Home | History | Annotate | Line # | Download | only in iconv
iconv.c revision 1.8
      1  1.8  christos /*	$NetBSD: iconv.c,v 1.8 2005/04/24 17:46:06 christos 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  1.8  christos __RCSID("$NetBSD: iconv.c,v 1.8 2005/04/24 17:46:06 christos Exp $");
     32  1.1  tshiozak #endif /* LIBC_SCCS and not lint */
     33  1.1  tshiozak 
     34  1.1  tshiozak #include <errno.h>
     35  1.1  tshiozak #include <stdio.h>
     36  1.1  tshiozak #include <stdlib.h>
     37  1.1  tshiozak #include <string.h>
     38  1.1  tshiozak #include <iconv.h>
     39  1.1  tshiozak #include <unistd.h>
     40  1.1  tshiozak #include <err.h>
     41  1.1  tshiozak 
     42  1.8  christos static void usage(void) __attribute__((__unused__));
     43  1.8  christos static char *estrdup(const char *);
     44  1.8  christos static int scmp(const void *, const void *);
     45  1.8  christos static void show_codesets(void);
     46  1.8  christos static void do_conv(const char *, FILE *, const char *, const char *, int, int);
     47  1.8  christos 
     48  1.1  tshiozak static void
     49  1.1  tshiozak usage(void)
     50  1.1  tshiozak {
     51  1.8  christos 	(void)fprintf(stderr, "Usage: %s [-cs] -f <from> -t <to> [file ...]\n"
     52  1.8  christos 	    "\t%s -l\n", getprogname(), getprogname());
     53  1.1  tshiozak 	exit(1);
     54  1.1  tshiozak }
     55  1.1  tshiozak 
     56  1.8  christos static char *
     57  1.8  christos estrdup(const char *str)
     58  1.8  christos {
     59  1.8  christos 	char *ptr = strdup(str);
     60  1.8  christos 	if (ptr == NULL)
     61  1.8  christos 		err(EXIT_FAILURE, "Cannot copy string");
     62  1.8  christos 	return ptr;
     63  1.8  christos }
     64  1.8  christos 
     65  1.5  jdolecek /*
     66  1.5  jdolecek  * qsort() helper function
     67  1.5  jdolecek  */
     68  1.5  jdolecek static int
     69  1.6      yamt scmp(const void *v1, const void *v2)
     70  1.5  jdolecek {
     71  1.6      yamt 	const char * const *s1 = v1;
     72  1.6      yamt 	const char * const *s2 = v2;
     73  1.6      yamt 
     74  1.6      yamt 	return strcasecmp(*s1, *s2);
     75  1.5  jdolecek }
     76  1.5  jdolecek 
     77  1.1  tshiozak static void
     78  1.1  tshiozak show_codesets(void)
     79  1.1  tshiozak {
     80  1.1  tshiozak 	char **list;
     81  1.1  tshiozak 	size_t sz, i;
     82  1.1  tshiozak 
     83  1.1  tshiozak 	if (__iconv_get_list(&list, &sz))
     84  1.1  tshiozak 		err(EXIT_FAILURE, "__iconv_get_list()");
     85  1.1  tshiozak 
     86  1.5  jdolecek 	qsort(list, sz, sizeof(char *), scmp);
     87  1.5  jdolecek 
     88  1.8  christos 	for (i = 0; i < sz; i++)
     89  1.8  christos 		(void)printf("%s\n", list[i]);
     90  1.1  tshiozak 
     91  1.1  tshiozak 	__iconv_free_list(list, sz);
     92  1.1  tshiozak }
     93  1.1  tshiozak 
     94  1.1  tshiozak #define INBUFSIZE 1024
     95  1.8  christos #define OUTBUFSIZE (INBUFSIZE * 2)
     96  1.1  tshiozak static void
     97  1.1  tshiozak do_conv(const char *fn, FILE *fp, const char *from, const char *to, int silent,
     98  1.8  christos     int hide_invalid)
     99  1.1  tshiozak {
    100  1.1  tshiozak 	char inbuf[INBUFSIZE], outbuf[OUTBUFSIZE], *out;
    101  1.1  tshiozak 	const char *in;
    102  1.1  tshiozak 	size_t inbytes, outbytes, ret, invalids;
    103  1.1  tshiozak 	iconv_t cd;
    104  1.1  tshiozak 	u_int32_t flags = 0;
    105  1.1  tshiozak 
    106  1.1  tshiozak 	if (hide_invalid)
    107  1.1  tshiozak 		flags |= __ICONV_F_HIDE_INVALID;
    108  1.1  tshiozak 	cd = iconv_open(to, from);
    109  1.1  tshiozak 	if (cd == (iconv_t)-1)
    110  1.3       wiz 		err(EXIT_FAILURE, "iconv_open(%s, %s)", to, from);
    111  1.1  tshiozak 
    112  1.1  tshiozak 	invalids = 0;
    113  1.1  tshiozak 	while ((inbytes = fread(inbuf, 1, INBUFSIZE, fp)) > 0) {
    114  1.1  tshiozak 		in = inbuf;
    115  1.8  christos 		while (inbytes > 0) {
    116  1.4      yamt 			size_t inval;
    117  1.4      yamt 
    118  1.1  tshiozak 			out = outbuf;
    119  1.1  tshiozak 			outbytes = OUTBUFSIZE;
    120  1.1  tshiozak 			ret = __iconv(cd, &in, &inbytes, &out, &outbytes,
    121  1.8  christos 			    flags, &inval);
    122  1.4      yamt 			invalids += inval;
    123  1.8  christos 			if (ret == (size_t)-1 && errno != E2BIG) {
    124  1.1  tshiozak 				/*
    125  1.1  tshiozak 				 * XXX: iconv(3) is bad interface.
    126  1.1  tshiozak 				 *   invalid character count is lost here.
    127  1.1  tshiozak 				 *   instead, we just provide __iconv function.
    128  1.1  tshiozak 				 */
    129  1.1  tshiozak 				if (errno != EINVAL || in == inbuf)
    130  1.1  tshiozak 					err(EXIT_FAILURE, "iconv()");
    131  1.1  tshiozak 
    132  1.1  tshiozak 				/* incomplete input character */
    133  1.8  christos 				(void)memmove(inbuf, in, inbytes);
    134  1.8  christos 				ret = fread(inbuf + inbytes, 1,
    135  1.8  christos 				    INBUFSIZE - inbytes, fp);
    136  1.1  tshiozak 				if (ret == 0) {
    137  1.8  christos 					if (feof(fp)) {
    138  1.8  christos 						errno = EINVAL;
    139  1.8  christos 						err(EXIT_FAILURE, "iconv()");
    140  1.8  christos 					} else
    141  1.1  tshiozak 						err(EXIT_FAILURE, "fread()");
    142  1.1  tshiozak 				}
    143  1.1  tshiozak 				in = inbuf;
    144  1.1  tshiozak 				inbytes += ret;
    145  1.1  tshiozak 			}
    146  1.1  tshiozak 			if (outbytes < OUTBUFSIZE)
    147  1.8  christos 				(void)fwrite(outbuf, 1, OUTBUFSIZE - outbytes,
    148  1.8  christos 				    stdout);
    149  1.1  tshiozak 		}
    150  1.1  tshiozak 	}
    151  1.1  tshiozak 	/* reset the shift state of the output buffer */
    152  1.1  tshiozak 	outbytes = OUTBUFSIZE;
    153  1.1  tshiozak 	out = outbuf;
    154  1.1  tshiozak 	ret = iconv(cd, NULL, NULL, &out, &outbytes);
    155  1.1  tshiozak 	if (ret == -1)
    156  1.1  tshiozak 		err(EXIT_FAILURE, "iconv()");
    157  1.1  tshiozak 	if (outbytes < OUTBUFSIZE)
    158  1.8  christos 		(void)fwrite(outbuf, 1, OUTBUFSIZE - outbytes, stdout);
    159  1.1  tshiozak 
    160  1.1  tshiozak 	if (invalids > 0 && !silent)
    161  1.2  tshiozak 		warnx("warning: invalid characters: %lu",
    162  1.8  christos 		    (unsigned long)invalids);
    163  1.1  tshiozak 
    164  1.1  tshiozak 	iconv_close(cd);
    165  1.1  tshiozak }
    166  1.1  tshiozak 
    167  1.1  tshiozak int
    168  1.1  tshiozak main(int argc, char **argv)
    169  1.1  tshiozak {
    170  1.1  tshiozak 	int ch, i;
    171  1.1  tshiozak 	int opt_l = 0, opt_s = 0, opt_c = 0;
    172  1.1  tshiozak 	char *opt_f = NULL, *opt_t = NULL;
    173  1.1  tshiozak 	FILE *fp;
    174  1.1  tshiozak 
    175  1.8  christos 	setprogname(argv[0]);
    176  1.8  christos 
    177  1.8  christos 	while ((ch = getopt(argc, argv, "cslf:t:")) != EOF) {
    178  1.1  tshiozak 		switch (ch) {
    179  1.1  tshiozak 		case 'c':
    180  1.1  tshiozak 			opt_c = 1;
    181  1.1  tshiozak 			break;
    182  1.1  tshiozak 		case 's':
    183  1.1  tshiozak 			opt_s = 1;
    184  1.1  tshiozak 			break;
    185  1.1  tshiozak 		case 'l':
    186  1.1  tshiozak 			/* list */
    187  1.1  tshiozak 			opt_l = 1;
    188  1.1  tshiozak 			break;
    189  1.1  tshiozak 		case 'f':
    190  1.1  tshiozak 			/* from */
    191  1.8  christos 			opt_f = estrdup(optarg);
    192  1.1  tshiozak 			break;
    193  1.1  tshiozak 		case 't':
    194  1.1  tshiozak 			/* to */
    195  1.8  christos 			opt_t = estrdup(optarg);
    196  1.1  tshiozak 			break;
    197  1.1  tshiozak 		default:
    198  1.1  tshiozak 			usage();
    199  1.1  tshiozak 		}
    200  1.1  tshiozak 	}
    201  1.8  christos 	argc -= optind;
    202  1.8  christos 	argv += optind;
    203  1.1  tshiozak 	if (opt_l) {
    204  1.8  christos 		if (argc > 0 || opt_s || opt_f != NULL || opt_t != NULL) {
    205  1.8  christos 			warnx("-l is not allowed with other flags.");
    206  1.1  tshiozak 			usage();
    207  1.1  tshiozak 		}
    208  1.1  tshiozak 		show_codesets();
    209  1.1  tshiozak 	} else {
    210  1.1  tshiozak 		if (opt_f == NULL || opt_t == NULL)
    211  1.1  tshiozak 			usage();
    212  1.1  tshiozak 
    213  1.1  tshiozak 		if (argc == 0)
    214  1.1  tshiozak 			do_conv("<stdin>", stdin, opt_f, opt_t, opt_s, opt_c);
    215  1.1  tshiozak 		else {
    216  1.8  christos 			for (i = 0; i < argc; i++) {
    217  1.1  tshiozak 				fp = fopen(argv[i], "r");
    218  1.1  tshiozak 				if (fp == NULL)
    219  1.8  christos 					err(EXIT_FAILURE, "Cannot open `%s'",
    220  1.8  christos 					    argv[i]);
    221  1.1  tshiozak 				do_conv(argv[i], fp, opt_f, opt_t, opt_s,
    222  1.8  christos 				    opt_c);
    223  1.8  christos 				(void)fclose(fp);
    224  1.1  tshiozak 			}
    225  1.1  tshiozak 		}
    226  1.1  tshiozak 	}
    227  1.1  tshiozak 	return EXIT_SUCCESS;
    228  1.1  tshiozak }
    229