libcitadel/lib/mime_parser.c

   1 // This is the MIME parser for Citadel.
   2 //
   3 // Copyright (c) 1998-2022 by the citadel.org development team.
   4 //
   5 // This program is open source software.  Use, duplication, or disclosure
   6 // is subject to the terms of the GNU General Public License, version 3.
   7
   8 #include <stdlib.h>
   9 #include <unistd.h>
  10 #include <stdio.h>
  11 #include <signal.h>
  12 #include <sys/types.h>
  13 #include <ctype.h>
  14 #include <string.h>
  15 #include <sys/stat.h>
  16 #include <sys/types.h>
  17 #include <dirent.h>
  18 #include <errno.h>
  19
  20 #include "xdgmime/xdgmime.h"
  21 #include "libcitadel.h"
  22 #include "libcitadellocal.h"
  23
  24 const unsigned char FromHexTable [256] = {
  25         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //  0
  26         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 10
  27         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 20
  28         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 30
  29         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x00, 0x01, // 40
  30         0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0xFF, 0xFF, // 50
  31         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, // 60
  32         0x0F, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 70
  33         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 80
  34         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0A, 0x0B, 0x0C, // 90
  35         0x0D, 0x0E, 0x0F, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //100
  36         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //110
  37         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //120
  38         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //130
  39         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //140
  40         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //150
  41         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //160
  42         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //170
  43         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //180
  44         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //190
  45         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //200
  46         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //210
  47         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //220
  48         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //230
  49         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //240
  50         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF                          //250
  51 };
  52
  53
  54 long extract_key(char *target, char *source, long sourcelen, char *key, long keylen, char KeyEnd) {
  55         char *sptr, *ptr = NULL;
  56         int double_quotes = 0;
  57         long RealKeyLen = keylen;
  58
  59         sptr = source;
  60
  61         while (sptr != NULL) {
  62                 ptr = bmstrcasestr_len(sptr, sourcelen - (sptr - source), key, keylen);
  63                 if (ptr != NULL) {
  64                         while (isspace(*(ptr + RealKeyLen)))
  65                                 RealKeyLen ++;
  66                         if (*(ptr + RealKeyLen) == KeyEnd) {
  67                                 sptr = NULL;
  68                                 RealKeyLen ++;
  69                         }
  70                         else {
  71                                 sptr = ptr + RealKeyLen + 1;
  72                         }
  73                 }
  74                 else
  75                         sptr = ptr;
  76         }
  77         if (ptr == NULL) {
  78                 *target = '\0';
  79                 return 0;
  80         }
  81         strcpy(target, (ptr + RealKeyLen));
  82
  83         for (ptr=target; (*ptr != 0); ptr++) {
  84
  85                 /* A semicolon means we've hit the end of the key, unless we're inside double quotes */
  86                 if ( (double_quotes != 1) && (*ptr == ';')) {
  87                         *ptr = 0;
  88                 }
  89
  90                 /* if we find double quotes, we've got a great set of string boundaries */
  91                 if (*ptr == '\"') {
  92                         ++double_quotes;
  93                         if (double_quotes == 1) {
  94                                 strcpy(ptr, ptr+1);
  95                         }
  96                         else {
  97                                 *ptr = 0;
  98                         }
  99                 }
 100         }
 101         *ptr = '\0';
 102         return ptr - target;
 103 }
 104
 105
 106 /*
 107  * For non-multipart messages, we need to generate a quickie partnum of "1"
 108  * to return to callback functions.  Some callbacks demand it.
 109  */
 110 char *fixed_partnum(char *supplied_partnum) {
 111         if (supplied_partnum == NULL) return "1";
 112         if (strlen(supplied_partnum)==0) return "1";
 113         return supplied_partnum;
 114 }
 115
 116
 117 static inline unsigned int _decode_hex(const char *Source) {
 118         unsigned int ret = '?';
 119         unsigned char LO_NIBBLE;
 120         unsigned char HI_NIBBLE;
 121
 122         HI_NIBBLE = FromHexTable[(unsigned char) *Source];
 123         LO_NIBBLE = FromHexTable[(unsigned char) *(Source+1)];
 124
 125         if ((LO_NIBBLE == 0xFF) || (LO_NIBBLE == 0xFF))
 126                 return ret;
 127         ret = HI_NIBBLE;
 128         ret = ret << 4;
 129         ret = ret | LO_NIBBLE;
 130         return ret;
 131 }
 132
 133 unsigned int decode_hex(char *Source) {return _decode_hex(Source);}
 134
 135 /*
 136  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
 137  * according to RFC2045 section 6.7
 138  */
 139 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
 140         unsigned int ch;
 141         int decoded_length = 0;
 142         int pos = 0;
 143
 144         while (pos < sourcelen) {
 145                 if (*(encoded + pos) == '=') {
 146                         pos ++;
 147                         if (*(encoded + pos) == '\n') {
 148                                 pos ++;
 149                         }
 150                         else if (*(encoded + pos) == '\r') {
 151                                 pos ++;
 152                                 if (*(encoded + pos) == '\n')
 153                                         pos++;
 154                         }
 155                         else {
 156                                 ch = _decode_hex(&encoded[pos]);
 157                                 pos += 2;
 158                                 decoded[decoded_length++] = ch;
 159                         }
 160                 }
 161                 else {
 162                         decoded[decoded_length++] = encoded[pos];
 163                         pos += 1;
 164                 }
 165         }
 166         decoded[decoded_length] = 0;
 167         return(decoded_length);
 168 }
 169
 170
 171 /*
 172  * Given a message or message-part body and a length, handle any necessary
 173  * decoding and pass the request up the stack.
 174  */
 175 void mime_decode(char *partnum,
 176                  char *part_start, size_t length,
 177                  char *content_type, char *charset, char *encoding,
 178                  char *disposition,
 179                  char *id,
 180                  char *name, char *filename,
 181                  MimeParserCallBackType CallBack,
 182                  MimeParserCallBackType PreMultiPartCallBack,
 183                  MimeParserCallBackType PostMultiPartCallBack,
 184                  void *userdata,
 185                  int dont_decode
 186 ) {
 187         char *decoded;
 188         size_t bytes_decoded = 0;
 189
 190         // Some encodings aren't really encodings
 191         if (!strcasecmp(encoding, "7bit"))
 192                 *encoding = '\0';
 193         if (!strcasecmp(encoding, "8bit"))
 194                 *encoding = '\0';
 195         if (!strcasecmp(encoding, "binary"))
 196                 *encoding = '\0';
 197         if (!strcasecmp(encoding, "ISO-8859-1"))
 198                 *encoding = '\0';
 199
 200         // If this part is not encoded, send as-is
 201         if ( (strlen(encoding) == 0) || (dont_decode)) {
 202                 if (CallBack != NULL) {
 203                         CallBack(name,
 204                                  filename,
 205                                  fixed_partnum(partnum),
 206                                  disposition,
 207                                  part_start,
 208                                  content_type,
 209                                  charset,
 210                                  length,
 211                                  encoding,
 212                                  id,
 213                                  userdata);
 214                         }
 215                 return;
 216         }
 217
 218         // Fail silently if we hit an unknown encoding.
 219         if ((strcasecmp(encoding, "base64")) && (strcasecmp(encoding, "quoted-printable"))) {
 220                 return;
 221         }
 222
 223         // Allocate a buffer for the decoded data.  The output buffer is slightly
 224         // larger than the input buffer; this assumes that the decoded data
 225         // will never be significantly larger than the encoded data.  This is a
 226         // safe assumption with base64, uuencode, and quoted-printable.
 227         decoded = malloc(length + 32768);
 228         if (decoded == NULL) {
 229                 return;
 230         }
 231
 232         if (!strcasecmp(encoding, "base64")) {
 233                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 234         }
 235         else if (!strcasecmp(encoding, "quoted-printable")) {
 236                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded, part_start, length);
 237         }
 238
 239         if (bytes_decoded > 0) if (CallBack != NULL) {
 240                         char encoding_buf[SIZ];
 241
 242                         strcpy(encoding_buf, "binary");
 243                         CallBack(name,
 244                                  filename,
 245                                  fixed_partnum(partnum),
 246                                  disposition,
 247                                  decoded,
 248                                  content_type,
 249                                  charset,
 250                                  bytes_decoded,
 251                                  encoding_buf,
 252                                  id,
 253                                  userdata);
 254         }
 255
 256         free(decoded);
 257 }
 258
 259 /*
 260  * this is the extract of mime_decode which can be called if 'dont_decode' was set;
 261  * to save the cpu intense process of decoding to the time when it realy wants the content.
 262  * returns:
 263  *   - > 0 we decoded something, its on *decoded, you need to free it.
 264  *   - = 0 no need to decode stuff. *decoded will be NULL.
 265  *   - < 0 an error occured, either an unknown encoding, or alloc failed. no need to free.
 266  */
 267 int mime_decode_now (char *part_start,
 268                      size_t length,
 269                      char *encoding,
 270                      char **decoded,
 271                      size_t *bytes_decoded)
 272 {
 273         *bytes_decoded = 0;
 274         *decoded = NULL;
 275         /* Some encodings aren't really encodings */
 276         if (!strcasecmp(encoding, "7bit"))
 277                 *encoding = '\0';
 278         if (!strcasecmp(encoding, "8bit"))
 279                 *encoding = '\0';
 280         if (!strcasecmp(encoding, "binary"))
 281                 *encoding = '\0';
 282
 283         /* If this part is not encoded, send as-is */
 284         if (strlen(encoding) == 0) {
 285                 return 0;
 286         }
 287
 288
 289         /* Fail if we hit an unknown encoding. */
 290         if ((strcasecmp(encoding, "base64"))
 291             && (strcasecmp(encoding, "quoted-printable"))) {
 292                 return -1;
 293         }
 294
 295         /*
 296          * Allocate a buffer for the decoded data.  The output buffer is slightly
 297          * larger than the input buffer; this assumes that the decoded data
 298          * will never be significantly larger than the encoded data.  This is a
 299          * safe assumption with base64, uuencode, and quoted-printable.
 300          */
 301         *decoded = malloc(length + 32768);
 302         if (decoded == NULL) {
 303                 return -1;
 304         }
 305
 306         if (!strcasecmp(encoding, "base64")) {
 307                 *bytes_decoded = CtdlDecodeBase64(*decoded, part_start, length);
 308                 return 1;
 309         }
 310         else if (!strcasecmp(encoding, "quoted-printable")) {
 311                 *bytes_decoded = CtdlDecodeQuotedPrintable(*decoded, part_start, length);
 312                 return 1;
 313         }
 314         return -1;
 315 }
 316
 317 typedef enum _eIntMimeHdrs {
 318         boundary,
 319         startary,
 320         endary,
 321         content_type,
 322         charset,
 323         encoding,
 324         content_type_name,
 325         content_disposition_name,
 326         filename,
 327         disposition,
 328         id,
 329         eMax /* don't move ! */
 330 } eIntMimeHdrs;
 331
 332 typedef struct _CBufStr {
 333         char Key[SIZ];
 334         long len;
 335 }CBufStr;
 336
 337 typedef struct _interesting_mime_headers {
 338         CBufStr b[eMax];
 339         long content_length;
 340         long is_multipart;
 341 } interesting_mime_headers;
 342
 343
 344 static void FlushInterestingMimes(interesting_mime_headers *m)
 345 {
 346         int i;
 347
 348         for (i = 0; i < eMax; i++) {
 349              m->b[i].Key[0] = '\0';
 350              m->b[i].len = 0;
 351         }
 352         m->content_length = -1;
 353 }
 354 static interesting_mime_headers *InitInterestingMimes(void)
 355 {
 356         interesting_mime_headers *m;
 357         m = (interesting_mime_headers*) malloc( sizeof(interesting_mime_headers));
 358
 359         FlushInterestingMimes(m);
 360
 361         return m;
 362 }
 363
 364
 365 static long parse_MimeHeaders(interesting_mime_headers *m,
 366                               char** pcontent_start,
 367                               char *content_end)
 368 {
 369         char buf[SIZ];
 370         char header[SIZ];
 371         long headerlen;
 372         char *ptr, *pch;
 373         int buflen = 0;
 374         int i;
 375
 376         /* Learn interesting things from the headers */
 377         ptr = *pcontent_start;
 378         *header = '\0';
 379         headerlen = 0;
 380         do {
 381                 ptr = memreadlinelen(ptr, buf, SIZ, &buflen);
 382
 383                 for (i = 0; i < buflen; ++i) {
 384                         if (isspace(buf[i])) {
 385                                 buf[i] = ' ';
 386                         }
 387                 }
 388
 389                 if (!isspace(buf[0]) && (headerlen > 0)) {
 390                         if (!strncasecmp(header, "Content-type:", 13)) {
 391                                 memcpy (m->b[content_type].Key, &header[13], headerlen - 12);
 392                                 m->b[content_type].Key[headerlen - 12] = '\0';
 393                                 m->b[content_type].len = striplt (m->b[content_type].Key);
 394
 395                                 m->b[content_type_name].len = extract_key(m->b[content_type_name].Key, CKEY(m->b[content_type]), HKEY("name"), '=');
 396                                 m->b[charset].len           = extract_key(m->b[charset].Key,           CKEY(m->b[content_type]), HKEY("charset"), '=');
 397                                 m->b[boundary].len          = extract_key(m->b[boundary].Key,          header,       headerlen,  HKEY("boundary"), '=');
 398
 399                                 /* Deal with weird headers */
 400                                 pch = strchr(m->b[content_type].Key, ' ');
 401                                 if (pch != NULL) {
 402                                         *pch = '\0';
 403                                         m->b[content_type].len = m->b[content_type].Key - pch;
 404                                 }
 405                                 pch = strchr(m->b[content_type].Key, ';');
 406                                 if (pch != NULL) {
 407                                         *pch = '\0';
 408                                         m->b[content_type].len = m->b[content_type].Key - pch;
 409                                 }
 410                         }
 411                         else if (!strncasecmp(header, "Content-Disposition:", 20)) {
 412                                 memcpy (m->b[disposition].Key, &header[20], headerlen - 19);
 413                                 m->b[disposition].Key[headerlen - 19] = '\0';
 414                                 m->b[disposition].len = striplt(m->b[disposition].Key);
 415
 416                                 m->b[content_disposition_name].len = extract_key(m->b[content_disposition_name].Key, CKEY(m->b[disposition]), HKEY("name"), '=');
 417                                 m->b[filename].len                 = extract_key(m->b[filename].Key,                 CKEY(m->b[disposition]), HKEY("filename"), '=');
 418                                 pch = strchr(m->b[disposition].Key, ';');
 419                                 if (pch != NULL) *pch = '\0';
 420                                 m->b[disposition].len = striplt(m->b[disposition].Key);
 421                         }
 422                         else if (!strncasecmp(header, "Content-ID:", 11)) {
 423                                 memcpy(m->b[id].Key, &header[11], headerlen - 11);
 424                                 m->b[id].Key[headerlen - 11] = '\0';
 425                                 striplt(m->b[id].Key);
 426                                 m->b[id].len = stripallbut(m->b[id].Key, '<', '>');
 427                         }
 428                         else if (!strncasecmp(header, "Content-length: ", 15)) {
 429                                 char *clbuf;
 430                                 clbuf = &header[15];
 431                                 while (isspace(*clbuf))
 432                                         clbuf ++;
 433                                 m->content_length = (size_t) atol(clbuf);
 434                         }
 435                         else if (!strncasecmp(header, "Content-transfer-encoding: ", 26)) {
 436                                 memcpy(m->b[encoding].Key, &header[26], headerlen - 26);
 437                                 m->b[encoding].Key[headerlen - 26] = '\0';
 438                                 m->b[encoding].len = striplt(m->b[encoding].Key);
 439                         }
 440                         *header = '\0';
 441                         headerlen = 0;
 442                 }
 443                 if ((headerlen + buflen + 2) < SIZ) {
 444                         memcpy(&header[headerlen], buf, buflen);
 445                         headerlen += buflen;
 446                         header[headerlen] = '\0';
 447                 }
 448                 if (ptr >= content_end) {
 449                         return -1;
 450                 }
 451         } while ((!IsEmptyStr(buf)) && (*ptr != 0));
 452
 453         m->is_multipart = m->b[boundary].len != 0;
 454         *pcontent_start = ptr;
 455
 456         return 0;
 457 }
 458
 459
 460 static int IsAsciiEncoding(interesting_mime_headers *m)
 461 {
 462
 463         if ((m->b[encoding].len != 0) &&
 464             (strcasecmp(m->b[encoding].Key, "base64") == 0))
 465                 return 1;
 466         if ((m->b[encoding].len != 0) &&
 467             (strcmp(m->b[encoding].Key, "quoted-printable") == 0))
 468                 return 1;
 469
 470         return 0;
 471 }
 472
 473 static char *FindNextContent(char *ptr,
 474                              char *content_end,
 475                              interesting_mime_headers *SubMimeHeaders,
 476                              interesting_mime_headers *m)
 477 {
 478         char *next_boundary;
 479         char  tmp;
 480
 481         if (IsAsciiEncoding(SubMimeHeaders)) {
 482                 tmp = *content_end;
 483                 *content_end = '\0';
 484
 485                 /**
 486                  * ok, if we have a content length of the mime part,
 487                  * try skipping the content on the search for the next
 488                  * boundary. since we don't trust the content_length
 489                  * to be all accurate, and suspect it to lose one digit
 490                  * per line with a line length of 80 chars, we need
 491                  * to start searching a little before..
 492                  */
 493
 494                 if ((SubMimeHeaders->content_length != -1) &&
 495                     (SubMimeHeaders->content_length > 10))
 496                 {
 497                         char *pptr;
 498                         long lines;
 499
 500                         lines = SubMimeHeaders->content_length / 80;
 501                         pptr = ptr + SubMimeHeaders->content_length - lines - 10;
 502                         if (pptr < content_end)
 503                                 ptr = pptr;
 504                 }
 505
 506                 next_boundary = strstr(ptr, m->b[startary].Key);
 507                 *content_end = tmp;
 508         }
 509         else {
 510                 char *srch;
 511                 /**
 512                  * ok, if we have a content length of the mime part,
 513                  * try skipping the content on the search for the next
 514                  * boundary. since we don't trust the content_length
 515                  * to be all accurate, start searching a little before..
 516                  */
 517
 518                 if ((SubMimeHeaders->content_length != -1) &&
 519                     (SubMimeHeaders->content_length > 10))
 520                 {
 521                         char *pptr;
 522                         pptr = ptr + SubMimeHeaders->content_length - 10;
 523                         if (pptr < content_end)
 524                                 ptr = pptr;
 525                 }
 526
 527
 528                 srch = next_boundary = NULL;
 529                 for (srch = memchr(ptr, '-',  content_end - ptr);
 530                      (srch != NULL) && (srch < content_end);
 531                      srch = memchr(srch, '-',  content_end - srch))
 532                 {
 533                         if (!memcmp(srch,
 534                                     m->b[startary].Key,
 535                                     m->b[startary].len))
 536                         {
 537                                 next_boundary = srch;
 538                                 srch = content_end;
 539                         }
 540                         else srch ++;
 541
 542                 }
 543
 544         }
 545         return next_boundary;
 546 }
 547
 548 /*
 549  * Break out the components of a multipart message
 550  * (This function expects to be fed HEADERS + CONTENT)
 551  * Note: NULL can be supplied as content_end; in this case, the message is
 552  * considered to have ended when the parser encounters a 0x00 byte.
 553  */
 554 static void recurseable_mime_parser(char *partnum,
 555                                     char *content_start, char *content_end,
 556                                     MimeParserCallBackType CallBack,
 557                                     MimeParserCallBackType PreMultiPartCallBack,
 558                                     MimeParserCallBackType PostMultiPartCallBack,
 559                                     void *userdata,
 560                                     int dont_decode,
 561                                     interesting_mime_headers *m)
 562 {
 563         interesting_mime_headers *SubMimeHeaders;
 564         char     *ptr;
 565         char     *part_start;
 566         char     *part_end = NULL;
 567         char     *evaluate_crlf_ptr = NULL;
 568         char     *next_boundary;
 569         char      nested_partnum[256];
 570         int       crlf_in_use = 0;
 571         int       part_seq = 0;
 572         CBufStr  *chosen_name;
 573
 574
 575         /* If this is a multipart message, then recursively process it */
 576         ptr = content_start;
 577         part_start = NULL;
 578         if (m->is_multipart) {
 579
 580                 /* Tell the client about this message's multipartedness */
 581                 if (PreMultiPartCallBack != NULL) {
 582                         PreMultiPartCallBack("",
 583                                              "",
 584                                              partnum,
 585                                              "",
 586                                              NULL,
 587                                              m->b[content_type].Key,
 588                                              m->b[charset].Key,
 589                                              0,
 590                                              m->b[encoding].Key,
 591                                              m->b[id].Key,
 592                                              userdata);
 593                 }
 594
 595                 /* Figure out where the boundaries are */
 596                 m->b[startary].len = snprintf(m->b[startary].Key, SIZ, "--%s", m->b[boundary].Key);
 597                 SubMimeHeaders = InitInterestingMimes ();
 598
 599                 while ((*ptr == '\r') || (*ptr == '\n')) ptr ++;
 600
 601                 if (strncmp(ptr, m->b[startary].Key, m->b[startary].len) == 0)
 602                         ptr += m->b[startary].len;
 603
 604                 while ((*ptr == '\r') || (*ptr == '\n')) ptr ++;
 605
 606                 part_start = NULL;
 607                 do {
 608                         char *optr;
 609
 610                         optr = ptr;
 611                         if (parse_MimeHeaders(SubMimeHeaders, &ptr, content_end) != 0)
 612                                 break;
 613                         if ((ptr - optr > 2) && (*(ptr - 2) == '\r')) {
 614                                 crlf_in_use = 1;
 615                         }
 616
 617                         part_start = ptr;
 618
 619                         next_boundary = FindNextContent(ptr, content_end, SubMimeHeaders, m);
 620                         if ((next_boundary != NULL) && (next_boundary - part_start < 3)) {
 621                                 FlushInterestingMimes(SubMimeHeaders);
 622                                 continue;
 623                         }
 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) == '-') &&
 658                                     (*(next_boundary + m->b[startary].len + 1) == '-') ){
 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') && (*(evaluate_crlf_ptr + 1) == '\n')) {
 669                                                 crlf_in_use = 1;
 670                                         }
 671                                         else {
 672                                                 crlf_in_use = 0;
 673                                         }
 674
 675                                         /* Advance past the LF ... now we're in the next part */
 676                                         ++part_start;
 677                                         ptr = part_start;
 678                                 }
 679                         }
 680                         else {
 681                                 /* Invalid end of multipart.  Bail out! */
 682                                 ptr = content_end;
 683                         }
 684                         FlushInterestingMimes(SubMimeHeaders);
 685                 } while ( (ptr < content_end) && (next_boundary != NULL) );
 686
 687                 free(SubMimeHeaders);
 688
 689                 if (PostMultiPartCallBack != NULL) {
 690                         PostMultiPartCallBack("",
 691                                               "",
 692                                               partnum,
 693                                               "",
 694                                               NULL,
 695                                               m->b[content_type].Key,
 696                                               m->b[charset].Key,
 697                                               0,
 698                                               m->b[encoding].Key,
 699                                               m->b[id].Key,
 700                                               userdata);
 701                 }
 702         } /* If it's not a multipart message, then do something with it */
 703         else {
 704                 size_t length;
 705                 part_start = ptr;
 706                 length = content_end - part_start;
 707                 ptr = part_end = content_end;
 708
 709                 /* The following code will truncate the MIME part to the size
 710                  * specified by the Content-length: header.   We have commented it
 711                  * out because these headers have a tendency to be wrong.
 712                  *
 713                  *      if ( (content_length > 0) && (length > content_length) ) {
 714                  *              length = content_length;
 715                  *      }
 716                  */
 717
 718                 /* Sometimes the "name" field is tacked on to Content-type,
 719                  * and sometimes it's tacked on to Content-disposition.  Use
 720                  * whichever one we have.
 721                  */
 722                 if (m->b[content_disposition_name].len > m->b[content_type_name].len) {
 723                         chosen_name = &m->b[content_disposition_name];
 724                 }
 725                 else {
 726                         chosen_name = &m->b[content_type_name];
 727                 }
 728
 729                 // Ok, we've got a non-multipart part here, so do something with it.
 730                 mime_decode(partnum,
 731                             part_start,
 732                             length,
 733                             m->b[content_type].Key,
 734                             m->b[charset].Key,
 735                             m->b[encoding].Key,
 736                             m->b[disposition].Key,
 737                             m->b[id].Key,
 738                             chosen_name->Key,
 739                             m->b[filename].Key,
 740                             CallBack,
 741                             NULL, NULL,
 742                             userdata,
 743                             dont_decode
 744                         );
 745
 746                 /*
 747                  * Now if it's an encapsulated message/rfc822 then we have to recurse into it
 748                  */
 749                 if (!strcasecmp(&m->b[content_type].Key[0], "message/rfc822")) {
 750
 751                         if (PreMultiPartCallBack != NULL) {
 752                                 PreMultiPartCallBack("",
 753                                                      "",
 754                                                      partnum,
 755                                                      "",
 756                                                      NULL,
 757                                                      m->b[content_type].Key,
 758                                                      m->b[charset].Key,
 759                                                      0,
 760                                                      m->b[encoding].Key,
 761                                                      m->b[id].Key,
 762                                                      userdata);
 763                         }
 764                         if (CallBack != NULL) {
 765                                 if (strlen(partnum) > 0) {
 766                                         snprintf(nested_partnum,
 767                                                  sizeof nested_partnum,
 768                                                  "%s.%d", partnum,
 769                                                  ++part_seq);
 770                                 }
 771                                 else {
 772                                         snprintf(nested_partnum,
 773                                                  sizeof nested_partnum,
 774                                                  "%d", ++part_seq);
 775                                 }
 776                                 the_mime_parser(nested_partnum,
 777                                                 part_start,
 778                                                 part_end,
 779                                                 CallBack,
 780                                                 PreMultiPartCallBack,
 781                                                 PostMultiPartCallBack,
 782                                                 userdata,
 783                                                 dont_decode
 784                                         );
 785                         }
 786                         if (PostMultiPartCallBack != NULL) {
 787                                 PostMultiPartCallBack("",
 788                                                       "",
 789                                                       partnum,
 790                                                       "",
 791                                                       NULL,
 792                                                       m->b[content_type].Key,
 793                                                       m->b[charset].Key,
 794                                                       0,
 795                                                       m->b[encoding].Key,
 796                                                       m->b[id].Key,
 797                                                       userdata);
 798                         }
 799
 800
 801                 }
 802
 803         }
 804
 805 }
 806
 807 /*
 808  * Break out the components of a multipart message
 809  * (This function expects to be fed HEADERS + CONTENT)
 810  * Note: NULL can be supplied as content_end; in this case, the message is
 811  * considered to have ended when the parser encounters a 0x00 byte.
 812  */
 813 void the_mime_parser(char *partnum,
 814                      char *content_start, char *content_end,
 815                      MimeParserCallBackType CallBack,
 816                      MimeParserCallBackType PreMultiPartCallBack,
 817                      MimeParserCallBackType PostMultiPartCallBack,
 818                      void *userdata,
 819                      int dont_decode)
 820 {
 821         interesting_mime_headers *m;
 822
 823         /* If the caller didn't supply an endpointer, generate one by measure */
 824         if (content_end == NULL) {
 825                 content_end = &content_start[strlen(content_start)];
 826         }
 827
 828         m = InitInterestingMimes();
 829
 830         if (!parse_MimeHeaders(m, &content_start, content_end))
 831         {
 832
 833                 recurseable_mime_parser(partnum,
 834                                         content_start, content_end,
 835                                         CallBack,
 836                                         PreMultiPartCallBack,
 837                                         PostMultiPartCallBack,
 838                                         userdata,
 839                                         dont_decode,
 840                                         m);
 841         }
 842         free(m);
 843 }
 844
 845
 846 /*
 847  * Entry point for the MIME parser.
 848  * (This function expects to be fed HEADERS + CONTENT)
 849  * Note: NULL can be supplied as content_end; in this case, the message is
 850  * considered to have ended when the parser encounters a 0x00 byte.
 851  */
 852 void mime_parser(char *content_start,
 853                  char *content_end,
 854                  MimeParserCallBackType CallBack,
 855                  MimeParserCallBackType PreMultiPartCallBack,
 856                  MimeParserCallBackType PostMultiPartCallBack,
 857                  void *userdata,
 858                  int dont_decode)
 859 {
 860
 861         the_mime_parser("", content_start, content_end,
 862                         CallBack,
 863                         PreMultiPartCallBack,
 864                         PostMultiPartCallBack,
 865                         userdata, dont_decode);
 866 }
 867
 868
 869
 870
 871
 872
 873 typedef struct _MimeGuess {
 874         const char *Pattern;
 875         size_t PatternLen;
 876         long PatternOffset;
 877         const char *MimeString;
 878 } MimeGuess;
 879
 880 MimeGuess MyMimes [] = {
 881         {
 882                 "GIF",
 883                 3,
 884                 0,
 885                 "image/gif"
 886         },
 887         {
 888                 "\xff\xd8",
 889                 2,
 890                 0,
 891                 "image/jpeg"
 892         },
 893         {
 894                 "\x89PNG",
 895                 4,
 896                 0,
 897                 "image/png"
 898         },
 899         { // last...
 900                 "",
 901                 0,
 902                 0,
 903                 ""
 904         }
 905 };
 906
 907
 908 const char *GuessMimeType(const char *data, size_t dlen)
 909 {
 910         int MimeIndex = 0;
 911
 912         while (MyMimes[MimeIndex].PatternLen != 0)
 913         {
 914                 if ((MyMimes[MimeIndex].PatternLen +
 915                      MyMimes[MimeIndex].PatternOffset < dlen) &&
 916                     strncmp(MyMimes[MimeIndex].Pattern,
 917                             &data[MyMimes[MimeIndex].PatternOffset],
 918                             MyMimes[MimeIndex].PatternLen) == 0)
 919                 {
 920                         return MyMimes[MimeIndex].MimeString;
 921                 }
 922                 MimeIndex ++;
 923         }
 924         /*
 925          * ok, our simple minded algorythm didn't find anything,
 926          * let the big chegger try it, he wil default to application/octet-stream
 927          */
 928         return (xdg_mime_get_mime_type_for_data(data, dlen));
 929 }
 930
 931
 932 const char* GuessMimeByFilename(const char *what, size_t len)
 933 {
 934         /* we know some hardcoded on our own, try them... */
 935         if ((len > 3) && !strncasecmp(&what[len - 4], ".gif", 4))
 936                 return "image/gif";
 937         else if ((len > 2) && !strncasecmp(&what[len - 3], ".js", 3))
 938                 return  "text/javascript";
 939         else if ((len > 3) && !strncasecmp(&what[len - 4], ".txt", 4))
 940                 return "text/plain";
 941         else if ((len > 3) && !strncasecmp(&what[len - 4], ".css", 4))
 942                 return "text/css";
 943         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htc", 4))
 944                 return "text/x-component";
 945         else if ((len > 3) && !strncasecmp(&what[len - 4], ".jpg", 4))
 946                 return "image/jpeg";
 947         else if ((len > 4) && !strncasecmp(&what[len - 5], ".jpeg", 5))
 948                 return "image/jpeg";
 949         else if ((len > 3) && !strncasecmp(&what[len - 4], ".png", 4))
 950                 return "image/png";
 951         else if ((len > 3) && !strncasecmp(&what[len - 4], ".ico", 4))
 952                 return "image/x-icon";
 953         else if ((len > 3) && !strncasecmp(&what[len - 4], ".vcf", 4))
 954                 return "text/x-vcard";
 955         else if ((len > 4) && !strncasecmp(&what[len - 5], ".html", 5))
 956                 return "text/html";
 957         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htm", 4))
 958                 return "text/html";
 959         else if ((len > 3) && !strncasecmp(&what[len - 4], ".wml", 4))
 960                 return "text/vnd.wap.wml";
 961         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmls", 5))
 962                 return "text/vnd.wap.wmlscript";
 963         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmlc", 5))
 964                 return "application/vnd.wap.wmlc";
 965         else if ((len > 5) && !strncasecmp(&what[len - 6], ".wmlsc", 6))
 966                 return "application/vnd.wap.wmlscriptc";
 967         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wbmp", 5))
 968                 return "image/vnd.wap.wbmp";
 969         else
 970                 /* and let xdgmime do the fallback. */
 971                 return xdg_mime_get_mime_type_from_file_name(what);
 972 }
 973
 974 static HashList *IconHash = NULL;
 975
 976 typedef struct IconName IconName;
 977
 978 struct IconName {
 979         char *FlatName;
 980         char *FileName;
 981 };
 982
 983 static void DeleteIcon(void *IconNamePtr)
 984 {
 985         IconName *Icon = (IconName*) IconNamePtr;
 986         free(Icon->FlatName);
 987         free(Icon->FileName);
 988         free(Icon);
 989 }
 990
 991 /*
 992 static const char *PrintFlat(void *IconNamePtr)
 993 {
 994         IconName *Icon = (IconName*) IconNamePtr;
 995         return Icon->FlatName;
 996 }
 997 static const char *PrintFile(void *IconNamePtr)
 998 {
 999         IconName *Icon = (IconName*) IconNamePtr;
1000         return Icon->FileName;
1001 }
1002 */
1003
1004 #define GENSTR "x-generic"
1005 #define IGNORE_PREFIX_1 "gnome-mime"
1006 int LoadIconDir(const char *DirName)
1007 {
1008         DIR *filedir = NULL;
1009         struct dirent *filedir_entry;
1010         int d_namelen;
1011         int d_without_ext;
1012         IconName *Icon;
1013
1014         filedir = opendir (DirName);
1015         IconHash = NewHash(1, NULL);
1016         if (filedir == NULL) {
1017                 return 0;
1018         }
1019
1020         while ((filedir_entry = readdir(filedir)))
1021         {
1022                 char *MinorPtr;
1023                 char *PStart;
1024 #ifdef _DIRENT_HAVE_D_NAMLEN
1025                 d_namelen = filedir_entry->d_namlen;
1026 #else
1027                 d_namelen = strlen(filedir_entry->d_name);
1028 #endif
1029                 d_without_ext = d_namelen;
1030                 while ((d_without_ext > 0) && (filedir_entry->d_name[d_without_ext] != '.'))
1031                         d_without_ext --;
1032                 if ((d_without_ext == 0) || (d_namelen < 3))
1033                         continue;
1034
1035                 if ((sizeof(IGNORE_PREFIX_1) < d_namelen) &&
1036                     (strncmp(IGNORE_PREFIX_1,
1037                              filedir_entry->d_name,
1038                              sizeof(IGNORE_PREFIX_1) - 1) == 0)) {
1039                         PStart = filedir_entry->d_name + sizeof(IGNORE_PREFIX_1);
1040                         d_without_ext -= sizeof(IGNORE_PREFIX_1);
1041                 }
1042                 else {
1043                         PStart = filedir_entry->d_name;
1044                 }
1045                 Icon = malloc(sizeof(IconName));
1046
1047                 Icon->FileName = malloc(d_namelen + 1);
1048                 memcpy(Icon->FileName, filedir_entry->d_name, d_namelen + 1);
1049
1050                 Icon->FlatName = malloc(d_without_ext + 1);
1051                 memcpy(Icon->FlatName, PStart, d_without_ext);
1052                 Icon->FlatName[d_without_ext] = '\0';
1053                 /* Try to find Minor type in image-jpeg */
1054                 MinorPtr = strchr(Icon->FlatName, '-');
1055                 if (MinorPtr != NULL) {
1056                         size_t MinorLen;
1057                         MinorLen = 1 + d_without_ext - (MinorPtr - Icon->FlatName + 1);
1058                         if ((MinorLen == sizeof(GENSTR)) &&
1059                             (strncmp(MinorPtr + 1, GENSTR, sizeof(GENSTR)) == 0)) {
1060                                 /* ok, we found a generic filename. cut the generic. */
1061                                 *MinorPtr = '\0';
1062                                 d_without_ext = d_without_ext - (MinorPtr - Icon->FlatName);
1063                         }
1064                         else { /* Map the major / minor separator to / */
1065                                 *MinorPtr = '/';
1066                         }
1067                 }
1068
1069 //              PrintHash(IconHash, PrintFlat, PrintFile);
1070 //              printf("%s - %s\n", Icon->FlatName, Icon->FileName);
1071                 Put(IconHash, Icon->FlatName, d_without_ext, Icon, DeleteIcon);
1072 //              PrintHash(IconHash, PrintFlat, PrintFile);
1073         }
1074         closedir(filedir);
1075         return 1;
1076 }
1077
1078 const char *GetIconFilename(char *MimeType, size_t len)
1079 {
1080         void *vIcon;
1081         IconName *Icon;
1082
1083         if(IconHash == NULL)
1084                 return NULL;
1085
1086         GetHash(IconHash, MimeType, len, &vIcon), Icon = (IconName*) vIcon;
1087         /* didn't find the exact mimetype? try major only. */
1088         if (Icon == NULL) {
1089                 char * pMinor;
1090                 pMinor = strchr(MimeType, '/');
1091                 if (pMinor != NULL) {
1092                         *pMinor = '\0';
1093                         GetHash(IconHash, MimeType, pMinor - MimeType, &vIcon),
1094                                 Icon = (IconName*) vIcon;
1095                 }
1096         }
1097         if (Icon == NULL) {
1098                 return NULL;
1099         }
1100
1101         /*printf("Getting: [%s] == [%s] -> [%s]\n", MimeType, Icon->FlatName, Icon->FileName);*/
1102         return Icon->FileName;
1103 }
1104
1105 void ShutDownLibCitadelMime(void)
1106 {
1107         DeleteHash(&IconHash);
1108 }