Home | History | Annotate | Line # | Download | only in iconv
iconv.c revision 1.15
      1  1.15      yamt /*	$NetBSD: iconv.c,v 1.15 2009/02/20 15:27:08 yamt Exp $ */
      2   1.1  tshiozak 
      3   1.1  tshiozak /*-
      4   1.1  tshiozak  * Copyright (c)2003 Citrus Project,
      5   1.1  tshiozak  * All rights reserved.
      6   1.1  tshiozak  *
      7   1.1  tshiozak  * Redistribution and use in source and binary forms, with or without
      8   1.1  tshiozak  * modification, are permitted provided that the following conditions
      9   1.1  tshiozak  * are met:
     10   1.1  tshiozak  * 1. Redistributions of source code must retain the above copyright
     11   1.1  tshiozak  *    notice, this list of conditions and the following disclaimer.
     12   1.1  tshiozak  * 2. Redistributions in binary form must reproduce the above copyright
     13   1.1  tshiozak  *    notice, this list of conditions and the following disclaimer in the
     14   1.1  tshiozak  *    documentation and/or other materials provided with the distribution.
     15   1.1  tshiozak  *
     16   1.1  tshiozak  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
     17   1.1  tshiozak  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     18   1.1  tshiozak  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     19   1.1  tshiozak  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
     20   1.1  tshiozak  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     21   1.1  tshiozak  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     22   1.1  tshiozak  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     23   1.1  tshiozak  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     24   1.1  tshiozak  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     25   1.1  tshiozak  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     26   1.1  tshiozak  * SUCH DAMAGE.
     27   1.1  tshiozak  */
     28   1.1  tshiozak 
     29   1.1  tshiozak #include <sys/cdefs.h>
     30   1.1  tshiozak #if defined(LIBC_SCCS) && !defined(lint)
     31  1.15      yamt __RCSID("$NetBSD: iconv.c,v 1.15 2009/02/20 15:27:08 yamt Exp $");
     32   1.1  tshiozak #endif /* LIBC_SCCS and not lint */
     33   1.1  tshiozak 
     34  1.12   tnozaki #include <err.h>
     35   1.1  tshiozak #include <errno.h>
     36  1.12   tnozaki #include <iconv.h>
     37  1.12   tnozaki #include <langinfo.h>
     38  1.13      yamt #include <locale.h>
     39   1.1  tshiozak #include <stdio.h>
     40   1.1  tshiozak #include <stdlib.h>
     41   1.1  tshiozak #include <string.h>
     42   1.1  tshiozak #include <unistd.h>
     43  1.10  christos #include <util.h>
     44   1.1  tshiozak 
     45  1.11     perry static void usage(void) __unused;
     46   1.8  christos static int scmp(const void *, const void *);
     47   1.8  christos static void show_codesets(void);
     48   1.8  christos static void do_conv(const char *, FILE *, const char *, const char *, int, int);
     49   1.8  christos 
     50   1.1  tshiozak static void
     51   1.1  tshiozak usage(void)
     52   1.1  tshiozak {
     53  1.12   tnozaki 	(void)fprintf(stderr,
     54  1.12   tnozaki 	    "Usage:\t%1$s [-cs] -f <from_code> -t <to_code> [file ...]\n"
     55  1.12   tnozaki 	    "\t%1$s -f <from_code> [-cs] [-t <to_code>] [file ...]\n"
     56  1.12   tnozaki 	    "\t%1$s -t <to_code> [-cs] [-f <from_code>] [file ...]\n"
     57  1.12   tnozaki 	    "\t%1$s -l\n", getprogname());
     58   1.1  tshiozak 	exit(1);
     59   1.1  tshiozak }
     60   1.1  tshiozak 
     61   1.5  jdolecek /*
     62   1.5  jdolecek  * qsort() helper function
     63   1.5  jdolecek  */
     64   1.5  jdolecek static int
     65   1.6      yamt scmp(const void *v1, const void *v2)
     66   1.5  jdolecek {
     67   1.6      yamt 	const char * const *s1 = v1;
     68   1.6      yamt 	const char * const *s2 = v2;
     69   1.6      yamt 
     70   1.6      yamt 	return strcasecmp(*s1, *s2);
     71   1.5  jdolecek }
     72   1.5  jdolecek 
     73   1.1  tshiozak static void
     74   1.1  tshiozak show_codesets(void)
     75   1.1  tshiozak {
     76   1.1  tshiozak 	char **list;
     77   1.1  tshiozak 	size_t sz, i;
     78   1.1  tshiozak 
     79   1.1  tshiozak 	if (__iconv_get_list(&list, &sz))
     80   1.1  tshiozak 		err(EXIT_FAILURE, "__iconv_get_list()");
     81   1.1  tshiozak 
     82   1.5  jdolecek 	qsort(list, sz, sizeof(char *), scmp);
     83   1.5  jdolecek 
     84   1.8  christos 	for (i = 0; i < sz; i++)
     85   1.8  christos 		(void)printf("%s\n", list[i]);
     86   1.1  tshiozak 
     87   1.1  tshiozak 	__iconv_free_list(list, sz);
     88   1.1  tshiozak }
     89   1.1  tshiozak 
     90   1.1  tshiozak #define INBUFSIZE 1024
     91   1.8  christos #define OUTBUFSIZE (INBUFSIZE * 2)
     92   1.1  tshiozak static void
     93   1.1  tshiozak do_conv(const char *fn, FILE *fp, const char *from, const char *to, int silent,
     94   1.8  christos     int hide_invalid)
     95   1.1  tshiozak {
     96   1.1  tshiozak 	char inbuf[INBUFSIZE], outbuf[OUTBUFSIZE], *out;
     97   1.1  tshiozak 	const char *in;
     98   1.1  tshiozak 	size_t inbytes, outbytes, ret, invalids;
     99   1.1  tshiozak 	iconv_t cd;
    100  1.14      yamt 	uint32_t flags = 0;
    101   1.1  tshiozak 
    102   1.1  tshiozak 	if (hide_invalid)
    103   1.1  tshiozak 		flags |= __ICONV_F_HIDE_INVALID;
    104   1.1  tshiozak 	cd = iconv_open(to, from);
    105   1.1  tshiozak 	if (cd == (iconv_t)-1)
    106   1.3       wiz 		err(EXIT_FAILURE, "iconv_open(%s, %s)", to, from);
    107   1.1  tshiozak 
    108   1.1  tshiozak 	invalids = 0;
    109   1.1  tshiozak 	while ((inbytes = fread(inbuf, 1, INBUFSIZE, fp)) > 0) {
    110   1.1  tshiozak 		in = inbuf;
    111   1.8  christos 		while (inbytes > 0) {
    112   1.4      yamt 			size_t inval;
    113   1.4      yamt 
    114   1.1  tshiozak 			out = outbuf;
    115   1.1  tshiozak 			outbytes = OUTBUFSIZE;
    116   1.1  tshiozak 			ret = __iconv(cd, &in, &inbytes, &out, &outbytes,
    117   1.8  christos 			    flags, &inval);
    118   1.4      yamt 			invalids += inval;
    119   1.8  christos 			if (ret == (size_t)-1 && errno != E2BIG) {
    120   1.1  tshiozak 				/*
    121   1.1  tshiozak 				 * XXX: iconv(3) is bad interface.
    122   1.1  tshiozak 				 *   invalid character count is lost here.
    123   1.1  tshiozak 				 *   instead, we just provide __iconv function.
    124   1.1  tshiozak 				 */
    125   1.1  tshiozak 				if (errno != EINVAL || in == inbuf)
    126   1.1  tshiozak 					err(EXIT_FAILURE, "iconv()");
    127   1.1  tshiozak 
    128   1.1  tshiozak 				/* incomplete input character */
    129   1.8  christos 				(void)memmove(inbuf, in, inbytes);
    130   1.8  christos 				ret = fread(inbuf + inbytes, 1,
    131   1.8  christos 				    INBUFSIZE - inbytes, fp);
    132   1.1  tshiozak 				if (ret == 0) {
    133   1.9  tshiozak 					if (feof(fp))
    134   1.9  tshiozak 						errx(EXIT_FAILURE,
    135   1.9  tshiozak 						     "unexpected end of file; "
    136   1.9  tshiozak 						     "the last character is "
    137   1.9  tshiozak 						     "incomplete.");
    138   1.9  tshiozak 					else
    139   1.1  tshiozak 						err(EXIT_FAILURE, "fread()");
    140   1.1  tshiozak 				}
    141   1.1  tshiozak 				in = inbuf;
    142   1.1  tshiozak 				inbytes += ret;
    143   1.1  tshiozak 			}
    144   1.1  tshiozak 			if (outbytes < OUTBUFSIZE)
    145   1.8  christos 				(void)fwrite(outbuf, 1, OUTBUFSIZE - outbytes,
    146   1.8  christos 				    stdout);
    147   1.1  tshiozak 		}
    148   1.1  tshiozak 	}
    149   1.1  tshiozak 	/* reset the shift state of the output buffer */
    150   1.1  tshiozak 	outbytes = OUTBUFSIZE;
    151   1.1  tshiozak 	out = outbuf;
    152   1.1  tshiozak 	ret = iconv(cd, NULL, NULL, &out, &outbytes);
    153  1.15      yamt 	if (ret == (size_t)-1)
    154   1.1  tshiozak 		err(EXIT_FAILURE, "iconv()");
    155   1.1  tshiozak 	if (outbytes < OUTBUFSIZE)
    156   1.8  christos 		(void)fwrite(outbuf, 1, OUTBUFSIZE - outbytes, stdout);
    157   1.1  tshiozak 
    158   1.1  tshiozak 	if (invalids > 0 && !silent)
    159   1.2  tshiozak 		warnx("warning: invalid characters: %lu",
    160   1.8  christos 		    (unsigned long)invalids);
    161   1.1  tshiozak 
    162   1.1  tshiozak 	iconv_close(cd);
    163   1.1  tshiozak }
    164   1.1  tshiozak 
    165   1.1  tshiozak int
    166   1.1  tshiozak main(int argc, char **argv)
    167   1.1  tshiozak {
    168   1.1  tshiozak 	int ch, i;
    169   1.1  tshiozak 	int opt_l = 0, opt_s = 0, opt_c = 0;
    170   1.1  tshiozak 	char *opt_f = NULL, *opt_t = NULL;
    171   1.1  tshiozak 	FILE *fp;
    172   1.1  tshiozak 
    173  1.13      yamt 	setlocale(LC_ALL, "");
    174   1.8  christos 	setprogname(argv[0]);
    175   1.8  christos 
    176   1.8  christos 	while ((ch = getopt(argc, argv, "cslf:t:")) != EOF) {
    177   1.1  tshiozak 		switch (ch) {
    178   1.1  tshiozak 		case 'c':
    179   1.1  tshiozak 			opt_c = 1;
    180   1.1  tshiozak 			break;
    181   1.1  tshiozak 		case 's':
    182   1.1  tshiozak 			opt_s = 1;
    183   1.1  tshiozak 			break;
    184   1.1  tshiozak 		case 'l':
    185   1.1  tshiozak 			/* list */
    186   1.1  tshiozak 			opt_l = 1;
    187   1.1  tshiozak 			break;
    188   1.1  tshiozak 		case 'f':
    189   1.1  tshiozak 			/* from */
    190   1.8  christos 			opt_f = estrdup(optarg);
    191   1.1  tshiozak 			break;
    192   1.1  tshiozak 		case 't':
    193   1.1  tshiozak 			/* to */
    194   1.8  christos 			opt_t = estrdup(optarg);
    195   1.1  tshiozak 			break;
    196   1.1  tshiozak 		default:
    197   1.1  tshiozak 			usage();
    198   1.1  tshiozak 		}
    199   1.1  tshiozak 	}
    200   1.8  christos 	argc -= optind;
    201   1.8  christos 	argv += optind;
    202   1.1  tshiozak 	if (opt_l) {
    203   1.8  christos 		if (argc > 0 || opt_s || opt_f != NULL || opt_t != NULL) {
    204   1.8  christos 			warnx("-l is not allowed with other flags.");
    205   1.1  tshiozak 			usage();
    206   1.1  tshiozak 		}
    207   1.1  tshiozak 		show_codesets();
    208   1.1  tshiozak 	} else {
    209  1.12   tnozaki 		if (opt_f == NULL) {
    210  1.12   tnozaki 			if (opt_t == NULL)
    211  1.12   tnozaki 				usage();
    212  1.12   tnozaki 			opt_f = nl_langinfo(CODESET);
    213  1.12   tnozaki 		} else if (opt_t == NULL)
    214  1.12   tnozaki 			opt_t = nl_langinfo(CODESET);
    215   1.1  tshiozak 
    216   1.1  tshiozak 		if (argc == 0)
    217   1.1  tshiozak 			do_conv("<stdin>", stdin, opt_f, opt_t, opt_s, opt_c);
    218   1.1  tshiozak 		else {
    219   1.8  christos 			for (i = 0; i < argc; i++) {
    220   1.1  tshiozak 				fp = fopen(argv[i], "r");
    221   1.1  tshiozak 				if (fp == NULL)
    222   1.8  christos 					err(EXIT_FAILURE, "Cannot open `%s'",
    223   1.8  christos 					    argv[i]);
    224   1.1  tshiozak 				do_conv(argv[i], fp, opt_f, opt_t, opt_s,
    225   1.8  christos 				    opt_c);
    226   1.8  christos 				(void)fclose(fp);
    227   1.1  tshiozak 			}
    228   1.1  tshiozak 		}
    229   1.1  tshiozak 	}
    230   1.1  tshiozak 	return EXIT_SUCCESS;
    231   1.1  tshiozak }
    232