xref: /netbsd-src/usr.bin/iconv/iconv.c (revision 77a1ad5f0039ea2bc6af76846debc20728bae498)
1*77a1ad5fSkamil /*	$NetBSD: iconv.c,v 1.20 2019/10/24 18:18:00 kamil Exp $ */
2ff5c3e68Stshiozak 
3ff5c3e68Stshiozak /*-
4ff5c3e68Stshiozak  * Copyright (c)2003 Citrus Project,
5ff5c3e68Stshiozak  * All rights reserved.
6ff5c3e68Stshiozak  *
7ff5c3e68Stshiozak  * Redistribution and use in source and binary forms, with or without
8ff5c3e68Stshiozak  * modification, are permitted provided that the following conditions
9ff5c3e68Stshiozak  * are met:
10ff5c3e68Stshiozak  * 1. Redistributions of source code must retain the above copyright
11ff5c3e68Stshiozak  *    notice, this list of conditions and the following disclaimer.
12ff5c3e68Stshiozak  * 2. Redistributions in binary form must reproduce the above copyright
13ff5c3e68Stshiozak  *    notice, this list of conditions and the following disclaimer in the
14ff5c3e68Stshiozak  *    documentation and/or other materials provided with the distribution.
15ff5c3e68Stshiozak  *
16ff5c3e68Stshiozak  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
17ff5c3e68Stshiozak  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
18ff5c3e68Stshiozak  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
19ff5c3e68Stshiozak  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
20ff5c3e68Stshiozak  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
21ff5c3e68Stshiozak  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
22ff5c3e68Stshiozak  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
23ff5c3e68Stshiozak  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
24ff5c3e68Stshiozak  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
25ff5c3e68Stshiozak  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
26ff5c3e68Stshiozak  * SUCH DAMAGE.
27ff5c3e68Stshiozak  */
28ff5c3e68Stshiozak 
29ff5c3e68Stshiozak #include <sys/cdefs.h>
30ff5c3e68Stshiozak #if defined(LIBC_SCCS) && !defined(lint)
31*77a1ad5fSkamil __RCSID("$NetBSD: iconv.c,v 1.20 2019/10/24 18:18:00 kamil Exp $");
32ff5c3e68Stshiozak #endif /* LIBC_SCCS and not lint */
33ff5c3e68Stshiozak 
340b9f4942Stnozaki #include <err.h>
35ff5c3e68Stshiozak #include <errno.h>
360b9f4942Stnozaki #include <iconv.h>
370b9f4942Stnozaki #include <langinfo.h>
38d92af407Syamt #include <locale.h>
39ff5c3e68Stshiozak #include <stdio.h>
40ff5c3e68Stshiozak #include <stdlib.h>
41ff5c3e68Stshiozak #include <string.h>
42ff5c3e68Stshiozak #include <unistd.h>
43cdab3a7aSchristos #include <util.h>
44ff5c3e68Stshiozak 
456818646aSjoerg static void usage(void) __dead;
46557621f7Schristos static int scmp(const void *, const void *);
47557621f7Schristos static void show_codesets(void);
48557621f7Schristos static void do_conv(const char *, FILE *, const char *, const char *, int, int);
49557621f7Schristos 
50ff5c3e68Stshiozak static void
usage(void)51ff5c3e68Stshiozak usage(void)
52ff5c3e68Stshiozak {
530b9f4942Stnozaki 	(void)fprintf(stderr,
540b9f4942Stnozaki 	    "Usage:\t%1$s [-cs] -f <from_code> -t <to_code> [file ...]\n"
550b9f4942Stnozaki 	    "\t%1$s -f <from_code> [-cs] [-t <to_code>] [file ...]\n"
560b9f4942Stnozaki 	    "\t%1$s -t <to_code> [-cs] [-f <from_code>] [file ...]\n"
570b9f4942Stnozaki 	    "\t%1$s -l\n", getprogname());
58ff5c3e68Stshiozak 	exit(1);
59ff5c3e68Stshiozak }
60ff5c3e68Stshiozak 
61b1427588Sjdolecek /*
62b1427588Sjdolecek  * qsort() helper function
63b1427588Sjdolecek  */
64b1427588Sjdolecek static int
scmp(const void * v1,const void * v2)658676c37cSyamt scmp(const void *v1, const void *v2)
66b1427588Sjdolecek {
678676c37cSyamt 	const char * const *s1 = v1;
688676c37cSyamt 	const char * const *s2 = v2;
698676c37cSyamt 
708676c37cSyamt 	return strcasecmp(*s1, *s2);
71b1427588Sjdolecek }
72b1427588Sjdolecek 
73ff5c3e68Stshiozak static void
show_codesets(void)74ff5c3e68Stshiozak show_codesets(void)
75ff5c3e68Stshiozak {
76ff5c3e68Stshiozak 	char **list;
77ff5c3e68Stshiozak 	size_t sz, i;
78ff5c3e68Stshiozak 
79ff5c3e68Stshiozak 	if (__iconv_get_list(&list, &sz))
80ff5c3e68Stshiozak 		err(EXIT_FAILURE, "__iconv_get_list()");
81ff5c3e68Stshiozak 
82b1427588Sjdolecek 	qsort(list, sz, sizeof(char *), scmp);
83b1427588Sjdolecek 
84557621f7Schristos 	for (i = 0; i < sz; i++)
85557621f7Schristos 		(void)printf("%s\n", list[i]);
86ff5c3e68Stshiozak 
87ff5c3e68Stshiozak 	__iconv_free_list(list, sz);
88ff5c3e68Stshiozak }
89ff5c3e68Stshiozak 
90ff5c3e68Stshiozak #define INBUFSIZE 1024
91ff5c3e68Stshiozak #define OUTBUFSIZE (INBUFSIZE * 2)
9229af9462Syamt /*ARGSUSED*/
93ff5c3e68Stshiozak static void
do_conv(const char * fn,FILE * fp,const char * from,const char * to,int silent,int hide_invalid)94ff5c3e68Stshiozak do_conv(const char *fn, FILE *fp, const char *from, const char *to, int silent,
95ff5c3e68Stshiozak     int hide_invalid)
96ff5c3e68Stshiozak {
97ff5c3e68Stshiozak 	char inbuf[INBUFSIZE], outbuf[OUTBUFSIZE], *out;
98ff5c3e68Stshiozak 	const char *in;
99ff5c3e68Stshiozak 	size_t inbytes, outbytes, ret, invalids;
100ff5c3e68Stshiozak 	iconv_t cd;
101288c1350Syamt 	uint32_t flags = 0;
10204fb4300Sdholland 	int serrno;
103ff5c3e68Stshiozak 
104ff5c3e68Stshiozak 	if (hide_invalid)
105ff5c3e68Stshiozak 		flags |= __ICONV_F_HIDE_INVALID;
106ff5c3e68Stshiozak 	cd = iconv_open(to, from);
107ff5c3e68Stshiozak 	if (cd == (iconv_t)-1)
1087d42d8e7Swiz 		err(EXIT_FAILURE, "iconv_open(%s, %s)", to, from);
109ff5c3e68Stshiozak 
110ff5c3e68Stshiozak 	invalids = 0;
111ff5c3e68Stshiozak 	while ((inbytes = fread(inbuf, 1, INBUFSIZE, fp)) > 0) {
112ff5c3e68Stshiozak 		in = inbuf;
113ff5c3e68Stshiozak 		while (inbytes > 0) {
114c046e5f6Syamt 			size_t inval;
115c046e5f6Syamt 
116ff5c3e68Stshiozak 			out = outbuf;
117ff5c3e68Stshiozak 			outbytes = OUTBUFSIZE;
118*77a1ad5fSkamil 			ret = __iconv(cd, __UNCONST(&in), &inbytes, &out,
119*77a1ad5fSkamil 			    &outbytes, flags, &inval);
12004fb4300Sdholland 			serrno = errno;
121c046e5f6Syamt 			invalids += inval;
12299b23c11Syamt 			if (outbytes < OUTBUFSIZE)
12399b23c11Syamt 				(void)fwrite(outbuf, 1, OUTBUFSIZE - outbytes,
12499b23c11Syamt 				    stdout);
12504fb4300Sdholland 			errno = serrno;
126ff5c3e68Stshiozak 			if (ret == (size_t)-1 && errno != E2BIG) {
127ff5c3e68Stshiozak 				/*
128ff5c3e68Stshiozak 				 * XXX: iconv(3) is bad interface.
129ff5c3e68Stshiozak 				 *   invalid character count is lost here.
130ff5c3e68Stshiozak 				 *   instead, we just provide __iconv function.
131ff5c3e68Stshiozak 				 */
132ff5c3e68Stshiozak 				if (errno != EINVAL || in == inbuf)
133ff5c3e68Stshiozak 					err(EXIT_FAILURE, "iconv()");
134ff5c3e68Stshiozak 
135ff5c3e68Stshiozak 				/* incomplete input character */
136557621f7Schristos 				(void)memmove(inbuf, in, inbytes);
137ff5c3e68Stshiozak 				ret = fread(inbuf + inbytes, 1,
138ff5c3e68Stshiozak 				    INBUFSIZE - inbytes, fp);
139ff5c3e68Stshiozak 				if (ret == 0) {
14099b23c11Syamt 					fflush(stdout);
1411cf148a9Stshiozak 					if (feof(fp))
1421cf148a9Stshiozak 						errx(EXIT_FAILURE,
1431cf148a9Stshiozak 						     "unexpected end of file; "
1441cf148a9Stshiozak 						     "the last character is "
1451cf148a9Stshiozak 						     "incomplete.");
1461cf148a9Stshiozak 					else
147ff5c3e68Stshiozak 						err(EXIT_FAILURE, "fread()");
148ff5c3e68Stshiozak 				}
149ff5c3e68Stshiozak 				in = inbuf;
150ff5c3e68Stshiozak 				inbytes += ret;
151ff5c3e68Stshiozak 			}
152ff5c3e68Stshiozak 		}
153ff5c3e68Stshiozak 	}
154ff5c3e68Stshiozak 	/* reset the shift state of the output buffer */
155ff5c3e68Stshiozak 	outbytes = OUTBUFSIZE;
156ff5c3e68Stshiozak 	out = outbuf;
157ff5c3e68Stshiozak 	ret = iconv(cd, NULL, NULL, &out, &outbytes);
1587abbdb7aSyamt 	if (ret == (size_t)-1)
159ff5c3e68Stshiozak 		err(EXIT_FAILURE, "iconv()");
160ff5c3e68Stshiozak 	if (outbytes < OUTBUFSIZE)
161557621f7Schristos 		(void)fwrite(outbuf, 1, OUTBUFSIZE - outbytes, stdout);
162ff5c3e68Stshiozak 
163ff5c3e68Stshiozak 	if (invalids > 0 && !silent)
164f0108a13Stshiozak 		warnx("warning: invalid characters: %lu",
165f0108a13Stshiozak 		    (unsigned long)invalids);
166ff5c3e68Stshiozak 
167ff5c3e68Stshiozak 	iconv_close(cd);
168ff5c3e68Stshiozak }
169ff5c3e68Stshiozak 
170ff5c3e68Stshiozak int
main(int argc,char ** argv)171ff5c3e68Stshiozak main(int argc, char **argv)
172ff5c3e68Stshiozak {
173ff5c3e68Stshiozak 	int ch, i;
174ff5c3e68Stshiozak 	int opt_l = 0, opt_s = 0, opt_c = 0;
175ff5c3e68Stshiozak 	char *opt_f = NULL, *opt_t = NULL;
176ff5c3e68Stshiozak 	FILE *fp;
177ff5c3e68Stshiozak 
178d92af407Syamt 	setlocale(LC_ALL, "");
179557621f7Schristos 	setprogname(argv[0]);
180557621f7Schristos 
181ff5c3e68Stshiozak 	while ((ch = getopt(argc, argv, "cslf:t:")) != EOF) {
182ff5c3e68Stshiozak 		switch (ch) {
183ff5c3e68Stshiozak 		case 'c':
184ff5c3e68Stshiozak 			opt_c = 1;
185ff5c3e68Stshiozak 			break;
186ff5c3e68Stshiozak 		case 's':
187ff5c3e68Stshiozak 			opt_s = 1;
188ff5c3e68Stshiozak 			break;
189ff5c3e68Stshiozak 		case 'l':
190ff5c3e68Stshiozak 			/* list */
191ff5c3e68Stshiozak 			opt_l = 1;
192ff5c3e68Stshiozak 			break;
193ff5c3e68Stshiozak 		case 'f':
194ff5c3e68Stshiozak 			/* from */
195557621f7Schristos 			opt_f = estrdup(optarg);
196ff5c3e68Stshiozak 			break;
197ff5c3e68Stshiozak 		case 't':
198ff5c3e68Stshiozak 			/* to */
199557621f7Schristos 			opt_t = estrdup(optarg);
200ff5c3e68Stshiozak 			break;
201ff5c3e68Stshiozak 		default:
202ff5c3e68Stshiozak 			usage();
203ff5c3e68Stshiozak 		}
204ff5c3e68Stshiozak 	}
205ff5c3e68Stshiozak 	argc -= optind;
206ff5c3e68Stshiozak 	argv += optind;
207ff5c3e68Stshiozak 	if (opt_l) {
208ff5c3e68Stshiozak 		if (argc > 0 || opt_s || opt_f != NULL || opt_t != NULL) {
209557621f7Schristos 			warnx("-l is not allowed with other flags.");
210ff5c3e68Stshiozak 			usage();
211ff5c3e68Stshiozak 		}
212ff5c3e68Stshiozak 		show_codesets();
213ff5c3e68Stshiozak 	} else {
2140b9f4942Stnozaki 		if (opt_f == NULL) {
2150b9f4942Stnozaki 			if (opt_t == NULL)
216ff5c3e68Stshiozak 				usage();
2170b9f4942Stnozaki 			opt_f = nl_langinfo(CODESET);
2180b9f4942Stnozaki 		} else if (opt_t == NULL)
2190b9f4942Stnozaki 			opt_t = nl_langinfo(CODESET);
220ff5c3e68Stshiozak 
221ff5c3e68Stshiozak 		if (argc == 0)
222ff5c3e68Stshiozak 			do_conv("<stdin>", stdin, opt_f, opt_t, opt_s, opt_c);
223ff5c3e68Stshiozak 		else {
224ff5c3e68Stshiozak 			for (i = 0; i < argc; i++) {
225ff5c3e68Stshiozak 				fp = fopen(argv[i], "r");
226ff5c3e68Stshiozak 				if (fp == NULL)
227557621f7Schristos 					err(EXIT_FAILURE, "Cannot open `%s'",
228557621f7Schristos 					    argv[i]);
229ff5c3e68Stshiozak 				do_conv(argv[i], fp, opt_f, opt_t, opt_s,
230ff5c3e68Stshiozak 				    opt_c);
231557621f7Schristos 				(void)fclose(fp);
232ff5c3e68Stshiozak 			}
233ff5c3e68Stshiozak 		}
234ff5c3e68Stshiozak 	}
235ff5c3e68Stshiozak 	return EXIT_SUCCESS;
236ff5c3e68Stshiozak }
237