iconv.c revision 1.13.10.1       1  1.13.10.1       jym /*	$NetBSD: iconv.c,v 1.13.10.1 2009/05/13 19:19:52 jym 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  1.13.10.1       jym __RCSID("$NetBSD: iconv.c,v 1.13.10.1 2009/05/13 19:19:52 jym 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  1.13.10.1       jym 	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  1.13.10.1       jym 			if (outbytes < OUTBUFSIZE)
    120  1.13.10.1       jym 				(void)fwrite(outbuf, 1, OUTBUFSIZE - outbytes,
    121  1.13.10.1       jym 				    stdout);
    122        1.8  christos 			if (ret == (size_t)-1 && errno != E2BIG) {
    123        1.1  tshiozak 				/*
    124        1.1  tshiozak 				 * XXX: iconv(3) is bad interface.
    125        1.1  tshiozak 				 *   invalid character count is lost here.
    126        1.1  tshiozak 				 *   instead, we just provide __iconv function.
    127        1.1  tshiozak 				 */
    128        1.1  tshiozak 				if (errno != EINVAL || in == inbuf)
    129        1.1  tshiozak 					err(EXIT_FAILURE, "iconv()");
    130        1.1  tshiozak 
    131        1.1  tshiozak 				/* incomplete input character */
    132        1.8  christos 				(void)memmove(inbuf, in, inbytes);
    133        1.8  christos 				ret = fread(inbuf + inbytes, 1,
    134        1.8  christos 				    INBUFSIZE - inbytes, fp);
    135        1.1  tshiozak 				if (ret == 0) {
    136  1.13.10.1       jym 					fflush(stdout);
    137        1.9  tshiozak 					if (feof(fp))
    138        1.9  tshiozak 						errx(EXIT_FAILURE,
    139        1.9  tshiozak 						     "unexpected end of file; "
    140        1.9  tshiozak 						     "the last character is "
    141        1.9  tshiozak 						     "incomplete.");
    142        1.9  tshiozak 					else
    143        1.1  tshiozak 						err(EXIT_FAILURE, "fread()");
    144        1.1  tshiozak 				}
    145        1.1  tshiozak 				in = inbuf;
    146        1.1  tshiozak 				inbytes += ret;
    147        1.1  tshiozak 			}
    148        1.1  tshiozak 		}
    149        1.1  tshiozak 	}
    150        1.1  tshiozak 	/* reset the shift state of the output buffer */
    151        1.1  tshiozak 	outbytes = OUTBUFSIZE;
    152        1.1  tshiozak 	out = outbuf;
    153        1.1  tshiozak 	ret = iconv(cd, NULL, NULL, &out, &outbytes);
    154  1.13.10.1       jym 	if (ret == (size_t)-1)
    155        1.1  tshiozak 		err(EXIT_FAILURE, "iconv()");
    156        1.1  tshiozak 	if (outbytes < OUTBUFSIZE)
    157        1.8  christos 		(void)fwrite(outbuf, 1, OUTBUFSIZE - outbytes, stdout);
    158        1.1  tshiozak 
    159        1.1  tshiozak 	if (invalids > 0 && !silent)
    160        1.2  tshiozak 		warnx("warning: invalid characters: %lu",
    161        1.8  christos 		    (unsigned long)invalids);
    162        1.1  tshiozak 
    163        1.1  tshiozak 	iconv_close(cd);
    164        1.1  tshiozak }
    165        1.1  tshiozak 
    166        1.1  tshiozak int
    167        1.1  tshiozak main(int argc, char **argv)
    168        1.1  tshiozak {
    169        1.1  tshiozak 	int ch, i;
    170        1.1  tshiozak 	int opt_l = 0, opt_s = 0, opt_c = 0;
    171        1.1  tshiozak 	char *opt_f = NULL, *opt_t = NULL;
    172        1.1  tshiozak 	FILE *fp;
    173        1.1  tshiozak 
    174       1.13      yamt 	setlocale(LC_ALL, "");
    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       1.12   tnozaki 		if (opt_f == NULL) {
    211       1.12   tnozaki 			if (opt_t == NULL)
    212       1.12   tnozaki 				usage();
    213       1.12   tnozaki 			opt_f = nl_langinfo(CODESET);
    214       1.12   tnozaki 		} else if (opt_t == NULL)
    215       1.12   tnozaki 			opt_t = nl_langinfo(CODESET);
    216        1.1  tshiozak 
    217        1.1  tshiozak 		if (argc == 0)
    218        1.1  tshiozak 			do_conv("<stdin>", stdin, opt_f, opt_t, opt_s, opt_c);
    219        1.1  tshiozak 		else {
    220        1.8  christos 			for (i = 0; i < argc; i++) {
    221        1.1  tshiozak 				fp = fopen(argv[i], "r");
    222        1.1  tshiozak 				if (fp == NULL)
    223        1.8  christos 					err(EXIT_FAILURE, "Cannot open `%s'",
    224        1.8  christos 					    argv[i]);
    225        1.1  tshiozak 				do_conv(argv[i], fp, opt_f, opt_t, opt_s,
    226        1.8  christos 				    opt_c);
    227        1.8  christos 				(void)fclose(fp);
    228        1.1  tshiozak 			}
    229        1.1  tshiozak 		}
    230        1.1  tshiozak 	}
    231        1.1  tshiozak 	return EXIT_SUCCESS;
    232        1.1  tshiozak }
    233