lib/unicodeio.c

   1 /* Unicode character output to streams with locale dependent encoding.
   2
   3    Copyright (C) 2000-2003, 2006, 2008-2020 Free Software Foundation, Inc.
   4
   5    This program is free software: you can redistribute it and/or modify
   6    it under the terms of the GNU General Public License as published by
   7    the Free Software Foundation; either version 3 of the License, or
   8    (at your option) any later version.
   9
  10    This program is distributed in the hope that it will be useful,
  11    but WITHOUT ANY WARRANTY; without even the implied warranty of
  12    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13    GNU General Public License for more details.
  14
  15    You should have received a copy of the GNU General Public License
  16    along with this program.  If not, see <https://www.gnu.org/licenses/>.  */
  17
  18 /* Written by Bruno Haible <haible@clisp.cons.org>.  */
  19
  20 #include <config.h>
  21
  22 /* Specification.  */
  23 #include "unicodeio.h"
  24
  25 #include <stdio.h>
  26 #include <string.h>
  27 #include <errno.h>
  28
  29 #if HAVE_ICONV
  30 # include <iconv.h>
  31 #endif
  32
  33 #include <error.h>
  34
  35 #include "gettext.h"
  36 #define _(msgid) gettext (msgid)
  37 #define N_(msgid) msgid
  38
  39 #include "localcharset.h"
  40 #include "unistr.h"
  41
  42 /* When we pass a Unicode character to iconv(), we must pass it in a
  43    suitable encoding. The standardized Unicode encodings are
  44    UTF-8, UCS-2, UCS-4, UTF-16, UTF-16BE, UTF-16LE, UTF-7.
  45    UCS-2 supports only characters up to \U0000FFFF.
  46    UTF-16 and variants support only characters up to \U0010FFFF.
  47    UTF-7 is way too complex and not supported by glibc-2.1.
  48    UCS-4 specification leaves doubts about endianness and byte order
  49    mark. glibc currently interprets it as big endian without byte order
  50    mark, but this is not backed by an RFC.
  51    So we use UTF-8. It supports characters up to \U7FFFFFFF and is
  52    unambiguously defined.  */
  53
  54 /* Luckily, the encoding's name is platform independent.  */
  55 #define UTF8_NAME "UTF-8"
  56
  57 /* Converts the Unicode character CODE to its multibyte representation
  58    in the current locale and calls the SUCCESS callback on the resulting
  59    byte sequence.  If an error occurs, invokes the FAILURE callback instead,
  60    passing it CODE and an English error string.
  61    Returns whatever the callback returned.
  62    Assumes that the locale doesn't change between two calls.  */
  63 long
  64 unicode_to_mb (unsigned int code,
  65                long (*success) (const char *buf, size_t buflen,
  66                                 void *callback_arg),
  67                long (*failure) (unsigned int code, const char *msg,
  68                                 void *callback_arg),
  69                void *callback_arg)
  70 {
  71   static int initialized;
  72   static int is_utf8;
  73 #if HAVE_ICONV
  74   static iconv_t utf8_to_local;
  75 #endif
  76
  77   char inbuf[6];
  78   int count;
  79
  80   if (!initialized)
  81     {
  82       const char *charset = locale_charset ();
  83
  84       is_utf8 = !strcmp (charset, UTF8_NAME);
  85 #if HAVE_ICONV
  86       if (!is_utf8)
  87         {
  88           utf8_to_local = iconv_open (charset, UTF8_NAME);
  89           if (utf8_to_local == (iconv_t)(-1))
  90             /* For an unknown encoding, assume ASCII.  */
  91             utf8_to_local = iconv_open ("ASCII", UTF8_NAME);
  92         }
  93 #endif
  94       initialized = 1;
  95     }
  96
  97   /* Test whether the utf8_to_local converter is available at all.  */
  98   if (!is_utf8)
  99     {
 100 #if HAVE_ICONV
 101       if (utf8_to_local == (iconv_t)(-1))
 102         return failure (code, N_("iconv function not usable"), callback_arg);
 103 #else
 104       return failure (code, N_("iconv function not available"), callback_arg);
 105 #endif
 106     }
 107
 108   /* Convert the character to UTF-8.  */
 109   count = u8_uctomb ((unsigned char *) inbuf, code, sizeof (inbuf));
 110   if (count < 0)
 111     return failure (code, N_("character out of range"), callback_arg);
 112
 113 #if HAVE_ICONV
 114   if (!is_utf8)
 115     {
 116       char outbuf[25];
 117       const char *inptr;
 118       size_t inbytesleft;
 119       char *outptr;
 120       size_t outbytesleft;
 121       size_t res;
 122
 123       inptr = inbuf;
 124       inbytesleft = count;
 125       outptr = outbuf;
 126       outbytesleft = sizeof (outbuf);
 127
 128       /* Convert the character from UTF-8 to the locale's charset.  */
 129       res = iconv (utf8_to_local,
 130                    (ICONV_CONST char **)&inptr, &inbytesleft,
 131                    &outptr, &outbytesleft);
 132       if (inbytesleft > 0 || res == (size_t)(-1)
 133           /* Irix iconv() inserts a NUL byte if it cannot convert. */
 134 # if !defined _LIBICONV_VERSION && (defined sgi || defined __sgi)
 135           || (res > 0 && code != 0 && outptr - outbuf == 1 && *outbuf == '\0')
 136 # endif
 137          )
 138         return failure (code, NULL, callback_arg);
 139
 140       /* Avoid glibc-2.1 bug and Solaris 7 bug.  */
 141 # if defined _LIBICONV_VERSION \
 142     || !(((__GLIBC__ - 0 == 2 && __GLIBC_MINOR__ - 0 <= 1) \
 143           && !defined __UCLIBC__) \
 144          || defined __sun)
 145
 146       /* Get back to the initial shift state.  */
 147       res = iconv (utf8_to_local, NULL, NULL, &outptr, &outbytesleft);
 148       if (res == (size_t)(-1))
 149         return failure (code, NULL, callback_arg);
 150 # endif
 151
 152       return success (outbuf, outptr - outbuf, callback_arg);
 153     }
 154 #endif
 155
 156   /* At this point, is_utf8 is true, so no conversion is needed.  */
 157   return success (inbuf, count, callback_arg);
 158 }
 159
 160 /* Simple success callback that outputs the converted string.
 161    The STREAM is passed as callback_arg.  */
 162 long
 163 fwrite_success_callback (const char *buf, size_t buflen, void *callback_arg)
 164 {
 165   FILE *stream = (FILE *) callback_arg;
 166
 167   /* The return value of fwrite can be ignored here, because under normal
 168      conditions (STREAM is an open stream and not wide-character oriented)
 169      when fwrite() returns a value != buflen it also sets STREAM's error
 170      indicator.  */
 171   fwrite (buf, 1, buflen, stream);
 172   return 0;
 173 }
 174
 175 /* Simple failure callback that displays an error and exits.  */
 176 static long
 177 exit_failure_callback (unsigned int code, const char *msg,
 178                        void *callback_arg _GL_UNUSED)
 179 {
 180   if (msg == NULL)
 181     error (1, 0, _("cannot convert U+%04X to local character set"), code);
 182   else
 183     error (1, 0, _("cannot convert U+%04X to local character set: %s"), code,
 184            gettext (msg));
 185   return -1;
 186 }
 187
 188 /* Simple failure callback that displays a fallback representation in plain
 189    ASCII, using the same notation as ISO C99 strings.  */
 190 static long
 191 fallback_failure_callback (unsigned int code,
 192                            const char *msg _GL_UNUSED,
 193                            void *callback_arg)
 194 {
 195   FILE *stream = (FILE *) callback_arg;
 196
 197   if (code < 0x10000)
 198     fprintf (stream, "\\u%04X", code);
 199   else
 200     fprintf (stream, "\\U%08X", code);
 201   return -1;
 202 }
 203
 204 /* Outputs the Unicode character CODE to the output stream STREAM.
 205    Upon failure, exit if exit_on_error is true, otherwise output a fallback
 206    notation.  */
 207 void
 208 print_unicode_char (FILE *stream, unsigned int code, int exit_on_error)
 209 {
 210   unicode_to_mb (code, fwrite_success_callback,
 211                  exit_on_error
 212                  ? exit_failure_callback
 213                  : fallback_failure_callback,
 214                  stream);
 215 }