usr.bin/iconv/iconv.c

   1 /* $FreeBSD$ */
   2 /* $NetBSD: iconv.c,v 1.16 2009/02/20 15:28:21 yamt Exp $ */
   3
   4 /*-
   5  * SPDX-License-Identifier: BSD-2-Clause
   6  *
   7  * Copyright (c)2003 Citrus Project,
   8  * All rights reserved.
   9  *
  10  * Redistribution and use in source and binary forms, with or without
  11  * modification, are permitted provided that the following conditions
  12  * are met:
  13  * 1. Redistributions of source code must retain the above copyright
  14  *    notice, this list of conditions and the following disclaimer.
  15  * 2. Redistributions in binary form must reproduce the above copyright
  16  *    notice, this list of conditions and the following disclaimer in the
  17  *    documentation and/or other materials provided with the distribution.
  18  *
  19  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
  20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
  23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  29  * SUCH DAMAGE.
  30  */
  31
  32 #include <sys/cdefs.h>
  33 #include <sys/capsicum.h>
  34
  35 #include <capsicum_helpers.h>
  36 #include <err.h>
  37 #include <errno.h>
  38 #include <getopt.h>
  39 #include <iconv.h>
  40 #include <limits.h>
  41 #include <locale.h>
  42 #include <stdbool.h>
  43 #include <stdio.h>
  44 #include <stdlib.h>
  45 #include <string.h>
  46 #include <unistd.h>
  47
  48 static int              do_conv(FILE *, iconv_t, bool, bool);
  49 static int              do_list(unsigned int, const char * const *, void *);
  50 static void             usage(void) __dead2;
  51
  52 static const struct option long_options[] = {
  53         {"from-code",           required_argument,      NULL, 'f'},
  54         {"list",                no_argument,            NULL, 'l'},
  55         {"silent",              no_argument,            NULL, 's'},
  56         {"to-code",             required_argument,      NULL, 't'},
  57         {NULL,                  no_argument,            NULL, 0}
  58 };
  59
  60 static void
  61 usage(void)
  62 {
  63         (void)fprintf(stderr,
  64             "Usage:\t%1$s [-cs] -f <from_code> -t <to_code> [file ...]\n"
  65             "\t%1$s -f <from_code> [-cs] [-t <to_code>] [file ...]\n"
  66             "\t%1$s -t <to_code> [-cs] [-f <from_code>] [file ...]\n"
  67             "\t%1$s -l\n", getprogname());
  68         exit(1);
  69 }
  70
  71 #define INBUFSIZE 1024
  72 #define OUTBUFSIZE (INBUFSIZE * 2)
  73 static int
  74 do_conv(FILE *fp, iconv_t cd, bool silent, bool hide_invalid)
  75 {
  76         char inbuf[INBUFSIZE], outbuf[OUTBUFSIZE], *in, *out;
  77         unsigned long long invalids;
  78         size_t inbytes, outbytes, ret;
  79
  80         int arg = (int)hide_invalid;
  81         if (iconvctl(cd, ICONV_SET_DISCARD_ILSEQ, (void *)&arg) == -1)
  82                 err(EXIT_FAILURE, "iconvctl(DISCARD_ILSEQ, %d)", arg);
  83
  84         invalids = 0;
  85         while ((inbytes = fread(inbuf, 1, INBUFSIZE, fp)) > 0) {
  86                 in = inbuf;
  87                 while (inbytes > 0) {
  88                         size_t inval;
  89
  90                         out = outbuf;
  91                         outbytes = OUTBUFSIZE;
  92                         ret = __iconv(cd, &in, &inbytes, &out, &outbytes,
  93                             0, &inval);
  94                         invalids += inval;
  95                         if (outbytes < OUTBUFSIZE)
  96                                 (void)fwrite(outbuf, 1, OUTBUFSIZE - outbytes,
  97                                     stdout);
  98                         if (ret == (size_t)-1 && errno != E2BIG) {
  99                                 if (errno != EINVAL || in == inbuf)
 100                                         err(EXIT_FAILURE, "iconv()");
 101
 102                                 /* incomplete input character */
 103                                 (void)memmove(inbuf, in, inbytes);
 104                                 ret = fread(inbuf + inbytes, 1,
 105                                     INBUFSIZE - inbytes, fp);
 106                                 if (ret == 0) {
 107                                         fflush(stdout);
 108                                         if (feof(fp))
 109                                                 errx(EXIT_FAILURE,
 110                                                     "unexpected end of file; "
 111                                                     "the last character is "
 112                                                     "incomplete.");
 113                                         else
 114                                                 err(EXIT_FAILURE, "fread()");
 115                                 }
 116                                 in = inbuf;
 117                                 inbytes += ret;
 118                         }
 119                 }
 120         }
 121         /* reset the shift state of the output buffer */
 122         outbytes = OUTBUFSIZE;
 123         out = outbuf;
 124         ret = iconv(cd, NULL, NULL, &out, &outbytes);
 125         if (ret == (size_t)-1)
 126                 err(EXIT_FAILURE, "iconv()");
 127         if (outbytes < OUTBUFSIZE)
 128                 (void)fwrite(outbuf, 1, OUTBUFSIZE - outbytes, stdout);
 129
 130         if (invalids > 0 && !silent)
 131                 warnx("warning: invalid characters: %llu", invalids);
 132
 133         return (invalids > 0);
 134 }
 135
 136 static int
 137 do_list(unsigned int n, const char * const *list, void *data __unused)
 138 {
 139         unsigned int i;
 140
 141         for(i = 0; i < n; i++) {
 142                 printf("%s", list[i]);
 143                 if (i < n - 1)
 144                         printf(" ");
 145         }
 146         printf("\n");
 147
 148         return (1);
 149 }
 150
 151 int
 152 main(int argc, char **argv)
 153 {
 154         iconv_t cd;
 155         FILE *fp;
 156         const char *opt_f, *opt_t;
 157         int ch, i, res;
 158         bool opt_c = false, opt_s = false;
 159
 160         opt_f = opt_t = "";
 161
 162         setlocale(LC_ALL, "");
 163         setprogname(argv[0]);
 164
 165         while ((ch = getopt_long(argc, argv, "csLlf:t:",
 166             long_options, NULL)) != -1) {
 167                 switch (ch) {
 168                 case 'c':
 169                         opt_c = true;
 170                         break;
 171                 case 's':
 172                         opt_s = true;
 173                         break;
 174                 case 'l':
 175                         /* list */
 176                         if (opt_s || opt_c || strcmp(opt_f, "") != 0 ||
 177                             strcmp(opt_t, "") != 0) {
 178                                 warnx("-l is not allowed with other flags.");
 179                                 usage();
 180                         }
 181                         iconvlist(do_list, NULL);
 182                         return (EXIT_SUCCESS);
 183                 case 'f':
 184                         /* from */
 185                         if (optarg != NULL)
 186                                 opt_f = optarg;
 187                         break;
 188                 case 't':
 189                         /* to */
 190                         if (optarg != NULL)
 191                                 opt_t = optarg;
 192                         break;
 193                 default:
 194                         usage();
 195                 }
 196         }
 197         argc -= optind;
 198         argv += optind;
 199         if ((strcmp(opt_f, "") == 0) && (strcmp(opt_t, "") == 0))
 200                 usage();
 201
 202         if (caph_limit_stdio() < 0)
 203                 err(EXIT_FAILURE, "capsicum");
 204
 205         /*
 206          * Cache NLS data, for strerror, for err(3), before entering capability
 207          * mode.
 208          */
 209         caph_cache_catpages();
 210
 211         /*
 212          * Cache iconv conversion handle before entering sandbox.
 213          */
 214         cd = iconv_open(opt_t, opt_f);
 215         if (cd == (iconv_t)-1)
 216                 err(EXIT_FAILURE, "iconv_open(%s, %s)", opt_t, opt_f);
 217
 218         if (argc == 0) {
 219                 if (caph_enter() < 0)
 220                         err(EXIT_FAILURE, "unable to enter capability mode");
 221                 res = do_conv(stdin, cd, opt_s, opt_c);
 222         } else {
 223                 res = 0;
 224                 for (i = 0; i < argc; i++) {
 225                         fp = (strcmp(argv[i], "-") != 0) ?
 226                             fopen(argv[i], "r") : stdin;
 227                         if (fp == NULL)
 228                                 err(EXIT_FAILURE, "Cannot open `%s'",
 229                                     argv[i]);
 230                         /* Enter Capsicum sandbox for final input file. */
 231                         if (i + 1 == argc && caph_enter() < 0)
 232                                 err(EXIT_FAILURE,
 233                                     "unable to enter capability mode");
 234                         res |= do_conv(fp, cd, opt_s, opt_c);
 235                         (void)fclose(fp);
 236
 237                         /* Reset iconv descriptor state. */
 238                         (void)iconv(cd, NULL, NULL, NULL, NULL);
 239                 }
 240         }
 241         iconv_close(cd);
 242         return (res == 0 ? EXIT_SUCCESS : EXIT_FAILURE);
 243 }