diff options
Diffstat (limited to 'lib/unicodeio.c')
| -rw-r--r-- | lib/unicodeio.c | 129 |
1 files changed, 70 insertions, 59 deletions
diff --git a/lib/unicodeio.c b/lib/unicodeio.c index c753e3d509..4f14adf459 100644 --- a/lib/unicodeio.c +++ b/lib/unicodeio.c @@ -39,15 +39,13 @@ extern int errno; #endif -#if HAVE_LIMITS_H -# include <limits.h> +#if HAVE_WCHAR_H +# include <wchar.h> #endif -/* MB_LEN_MAX is incorrectly defined to be 1 in at least one GCC - installation; work around this configuration error. */ -#if MB_LEN_MAX < 6 -# undef MB_LEN_MAX -# define MB_LEN_MAX 6 +/* Some systems, like BeOS, have multibyte encodings but lack mbstate_t. */ +#if HAVE_WCRTOMB && defined mbstate_t +# define wcrtomb(s, wc, ps) (wcrtomb) (s, wc, 0) #endif #if HAVE_ICONV @@ -65,21 +63,35 @@ extern int errno; #include "unicodeio.h" -#if __STDC_ISO_10646__ && HAVE_WCTOMB +#if __STDC_ISO_10646__ && HAVE_WCRTOMB /* Values of type wchar_t are Unicode code points. */ -/* Place into BUF the locale-dependent representation of the character - CODE. Return the size of the result. If there is a conversion - error, return -1, setting errno appropriately. Assumes that the - locale doesn't change between two calls. */ -static size_t -convert_unicode_char (char buf[MB_LEN_MAX], unsigned int code) +/* Outputs the Unicode character CODE to the output stream STREAM. + Assumes that the locale doesn't change between two calls. */ +void +print_unicode_char (FILE *stream, unsigned int code) { - wchar_t wc = code; - errno = 0; - /* Test for truncation before invoking wctomb. */ - return wc == code ? wctomb (buf, wc) : -1; + wchar_t wc = (wchar_t) code; + + /* Test for truncation. */ + if (wc == code) + { + /* Convert from wide character to multibyte representation. */ + char buf[64]; /* Assume MB_LEN_MAX <= 64. */ + mbstate_t state; + size_t res; + + memset (&state, 0, sizeof (mbstate_t)); + res = wcrtomb (buf, wc, &state); + if (res != (size_t)(-1)) + fwrite (buf, 1, res, stream); + else + error (1, errno, + _("cannot convert U+%04X to local character set"), code); + } + else + error (1, 0, _("cannot convert U+%04X to local character set"), code); } #else @@ -135,12 +147,10 @@ utf8_wctomb (unsigned char *r, unsigned int wc) /* Luckily, the encoding's name is platform independent. */ # define UTF8_NAME "UTF-8" -/* Place into BUF the locale-dependent representation of the character - CODE. Return the size of the result. If there is a conversion - error, return -1, setting errno appropriately. Assumes that the - locale doesn't change between two calls. */ -static size_t -convert_unicode_char (char buf[MB_LEN_MAX], unsigned int code) +/* Outputs the Unicode character CODE to the output stream STREAM. + Assumes that the locale doesn't change between two calls. */ +void +print_unicode_char (FILE *stream, unsigned int code) { static int initialized; static int is_utf8; @@ -148,6 +158,9 @@ convert_unicode_char (char buf[MB_LEN_MAX], unsigned int code) static iconv_t utf8_to_local; # endif + char inbuf[6]; + int count; + if (!initialized) { extern const char *locale_charset PARAMS ((void)); @@ -159,16 +172,15 @@ convert_unicode_char (char buf[MB_LEN_MAX], unsigned int code) { utf8_to_local = (charset != NULL ? iconv_open (charset, UTF8_NAME) - : (iconv_t) -1); - if (utf8_to_local == (iconv_t) -1) + : (iconv_t)(-1)); + if (utf8_to_local == (iconv_t)(-1)) { /* For an unknown encoding, assume ASCII. */ utf8_to_local = iconv_open ("ASCII", UTF8_NAME); - if (utf8_to_local == (iconv_t) -1) - { - errno = ENOTSUP; - return -1; - } + if (utf8_to_local == (iconv_t)(-1)) + error (1, 0, + _("cannot output U+%04X: iconv function not usable"), + code); } } # endif @@ -176,58 +188,57 @@ convert_unicode_char (char buf[MB_LEN_MAX], unsigned int code) } /* Convert the character to UTF-8. */ + count = utf8_wctomb ((unsigned char *) inbuf, code); + if (count < 0) + error (1, 0, _("U+%04X: character out of range"), code); + if (is_utf8) - return utf8_wctomb ((unsigned char *) buf, code); + { + fwrite (inbuf, 1, count, stream); + } else { # if HAVE_ICONV - char inbuf[6]; - const char *inptr = inbuf; - size_t inbytesleft = utf8_wctomb ((unsigned char *) inbuf, code); - char *outptr = buf; - size_t outbytesleft = MB_LEN_MAX; + char outbuf[25]; + const char *inptr; + size_t inbytesleft; + char *outptr; + size_t outbytesleft; size_t res; - if (inbytesleft == (size_t) -1) - return -1; + inptr = inbuf; + inbytesleft = count; + outptr = outbuf; + outbytesleft = sizeof (outbuf); /* Convert the character from UTF-8 to the locale's charset. */ res = iconv (utf8_to_local, &inptr, &inbytesleft, &outptr, &outbytesleft); - if (inbytesleft > 0 || res == (size_t) -1 + if (inbytesleft > 0 || res == (size_t)(-1) /* Irix iconv() inserts a NUL byte if it cannot convert. */ # if !defined _LIBICONV_VERSION && (defined sgi || defined __sgi) || (res > 0 && code != 0 && outptr - outbuf == 1 && *outbuf == '\0') # endif - ) - return -1; + ) + error (1, res == (size_t)(-1) ? errno : 0, + _("cannot convert U+%04X to local character set"), code); /* Avoid glibc-2.1 bug and Solaris 2.7 bug. */ # if defined _LIBICONV_VERSION \ || !((__GLIBC__ - 0 == 2 && __GLIBC_MINOR__ - 0 <= 1) || defined __sun) /* Get back to the initial shift state. */ - return iconv (utf8_to_local, NULL, NULL, &outptr, &outbytesleft); + res = iconv (utf8_to_local, NULL, NULL, &outptr, &outbytesleft); + if (res == (size_t)(-1)) + error (1, errno, _("cannot convert U+%04X to local character set"), + code); # endif - return outptr - buf; + fwrite (outbuf, 1, outptr - outbuf, stream); # else - errno = ENOTSUP; - return -1; + error (1, 0, _("cannot output U+%04X: iconv function not available"), + code); # endif } } #endif - -/* Output the Unicode character CODE to the output stream STREAM. */ -void -print_unicode_char (FILE *stream, unsigned int code) -{ - char buf[MB_LEN_MAX]; - size_t s = convert_unicode_char (buf, code); - - if (s == (size_t) -1) - error (1, errno, _("cannot convert U+%04X to local character set"), code); - else - fwrite (buf, 1, s, stream); -} |