util/siconv.c

   1 /*
   2  * Copyright (c) 1997-2002, Index Data
   3  * See the file LICENSE for details.
   4  *
   5  * $Id: siconv.c,v 1.1 2002-08-27 14:02:13 adam Exp $
   6  */
   7
   8 #if HAVE_CONFIG_H
   9 #include <config.h>
  10 #endif
  11
  12 #include <errno.h>
  13 #include <string.h>
  14 #include <ctype.h>
  15
  16 #if HAVE_ICONV_H
  17 #include <iconv.h>
  18 #endif
  19
  20 #include <yaz/yaz-util.h>
  21
  22 struct yaz_iconv_struct {
  23     int my_errno;
  24     unsigned long (*read_handle)(yaz_iconv_t cd, char **inbuf,
  25                                  size_t *inbytesleft);
  26     size_t (*write_handle)(yaz_iconv_t cd, unsigned long x,
  27                            char **outbuf, size_t *outbytesleft);
  28 #if HAVE_ICONV_H
  29     iconv_t iconv_cd;
  30 #endif
  31 };
  32
  33
  34 static unsigned long yaz_read_ISO8859_1 (yaz_iconv_t cd,
  35                                          char **inbuf, size_t *inbytesleft)
  36 {
  37     unsigned char *inp = *inbuf;
  38     unsigned long x = 0;
  39     x = inp[0];
  40     (*inbytesleft)--;
  41     inp++;
  42     *inbuf = inp;
  43     return x;
  44 }
  45
  46 static unsigned long yaz_read_UTF8 (yaz_iconv_t cd,
  47                                     char **inbuf, size_t *inbytesleft)
  48 {
  49     unsigned char *inp = *inbuf;
  50     unsigned long x = 0;
  51     if (inp[0] <= 0x7f)
  52     {
  53         x = inp[0];
  54
  55         (*inbytesleft)--;
  56         inp++;
  57     }
  58     else if (inp[0] <= 0xdf && *inbytesleft >= 2)
  59     {
  60         x = ((inp[0] & 0x1f) << 6) + (inp[1] & 0x3f);
  61
  62         (*inbytesleft) -= 2;
  63         inp += 2;
  64     }
  65     else if (inp[0] <= 0xef && *inbytesleft >= 3)
  66     {
  67         x =  ((inp[0] & 0x0f) << 12) +
  68             ((inp[1] & 0x3f) << 6) +  (inp[1] & 0x3f);
  69
  70         (*inbytesleft) -= 3;
  71         inp += 3;
  72     }
  73     else if (inp[0] <= 0xef && *inbytesleft >= 4)
  74     {
  75         x =  ((inp[0] & 0x07) << 18) +
  76             ((inp[1] & 0x3f) << 12) + ((inp[2] & 0x3f) << 6) +
  77             (inp[3] & 0x3f);
  78
  79         (*inbytesleft) -= 4;
  80         inp += 4;
  81     }
  82     else
  83     {
  84         cd->my_errno = YAZ_ICONV_EINVAL;
  85     }
  86     *inbuf = inp;
  87     return x;
  88 }
  89
  90 static unsigned long yaz_read_UCS4 (yaz_iconv_t cd,
  91                                     char **inbuf, size_t *inbytesleft)
  92 {
  93     unsigned char *inp = *inbuf;
  94     unsigned long x = 0;
  95
  96     if (*inbytesleft < 4)
  97     {
  98         cd->my_errno = YAZ_ICONV_EINVAL; /* incomplete input */
  99         return 0;
 100     }
 101     memcpy (&x, inp, sizeof(x));
 102     (*inbytesleft) -= 4;
 103     inp += 4;
 104     *inbuf = inp;
 105     return x;
 106 }
 107
 108 static size_t yaz_write_UTF8 (yaz_iconv_t cd, unsigned long x,
 109                               char **outbuf, size_t *outbytesleft)
 110 {
 111     unsigned char *outp = *outbuf;
 112     if (x <= 0x7f && *outbytesleft >= 1)
 113     {
 114         *outp++ = x;
 115         (*outbytesleft)--;
 116     }
 117     else if (x <= 0x7ff && *outbytesleft >= 2)
 118     {
 119         *outp++ = (x >> 6) | 0xc0;
 120         *outp++ = (x & 0x3f) | 0x80;
 121         (*outbytesleft) -= 2;
 122     }
 123     else if (x <= 0xffff && *outbytesleft >= 3)
 124     {
 125         *outp++ = (x >> 12) | 0xe0;
 126         *outp++ = ((x >> 6) & 0x3f) | 0x80;
 127         *outp++ = (x & 0x3f) | 0x80;
 128         (*outbytesleft) -= 3;
 129     }
 130     else if (x <= 0x1fffff && *outbytesleft >= 4)
 131     {
 132         *outp++ = (x >> 18) | 0xf0;
 133         *outp++ = ((x >> 12) & 0x3f) | 0x80;
 134         *outp++ = ((x >> 6) & 0x3f) | 0x80;
 135         *outp++ = (x & 0x3f) | 0x80;
 136         (*outbytesleft) -= 4;
 137     }
 138     else if (x > 0x1fffff)
 139     {
 140         cd->my_errno = YAZ_ICONV_EILSEQ;  /* invalid sequence */
 141         return (size_t)(-1);
 142     }
 143     else
 144     {
 145         cd->my_errno = YAZ_ICONV_E2BIG;  /* not room for output */
 146         return (size_t)(-1);
 147     }
 148     *outbuf = outp;
 149     return 0;
 150 }
 151
 152 static size_t yaz_write_ISO8859_1 (yaz_iconv_t cd, unsigned long x,
 153                                    char **outbuf, size_t *outbytesleft)
 154 {
 155     unsigned char *outp = *outbuf;
 156     if (x > 255 || x < 1)
 157     {
 158         cd->my_errno = YAZ_ICONV_EILSEQ;
 159         return (size_t) -1;
 160     }
 161     else if (*outbytesleft >= 1)
 162     {
 163         *outp++ = x;
 164         (*outbytesleft)--;
 165     }
 166     else
 167     {
 168         cd->my_errno = YAZ_ICONV_E2BIG;
 169         return (size_t)(-1);
 170     }
 171     *outbuf = outp;
 172     return 0;
 173 }
 174
 175
 176 static size_t yaz_write_UCS4 (yaz_iconv_t cd, unsigned long x,
 177                               char **outbuf, size_t *outbytesleft)
 178 {
 179     unsigned char *outp = *outbuf;
 180     if (x < 1 || x > 0x1fffff)
 181     {
 182         cd->my_errno = YAZ_ICONV_EILSEQ;
 183         return (size_t)(-1);
 184     }
 185     else if (*outbytesleft >= 4)
 186     {
 187         memcpy (outp, &x, sizeof(x));
 188         outp += 4;
 189         (*outbytesleft) -= 4;
 190     }
 191     else
 192     {
 193         cd->my_errno = YAZ_ICONV_E2BIG;
 194         return (size_t)(-1);
 195     }
 196     *outbuf = outp;
 197     return 0;
 198 }
 199
 200 yaz_iconv_t yaz_iconv_open (const char *tocode, const char *fromcode)
 201 {
 202     yaz_iconv_t cd = xmalloc (sizeof(*cd));
 203
 204     cd->write_handle = 0;
 205     cd->read_handle = 0;
 206     cd->my_errno = YAZ_ICONV_UNKNOWN;
 207
 208     if (!strcmp(fromcode, "UTF-8"))
 209         cd->read_handle = yaz_read_UTF8;
 210     else if (!strcmp(fromcode, "ISO-8859-1"))
 211         cd->read_handle = yaz_read_ISO8859_1;
 212     else if (!strcmp(fromcode, "UCS-4"))
 213         cd->read_handle = yaz_read_UCS4;
 214
 215
 216     if (!strcmp(tocode, "UTF-8"))
 217         cd->write_handle = yaz_write_UTF8;
 218     else if (!strcmp (tocode, "ISO-8859-1"))
 219         cd->write_handle = yaz_write_ISO8859_1;
 220     else if (!strcmp (tocode, "UCS-4"))
 221         cd->write_handle = yaz_write_UCS4;
 222
 223 #if HAVE_ICONV_H
 224     cd->iconv_cd = 0;
 225     if (!cd->read_handle || !cd->write_handle)
 226     {
 227         cd->iconv_cd = iconv_open (tocode, fromcode);
 228         if (cd->iconv_cd == (iconv_t) (-1))
 229         {
 230             xfree (cd);
 231             return 0;
 232         }
 233     }
 234 #else
 235     if (!cd->to_UCS4 || !cd->from_UCS4)
 236     {
 237         xfree (cd);
 238         return 0;
 239     }
 240 #endif
 241     return cd;
 242 }
 243
 244 size_t yaz_iconv (yaz_iconv_t cd, char **inbuf, size_t *inbytesleft,
 245                   char **outbuf, size_t *outbytesleft)
 246 {
 247     char *inbuf0;
 248     size_t r = 0;
 249 #if HAVE_ICONV_H
 250     if (cd->iconv_cd)
 251     {
 252         size_t r =
 253             iconv(cd->iconv_cd, inbuf, inbytesleft, outbuf, outbytesleft);
 254         if (r == (size_t)(-1))
 255         {
 256             switch (errno)
 257             {
 258             case E2BIG:
 259                 cd->my_errno = YAZ_ICONV_E2BIG;
 260                 break;
 261             case EINVAL:
 262                 cd->my_errno = YAZ_ICONV_EINVAL;
 263                 break;
 264             case EILSEQ:
 265                 cd->my_errno = YAZ_ICONV_EILSEQ;
 266                 break;
 267             default:
 268                 cd->my_errno = YAZ_ICONV_UNKNOWN;
 269             }
 270         }
 271         return r;
 272     }
 273 #endif
 274     if (inbuf == 0 || *inbuf == 0)
 275         return 0;
 276     inbuf0 = *inbuf;
 277     while (1)
 278     {
 279         unsigned long x;
 280
 281         if (*inbytesleft == 0)
 282         {
 283             r = *inbuf - inbuf0;
 284             break;
 285         }
 286
 287         x = (cd->read_handle)(cd, inbuf, inbytesleft);
 288         if (x == 0)
 289         {
 290             r = (size_t)(-1);
 291             break;
 292         }
 293         r = (cd->write_handle)(cd, x, outbuf, outbytesleft);
 294         if (r)
 295             break;
 296     }
 297     return r;
 298 }
 299
 300 int yaz_iconv_error (yaz_iconv_t cd)
 301 {
 302     return cd->my_errno;
 303 }
 304
 305 int yaz_iconv_close (yaz_iconv_t cd)
 306 {
 307 #if HAVE_ICONV_H
 308     if (cd->iconv_cd)
 309         iconv_close (cd->iconv_cd);
 310 #endif
 311     xfree (cd);
 312     return 0;
 313 }
 314
 315