 + _ucdcmp_unload();
1.1009 + if (masks & UCDATA_CMBCL)
1.1010 + _uccmcl_unload();
1.1011 + if (masks & UCDATA_NUM)
1.1012 + _ucnumb_unload();
1.1013 +}
1.1014 +
1.1015 +void
1.1016 +#ifdef __STDC__
1.1017 +ucdata_reload(char *paths, int masks)
1.1018 +#else
1.1019 +ucdata_reload(paths, masks)
1.1020 +char *paths;
1.1021 +int masks;
1.1022 +#endif
1.1023 +{
1.1024 + if (masks & UCDATA_CTYPE)
1.1025 + _ucprop_load(paths, 1);
1.1026 + if (masks & UCDATA_CASE)
1.1027 + _uccase_load(paths, 1);
1.1028 + if (masks & UCDATA_DECOMP)
1.1029 + _ucdcmp_load(paths, 1);
1.1030 + if (masks & UCDATA_CMBCL)
1.1031 + _uccmcl_load(paths, 1);
1.1032 + if (masks & UCDATA_NUM)
1.1033 + _ucnumb_load(paths, 1);
1.1034 +}
1.1035 +
1.1036 +#ifdef TEST
1.1037 +
1.1038 +void
1.1039 +#ifdef __STDC__
1.1040 +main(void)
1.1041 +#else
1.1042 +main()
1.1043 +#endif
1.1044 +{
1.1045 + int dig;
1.1046 + unsigned long i, lo, *dec;
1.1047 + struct ucnumber num;
1.1048 +
1.1049 + ucdata_setup(".");
1.1050 +
1.1051 + if (ucisweak(0x30))
1.1052 + printf("WEAK\n");
1.1053 + else
1.1054 + printf("NOT WEAK\n");
1.1055 +
1.1056 + printf("LOWER 0x%04lX\n", uctolower(0xff3a));
1.1057 + printf("UPPER 0x%04lX\n", uctoupper(0xff5a));
1.1058 +
1.1059 + if (ucisalpha(0x1d5))
1.1060 + printf("ALPHA\n");
1.1061 + else
1.1062 + printf("NOT ALPHA\n");
1.1063 +
1.1064 + if (ucisupper(0x1d5)) {
1.1065 + printf("UPPER\n");
1.1066 + lo = uctolower(0x1d5);
1.1067 + printf("0x%04lx\n", lo);
1.1068 + lo = uctotitle(0x1d5);
1.1069 + printf("0x%04lx\n", lo);
1.1070 + } else
1.1071 + printf("NOT UPPER\n");
1.1072 +
1.1073 + if (ucistitle(0x1d5))
1.1074 + printf("TITLE\n");
1.1075 + else
1.1076 + printf("NOT TITLE\n");
1.1077 +
1.1078 + if (uciscomposite(0x1d5))
1.1079 + printf("COMPOSITE\n");
1.1080 + else
1.1081 + printf("NOT COMPOSITE\n");
1.1082 +
1.1083 + if (ucdecomp(0x1d5, &lo, &dec)) {
1.1084 + for (i = 0; i < lo; i++)
1.1085 + printf("0x%04lx ", dec[i]);
1.1086 + putchar('\n');
1.1087 + }
1.1088 +
1.1089 + if ((lo = uccombining_class(0x41)) != 0)
1.1090 + printf("0x41 CCL %ld\n", lo);
1.1091 +
1.1092 + if (ucisxdigit(0xfeff))
1.1093 + printf("0xFEFF HEX DIGIT\n");
1.1094 + else
1.1095 + printf("0xFEFF NOT HEX DIGIT\n");
1.1096 +
1.1097 + if (ucisdefined(0x10000))
1.1098 + printf("0x10000 DEFINED\n");
1.1099 + else
1.1100 + printf("0x10000 NOT DEFINED\n");
1.1101 +
1.1102 + if (ucnumber_lookup(0x30, &num)) {
1.1103 + if (num.numerator != num.denominator)
1.1104 + printf("UCNUMBER: 0x30 = %d/%d\n", num.numerator, num.denominator);
1.1105 + else
1.1106 + printf("UCNUMBER: 0x30 = %d\n", num.numerator);
1.1107 + } else
1.1108 + printf("UCNUMBER: 0x30 NOT A NUMBER\n");
1.1109 +
1.1110 + if (ucnumber_lookup(0xbc, &num)) {
1.1111 + if (num.numerator != num.denominator)
1.1112 + printf("UCNUMBER: 0xbc = %d/%d\n", num.numerator, num.denominator);
1.1113 + else
1.1114 + printf("UCNUMBER: 0xbc = %d\n", num.numerator);
1.1115 + } else
1.1116 + printf("UCNUMBER: 0xbc NOT A NUMBER\n");
1.1117 +
1.1118 +
1.1119 + if (ucnumber_lookup(0xff19, &num)) {
1.1120 + if (num.numerator != num.denominator)
1.1121 + printf("UCNUMBER: 0xff19 = %d/%d\n", num.numerator, num.denominator);
1.1122 + else
1.1123 + printf("UCNUMBER: 0xff19 = %d\n", num.numerator);
1.1124 + } else
1.1125 + printf("UCNUMBER: 0xff19 NOT A NUMBER\n");
1.1126 +
1.1127 + if (ucnumber_lookup(0x4e00, &num)) {
1.1128 + if (num.numerator != num.denominator)
1.1129 + printf("UCNUMBER: 0x4e00 = %d/%d\n", num.numerator, num.denominator);
1.1130 + else
1.1131 + printf("UCNUMBER: 0x4e00 = %d\n", num.numerator);
1.1132 + } else
1.1133 + printf("UCNUMBER: 0x4e00 NOT A NUMBER\n");
1.1134 +
1.1135 + if (ucdigit_lookup(0x06f9, &dig))
1.1136 + printf("UCDIGIT: 0x6f9 = %d\n", dig);
1.1137 + else
1.1138 + printf("UCDIGIT: 0x6f9 NOT A NUMBER\n");
1.1139 +
1.1140 + dig = ucgetdigit(0x0969);
1.1141 + printf("UCGETDIGIT: 0x969 = %d\n", dig);
1.1142 +
1.1143 + num = ucgetnumber(0x30);
1.1144 + if (num.numerator != num.denominator)
1.1145 + printf("UCGETNUMBER: 0x30 = %d/%d\n", num.numerator, num.denominator);
1.1146 + else
1.1147 + printf("UCGETNUMBER: 0x30 = %d\n", num.numerator);
1.1148 +
1.1149 + num = ucgetnumber(0xbc);
1.1150 + if (num.numerator != num.denominator)
1.1151 + printf("UCGETNUMBER: 0xbc = %d/%d\n", num.numerator, num.denominator);
1.1152 + else
1.1153 + printf("UCGETNUMBER: 0xbc = %d\n", num.numerator);
1.1154 +
1.1155 + num = ucgetnumber(0xff19);
1.1156 + if (num.numerator != num.denominator)
1.1157 + printf("UCGETNUMBER: 0xff19 = %d/%d\n", num.numerator, num.denominator);
1.1158 + else
1.1159 + printf("UCGETNUMBER: 0xff19 = %d\n", num.numerator);
1.1160 +
1.1161 + ucdata_cleanup();
1.1162 + exit(0);
1.1163 +}
1.1164 +
1.1165 +#endif /* TEST */
