|
 |
3a5018 |
/*
|
|
|
3a5018 |
* iconv.c
|
|
|
3a5018 |
* Implementation of SUSv4 XCU iconv utility
|
|
|
3a5018 |
* Copyright © 2011 Rich Felker
|
|
|
3a5018 |
* Licensed under the terms of the GNU General Public License, v2 or later
|
|
|
3a5018 |
*/
|
|
|
3a5018 |
|
|
|
3a5018 |
#include <stdlib.h>
|
|
|
3a5018 |
#include <stdio.h>
|
|
|
3a5018 |
#include <iconv.h>
|
|
|
3a5018 |
#include <locale.h>
|
|
|
3a5018 |
#include <langinfo.h>
|
|
|
3a5018 |
#include <unistd.h>
|
|
|
3a5018 |
#include <errno.h>
|
|
|
3a5018 |
#include <string.h>
|
|
|
3a5018 |
|
|
|
3a5018 |
int main(int argc, char **argv)
|
|
|
3a5018 |
{
|
|
|
3a5018 |
const char *from=0, *to=0;
|
|
|
3a5018 |
int b;
|
|
|
3a5018 |
iconv_t cd;
|
|
|
3a5018 |
char buf[BUFSIZ];
|
|
|
3a5018 |
char outbuf[BUFSIZ*4];
|
|
|
3a5018 |
char *in, *out;
|
|
|
3a5018 |
size_t inb;
|
|
|
3a5018 |
size_t l;
|
|
|
3a5018 |
size_t unitsize=0;
|
|
|
3a5018 |
int err=0;
|
|
|
3a5018 |
FILE *f;
|
|
|
3a5018 |
|
|
|
3a5018 |
while ((b = getopt(argc, argv, "f:t:csl")) != EOF) switch(b) {
|
|
|
3a5018 |
case 'l':
|
|
|
3a5018 |
puts("UTF-8, UTF-16BE, UTF-16LE, UTF-32BE, UTF32-LE, UCS-2BE, UCS-2LE, WCHAR_T,\n"
|
|
|
3a5018 |
"US_ASCII, ISO8859-1, ISO8859-2, ISO8859-3, ISO8859-4, ISO8859-5,\n"
|
|
|
3a5018 |
"ISO8859-6, ISO8859-7, ...");
|
|
|
3a5018 |
exit(0);
|
|
|
3a5018 |
case 'c': case 's': break;
|
|
|
3a5018 |
case 'f': from=optarg; break;
|
|
|
3a5018 |
case 't': to=optarg; break;
|
|
|
3a5018 |
default: exit(1);
|
|
|
3a5018 |
}
|
|
|
3a5018 |
|
|
|
3a5018 |
if (!from || !to) {
|
|
|
3a5018 |
setlocale(LC_CTYPE, "");
|
|
|
3a5018 |
if (!to) to = nl_langinfo(CODESET);
|
|
|
3a5018 |
if (!from) from = nl_langinfo(CODESET);
|
|
|
3a5018 |
}
|
|
|
3a5018 |
cd = iconv_open(to, from);
|
|
|
3a5018 |
if (cd == (iconv_t)-1) {
|
|
|
3a5018 |
if (iconv_open(to, "WCHAR_T") == (iconv_t)-1)
|
|
|
3a5018 |
fprintf(stderr, "iconv: destination charset %s: ", to);
|
|
|
3a5018 |
else
|
|
|
3a5018 |
fprintf(stderr, "iconv: source charset %s: ", from);
|
|
|
3a5018 |
perror("");
|
|
|
3a5018 |
exit(1);
|
|
|
3a5018 |
}
|
|
|
3a5018 |
if (optind == argc) argv[argc++] = "-";
|
|
|
3a5018 |
|
|
|
3a5018 |
for (; optind < argc; optind++) {
|
|
|
3a5018 |
if (argv[optind][0]=='-' && !argv[optind][1]) {
|
|
|
3a5018 |
f = stdin;
|
|
|
3a5018 |
argv[optind] = "(stdin)";
|
|
|
3a5018 |
} else if (!(f = fopen(argv[optind], "rb"))) {
|
|
|
3a5018 |
fprintf(stderr, "iconv: %s: ", argv[optind]);
|
|
|
3a5018 |
perror("");
|
|
|
3a5018 |
err = 1;
|
|
|
3a5018 |
continue;
|
|
|
3a5018 |
}
|
|
|
3a5018 |
inb = 0;
|
|
|
3a5018 |
for (;;) {
|
|
|
3a5018 |
in = buf;
|
|
|
3a5018 |
out = outbuf;
|
|
|
3a5018 |
l = fread(buf+inb, 1, sizeof(buf)-inb, f);
|
|
|
3a5018 |
inb += l;
|
|
|
3a5018 |
if (!inb) break;
|
|
|
3a5018 |
if (iconv(cd, &in, &inb, &out, (size_t [1]){sizeof outbuf})==-1
|
|
|
3a5018 |
&& errno == EILSEQ) {
|
|
|
3a5018 |
if (!unitsize) {
|
|
|
3a5018 |
wchar_t wc='0';
|
|
|
3a5018 |
char dummy[4], *dummyp=dummy;
|
|
|
3a5018 |
iconv_t cd2 = iconv_open(from, "WCHAR_T");
|
|
|
3a5018 |
if (cd == (iconv_t)-1) {
|
|
|
3a5018 |
unitsize = 1;
|
|
|
3a5018 |
} else {
|
|
|
3a5018 |
iconv(cd2,
|
|
|
3a5018 |
(char *[1]){(char *)&wc},
|
|
|
3a5018 |
(size_t[1]){1},
|
|
|
3a5018 |
&dummyp, (size_t[1]){4});
|
|
|
3a5018 |
unitsize = dummyp-dummy;
|
|
|
3a5018 |
if (!unitsize) unitsize=1;
|
|
|
3a5018 |
}
|
|
|
3a5018 |
}
|
|
|
3a5018 |
inb-=unitsize;
|
|
|
3a5018 |
in+=unitsize;
|
|
|
3a5018 |
}
|
|
|
3a5018 |
if (inb && !l && errno==EINVAL) break;
|
|
|
3a5018 |
if (out>outbuf && !fwrite(outbuf, out-outbuf, 1, stdout)) {
|
|
|
3a5018 |
perror("iconv: write error");
|
|
|
3a5018 |
exit(1);
|
|
|
3a5018 |
}
|
|
|
3a5018 |
if (inb) memmove(buf, in, inb);
|
|
|
3a5018 |
}
|
|
|
3a5018 |
if (ferror(f)) {
|
|
|
3a5018 |
fprintf(stderr, "iconv: %s: ", argv[optind]);
|
|
|
3a5018 |
perror("");
|
|
|
3a5018 |
err = 1;
|
|
|
3a5018 |
}
|
|
|
3a5018 |
}
|
|
|
3a5018 |
return err;
|
|
|
3a5018 |
}
|