citrus_euc module accept wrong byte sequence.

[Takehiko NOZAKI]

citrus_euc module doesn't check second byte range, so illegal byte sequence (ex: "\xA4\x0") may wrongly converted to wchar_t(by mbrtowc) and other encoding(by iconv). this bug derived from 4.4BSD rune, it's 20 years old!

[Takehiko NOZAKI]

minimal test case:

#include <assert.h>
#include <errno.h>
#include <locale.h>
#include <stdio.h>
#include <string.h>
#include <wchar.h>

int
main(void)
{
        char *s = "\xa4";
        mbstate_t st;
        wchar_t wc;
        size_t ret;
        int e;
        setlocale(LC_CTYPE, "ja_JP.eucJP");
        memset(&st, 0, sizeof(st));
        ret = mbrtowc(&wc, s, 2, &st);
        e = errno;
        printf("%zd\n", ret);
        printf("%s\n", strerror(e));
        assert(ret == (size_t)-1 && e == EILSEQ);
}

[Takehiko NOZAKI]

BUGFIX: Issue #12 -- citrus_euc module accept wrong byte sequence.

→ <<cset 19be40a6c09e>>