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].Key[headerlen - 12] = '\0';
 410                                 m->b[content_type].len = striplt (m->b[content_type].Key);
 411
 412                                 m->b[content_type_name].len = extract_key(m->b[content_type_name].Key, CKEY(m->b[content_type]), HKEY("name"), '=');
 413                                 m->b[charset].len           = extract_key(m->b[charset].Key,           CKEY(m->b[content_type]), HKEY("charset"), '=');
 414                                 m->b[boundary].len          = extract_key(m->b[boundary].Key,          header,       headerlen,  HKEY("boundary"), '=');
 415
 416                                 /* Deal with weird headers */
 417                                 pch = strchr(m->b[content_type].Key, ' ');
 418                                 if (pch != NULL) {
 419                                         *pch = '\0';
 420                                         m->b[content_type].len = m->b[content_type].Key - pch;
 421                                 }
 422                                 pch = strchr(m->b[content_type].Key, ';');
 423                                 if (pch != NULL) {
 424                                         *pch = '\0';
 425                                         m->b[content_type].len = m->b[content_type].Key - pch;
 426                                 }
 427                         }
 428                         else if (!strncasecmp(header, "Content-Disposition:", 20)) {
 429                                 memcpy (m->b[disposition].Key, &header[20], headerlen - 19);
 430                                 m->b[disposition].Key[headerlen - 19] = '\0';
 431                                 m->b[disposition].len = striplt(m->b[disposition].Key);
 432
 433                                 m->b[content_disposition_name].len = extract_key(m->b[content_disposition_name].Key, CKEY(m->b[disposition]), HKEY("name"), '=');
 434                                 m->b[filename].len                 = extract_key(m->b[filename].Key,                 CKEY(m->b[disposition]), HKEY("filename"), '=');
 435                                 pch = strchr(m->b[disposition].Key, ';');
 436                                 if (pch != NULL) *pch = '\0';
 437                                 m->b[disposition].len = striplt(m->b[disposition].Key);
 438                         }
 439                         else if (!strncasecmp(header, "Content-ID:", 11)) {
 440                                 memcpy(m->b[id].Key, &header[11], headerlen - 11);
 441                                 m->b[id].Key[headerlen - 11] = '\0';
 442                                 striplt(m->b[id].Key);
 443                                 m->b[id].len = stripallbut(m->b[id].Key, '<', '>');
 444                         }
 445                         else if (!strncasecmp(header, "Content-length: ", 15)) {
 446                                 char *clbuf;
 447                                 clbuf = &header[15];
 448                                 while (isspace(*clbuf))
 449                                         clbuf ++;
 450                                 m->content_length = (size_t) atol(clbuf);
 451                         }
 452                         else if (!strncasecmp(header, "Content-transfer-encoding: ", 26)) {
 453                                 memcpy(m->b[encoding].Key, &header[26], headerlen - 26);
 454                                 m->b[encoding].Key[headerlen - 26] = '\0';
 455                                 m->b[encoding].len = striplt(m->b[encoding].Key);
 456                         }
 457                         *header = '\0';
 458                         headerlen = 0;
 459                 }
 460                 if ((headerlen + buflen + 2) < SIZ) {
 461                         memcpy(&header[headerlen], buf, buflen);
 462                         headerlen += buflen;
 463                         header[headerlen] = '\0';
 464                 }
 465                 if (ptr >= content_end) {
 466                         return -1;
 467                 }
 468         } while ((!IsEmptyStr(buf)) && (*ptr != 0));
 469
 470         m->is_multipart = m->b[boundary].len != 0;
 471         *pcontent_start = ptr;
 472
 473         return 0;
 474 }
 475
 476
 477 static int IsAsciiEncoding(interesting_mime_headers *m)
 478 {
 479
 480         if ((m->b[encoding].len != 0) &&
 481             (strcasecmp(m->b[encoding].Key, "base64") == 0))
 482                 return 1;
 483         if ((m->b[encoding].len != 0) &&
 484             (strcmp(m->b[encoding].Key, "quoted-printable") == 0))
 485                 return 1;
 486
 487         return 0;
 488 }
 489
 490 static char *FindNextContent(char *ptr,
 491                              char *content_end,
 492                              interesting_mime_headers *SubMimeHeaders,
 493                              interesting_mime_headers *m)
 494 {
 495         char *next_boundary;
 496         char  tmp;
 497
 498         if (IsAsciiEncoding(SubMimeHeaders)) {
 499                 tmp = *content_end;
 500                 *content_end = '\0';
 501
 502                 /**
 503                  * ok, if we have a content length of the mime part,
 504                  * try skipping the content on the search for the next
 505                  * boundary. since we don't trust the content_length
 506                  * to be all accurate, and suspect it to lose one digit
 507                  * per line with a line length of 80 chars, we need
 508                  * to start searching a little before..
 509                  */
 510
 511                 if ((SubMimeHeaders->content_length != -1) &&
 512                     (SubMimeHeaders->content_length > 10))
 513                 {
 514                         char *pptr;
 515                         long lines;
 516
 517                         lines = SubMimeHeaders->content_length / 80;
 518                         pptr = ptr + SubMimeHeaders->content_length - lines - 10;
 519                         if (pptr < content_end)
 520                                 ptr = pptr;
 521                 }
 522
 523                 next_boundary = strstr(ptr, m->b[startary].Key);
 524                 *content_end = tmp;
 525         }
 526         else {
 527                 char *srch;
 528                 /**
 529                  * ok, if we have a content length of the mime part,
 530                  * try skipping the content on the search for the next
 531                  * boundary. since we don't trust the content_length
 532                  * to be all accurate, start searching a little before..
 533                  */
 534
 535                 if ((SubMimeHeaders->content_length != -1) &&
 536                     (SubMimeHeaders->content_length > 10))
 537                 {
 538                         char *pptr;
 539                         pptr = ptr + SubMimeHeaders->content_length - 10;
 540                         if (pptr < content_end)
 541                                 ptr = pptr;
 542                 }
 543
 544
 545                 srch = next_boundary = NULL;
 546                 for (srch = memchr(ptr, '-',  content_end - ptr);
 547                      (srch != NULL) && (srch < content_end);
 548                      srch = memchr(srch, '-',  content_end - srch))
 549                 {
 550                         if (!memcmp(srch,
 551                                     m->b[startary].Key,
 552                                     m->b[startary].len))
 553                         {
 554                                 next_boundary = srch;
 555                                 srch = content_end;
 556                         }
 557                         else srch ++;
 558
 559                 }
 560
 561         }
 562         return next_boundary;
 563 }
 564
 565 /*
 566  * Break out the components of a multipart message
 567  * (This function expects to be fed HEADERS + CONTENT)
 568  * Note: NULL can be supplied as content_end; in this case, the message is
 569  * considered to have ended when the parser encounters a 0x00 byte.
 570  */
 571 static void recurseable_mime_parser(char *partnum,
 572                                     char *content_start, char *content_end,
 573                                     MimeParserCallBackType CallBack,
 574                                     MimeParserCallBackType PreMultiPartCallBack,
 575                                     MimeParserCallBackType PostMultiPartCallBack,
 576                                     void *userdata,
 577                                     int dont_decode,
 578                                     interesting_mime_headers *m)
 579 {
 580         interesting_mime_headers *SubMimeHeaders;
 581         char     *ptr;
 582         char     *part_start;
 583         char     *part_end = NULL;
 584         char     *evaluate_crlf_ptr = NULL;
 585         char     *next_boundary;
 586         char      nested_partnum[256];
 587         int       crlf_in_use = 0;
 588         int       part_seq = 0;
 589         CBufStr  *chosen_name;
 590
 591
 592         /* If this is a multipart message, then recursively process it */
 593         ptr = content_start;
 594         part_start = NULL;
 595         if (m->is_multipart) {
 596
 597                 /* Tell the client about this message's multipartedness */
 598                 if (PreMultiPartCallBack != NULL) {
 599                         PreMultiPartCallBack("",
 600                                              "",
 601                                              partnum,
 602                                              "",
 603                                              NULL,
 604                                              m->b[content_type].Key,
 605                                              m->b[charset].Key,
 606                                              0,
 607                                              m->b[encoding].Key,
 608                                              m->b[id].Key,
 609                                              userdata);
 610                 }
 611
 612                 /* Figure out where the boundaries are */
 613                 m->b[startary].len = snprintf(m->b[startary].Key, SIZ, "--%s", m->b[boundary].Key);
 614                 SubMimeHeaders = InitInterestingMimes ();
 615                 if (*ptr == '\r')
 616                         ptr ++;
 617                 if (*ptr == '\n')
 618                         ptr ++;
 619                 if (strncmp(ptr, m->b[startary].Key, m->b[startary].len) == 0)
 620                         ptr += m->b[startary].len;
 621                 if (*ptr == '\r')
 622                         ptr ++;
 623                 if (*ptr == '\n')
 624                         ptr ++;
 625                 part_start = NULL;
 626                 do {
 627                         char *optr;
 628
 629                         optr = ptr;
 630                         if (parse_MimeHeaders(SubMimeHeaders, &ptr, content_end) != 0)
 631                                 break;
 632                         if ((ptr - optr > 2) &&
 633                             (*(ptr - 2) == '\r'))
 634                                 crlf_in_use = 1;
 635
 636                         part_start = ptr;
 637
 638                         next_boundary = FindNextContent(ptr,
 639                                                         content_end,
 640                                                         SubMimeHeaders,
 641                                                         m);
 642                         if ((next_boundary != NULL) &&
 643                             (next_boundary - part_start < 3)) {
 644                                 FlushInterestingMimes(SubMimeHeaders);
 645
 646                                 continue;
 647                         }
 648
 649                         if ( (part_start != NULL) && (next_boundary != NULL) ) {
 650                                 part_end = next_boundary;
 651                                 --part_end;             /* omit the trailing LF */
 652                                 if (crlf_in_use) {
 653                                         --part_end;     /* omit the trailing CR */
 654                                 }
 655
 656                                 if (!IsEmptyStr(partnum)) {
 657                                         snprintf(nested_partnum,
 658                                                  sizeof nested_partnum,
 659                                                  "%s.%d", partnum,
 660                                                  ++part_seq);
 661                                 }
 662                                 else {
 663                                         snprintf(nested_partnum,
 664                                                  sizeof nested_partnum,
 665                                                  "%d", ++part_seq);
 666                                 }
 667                                 recurseable_mime_parser(nested_partnum,
 668                                                         part_start,
 669                                                         part_end,
 670                                                         CallBack,
 671                                                         PreMultiPartCallBack,
 672                                                         PostMultiPartCallBack,
 673                                                         userdata,
 674                                                         dont_decode,
 675                                                         SubMimeHeaders);
 676                         }
 677
 678                         if (next_boundary != NULL) {
 679                                 /* If we pass out of scope, don't attempt to
 680                                  * read past the end boundary. */
 681                                 if ((*(next_boundary + m->b[startary].len + 1) == '-') &&
 682                                     (*(next_boundary + m->b[startary].len + 2) == '-') ){
 683                                         ptr = content_end;
 684                                 }
 685                                 else {
 686                                         /* Set up for the next part. */
 687                                         part_start = strstr(next_boundary, "\n");
 688
 689                                         /* Determine whether newlines are LF or CRLF */
 690                                         evaluate_crlf_ptr = part_start;
 691                                         --evaluate_crlf_ptr;
 692                                         if ((*evaluate_crlf_ptr == '\r') &&
 693                                             (*(evaluate_crlf_ptr + 1) == '\n'))
 694                                         {
 695                                                 crlf_in_use = 1;
 696                                         }
 697                                         else {
 698                                                 crlf_in_use = 0;
 699                                         }
 700
 701                                         /* Advance past the LF ... now we're in the next part */
 702                                         ++part_start;
 703                                         ptr = part_start;
 704                                 }
 705                         }
 706                         else {
 707                                 /* Invalid end of multipart.  Bail out! */
 708                                 ptr = content_end;
 709                         }
 710                         FlushInterestingMimes(SubMimeHeaders);
 711                 } while ( (ptr < content_end) && (next_boundary != NULL) );
 712
 713                 free(SubMimeHeaders);
 714
 715                 if (PostMultiPartCallBack != NULL) {
 716                         PostMultiPartCallBack("",
 717                                               "",
 718                                               partnum,
 719                                               "",
 720                                               NULL,
 721                                               m->b[content_type].Key,
 722                                               m->b[charset].Key,
 723                                               0,
 724                                               m->b[encoding].Key,
 725                                               m->b[id].Key,
 726                                               userdata);
 727                 }
 728         } /* If it's not a multipart message, then do something with it */
 729         else {
 730                 size_t length;
 731                 part_start = ptr;
 732                 length = content_end - part_start;
 733                 ptr = part_end = content_end;
 734
 735
 736                 /* The following code will truncate the MIME part to the size
 737                  * specified by the Content-length: header.   We have commented it
 738                  * out because these headers have a tendency to be wrong.
 739                  *
 740                  *      if ( (content_length > 0) && (length > content_length) ) {
 741                  *              length = content_length;
 742                  *      }
 743                  */
 744
 745                 /* Sometimes the "name" field is tacked on to Content-type,
 746                  * and sometimes it's tacked on to Content-disposition.  Use
 747                  * whichever one we have.
 748                  */
 749                 if (m->b[content_disposition_name].len > m->b[content_type_name].len) {
 750                         chosen_name = &m->b[content_disposition_name];
 751                 }
 752                 else {
 753                         chosen_name = &m->b[content_type_name];
 754                 }
 755
 756                 /* Ok, we've got a non-multipart part here, so do something with it.
 757                  */
 758                 mime_decode(partnum,
 759                             part_start,
 760                             length,
 761                             m->b[content_type].Key,
 762                             m->b[charset].Key,
 763                             m->b[encoding].Key,
 764                             m->b[disposition].Key,
 765                             m->b[id].Key,
 766                             chosen_name->Key,
 767                             m->b[filename].Key,
 768                             CallBack,
 769                             NULL, NULL,
 770                             userdata,
 771                             dont_decode
 772                         );
 773
 774                 /*
 775                  * Now if it's an encapsulated message/rfc822 then we have to recurse into it
 776                  */
 777                 if (!strcasecmp(&m->b[content_type].Key[0], "message/rfc822")) {
 778
 779                         if (PreMultiPartCallBack != NULL) {
 780                                 PreMultiPartCallBack("",
 781                                                      "",
 782                                                      partnum,
 783                                                      "",
 784                                                      NULL,
 785                                                      m->b[content_type].Key,
 786                                                      m->b[charset].Key,
 787                                                      0,
 788                                                      m->b[encoding].Key,
 789                                                      m->b[id].Key,
 790                                                      userdata);
 791                         }
 792                         if (CallBack != NULL) {
 793                                 if (strlen(partnum) > 0) {
 794                                         snprintf(nested_partnum,
 795                                                  sizeof nested_partnum,
 796                                                  "%s.%d", partnum,
 797                                                  ++part_seq);
 798                                 }
 799                                 else {
 800                                         snprintf(nested_partnum,
 801                                                  sizeof nested_partnum,
 802                                                  "%d", ++part_seq);
 803                                 }
 804                                 the_mime_parser(nested_partnum,
 805                                                 part_start,
 806                                                 part_end,
 807                                                 CallBack,
 808                                                 PreMultiPartCallBack,
 809                                                 PostMultiPartCallBack,
 810                                                 userdata,
 811                                                 dont_decode
 812                                         );
 813                         }
 814                         if (PostMultiPartCallBack != NULL) {
 815                                 PostMultiPartCallBack("",
 816                                                       "",
 817                                                       partnum,
 818                                                       "",
 819                                                       NULL,
 820                                                       m->b[content_type].Key,
 821                                                       m->b[charset].Key,
 822                                                       0,
 823                                                       m->b[encoding].Key,
 824                                                       m->b[id].Key,
 825                                                       userdata);
 826                         }
 827
 828
 829                 }
 830
 831         }
 832
 833 }
 834
 835 /*
 836  * Break out the components of a multipart message
 837  * (This function expects to be fed HEADERS + CONTENT)
 838  * Note: NULL can be supplied as content_end; in this case, the message is
 839  * considered to have ended when the parser encounters a 0x00 byte.
 840  */
 841 void the_mime_parser(char *partnum,
 842                      char *content_start, char *content_end,
 843                      MimeParserCallBackType CallBack,
 844                      MimeParserCallBackType PreMultiPartCallBack,
 845                      MimeParserCallBackType PostMultiPartCallBack,
 846                      void *userdata,
 847                      int dont_decode)
 848 {
 849         interesting_mime_headers *m;
 850
 851         /* If the caller didn't supply an endpointer, generate one by measure */
 852         if (content_end == NULL) {
 853                 content_end = &content_start[strlen(content_start)];
 854         }
 855
 856         m = InitInterestingMimes();
 857
 858         if (!parse_MimeHeaders(m, &content_start, content_end))
 859         {
 860
 861                 recurseable_mime_parser(partnum,
 862                                         content_start, content_end,
 863                                         CallBack,
 864                                         PreMultiPartCallBack,
 865                                         PostMultiPartCallBack,
 866                                         userdata,
 867                                         dont_decode,
 868                                         m);
 869         }
 870         free(m);
 871 }
 872
 873 /*
 874  * Entry point for the MIME parser.
 875  * (This function expects to be fed HEADERS + CONTENT)
 876  * Note: NULL can be supplied as content_end; in this case, the message is
 877  * considered to have ended when the parser encounters a 0x00 byte.
 878  */
 879 void mime_parser(char *content_start,
 880                  char *content_end,
 881                  MimeParserCallBackType CallBack,
 882                  MimeParserCallBackType PreMultiPartCallBack,
 883                  MimeParserCallBackType PostMultiPartCallBack,
 884                  void *userdata,
 885                  int dont_decode)
 886 {
 887
 888         the_mime_parser("", content_start, content_end,
 889                         CallBack,
 890                         PreMultiPartCallBack,
 891                         PostMultiPartCallBack,
 892                         userdata, dont_decode);
 893 }
 894
 895
 896
 897
 898
 899
 900 typedef struct _MimeGuess {
 901         const char *Pattern;
 902         size_t PatternLen;
 903         long PatternOffset;
 904         const char *MimeString;
 905 } MimeGuess;
 906
 907 MimeGuess MyMimes [] = {
 908         {
 909                 "GIF",
 910                 3,
 911                 0,
 912                 "image/gif"
 913         },
 914         {
 915                 "\xff\xd8",
 916                 2,
 917                 0,
 918                 "image/jpeg"
 919         },
 920         {
 921                 "\x89PNG",
 922                 4,
 923                 0,
 924                 "image/png"
 925         },
 926         { // last...
 927                 "",
 928                 0,
 929                 0,
 930                 ""
 931         }
 932 };
 933
 934
 935 const char *GuessMimeType(const char *data, size_t dlen)
 936 {
 937         int MimeIndex = 0;
 938
 939         while (MyMimes[MimeIndex].PatternLen != 0)
 940         {
 941                 if ((MyMimes[MimeIndex].PatternLen +
 942                      MyMimes[MimeIndex].PatternOffset < dlen) &&
 943                     strncmp(MyMimes[MimeIndex].Pattern,
 944                             &data[MyMimes[MimeIndex].PatternOffset],
 945                             MyMimes[MimeIndex].PatternLen) == 0)
 946                 {
 947                         return MyMimes[MimeIndex].MimeString;
 948                 }
 949                 MimeIndex ++;
 950         }
 951         /*
 952          * ok, our simple minded algorythm didn't find anything,
 953          * let the big chegger try it, he wil default to application/octet-stream
 954          */
 955         return (xdg_mime_get_mime_type_for_data(data, dlen));
 956 }
 957
 958
 959 const char* GuessMimeByFilename(const char *what, size_t len)
 960 {
 961         /* we know some hardcoded on our own, try them... */
 962         if ((len > 3) && !strncasecmp(&what[len - 4], ".gif", 4))
 963                 return "image/gif";
 964         else if ((len > 2) && !strncasecmp(&what[len - 3], ".js", 3))
 965                 return  "text/javascript";
 966         else if ((len > 3) && !strncasecmp(&what[len - 4], ".txt", 4))
 967                 return "text/plain";
 968         else if ((len > 3) && !strncasecmp(&what[len - 4], ".css", 4))
 969                 return "text/css";
 970         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htc", 4))
 971                 return "text/x-component";
 972         else if ((len > 3) && !strncasecmp(&what[len - 4], ".jpg", 4))
 973                 return "image/jpeg";
 974         else if ((len > 3) && !strncasecmp(&what[len - 4], ".png", 4))
 975                 return "image/png";
 976         else if ((len > 3) && !strncasecmp(&what[len - 4], ".ico", 4))
 977                 return "image/x-icon";
 978         else if ((len > 3) && !strncasecmp(&what[len - 4], ".vcf", 4))
 979                 return "text/x-vcard";
 980         else if ((len > 4) && !strncasecmp(&what[len - 5], ".html", 5))
 981                 return "text/html";
 982         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htm", 4))
 983                 return "text/html";
 984         else if ((len > 3) && !strncasecmp(&what[len - 4], ".wml", 4))
 985                 return "text/vnd.wap.wml";
 986         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmls", 5))
 987                 return "text/vnd.wap.wmlscript";
 988         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmlc", 5))
 989                 return "application/vnd.wap.wmlc";
 990         else if ((len > 5) && !strncasecmp(&what[len - 6], ".wmlsc", 6))
 991                 return "application/vnd.wap.wmlscriptc";
 992         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wbmp", 5))
 993                 return "image/vnd.wap.wbmp";
 994         else
 995                 /* and let xdgmime do the fallback. */
 996                 return xdg_mime_get_mime_type_from_file_name(what);
 997 }
 998
 999 static HashList *IconHash = NULL;
1000
1001 typedef struct IconName IconName;
1002
1003 struct IconName {
1004         char *FlatName;
1005         char *FileName;
1006 };
1007
1008 static void DeleteIcon(void *IconNamePtr)
1009 {
1010         IconName *Icon = (IconName*) IconNamePtr;
1011         free(Icon->FlatName);
1012         free(Icon->FileName);
1013         free(Icon);
1014 }
1015
1016 /*
1017 static const char *PrintFlat(void *IconNamePtr)
1018 {
1019         IconName *Icon = (IconName*) IconNamePtr;
1020         return Icon->FlatName;
1021 }
1022 static const char *PrintFile(void *IconNamePtr)
1023 {
1024         IconName *Icon = (IconName*) IconNamePtr;
1025         return Icon->FileName;
1026 }
1027 */
1028
1029 #define GENSTR "x-generic"
1030 #define IGNORE_PREFIX_1 "gnome-mime"
1031 int LoadIconDir(const char *DirName)
1032 {
1033         DIR *filedir = NULL;
1034         struct dirent *filedir_entry;
1035         int d_namelen;
1036         int d_without_ext;
1037         IconName *Icon;
1038
1039         filedir = opendir (DirName);
1040         IconHash = NewHash(1, NULL);
1041         if (filedir == NULL) {
1042                 return 0;
1043         }
1044
1045         while ((filedir_entry = readdir(filedir)))
1046         {
1047                 char *MinorPtr;
1048                 char *PStart;
1049 #ifdef _DIRENT_HAVE_D_NAMELEN
1050                 d_namelen = filedir_entry->d_namelen;
1051 #else
1052                 d_namelen = strlen(filedir_entry->d_name);
1053 #endif
1054                 d_without_ext = d_namelen;
1055                 while ((d_without_ext > 0) && (filedir_entry->d_name[d_without_ext] != '.'))
1056                         d_without_ext --;
1057                 if ((d_without_ext == 0) || (d_namelen < 3))
1058                         continue;
1059
1060                 if ((sizeof(IGNORE_PREFIX_1) < d_namelen) &&
1061                     (strncmp(IGNORE_PREFIX_1,
1062                              filedir_entry->d_name,
1063                              sizeof(IGNORE_PREFIX_1) - 1) == 0)) {
1064                         PStart = filedir_entry->d_name + sizeof(IGNORE_PREFIX_1);
1065                         d_without_ext -= sizeof(IGNORE_PREFIX_1);
1066                 }
1067                 else {
1068                         PStart = filedir_entry->d_name;
1069                 }
1070                 Icon = malloc(sizeof(IconName));
1071
1072                 Icon->FileName = malloc(d_namelen + 1);
1073                 memcpy(Icon->FileName, filedir_entry->d_name, d_namelen + 1);
1074
1075                 Icon->FlatName = malloc(d_without_ext + 1);
1076                 memcpy(Icon->FlatName, PStart, d_without_ext);
1077                 Icon->FlatName[d_without_ext] = '\0';
1078                 /* Try to find Minor type in image-jpeg */
1079                 MinorPtr = strchr(Icon->FlatName, '-');
1080                 if (MinorPtr != NULL) {
1081                         size_t MinorLen;
1082                         MinorLen = 1 + d_without_ext - (MinorPtr - Icon->FlatName + 1);
1083                         if ((MinorLen == sizeof(GENSTR)) &&
1084                             (strncmp(MinorPtr + 1, GENSTR, sizeof(GENSTR)) == 0)) {
1085                                 /* ok, we found a generic filename. cut the generic. */
1086                                 *MinorPtr = '\0';
1087                                 d_without_ext = d_without_ext - (MinorPtr - Icon->FlatName);
1088                         }
1089                         else { /* Map the major / minor separator to / */
1090                                 *MinorPtr = '/';
1091                         }
1092                 }
1093
1094 //              PrintHash(IconHash, PrintFlat, PrintFile);
1095 //              printf("%s - %s\n", Icon->FlatName, Icon->FileName);
1096                 Put(IconHash, Icon->FlatName, d_without_ext, Icon, DeleteIcon);
1097 //              PrintHash(IconHash, PrintFlat, PrintFile);
1098         }
1099         closedir(filedir);
1100         return 1;
1101 }
1102
1103 const char *GetIconFilename(char *MimeType, size_t len)
1104 {
1105         void *vIcon;
1106         IconName *Icon;
1107
1108         if(IconHash == NULL)
1109                 return NULL;
1110
1111         GetHash(IconHash, MimeType, len, &vIcon), Icon = (IconName*) vIcon;
1112         /* didn't find the exact mimetype? try major only. */
1113         if (Icon == NULL) {
1114                 char * pMinor;
1115                 pMinor = strchr(MimeType, '/');
1116                 if (pMinor != NULL) {
1117                         *pMinor = '\0';
1118                         GetHash(IconHash, MimeType, pMinor - MimeType, &vIcon),
1119                                 Icon = (IconName*) vIcon;
1120                 }
1121         }
1122         if (Icon == NULL) {
1123                 return NULL;
1124         }
1125
1126         /*printf("Getting: [%s] == [%s] -> [%s]\n", MimeType, Icon->FlatName, Icon->FileName);*/
1127         return Icon->FileName;
1128 }
1129
1130 void ShutDownLibCitadelMime(void)
1131 {
1132         DeleteHash(&IconHash);
1133 }