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