libcitadel/lib/mime_parser.c

   1 /*
   2  * This is the MIME parser for Citadel.
   3  *
   4  * Copyright (c) 1998-2010 by the citadel.org development team.
   5  * This code is distributed under the GNU General Public License v3.
   6  *
   7  */
   8
   9 #include <stdlib.h>
  10 #include <unistd.h>
  11 #include <stdio.h>
  12 #include <signal.h>
  13 #include <sys/types.h>
  14 #include <ctype.h>
  15 #include <string.h>
  16 #include <sys/stat.h>
  17 #include <sys/types.h>
  18 #include <dirent.h>
  19 #include <errno.h>
  20
  21 #include "xdgmime/xdgmime.h"
  22 #include "libcitadel.h"
  23 #include "libcitadellocal.h"
  24
  25 const unsigned char FromHexTable [256] = {
  26         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //  0
  27         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 10
  28         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 20
  29         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 30
  30         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x00, 0x01, // 40
  31         0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0xFF, 0xFF, // 50
  32         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, // 60
  33         0x0F, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 70
  34         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 80
  35         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0A, 0x0B, 0x0C, // 90
  36         0x0D, 0x0E, 0x0F, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //100
  37         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //110
  38         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //120
  39         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //130
  40         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //140
  41         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //150
  42         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //160
  43         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //170
  44         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //180
  45         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //190
  46         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //200
  47         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //210
  48         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //220
  49         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //230
  50         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //240
  51         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF                          //250
  52 };
  53
  54
  55 long extract_key(char *target, char *source, long sourcelen, char *key, long keylen, char KeyEnd)
  56 {
  57         char *sptr, *ptr = NULL;
  58         int double_quotes = 0;
  59         long RealKeyLen = keylen;
  60
  61         sptr = source;
  62
  63         while (sptr != NULL)
  64         {
  65                 ptr = bmstrcasestr_len(sptr, sourcelen - (sptr - source),
  66                                        key, keylen);
  67                 if(ptr != NULL)
  68                 {
  69                         while (isspace(*(ptr + RealKeyLen)))
  70                                 RealKeyLen ++;
  71                         if (*(ptr + RealKeyLen) == KeyEnd)
  72                         {
  73                                 sptr = NULL;
  74                                 RealKeyLen ++;
  75                         }
  76                         else
  77                         {
  78                                 sptr = ptr + RealKeyLen + 1;
  79                         }
  80                 }
  81                 else
  82                         sptr = ptr;
  83         }
  84         if (ptr == NULL) {
  85                 *target = '\0';
  86                 return 0;
  87         }
  88         strcpy(target, (ptr + RealKeyLen));
  89
  90         for (ptr=target; (*ptr != 0); ptr++) {
  91
  92                 /* A semicolon means we've hit the end of the key, unless we're inside double quotes */
  93                 if ( (double_quotes != 1) && (*ptr == ';')) {
  94                         *ptr = 0;
  95                 }
  96
  97                 /* if we find double quotes, we've got a great set of string boundaries */
  98                 if (*ptr == '\"') {
  99                         ++double_quotes;
 100                         if (double_quotes == 1) {
 101                                 strcpy(ptr, ptr+1);
 102                         }
 103                         else {
 104                                 *ptr = 0;
 105                         }
 106                 }
 107         }
 108         *ptr = '\0';
 109         return ptr - target;
 110 }
 111
 112
 113 /*
 114  * For non-multipart messages, we need to generate a quickie partnum of "1"
 115  * to return to callback functions.  Some callbacks demand it.
 116  */
 117 char *fixed_partnum(char *supplied_partnum) {
 118         if (supplied_partnum == NULL) return "1";
 119         if (strlen(supplied_partnum)==0) return "1";
 120         return supplied_partnum;
 121 }
 122
 123
 124 static inline unsigned int _decode_hex(const char *Source)
 125 {
 126         int ret = '?';
 127         unsigned char LO_NIBBLE;
 128         unsigned char HI_NIBBLE;
 129
 130         HI_NIBBLE = FromHexTable[(unsigned char) *Source];
 131         LO_NIBBLE = FromHexTable[(unsigned char) *(Source+1)];
 132
 133         if ((LO_NIBBLE == 0xFF) || (LO_NIBBLE == 0xFF))
 134                 return ret;
 135         ret = HI_NIBBLE;
 136         ret = ret << 4;
 137         ret = ret | LO_NIBBLE;
 138         return ret;
 139 }
 140
 141 unsigned int decode_hex(char *Source) {return _decode_hex(Source);}
 142
 143 /*
 144  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
 145  * according to RFC2045 section 6.7
 146  */
 147 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
 148         unsigned int ch;
 149         int decoded_length = 0;
 150         int pos = 0;
 151
 152         while (pos < sourcelen)
 153         {
 154                 if (*(encoded + pos) == '=')
 155                 {
 156                         pos ++;
 157                         if (*(encoded + pos) == '\n')
 158                         {
 159                                 pos ++;
 160                         }
 161                         else if (*(encoded + pos) == '\r')
 162                         {
 163                                 pos ++;
 164                                 if (*(encoded + pos) == '\n')
 165                                         pos++;
 166                         }
 167                         else
 168                         {
 169                                 ch = 0;
 170                                 ch = _decode_hex(&encoded[pos]);
 171                                 pos += 2;
 172                                 decoded[decoded_length++] = ch;
 173                         }
 174                 }
 175                 else
 176                 {
 177                         decoded[decoded_length++] = encoded[pos];
 178                         pos += 1;
 179                 }
 180         }
 181         decoded[decoded_length] = 0;
 182         return(decoded_length);
 183 }
 184
 185
 186 /*
 187  * Given a message or message-part body and a length, handle any necessary
 188  * decoding and pass the request up the stack.
 189  */
 190 void mime_decode(char *partnum,
 191                  char *part_start, size_t length,
 192                  char *content_type, char *charset, char *encoding,
 193                  char *disposition,
 194                  char *id,
 195                  char *name, char *filename,
 196                  MimeParserCallBackType CallBack,
 197                  MimeParserCallBackType PreMultiPartCallBack,
 198                  MimeParserCallBackType PostMultiPartCallBack,
 199                  void *userdata,
 200                  int dont_decode)
 201 {
 202
 203         char *decoded;
 204         size_t bytes_decoded = 0;
 205
 206         /* Some encodings aren't really encodings */
 207         if (!strcasecmp(encoding, "7bit"))
 208                 strcpy(encoding, "");
 209         if (!strcasecmp(encoding, "8bit"))
 210                 strcpy(encoding, "");
 211         if (!strcasecmp(encoding, "binary"))
 212                 strcpy(encoding, "");
 213
 214         /* If this part is not encoded, send as-is */
 215         if ( (strlen(encoding) == 0) || (dont_decode)) {
 216                 if (CallBack != NULL) {
 217                         CallBack(name,
 218                                  filename,
 219                                  fixed_partnum(partnum),
 220                                  disposition,
 221                                  part_start,
 222                                  content_type,
 223                                  charset,
 224                                  length,
 225                                  encoding,
 226                                  id,
 227                                  userdata);
 228                         }
 229                 return;
 230         }
 231
 232         /* Fail silently if we hit an unknown encoding. */
 233         if ((strcasecmp(encoding, "base64"))
 234             && (strcasecmp(encoding, "quoted-printable"))) {
 235                 return;
 236         }
 237
 238         /*
 239          * Allocate a buffer for the decoded data.  The output buffer is slightly
 240          * larger than the input buffer; this assumes that the decoded data
 241          * will never be significantly larger than the encoded data.  This is a
 242          * safe assumption with base64, uuencode, and quoted-printable.
 243          */
 244         decoded = malloc(length + 32768);
 245         if (decoded == NULL) {
 246                 return;
 247         }
 248
 249         if (!strcasecmp(encoding, "base64")) {
 250                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 251         }
 252         else if (!strcasecmp(encoding, "quoted-printable")) {
 253                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded, part_start, length);
 254         }
 255
 256         if (bytes_decoded > 0) if (CallBack != NULL) {
 257                         char encoding_buf[SIZ];
 258
 259                         strcpy(encoding_buf, "binary");
 260                         CallBack(name,
 261                                  filename,
 262                                  fixed_partnum(partnum),
 263                                  disposition,
 264                                  decoded,
 265                                  content_type,
 266                                  charset,
 267                                  bytes_decoded,
 268                                  encoding_buf,
 269                                  id,
 270                                  userdata);
 271         }
 272
 273         free(decoded);
 274 }
 275
 276 /*
 277  * this is the extract of mime_decode which can be called if 'dont_decode' was set;
 278  * to save the cpu intense process of decoding to the time when it realy wants the content.
 279  * returns:
 280  *   - > 0 we decoded something, its on *decoded, you need to free it.
 281  *   - = 0 no need to decode stuff. *decoded will be NULL.
 282  *   - < 0 an error occured, either an unknown encoding, or alloc failed. no need to free.
 283  */
 284 int mime_decode_now (char *part_start,
 285                      size_t length,
 286                      char *encoding,
 287                      char **decoded,
 288                      size_t *bytes_decoded)
 289 {
 290         *bytes_decoded = 0;
 291         *decoded = NULL;
 292         /* Some encodings aren't really encodings */
 293         if (!strcasecmp(encoding, "7bit"))
 294                 strcpy(encoding, "");
 295         if (!strcasecmp(encoding, "8bit"))
 296                 strcpy(encoding, "");
 297         if (!strcasecmp(encoding, "binary"))
 298                 strcpy(encoding, "");
 299
 300         /* If this part is not encoded, send as-is */
 301         if (strlen(encoding) == 0) {
 302                 return 0;
 303         }
 304
 305
 306         /* Fail if we hit an unknown encoding. */
 307         if ((strcasecmp(encoding, "base64"))
 308             && (strcasecmp(encoding, "quoted-printable"))) {
 309                 return -1;
 310         }
 311
 312         /*
 313          * Allocate a buffer for the decoded data.  The output buffer is slightly
 314          * larger than the input buffer; this assumes that the decoded data
 315          * will never be significantly larger than the encoded data.  This is a
 316          * safe assumption with base64, uuencode, and quoted-printable.
 317          */
 318         *decoded = malloc(length + 32768);
 319         if (decoded == NULL) {
 320                 return -1;
 321         }
 322
 323         if (!strcasecmp(encoding, "base64")) {
 324                 *bytes_decoded = CtdlDecodeBase64(*decoded, part_start, length);
 325                 return 1;
 326         }
 327         else if (!strcasecmp(encoding, "quoted-printable")) {
 328                 *bytes_decoded = CtdlDecodeQuotedPrintable(*decoded, part_start, length);
 329                 return 1;
 330         }
 331         return -1;
 332 }
 333
 334 typedef enum _eIntMimeHdrs {
 335         boundary,
 336         startary,
 337         endary,
 338         content_type,
 339         charset,
 340         encoding,
 341         content_type_name,
 342         content_disposition_name,
 343         filename,
 344         disposition,
 345         id,
 346         eMax /* don't move ! */
 347 } eIntMimeHdrs;
 348
 349 typedef struct _CBufStr {
 350         char Key[SIZ];
 351         long len;
 352 }CBufStr;
 353
 354 typedef struct _interesting_mime_headers {
 355         CBufStr b[eMax];
 356         long content_length;
 357         long is_multipart;
 358 } interesting_mime_headers;
 359
 360
 361 static void FlushInterestingMimes(interesting_mime_headers *m)
 362 {
 363         int i;
 364
 365         for (i = 0; i < eMax; i++) {
 366              m->b[i].Key[0] = '\0';
 367              m->b[i].len = 0;
 368         }
 369         m->content_length = -1;
 370 }
 371 static interesting_mime_headers *InitInterestingMimes(void)
 372 {
 373         interesting_mime_headers *m;
 374         m = (interesting_mime_headers*) malloc( sizeof(interesting_mime_headers));
 375
 376         FlushInterestingMimes(m);
 377
 378         return m;
 379 }
 380
 381
 382 static long parse_MimeHeaders(interesting_mime_headers *m,
 383                               char** pcontent_start,
 384                               char *content_end)
 385 {
 386         char buf[SIZ];
 387         char header[SIZ];
 388         long headerlen;
 389         char *ptr, *pch;
 390         int buflen = 0;
 391         int i;
 392
 393         /* Learn interesting things from the headers */
 394         ptr = *pcontent_start;
 395         *header = '\0';
 396         headerlen = 0;
 397         do {
 398                 ptr = memreadlinelen(ptr, buf, SIZ, &buflen);
 399
 400                 for (i = 0; i < buflen; ++i) {
 401                         if (isspace(buf[i])) {
 402                                 buf[i] = ' ';
 403                         }
 404                 }
 405
 406                 if (!isspace(buf[0]) && (headerlen > 0)) {
 407                         if (!strncasecmp(header, "Content-type:", 13)) {
 408                                 memcpy (m->b[content_type].Key, &header[13], headerlen - 12);
 409                                 m->b[content_type].len = striplt (m->b[content_type].Key);
 410
 411                                 m->b[content_type_name].len = extract_key(m->b[content_type_name].Key, CKEY(m->b[content_type]), HKEY("name"), '=');
 412                                 m->b[charset].len           = extract_key(m->b[charset].Key,           CKEY(m->b[content_type]), HKEY("charset"), '=');
 413                                 m->b[boundary].len          = extract_key(m->b[boundary].Key,          header,       headerlen,  HKEY("boundary"), '=');
 414
 415                                 /* Deal with weird headers */
 416                                 pch = strchr(m->b[content_type].Key, ' ');
 417                                 if (pch != NULL) {
 418                                         *pch = '\0';
 419                                         m->b[content_type].len = m->b[content_type].Key - pch;
 420                                 }
 421                                 pch = strchr(m->b[content_type].Key, ';');
 422                                 if (pch != NULL) {
 423                                         *pch = '\0';
 424                                         m->b[content_type].len = m->b[content_type].Key - pch;
 425                                 }
 426                         }
 427                         else if (!strncasecmp(header, "Content-Disposition:", 20)) {
 428                                 memcpy (m->b[disposition].Key, &header[20], headerlen - 19);
 429                                 m->b[disposition].len = striplt(m->b[disposition].Key);
 430
 431                                 m->b[content_disposition_name].len = extract_key(m->b[content_disposition_name].Key, CKEY(m->b[disposition]), HKEY("name"), '=');
 432                                 m->b[filename].len                 = extract_key(m->b[filename].Key,                 CKEY(m->b[disposition]), HKEY("filename"), '=');
 433                                 pch = strchr(m->b[disposition].Key, ';');
 434                                 if (pch != NULL) *pch = '\0';
 435                                 m->b[disposition].len = striplt(m->b[disposition].Key);
 436                         }
 437                         else if (!strncasecmp(header, "Content-ID:", 11)) {
 438                                 memcpy(m->b[id].Key, &header[11], headerlen);
 439                                 striplt(m->b[id].Key);
 440                                 m->b[id].len = stripallbut(m->b[id].Key, '<', '>');
 441                         }
 442                         else if (!strncasecmp(header, "Content-length: ", 15)) {
 443                                 char *clbuf;
 444                                 clbuf = &header[15];
 445                                 while (isspace(*clbuf))
 446                                         clbuf ++;
 447                                 m->content_length = (size_t) atol(clbuf);
 448                         }
 449                         else if (!strncasecmp(header, "Content-transfer-encoding: ", 26)) {
 450                                 memcpy(m->b[encoding].Key, &header[26], headerlen - 26);
 451                                 m->b[encoding].len = striplt(m->b[encoding].Key);
 452                         }
 453                         *header = '\0';
 454                         headerlen = 0;
 455                 }
 456                 if ((headerlen + buflen + 2) < SIZ) {
 457                         memcpy(&header[headerlen], buf, buflen);
 458                         headerlen += buflen;
 459                         header[headerlen] = '\0';
 460                 }
 461                 if (ptr >= content_end) {
 462                         return -1;
 463                 }
 464         } while ((!IsEmptyStr(buf)) && (*ptr != 0));
 465
 466         m->is_multipart = m->b[boundary].len != 0;
 467         *pcontent_start = ptr;
 468
 469         return 0;
 470 }
 471
 472
 473 static int IsAsciiEncoding(interesting_mime_headers *m)
 474 {
 475         if ((m->b[encoding].len != 0) &&
 476             (strcmp(m->b[encoding].Key, "binary") == 0))
 477                 return 0;
 478         else
 479                 return 1;
 480 }
 481
 482 static char *FindNextContent(char *ptr,
 483                              char *content_end,
 484                              interesting_mime_headers *SubMimeHeaders,
 485                              interesting_mime_headers *m)
 486 {
 487         char *next_boundary;
 488         char  tmp;
 489
 490         if (IsAsciiEncoding(SubMimeHeaders)) {
 491                 tmp = *content_end;
 492                 *content_end = '\0';
 493
 494                 /**
 495                  * ok, if we have a content length of the mime part,
 496                  * try skipping the content on the search for the next
 497                  * boundary. since we don't trust the content_length
 498                  * to be all accurate, and suspect it to lose one digit
 499                  * per line with a line length of 80 chars, we need
 500                  * to start searching a little before..
 501                  */
 502
 503                 if ((SubMimeHeaders->content_length != -1) &&
 504                     (SubMimeHeaders->content_length > 10))
 505                 {
 506                         char *pptr;
 507                         long lines;
 508
 509                         lines = SubMimeHeaders->content_length / 80;
 510                         pptr = ptr + SubMimeHeaders->content_length - lines - 10;
 511                         if (pptr < content_end)
 512                                 ptr = pptr;
 513                 }
 514
 515                 next_boundary = strstr(ptr, m->b[startary].Key);
 516                 *content_end = tmp;
 517         }
 518         else {
 519                 char *srch;
 520                 /**
 521                  * ok, if we have a content length of the mime part,
 522                  * try skipping the content on the search for the next
 523                  * boundary. since we don't trust the content_length
 524                  * to be all accurate, start searching a little before..
 525                  */
 526
 527                 if ((SubMimeHeaders->content_length != -1) &&
 528                     (SubMimeHeaders->content_length > 10))
 529                 {
 530                         char *pptr;
 531                         pptr = ptr + SubMimeHeaders->content_length - 10;
 532                         if (pptr < content_end)
 533                                 ptr = pptr;
 534                 }
 535                 next_boundary = NULL;
 536                 for (srch=ptr; srch<content_end; ++srch) {
 537                         if (!memcmp(srch,
 538                                     m->b[startary].Key,
 539                                     m->b[startary].len))
 540                         {
 541                                 next_boundary = srch;
 542                                 srch = content_end;
 543                         }
 544                 }
 545
 546         }
 547         return next_boundary;
 548 }
 549
 550 /*
 551  * Break out the components of a multipart message
 552  * (This function expects to be fed HEADERS + CONTENT)
 553  * Note: NULL can be supplied as content_end; in this case, the message is
 554  * considered to have ended when the parser encounters a 0x00 byte.
 555  */
 556 static void recurseable_mime_parser(char *partnum,
 557                                     char *content_start, char *content_end,
 558                                     MimeParserCallBackType CallBack,
 559                                     MimeParserCallBackType PreMultiPartCallBack,
 560                                     MimeParserCallBackType PostMultiPartCallBack,
 561                                     void *userdata,
 562                                     int dont_decode,
 563                                     interesting_mime_headers *m)
 564 {
 565         interesting_mime_headers *SubMimeHeaders;
 566         char     *ptr;
 567         char     *part_start;
 568         char     *part_end = NULL;
 569         char     *evaluate_crlf_ptr = NULL;
 570         char     *next_boundary;
 571         char      nested_partnum[256];
 572         int       crlf_in_use = 0;
 573         int       part_seq = 0;
 574         CBufStr  *chosen_name;
 575
 576
 577         /* If this is a multipart message, then recursively process it */
 578         ptr = content_start;
 579         part_start = NULL;
 580         if (m->is_multipart) {
 581
 582                 /* Tell the client about this message's multipartedness */
 583                 if (PreMultiPartCallBack != NULL) {
 584                         PreMultiPartCallBack("",
 585                                              "",
 586                                              partnum,
 587                                              "",
 588                                              NULL,
 589                                              m->b[content_type].Key,
 590                                              m->b[charset].Key,
 591                                              0,
 592                                              m->b[encoding].Key,
 593                                              m->b[id].Key,
 594                                              userdata);
 595                 }
 596
 597                 /* Figure out where the boundaries are */
 598                 m->b[startary].len = snprintf(m->b[startary].Key, SIZ, "--%s", m->b[boundary].Key);
 599                 SubMimeHeaders = InitInterestingMimes ();
 600                 if (*ptr == '\r')
 601                         ptr ++;
 602                 if (*ptr == '\n')
 603                         ptr ++;
 604                 if (strncmp(ptr, m->b[startary].Key, m->b[startary].len) == 0)
 605                         ptr += m->b[startary].len;
 606                 if (*ptr == '\r')
 607                         ptr ++;
 608                 if (*ptr == '\n')
 609                         ptr ++;
 610                 part_start = NULL;
 611                 do {
 612
 613                         if (parse_MimeHeaders(SubMimeHeaders, &ptr, content_end) != 0)
 614                                 break;
 615                         part_start = ptr;
 616
 617                         next_boundary = FindNextContent(ptr,
 618                                                         content_end,
 619                                                         SubMimeHeaders,
 620                                                         m);
 621                         if ((next_boundary != NULL) &&
 622                             (next_boundary - part_start < 3))
 623                                 continue;
 624
 625                         if ( (part_start != NULL) && (next_boundary != NULL) ) {
 626                                 part_end = next_boundary;
 627                                 --part_end;             /* omit the trailing LF */
 628                                 if (crlf_in_use) {
 629                                         --part_end;     /* omit the trailing CR */
 630                                 }
 631
 632                                 if (!IsEmptyStr(partnum)) {
 633                                         snprintf(nested_partnum,
 634                                                  sizeof nested_partnum,
 635                                                  "%s.%d", partnum,
 636                                                  ++part_seq);
 637                                 }
 638                                 else {
 639                                         snprintf(nested_partnum,
 640                                                  sizeof nested_partnum,
 641                                                  "%d", ++part_seq);
 642                                 }
 643                                 recurseable_mime_parser(nested_partnum,
 644                                                         part_start,
 645                                                         part_end,
 646                                                         CallBack,
 647                                                         PreMultiPartCallBack,
 648                                                         PostMultiPartCallBack,
 649                                                         userdata,
 650                                                         dont_decode,
 651                                                         SubMimeHeaders);
 652                         }
 653
 654                         if (next_boundary != NULL) {
 655                                 /* If we pass out of scope, don't attempt to
 656                                  * read past the end boundary. */
 657                                 if ((*(next_boundary + m->b[startary].len + 1) == '-') &&
 658                                     (*(next_boundary + m->b[startary].len + 2) == '-') ){
 659                                         ptr = content_end;
 660                                 }
 661                                 else {
 662                                         /* Set up for the next part. */
 663                                         part_start = strstr(next_boundary, "\n");
 664
 665                                         /* Determine whether newlines are LF or CRLF */
 666                                         evaluate_crlf_ptr = part_start;
 667                                         --evaluate_crlf_ptr;
 668                                         if ((*evaluate_crlf_ptr == '\r') &&
 669                                             (*(evaluate_crlf_ptr + 1) == '\n'))
 670                                         {
 671                                                 crlf_in_use = 1;
 672                                         }
 673                                         else {
 674                                                 crlf_in_use = 0;
 675                                         }
 676
 677                                         /* Advance past the LF ... now we're in the next part */
 678                                         ++part_start;
 679                                         ptr = part_start;
 680                                 }
 681                         }
 682                         else {
 683                                 /* Invalid end of multipart.  Bail out! */
 684                                 ptr = content_end;
 685                         }
 686                         FlushInterestingMimes(SubMimeHeaders);
 687                 } while ( (ptr < content_end) && (next_boundary != NULL) );
 688
 689                 free(SubMimeHeaders);
 690
 691                 if (PostMultiPartCallBack != NULL) {
 692                         PostMultiPartCallBack("",
 693                                               "",
 694                                               partnum,
 695                                               "",
 696                                               NULL,
 697                                               m->b[content_type].Key,
 698                                               m->b[charset].Key,
 699                                               0,
 700                                               m->b[encoding].Key,
 701                                               m->b[id].Key,
 702                                               userdata);
 703                 }
 704         } /* If it's not a multipart message, then do something with it */
 705         else {
 706                 size_t length;
 707                 part_start = ptr;
 708                 length = content_end - part_start;
 709                 ptr = part_end = content_end;
 710
 711
 712                 /* The following code will truncate the MIME part to the size
 713                  * specified by the Content-length: header.   We have commented it
 714                  * out because these headers have a tendency to be wrong.
 715                  *
 716                  *      if ( (content_length > 0) && (length > content_length) ) {
 717                  *              length = content_length;
 718                  *      }
 719                  */
 720
 721                 /* Sometimes the "name" field is tacked on to Content-type,
 722                  * and sometimes it's tacked on to Content-disposition.  Use
 723                  * whichever one we have.
 724                  */
 725                 if (m->b[content_disposition_name].len > m->b[content_type_name].len) {
 726                         chosen_name = &m->b[content_disposition_name];
 727                 }
 728                 else {
 729                         chosen_name = &m->b[content_type_name];
 730                 }
 731
 732                 /* Ok, we've got a non-multipart part here, so do something with it.
 733                  */
 734                 mime_decode(partnum,
 735                             part_start,
 736                             length,
 737                             m->b[content_type].Key,
 738                             m->b[charset].Key,
 739                             m->b[encoding].Key,
 740                             m->b[disposition].Key,
 741                             m->b[id].Key,
 742                             chosen_name->Key,
 743                             m->b[filename].Key,
 744                             CallBack,
 745                             NULL, NULL,
 746                             userdata,
 747                             dont_decode
 748                         );
 749
 750                 /*
 751                  * Now if it's an encapsulated message/rfc822 then we have to recurse into it
 752                  */
 753                 if (!strcasecmp(&m->b[content_type].Key[0], "message/rfc822")) {
 754
 755                         if (PreMultiPartCallBack != NULL) {
 756                                 PreMultiPartCallBack("",
 757                                                      "",
 758                                                      partnum,
 759                                                      "",
 760                                                      NULL,
 761                                                      m->b[content_type].Key,
 762                                                      m->b[charset].Key,
 763                                                      0,
 764                                                      m->b[encoding].Key,
 765                                                      m->b[id].Key,
 766                                                      userdata);
 767                         }
 768                         if (CallBack != NULL) {
 769                                 if (strlen(partnum) > 0) {
 770                                         snprintf(nested_partnum,
 771                                                  sizeof nested_partnum,
 772                                                  "%s.%d", partnum,
 773                                                  ++part_seq);
 774                                 }
 775                                 else {
 776                                         snprintf(nested_partnum,
 777                                                  sizeof nested_partnum,
 778                                                  "%d", ++part_seq);
 779                                 }
 780                                 the_mime_parser(nested_partnum,
 781                                                 part_start,
 782                                                 part_end,
 783                                                 CallBack,
 784                                                 PreMultiPartCallBack,
 785                                                 PostMultiPartCallBack,
 786                                                 userdata,
 787                                                 dont_decode
 788                                         );
 789                         }
 790                         if (PostMultiPartCallBack != NULL) {
 791                                 PostMultiPartCallBack("",
 792                                                       "",
 793                                                       partnum,
 794                                                       "",
 795                                                       NULL,
 796                                                       m->b[content_type].Key,
 797                                                       m->b[charset].Key,
 798                                                       0,
 799                                                       m->b[encoding].Key,
 800                                                       m->b[id].Key,
 801                                                       userdata);
 802                         }
 803
 804
 805                 }
 806
 807         }
 808
 809 }
 810
 811 /*
 812  * Break out the components of a multipart message
 813  * (This function expects to be fed HEADERS + CONTENT)
 814  * Note: NULL can be supplied as content_end; in this case, the message is
 815  * considered to have ended when the parser encounters a 0x00 byte.
 816  */
 817 void the_mime_parser(char *partnum,
 818                      char *content_start, char *content_end,
 819                      MimeParserCallBackType CallBack,
 820                      MimeParserCallBackType PreMultiPartCallBack,
 821                      MimeParserCallBackType PostMultiPartCallBack,
 822                      void *userdata,
 823                      int dont_decode)
 824 {
 825         interesting_mime_headers *m;
 826
 827         /* If the caller didn't supply an endpointer, generate one by measure */
 828         if (content_end == NULL) {
 829                 content_end = &content_start[strlen(content_start)];
 830         }
 831
 832         m = InitInterestingMimes();
 833
 834         if (!parse_MimeHeaders(m, &content_start, content_end))
 835         {
 836
 837                 recurseable_mime_parser(partnum,
 838                                         content_start, content_end,
 839                                         CallBack,
 840                                         PreMultiPartCallBack,
 841                                         PostMultiPartCallBack,
 842                                         userdata,
 843                                         dont_decode,
 844                                         m);
 845         }
 846         free(m);
 847 }
 848
 849 /*
 850  * Entry point for the MIME parser.
 851  * (This function expects to be fed HEADERS + CONTENT)
 852  * Note: NULL can be supplied as content_end; in this case, the message is
 853  * considered to have ended when the parser encounters a 0x00 byte.
 854  */
 855 void mime_parser(char *content_start,
 856                  char *content_end,
 857                  MimeParserCallBackType CallBack,
 858                  MimeParserCallBackType PreMultiPartCallBack,
 859                  MimeParserCallBackType PostMultiPartCallBack,
 860                  void *userdata,
 861                  int dont_decode)
 862 {
 863
 864         the_mime_parser("", content_start, content_end,
 865                         CallBack,
 866                         PreMultiPartCallBack,
 867                         PostMultiPartCallBack,
 868                         userdata, dont_decode);
 869 }
 870
 871
 872
 873
 874
 875
 876 typedef struct _MimeGuess {
 877         const char *Pattern;
 878         size_t PatternLen;
 879         long PatternOffset;
 880         const char *MimeString;
 881 } MimeGuess;
 882
 883 MimeGuess MyMimes [] = {
 884         {
 885                 "GIF",
 886                 3,
 887                 0,
 888                 "image/gif"
 889         },
 890         {
 891                 "\xff\xd8",
 892                 2,
 893                 0,
 894                 "image/jpeg"
 895         },
 896         {
 897                 "\x89PNG",
 898                 4,
 899                 0,
 900                 "image/png"
 901         },
 902         { // last...
 903                 "",
 904                 0,
 905                 0,
 906                 ""
 907         }
 908 };
 909
 910
 911 const char *GuessMimeType(const char *data, size_t dlen)
 912 {
 913         int MimeIndex = 0;
 914
 915         while (MyMimes[MimeIndex].PatternLen != 0)
 916         {
 917                 if ((MyMimes[MimeIndex].PatternLen +
 918                      MyMimes[MimeIndex].PatternOffset < dlen) &&
 919                     strncmp(MyMimes[MimeIndex].Pattern,
 920                             &data[MyMimes[MimeIndex].PatternOffset],
 921                             MyMimes[MimeIndex].PatternLen) == 0)
 922                 {
 923                         return MyMimes[MimeIndex].MimeString;
 924                 }
 925                 MimeIndex ++;
 926         }
 927         /*
 928          * ok, our simple minded algorythm didn't find anything,
 929          * let the big chegger try it, he wil default to application/octet-stream
 930          */
 931         return (xdg_mime_get_mime_type_for_data(data, dlen));
 932 }
 933
 934
 935 const char* GuessMimeByFilename(const char *what, size_t len)
 936 {
 937         /* we know some hardcoded on our own, try them... */
 938         if ((len > 3) && !strncasecmp(&what[len - 4], ".gif", 4))
 939                 return "image/gif";
 940         else if ((len > 2) && !strncasecmp(&what[len - 3], ".js", 3))
 941                 return  "text/javascript";
 942         else if ((len > 3) && !strncasecmp(&what[len - 4], ".txt", 4))
 943                 return "text/plain";
 944         else if ((len > 3) && !strncasecmp(&what[len - 4], ".css", 4))
 945                 return "text/css";
 946         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htc", 4))
 947                 return "text/x-component";
 948         else if ((len > 3) && !strncasecmp(&what[len - 4], ".jpg", 4))
 949                 return "image/jpeg";
 950         else if ((len > 3) && !strncasecmp(&what[len - 4], ".png", 4))
 951                 return "image/png";
 952         else if ((len > 3) && !strncasecmp(&what[len - 4], ".ico", 4))
 953                 return "image/x-icon";
 954         else if ((len > 3) && !strncasecmp(&what[len - 4], ".vcf", 4))
 955                 return "text/x-vcard";
 956         else if ((len > 4) && !strncasecmp(&what[len - 5], ".html", 5))
 957                 return "text/html";
 958         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htm", 4))
 959                 return "text/html";
 960         else if ((len > 3) && !strncasecmp(&what[len - 4], ".wml", 4))
 961                 return "text/vnd.wap.wml";
 962         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmls", 5))
 963                 return "text/vnd.wap.wmlscript";
 964         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmlc", 5))
 965                 return "application/vnd.wap.wmlc";
 966         else if ((len > 5) && !strncasecmp(&what[len - 6], ".wmlsc", 6))
 967                 return "application/vnd.wap.wmlscriptc";
 968         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wbmp", 5))
 969                 return "image/vnd.wap.wbmp";
 970         else
 971                 /* and let xdgmime do the fallback. */
 972                 return xdg_mime_get_mime_type_from_file_name(what);
 973 }
 974
 975 static HashList *IconHash = NULL;
 976
 977 typedef struct IconName IconName;
 978
 979 struct IconName {
 980         char *FlatName;
 981         char *FileName;
 982 };
 983
 984 static void DeleteIcon(void *IconNamePtr)
 985 {
 986         IconName *Icon = (IconName*) IconNamePtr;
 987         free(Icon->FlatName);
 988         free(Icon->FileName);
 989         free(Icon);
 990 }
 991
 992 /*
 993 static const char *PrintFlat(void *IconNamePtr)
 994 {
 995         IconName *Icon = (IconName*) IconNamePtr;
 996         return Icon->FlatName;
 997 }
 998 static const char *PrintFile(void *IconNamePtr)
 999 {
1000         IconName *Icon = (IconName*) IconNamePtr;
1001         return Icon->FileName;
1002 }
1003 */
1004
1005 #define GENSTR "x-generic"
1006 #define IGNORE_PREFIX_1 "gnome-mime"
1007 int LoadIconDir(const char *DirName)
1008 {
1009         DIR *filedir = NULL;
1010         struct dirent *filedir_entry;
1011         int d_namelen;
1012         int d_without_ext;
1013         IconName *Icon;
1014
1015         filedir = opendir (DirName);
1016         IconHash = NewHash(1, NULL);
1017         if (filedir == NULL) {
1018                 return 0;
1019         }
1020
1021         while ((filedir_entry = readdir(filedir)))
1022         {
1023                 char *MinorPtr;
1024                 char *PStart;
1025 #ifdef _DIRENT_HAVE_D_NAMELEN
1026                 d_namelen = filedir_entry->d_namelen;
1027 #else
1028                 d_namelen = strlen(filedir_entry->d_name);
1029 #endif
1030                 d_without_ext = d_namelen;
1031                 while ((d_without_ext > 0) && (filedir_entry->d_name[d_without_ext] != '.'))
1032                         d_without_ext --;
1033                 if ((d_without_ext == 0) || (d_namelen < 3))
1034                         continue;
1035
1036                 if ((sizeof(IGNORE_PREFIX_1) < d_namelen) &&
1037                     (strncmp(IGNORE_PREFIX_1,
1038                              filedir_entry->d_name,
1039                              sizeof(IGNORE_PREFIX_1) - 1) == 0)) {
1040                         PStart = filedir_entry->d_name + sizeof(IGNORE_PREFIX_1);
1041                         d_without_ext -= sizeof(IGNORE_PREFIX_1);
1042                 }
1043                 else {
1044                         PStart = filedir_entry->d_name;
1045                 }
1046                 Icon = malloc(sizeof(IconName));
1047
1048                 Icon->FileName = malloc(d_namelen + 1);
1049                 memcpy(Icon->FileName, filedir_entry->d_name, d_namelen + 1);
1050
1051                 Icon->FlatName = malloc(d_without_ext + 1);
1052                 memcpy(Icon->FlatName, PStart, d_without_ext);
1053                 Icon->FlatName[d_without_ext] = '\0';
1054                 /* Try to find Minor type in image-jpeg */
1055                 MinorPtr = strchr(Icon->FlatName, '-');
1056                 if (MinorPtr != NULL) {
1057                         size_t MinorLen;
1058                         MinorLen = 1 + d_without_ext - (MinorPtr - Icon->FlatName + 1);
1059                         if ((MinorLen == sizeof(GENSTR)) &&
1060                             (strncmp(MinorPtr + 1, GENSTR, sizeof(GENSTR)) == 0)) {
1061                                 /* ok, we found a generic filename. cut the generic. */
1062                                 *MinorPtr = '\0';
1063                                 d_without_ext = d_without_ext - (MinorPtr - Icon->FlatName);
1064                         }
1065                         else { /* Map the major / minor separator to / */
1066                                 *MinorPtr = '/';
1067                         }
1068                 }
1069
1070 //              PrintHash(IconHash, PrintFlat, PrintFile);
1071 //              printf("%s - %s\n", Icon->FlatName, Icon->FileName);
1072                 Put(IconHash, Icon->FlatName, d_without_ext, Icon, DeleteIcon);
1073 //              PrintHash(IconHash, PrintFlat, PrintFile);
1074         }
1075         closedir(filedir);
1076         return 1;
1077 }
1078
1079 const char *GetIconFilename(char *MimeType, size_t len)
1080 {
1081         void *vIcon;
1082         IconName *Icon;
1083
1084         if(IconHash == NULL)
1085                 return NULL;
1086
1087         GetHash(IconHash, MimeType, len, &vIcon), Icon = (IconName*) vIcon;
1088         /* didn't find the exact mimetype? try major only. */
1089         if (Icon == NULL) {
1090                 char * pMinor;
1091                 pMinor = strchr(MimeType, '/');
1092                 if (pMinor != NULL) {
1093                         *pMinor = '\0';
1094                         GetHash(IconHash, MimeType, pMinor - MimeType, &vIcon),
1095                                 Icon = (IconName*) vIcon;
1096                 }
1097         }
1098         if (Icon == NULL) {
1099                 return NULL;
1100         }
1101
1102         /*printf("Getting: [%s] == [%s] -> [%s]\n", MimeType, Icon->FlatName, Icon->FileName);*/
1103         return Icon->FileName;
1104 }
1105
1106 void ShutDownLibCitadelMime(void)
1107 {
1108         DeleteHash(&IconHash);
1109 }