lib/libkiconv/xlat16_iconv.c

   1 /*-
   2  * Copyright (c) 2003, 2005 Ryuichiro Imura
   3  * All rights reserved.
   4  *
   5  * Redistribution and use in source and binary forms, with or without
   6  * modification, are permitted provided that the following conditions
   7  * are met:
   8  * 1. Redistributions of source code must retain the above copyright
   9  *    notice, this list of conditions and the following disclaimer.
  10  * 2. Redistributions in binary form must reproduce the above copyright
  11  *    notice, this list of conditions and the following disclaimer in the
  12  *    documentation and/or other materials provided with the distribution.
  13  *
  14  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  15  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  16  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  17  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  18  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  19  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  20  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  21  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  22  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  23  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  24  * SUCH DAMAGE.
  25  *
  26  * $FreeBSD$
  27  */
  28
  29 /*
  30  * kiconv(3) requires shared linked, and reduce module size
  31  * when statically linked.
  32  */
  33
  34 #ifdef PIC
  35
  36 #include <sys/types.h>
  37 #include <sys/iconv.h>
  38 #include <sys/sysctl.h>
  39
  40 #include <ctype.h>
  41 #include <dlfcn.h>
  42 #include <err.h>
  43 #include <errno.h>
  44 #include <locale.h>
  45 #include <stdio.h>
  46 #include <stdlib.h>
  47 #include <string.h>
  48 #include <wctype.h>
  49
  50 #include "quirks.h"
  51
  52 typedef void *iconv_t;
  53
  54 struct xlat16_table {
  55         uint32_t *      idx[0x200];
  56         void *          data;
  57         size_t          size;
  58 };
  59
  60 static struct xlat16_table kiconv_xlat16_open(const char *, const char *, int);
  61 static int chklocale(int, const char *);
  62
  63 static int my_iconv_init(void);
  64 static iconv_t (*my_iconv_open)(const char *, const char *);
  65 static size_t (*my_iconv)(iconv_t, const char **, size_t *, char **, size_t *);
  66 static int (*my_iconv_close)(iconv_t);
  67 static size_t my_iconv_char(iconv_t, const u_char **, size_t *, u_char **, size_t *);
  68
  69 int
  70 kiconv_add_xlat16_cspair(const char *tocode, const char *fromcode, int flag)
  71 {
  72         int error;
  73         size_t idxsize;
  74         struct xlat16_table xt;
  75         void *data;
  76         char *p;
  77         const char unicode[] = ENCODING_UNICODE;
  78
  79         if ((flag & KICONV_WCTYPE) == 0 &&
  80             strcmp(unicode, tocode) != 0 &&
  81             strcmp(unicode, fromcode) != 0 &&
  82             kiconv_lookupconv(unicode) == 0) {
  83                 error = kiconv_add_xlat16_cspair(unicode, fromcode, flag);
  84                 if (error)
  85                         return (-1);
  86                 error = kiconv_add_xlat16_cspair(tocode, unicode, flag);
  87                 return (error);
  88         }
  89
  90         if (kiconv_lookupcs(tocode, fromcode) == 0)
  91                 return (0);
  92
  93         if (flag & KICONV_WCTYPE)
  94                 xt = kiconv_xlat16_open(fromcode, fromcode, flag);
  95         else
  96                 xt = kiconv_xlat16_open(tocode, fromcode, flag);
  97         if (xt.size == 0)
  98                 return (-1);
  99
 100         idxsize = sizeof(xt.idx);
 101
 102         if ((idxsize + xt.size) > ICONV_CSMAXDATALEN) {
 103                 errno = E2BIG;
 104                 return (-1);
 105         }
 106
 107         if ((data = malloc(idxsize + xt.size)) != NULL) {
 108                 p = data;
 109                 memcpy(p, xt.idx, idxsize);
 110                 p += idxsize;
 111                 memcpy(p, xt.data, xt.size);
 112                 error = kiconv_add_xlat16_table(tocode, fromcode, data,
 113                     (int)(idxsize + xt.size));
 114                 return (error);
 115         }
 116
 117         return (-1);
 118 }
 119
 120 int
 121 kiconv_add_xlat16_cspairs(const char *foreigncode, const char *localcode)
 122 {
 123         int error, locale;
 124
 125         error = kiconv_add_xlat16_cspair(foreigncode, localcode,
 126             KICONV_FROM_LOWER | KICONV_FROM_UPPER);
 127         if (error)
 128                 return (error);
 129         error = kiconv_add_xlat16_cspair(localcode, foreigncode,
 130             KICONV_LOWER | KICONV_UPPER);
 131         if (error)
 132                 return (error);
 133         locale = chklocale(LC_CTYPE, localcode);
 134         if (locale == 0) {
 135                 error = kiconv_add_xlat16_cspair(KICONV_WCTYPE_NAME, localcode,
 136                     KICONV_WCTYPE);
 137                 if (error)
 138                         return (error);
 139         }
 140
 141         return (0);
 142 }
 143
 144 static struct xlat16_table
 145 kiconv_xlat16_open(const char *tocode, const char *fromcode, int lcase)
 146 {
 147         u_char src[3], dst[4], *srcp, *dstp, ud, ld;
 148         int us, ls, ret;
 149         uint16_t c;
 150         uint32_t table[0x80];
 151         size_t inbytesleft, outbytesleft, pre_q_size, post_q_size;
 152         struct xlat16_table xt;
 153         struct quirk_replace_list *pre_q_list, *post_q_list;
 154         iconv_t cd;
 155         char *p;
 156
 157         xt.data = NULL;
 158         xt.size = 0;
 159
 160         src[2] = '\0';
 161         dst[3] = '\0';
 162
 163         ret = my_iconv_init();
 164         if (ret)
 165                 return (xt);
 166
 167         cd = my_iconv_open(search_quirk(tocode, fromcode, &pre_q_list, &pre_q_size),
 168             search_quirk(fromcode, tocode, &post_q_list, &post_q_size));
 169         if (cd == (iconv_t) (-1))
 170                 return (xt);
 171
 172         if ((xt.data = malloc(0x200 * 0x80 * sizeof(uint32_t))) == NULL)
 173                 return (xt);
 174
 175         p = xt.data;
 176
 177         for (ls = 0 ; ls < 0x200 ; ls++) {
 178                 xt.idx[ls] = NULL;
 179                 for (us = 0 ; us < 0x80 ; us++) {
 180                         srcp = src;
 181                         dstp = dst;
 182
 183                         inbytesleft = 2;
 184                         outbytesleft = 3;
 185                         bzero(dst, outbytesleft);
 186
 187                         c = ((ls & 0x100 ? us | 0x80 : us) << 8) | (u_char)ls;
 188
 189                         if (lcase & KICONV_WCTYPE) {
 190                                 if ((c & 0xff) == 0)
 191                                         c >>= 8;
 192                                 if (iswupper(c)) {
 193                                         c = towlower(c);
 194                                         if ((c & 0xff00) == 0)
 195                                                 c <<= 8;
 196                                         table[us] = c | XLAT16_HAS_LOWER_CASE;
 197                                 } else if (iswlower(c)) {
 198                                         c = towupper(c);
 199                                         if ((c & 0xff00) == 0)
 200                                                 c <<= 8;
 201                                         table[us] = c | XLAT16_HAS_UPPER_CASE;
 202                                 } else
 203                                         table[us] = 0;
 204                                 /*
 205                                  * store not NULL
 206                                  */
 207                                 if (table[us])
 208                                         xt.idx[ls] = table;
 209
 210                                 continue;
 211                         }
 212
 213                         c = quirk_vendor2unix(c, pre_q_list, pre_q_size);
 214                         src[0] = (u_char)(c >> 8);
 215                         src[1] = (u_char)c;
 216
 217                         ret = my_iconv_char(cd, (const u_char **)&srcp,
 218                             &inbytesleft, &dstp, &outbytesleft);
 219                         if (ret == -1) {
 220                                 table[us] = 0;
 221                                 continue;
 222                         }
 223
 224                         ud = (u_char)dst[0];
 225                         ld = (u_char)dst[1];
 226
 227                         switch(outbytesleft) {
 228                         case 0:
 229 #ifdef XLAT16_ACCEPT_3BYTE_CHR
 230                                 table[us] = (ud << 8) | ld;
 231                                 table[us] |= (u_char)dst[2] << 16;
 232                                 table[us] |= XLAT16_IS_3BYTE_CHR;
 233 #else
 234                                 table[us] = 0;
 235                                 continue;
 236 #endif
 237                                 break;
 238                         case 1:
 239                                 table[us] = quirk_unix2vendor((ud << 8) | ld,
 240                                     post_q_list, post_q_size);
 241                                 if ((table[us] >> 8) == 0)
 242                                         table[us] |= XLAT16_ACCEPT_NULL_OUT;
 243                                 break;
 244                         case 2:
 245                                 table[us] = ud;
 246                                 if (lcase & KICONV_LOWER && ud != tolower(ud)) {
 247                                         table[us] |= (u_char)tolower(ud) << 16;
 248                                         table[us] |= XLAT16_HAS_LOWER_CASE;
 249                                 }
 250                                 if (lcase & KICONV_UPPER && ud != toupper(ud)) {
 251                                         table[us] |= (u_char)toupper(ud) << 16;
 252                                         table[us] |= XLAT16_HAS_UPPER_CASE;
 253                                 }
 254                                 break;
 255                         }
 256
 257                         switch(inbytesleft) {
 258                         case 0:
 259                                 if ((ls & 0xff) == 0)
 260                                         table[us] |= XLAT16_ACCEPT_NULL_IN;
 261                                 break;
 262                         case 1:
 263                                 c = ls > 0xff ? us | 0x80 : us;
 264                                 if (lcase & KICONV_FROM_LOWER && c != tolower(c)) {
 265                                         table[us] |= (u_char)tolower(c) << 16;
 266                                         table[us] |= XLAT16_HAS_FROM_LOWER_CASE;
 267                                 }
 268                                 if (lcase & KICONV_FROM_UPPER && c != toupper(c)) {
 269                                         table[us] |= (u_char)toupper(c) << 16;
 270                                         table[us] |= XLAT16_HAS_FROM_UPPER_CASE;
 271                                 }
 272                                 break;
 273                         }
 274
 275                         if (table[us] == 0)
 276                                 continue;
 277
 278                         /*
 279                          * store not NULL
 280                          */
 281                         xt.idx[ls] = table;
 282                 }
 283                 if (xt.idx[ls]) {
 284                         memcpy(p, table, sizeof(table));
 285                         p += sizeof(table);
 286                 }
 287         }
 288         my_iconv_close(cd);
 289
 290         xt.size = p - (char *)xt.data;
 291         xt.data = realloc(xt.data, xt.size);
 292         return (xt);
 293 }
 294
 295 static int
 296 chklocale(int category, const char *code)
 297 {
 298         char *p;
 299         int error = -1;
 300
 301         p = strchr(setlocale(category, NULL), '.');
 302         if (p++) {
 303                 error = strcasecmp(code, p);
 304                 if (error) {
 305                         /* XXX - can't avoid calling quirk here... */
 306                         error = strcasecmp(code, kiconv_quirkcs(p,
 307                             KICONV_VENDOR_MICSFT));
 308                 }
 309         }
 310         return (error);
 311 }
 312
 313 static int
 314 my_iconv_init(void)
 315 {
 316         void *iconv_lib;
 317
 318         iconv_lib = dlopen("libiconv.so", RTLD_LAZY | RTLD_GLOBAL);
 319         if (iconv_lib == NULL) {
 320                 warn("Unable to load iconv library: %s\n", dlerror());
 321                 errno = ENOENT;
 322                 return (-1);
 323         }
 324         my_iconv_open = dlsym(iconv_lib, "iconv_open");
 325         my_iconv = dlsym(iconv_lib, "iconv");
 326         my_iconv_close = dlsym(iconv_lib, "iconv_close");
 327
 328         return (0);
 329 }
 330
 331 static size_t
 332 my_iconv_char(iconv_t cd, const u_char **ibuf, size_t * ilen, u_char **obuf,
 333         size_t * olen)
 334 {
 335         const u_char *sp;
 336         u_char *dp, ilocal[3], olocal[3];
 337         u_char c1, c2;
 338         int ret;
 339         size_t ir, or;
 340
 341         sp = *ibuf;
 342         dp = *obuf;
 343         ir = *ilen;
 344
 345         bzero(*obuf, *olen);
 346         ret = my_iconv(cd, (const char **)&sp, ilen, (char **)&dp, olen);
 347         c1 = (*obuf)[0];
 348         c2 = (*obuf)[1];
 349
 350         if (ret == -1) {
 351                 if (*ilen == ir - 1 && (*ibuf)[1] == '\0' && (c1 || c2))
 352                         return (0);
 353                 else
 354                         return (-1);
 355         }
 356
 357         /*
 358          * We must judge if inbuf is a single byte char or double byte char.
 359          * Here, to judge, try first byte(*sp) conversion and compare.
 360          */
 361         ir = 1;
 362         or = 3;
 363
 364         bzero(olocal, or);
 365         memcpy(ilocal, *ibuf, sizeof(ilocal));
 366         sp = ilocal;
 367         dp = olocal;
 368
 369         if ((my_iconv(cd,(const char **)&sp, &ir, (char **)&dp, &or)) != -1) {
 370                 if (olocal[0] != c1)
 371                         return (ret);
 372
 373                 if (olocal[1] == c2 && (*ibuf)[1] == '\0') {
 374                         /*
 375                          * inbuf is a single byte char
 376                          */
 377                         *ilen = 1;
 378                         *olen = or;
 379                         return (ret);
 380                 }
 381
 382                 switch(or) {
 383                 case 0:
 384                 case 1:
 385                         if (olocal[1] == c2) {
 386                                 /*
 387                                  * inbuf is a single byte char,
 388                                  * so return false here.
 389                                  */
 390                                 return (-1);
 391                         } else {
 392                                 /*
 393                                  * inbuf is a double byte char
 394                                  */
 395                                 return (ret);
 396                         }
 397                         break;
 398                 case 2:
 399                         /*
 400                          * should compare second byte of inbuf
 401                          */
 402                         break;
 403                 }
 404         } else {
 405                 /*
 406                  * inbuf clould not be splitted, so inbuf is
 407                  * a double byte char.
 408                  */
 409                 return (ret);
 410         }
 411
 412         /*
 413          * try second byte(*(sp+1)) conversion, and compare
 414          */
 415         ir = 1;
 416         or = 3;
 417
 418         bzero(olocal, or);
 419
 420         sp = ilocal + 1;
 421         dp = olocal;
 422
 423         if ((my_iconv(cd,(const char **)&sp, &ir, (char **)&dp, &or)) != -1) {
 424                 if (olocal[0] == c2)
 425                         /*
 426                          * inbuf is a single byte char
 427                          */
 428                         return (-1);
 429         }
 430
 431         return (ret);
 432 }
 433
 434 #else /* statically linked */
 435
 436 #include <sys/types.h>
 437 #include <sys/iconv.h>
 438 #include <errno.h>
 439
 440 int
 441 kiconv_add_xlat16_cspair(const char *tocode __unused, const char *fromcode __unused,
 442     int flag __unused)
 443 {
 444
 445         errno = EINVAL;
 446         return (-1);
 447 }
 448
 449 int
 450 kiconv_add_xlat16_cspairs(const char *tocode __unused, const char *fromcode __unused)
 451 {
 452         errno = EINVAL;
 453         return (-1);
 454 }
 455
 456 #endif /* PIC */