diplodocus.org Git - nmh/blob - sbr/fmt_rfc2047.c

   1
   2 /*
   3  * fmt_rfc2047.c -- decode RFC-2047 header format
   4  *
   5  * This code is Copyright (c) 2002, by the authors of nmh.  See the
   6  * COPYRIGHT file in the root directory of the nmh distribution for
   7  * complete copyright information.
   8  */
   9
  10 #include <h/mh.h>
  11 #include <h/utils.h>
  12 #ifdef HAVE_ICONV
  13 #  include <iconv.h>
  14 #endif
  15
  16 static signed char hexindex[] = {
  17     -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
  18     -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
  19     -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
  20      0, 1, 2, 3, 4, 5, 6, 7, 8, 9,-1,-1,-1,-1,-1,-1,
  21     -1,10,11,12,13,14,15,-1,-1,-1,-1,-1,-1,-1,-1,-1,
  22     -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,
  23     -1,10,11,12,13,14,15,-1,-1,-1,-1,-1,-1,-1,-1,-1,
  24     -1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1,-1
  25 };
  26
  27 static signed char index_64[128] = {
  28     -1,-1,-1,-1, -1,-1,-1,-1, -1,-1,-1,-1, -1,-1,-1,-1,
  29     -1,-1,-1,-1, -1,-1,-1,-1, -1,-1,-1,-1, -1,-1,-1,-1,
  30     -1,-1,-1,-1, -1,-1,-1,-1, -1,-1,-1,62, -1,-1,-1,63,
  31     52,53,54,55, 56,57,58,59, 60,61,-1,-1, -1,-1,-1,-1,
  32     -1, 0, 1, 2,  3, 4, 5, 6,  7, 8, 9,10, 11,12,13,14,
  33     15,16,17,18, 19,20,21,22, 23,24,25,-1, -1,-1,-1,-1,
  34     -1,26,27,28, 29,30,31,32, 33,34,35,36, 37,38,39,40,
  35     41,42,43,44, 45,46,47,48, 49,50,51,-1, -1,-1,-1,-1
  36 };
  37
  38 #define char64(c) (((unsigned char) (c) > 127) ? -1 : index_64[(unsigned char) (c)])
  39
  40 static int
  41 unqp (unsigned char byte1, unsigned char byte2)
  42 {
  43     if (hexindex[byte1] == -1 || hexindex[byte2] == -1)
  44         return -1;
  45     return (hexindex[byte1] << 4 | hexindex[byte2]);
  46 }
  47
  48 /* Check if character is linear whitespace */
  49 #define is_lws(c)  ((c) == ' ' || (c) == '\t' || (c) == '\n')
  50
  51
  52 /*
  53  * Decode the string as a RFC-2047 header field
  54  */
  55
  56 /* Add character to the destination buffer, and bomb out if it fills up */
  57 #define ADDCHR(C) do { *q++ = (C); dstlen--; if (!dstlen) goto buffull; } while (0)
  58
  59 int
  60 decode_rfc2047 (char *str, char *dst, size_t dstlen)
  61 {
  62     char *p, *q, *pp;
  63     char *startofmime, *endofmime, *endofcharset;
  64     int c, quoted_printable;
  65     int encoding_found = 0;     /* did we decode anything?                */
  66     int between_encodings = 0;  /* are we between two encodings?          */
  67     int equals_pending = 0;     /* is there a '=' pending?                */
  68     int whitespace = 0;         /* how much whitespace between encodings? */
  69 #ifdef HAVE_ICONV
  70     int use_iconv = 0;          /* are we converting encoding with iconv? */
  71     iconv_t cd = NULL;
  72     int fromutf8 = 0;
  73     char *saveq, *convbuf = NULL;
  74     size_t savedstlen;
  75 #endif
  76
  77     if (!str)
  78         return 0;
  79
  80     /*
  81      * Do a quick and dirty check for the '=' character.
  82      * This should quickly eliminate many cases.
  83      */
  84     if (!strchr (str, '='))
  85         return 0;
  86
  87     for (p = str, q = dst; *p; p++) {
  88
  89         /* reset iconv */
  90 #ifdef HAVE_ICONV
  91         if (use_iconv) {
  92             iconv_close(cd);
  93             use_iconv = 0;
  94         }
  95 #endif
  96         /*
  97          * If we had an '=' character pending from
  98          * last iteration, then add it first.
  99          */
 100         if (equals_pending) {
 101             ADDCHR('=');
 102             equals_pending = 0;
 103             between_encodings = 0;      /* we have added non-whitespace text */
 104         }
 105
 106         if (*p != '=') {
 107             /* count linear whitespace while between encodings */
 108             if (between_encodings && is_lws(*p))
 109                 whitespace++;
 110             else
 111                 between_encodings = 0;  /* we have added non-whitespace text */
 112             ADDCHR(*p);
 113             continue;
 114         }
 115
 116         equals_pending = 1;     /* we have a '=' pending */
 117
 118         /* Check for initial =? */
 119         if (*p == '=' && p[1] && p[1] == '?' && p[2]) {
 120             startofmime = p + 2;
 121
 122             /* Scan ahead for the next '?' character */
 123             for (pp = startofmime; *pp && *pp != '?'; pp++)
 124                 ;
 125
 126             if (!*pp)
 127                 continue;
 128
 129             /*
 130              * RFC 2231 specifies that language information can appear
 131              * in a charset specification like so:
 132              *
 133              * =?us-ascii*en?Q?Foo?=
 134              *
 135              * Right now we don't use language information, so ignore it.
 136              */
 137
 138             for (endofcharset = startofmime;
 139                         *endofcharset != '*' && endofcharset < pp;
 140                                                         endofcharset++)
 141                 ;
 142
 143             /* Check if character set can be handled natively */
 144             if (!check_charset(startofmime, endofcharset - startofmime)) {
 145 #ifdef HAVE_ICONV
 146                 /* .. it can't. We'll use iconv then. */
 147                 *endofcharset = '\0';
 148                 cd = iconv_open(get_charset(), startofmime);
 149                 fromutf8 = !strcasecmp(startofmime, "UTF-8");
 150                 *pp = '?';
 151                 if (cd == (iconv_t)-1) continue;
 152                 use_iconv = 1;
 153 #else
 154                 continue;
 155 #endif
 156             }
 157
 158             startofmime = pp + 1;
 159
 160             /* Check for valid encoding type */
 161             if (*startofmime != 'B' && *startofmime != 'b' &&
 162                 *startofmime != 'Q' && *startofmime != 'q')
 163                 continue;
 164
 165             /* Is encoding quoted printable or base64? */
 166             quoted_printable = (*startofmime == 'Q' || *startofmime == 'q');
 167             startofmime++;
 168
 169             /* Check for next '?' character */
 170             if (*startofmime != '?')
 171                 continue;
 172             startofmime++;
 173
 174             /*
 175              * Scan ahead for the ending ?=
 176              *
 177              * While doing this, we will also check if encoded
 178              * word has any embedded linear whitespace.
 179              */
 180             endofmime = NULL;
 181             for (pp = startofmime; *pp && *(pp+1); pp++) {
 182                 if (is_lws(*pp)) {
 183                     break;
 184                 } else if (*pp == '?' && pp[1] == '=') {
 185                     endofmime = pp;
 186                     break;
 187                 }
 188             }
 189             if (is_lws(*pp) || endofmime == NULL)
 190                 continue;
 191
 192             /*
 193              * We've found an encoded word, so we can drop
 194              * the '=' that was pending
 195              */
 196             equals_pending = 0;
 197
 198             /*
 199              * If we are between two encoded words separated only by
 200              * linear whitespace, then we ignore the whitespace.
 201              * We will roll back the buffer the number of whitespace
 202              * characters we've seen since last encoded word.
 203              */
 204             if (between_encodings) {
 205                 q -= whitespace;
 206                 dstlen += whitespace;
 207             }
 208
 209 #ifdef HAVE_ICONV
 210             /*
 211              * empty encoded text. This ensures that we don't
 212              * malloc 0 bytes but skip on to the end
 213              */
 214             if (endofmime == startofmime && use_iconv) {
 215                 use_iconv = 0;
 216                 iconv_close(cd);
 217             }
 218
 219             if (use_iconv) {
 220                 saveq = q;
 221                 savedstlen = dstlen;
 222                 q = convbuf = (char *) mh_xmalloc(endofmime - startofmime);
 223             }
 224 /* ADDCHR2 is for adding characters when q is or might be convbuf:
 225  * in this case on buffer-full we want to run iconv before returning.
 226  * I apologise for the dreadful name.
 227  */
 228 #define ADDCHR2(C) do { *q++ = (C); dstlen--; if (!dstlen) goto iconvbuffull; } while (0)
 229 #else
 230 #define ADDCHR2(C) ADDCHR(C)
 231 #endif
 232
 233             /* Now decode the text */
 234             if (quoted_printable) {
 235                 for (pp = startofmime; pp < endofmime; pp++) {
 236                     if (*pp == '=') {
 237                         c = unqp (pp[1], pp[2]);
 238                         if (c == -1)
 239                             continue;
 240                         if (c != 0)
 241                             *q++ = c;
 242                         pp += 2;
 243                     } else if (*pp == '_') {
 244                         ADDCHR2(' ');
 245                     } else {
 246                         ADDCHR2(*pp);
 247                     }
 248                 }
 249             } else {
 250                 /* base64 */
 251                 int c1, c2, c3, c4;
 252                 c1 = c2 = c3 = c4 = -1;
 253
 254                 pp = startofmime;
 255                 while (pp < endofmime) {
 256                     /* 6 + 2 bits */
 257                     while ((pp < endofmime) &&
 258                            ((c1 = char64(*pp)) == -1)) {
 259                         pp++;
 260                     }
 261                     if (pp < endofmime) {
 262                         pp++;
 263                     }
 264                     while ((pp < endofmime) &&
 265                            ((c2 = char64(*pp)) == -1)) {
 266                         pp++;
 267                     }
 268                     if (pp < endofmime && c1 != -1 && c2 != -1) {
 269                         ADDCHR2((c1 << 2) | (c2 >> 4));
 270                         pp++;
 271                     }
 272                     /* 4 + 4 bits */
 273                     while ((pp < endofmime) &&
 274                            ((c3 = char64(*pp)) == -1)) {
 275                         pp++;
 276                     }
 277                     if (pp < endofmime && c2 != -1 && c3 != -1) {
 278                         ADDCHR2(((c2 & 0xF) << 4) | (c3 >> 2));
 279                         pp++;
 280                     }
 281                     /* 2 + 6 bits */
 282                     while ((pp < endofmime) &&
 283                            ((c4 = char64(*pp)) == -1)) {
 284                         pp++;
 285                     }
 286                     if (pp < endofmime && c3 != -1 && c4 != -1) {
 287                         ADDCHR2(((c3 & 0x3) << 6) | (c4));
 288                         pp++;
 289                     }
 290                 }
 291             }
 292
 293 #ifdef HAVE_ICONV
 294         iconvbuffull:
 295             /* NB that the string at convbuf is not necessarily NUL terminated here:
 296              * q points to the first byte after the valid part.
 297              */
 298             /* Convert to native character set */
 299             if (use_iconv) {
 300                 size_t inbytes = q - convbuf;
 301                 ICONV_CONST char *start = convbuf;
 302
 303                 while (inbytes) {
 304                     if (iconv(cd, &start, &inbytes, &saveq, &savedstlen) ==
 305                             (size_t)-1) {
 306                         if (errno != EILSEQ) break;
 307                         /* character couldn't be converted. we output a `?'
 308                          * and try to carry on which won't work if
 309                          * either encoding was stateful */
 310                         iconv (cd, 0, 0, &saveq, &savedstlen);
 311                         if (!savedstlen)
 312                             break;
 313                         *saveq++ = '?';
 314                         savedstlen--;
 315                         if (!savedstlen)
 316                             break;
 317                         /* skip to next input character */
 318                         if (fromutf8) {
 319                             for (++start, --inbytes;
 320                                  start < q  &&  (*start & 192) == 128;
 321                                  ++start, --inbytes)
 322                                 continue;
 323                         } else
 324                             start++, inbytes--;
 325                         if (start >= q)
 326                             break;
 327                     }
 328                 }
 329                 q = saveq;
 330                 /* Stop now if (1) we hit the end of the buffer trying to do
 331                  * MIME decoding and have just iconv-converted a partial string
 332                  * or (2) our iconv-conversion hit the end of the buffer.
 333                  */
 334                 if (!dstlen || !savedstlen)
 335                     goto buffull;
 336                 dstlen = savedstlen;
 337                 free(convbuf);
 338             }
 339 #endif
 340
 341             /*
 342              * Now that we are done decoding this particular
 343              * encoded word, advance string to trailing '='.
 344              */
 345             p = endofmime + 1;
 346
 347             encoding_found = 1;         /* we found (at least 1) encoded word */
 348             between_encodings = 1;      /* we have just decoded something     */
 349             whitespace = 0;             /* re-initialize amount of whitespace */
 350         }
 351     }
 352 #ifdef HAVE_ICONV
 353     if (use_iconv) iconv_close(cd);
 354 #endif
 355
 356     /* If an equals was pending at end of string, add it now. */
 357     if (equals_pending)
 358         ADDCHR('=');
 359     *q = '\0';
 360
 361     return encoding_found;
 362
 363   buffull:
 364     /* q is currently just off the end of the buffer, so rewind to NUL terminate */
 365     q--;
 366     *q = '\0';
 367     return encoding_found;
 368 }