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; you can redistribute it and/or modify
   6 // it under the terms of the GNU General Public License as published by
   7 // the Free Software Foundation; either version 3 of the License, or
   8 // (at your option) any later version.
   9 //
  10 // This program is distributed in the hope that it will be useful,
  11 // but WITHOUT ANY WARRANTY; without even the implied warranty of
  12 // MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13 // GNU General Public License for more details.
  14 //
  15 // You should have received a copy of the GNU General Public License
  16 // along with this program; if not, write to the Free Software
  17 // Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  18
  19 #include <stdlib.h>
  20 #include <unistd.h>
  21 #include <stdio.h>
  22 #include <signal.h>
  23 #include <sys/types.h>
  24 #include <ctype.h>
  25 #include <string.h>
  26 #include <sys/stat.h>
  27 #include <sys/types.h>
  28 #include <dirent.h>
  29 #include <errno.h>
  30
  31 #include "xdgmime/xdgmime.h"
  32 #include "libcitadel.h"
  33 #include "libcitadellocal.h"
  34
  35 const unsigned char FromHexTable [256] = {
  36         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //  0
  37         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 10
  38         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 20
  39         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 30
  40         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x00, 0x01, // 40
  41         0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0xFF, 0xFF, // 50
  42         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, // 60
  43         0x0F, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 70
  44         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 80
  45         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0A, 0x0B, 0x0C, // 90
  46         0x0D, 0x0E, 0x0F, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //100
  47         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //110
  48         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //120
  49         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //130
  50         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //140
  51         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //150
  52         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //160
  53         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //170
  54         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //180
  55         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //190
  56         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //200
  57         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //210
  58         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //220
  59         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //230
  60         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //240
  61         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF                          //250
  62 };
  63
  64
  65 long extract_key(char *target, char *source, long sourcelen, char *key, long keylen, char KeyEnd) {
  66         char *sptr, *ptr = NULL;
  67         int double_quotes = 0;
  68         long RealKeyLen = keylen;
  69
  70         sptr = source;
  71
  72         while (sptr != NULL) {
  73                 ptr = bmstrcasestr_len(sptr, sourcelen - (sptr - source), key, keylen);
  74                 if (ptr != NULL) {
  75                         while (isspace(*(ptr + RealKeyLen)))
  76                                 RealKeyLen ++;
  77                         if (*(ptr + RealKeyLen) == KeyEnd) {
  78                                 sptr = NULL;
  79                                 RealKeyLen ++;
  80                         }
  81                         else {
  82                                 sptr = ptr + RealKeyLen + 1;
  83                         }
  84                 }
  85                 else
  86                         sptr = ptr;
  87         }
  88         if (ptr == NULL) {
  89                 *target = '\0';
  90                 return 0;
  91         }
  92         strcpy(target, (ptr + RealKeyLen));
  93
  94         for (ptr=target; (*ptr != 0); ptr++) {
  95
  96                 /* A semicolon means we've hit the end of the key, unless we're inside double quotes */
  97                 if ( (double_quotes != 1) && (*ptr == ';')) {
  98                         *ptr = 0;
  99                 }
 100
 101                 /* if we find double quotes, we've got a great set of string boundaries */
 102                 if (*ptr == '\"') {
 103                         ++double_quotes;
 104                         if (double_quotes == 1) {
 105                                 strcpy(ptr, ptr+1);
 106                         }
 107                         else {
 108                                 *ptr = 0;
 109                         }
 110                 }
 111         }
 112         *ptr = '\0';
 113         return ptr - target;
 114 }
 115
 116
 117 /*
 118  * For non-multipart messages, we need to generate a quickie partnum of "1"
 119  * to return to callback functions.  Some callbacks demand it.
 120  */
 121 char *fixed_partnum(char *supplied_partnum) {
 122         if (supplied_partnum == NULL) return "1";
 123         if (strlen(supplied_partnum)==0) return "1";
 124         return supplied_partnum;
 125 }
 126
 127
 128 static inline unsigned int _decode_hex(const char *Source) {
 129         unsigned int ret = '?';
 130         unsigned char LO_NIBBLE;
 131         unsigned char HI_NIBBLE;
 132
 133         HI_NIBBLE = FromHexTable[(unsigned char) *Source];
 134         LO_NIBBLE = FromHexTable[(unsigned char) *(Source+1)];
 135
 136         if ((LO_NIBBLE == 0xFF) || (LO_NIBBLE == 0xFF))
 137                 return ret;
 138         ret = HI_NIBBLE;
 139         ret = ret << 4;
 140         ret = ret | LO_NIBBLE;
 141         return ret;
 142 }
 143
 144 unsigned int decode_hex(char *Source) {return _decode_hex(Source);}
 145
 146 /*
 147  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
 148  * according to RFC2045 section 6.7
 149  */
 150 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
 151         unsigned int ch;
 152         int decoded_length = 0;
 153         int pos = 0;
 154
 155         while (pos < sourcelen) {
 156                 if (*(encoded + pos) == '=') {
 157                         pos ++;
 158                         if (*(encoded + pos) == '\n') {
 159                                 pos ++;
 160                         }
 161                         else if (*(encoded + pos) == '\r') {
 162                                 pos ++;
 163                                 if (*(encoded + pos) == '\n')
 164                                         pos++;
 165                         }
 166                         else {
 167                                 ch = _decode_hex(&encoded[pos]);
 168                                 pos += 2;
 169                                 decoded[decoded_length++] = ch;
 170                         }
 171                 }
 172                 else {
 173                         decoded[decoded_length++] = encoded[pos];
 174                         pos += 1;
 175                 }
 176         }
 177         decoded[decoded_length] = 0;
 178         return(decoded_length);
 179 }
 180
 181
 182 /*
 183  * Given a message or message-part body and a length, handle any necessary
 184  * decoding and pass the request up the stack.
 185  */
 186 void mime_decode(char *partnum,
 187                  char *part_start, size_t length,
 188                  char *content_type, char *charset, char *encoding,
 189                  char *disposition,
 190                  char *id,
 191                  char *name, char *filename,
 192                  MimeParserCallBackType CallBack,
 193                  MimeParserCallBackType PreMultiPartCallBack,
 194                  MimeParserCallBackType PostMultiPartCallBack,
 195                  void *userdata,
 196                  int dont_decode
 197 ) {
 198         char *decoded;
 199         size_t bytes_decoded = 0;
 200
 201         // Some encodings aren't really encodings
 202         if (!strcasecmp(encoding, "7bit"))
 203                 *encoding = '\0';
 204         if (!strcasecmp(encoding, "8bit"))
 205                 *encoding = '\0';
 206         if (!strcasecmp(encoding, "binary"))
 207                 *encoding = '\0';
 208         if (!strcasecmp(encoding, "ISO-8859-1"))
 209                 *encoding = '\0';
 210
 211         // If this part is not encoded, send as-is
 212         if ( (strlen(encoding) == 0) || (dont_decode)) {
 213                 if (CallBack != NULL) {
 214                         CallBack(name,
 215                                  filename,
 216                                  fixed_partnum(partnum),
 217                                  disposition,
 218                                  part_start,
 219                                  content_type,
 220                                  charset,
 221                                  length,
 222                                  encoding,
 223                                  id,
 224                                  userdata);
 225                         }
 226                 return;
 227         }
 228
 229         // Fail silently if we hit an unknown encoding.
 230         if ((strcasecmp(encoding, "base64")) && (strcasecmp(encoding, "quoted-printable"))) {
 231                 return;
 232         }
 233         fprintf(stderr, "\033[33mSource encoded length: %d\033[0m\n", length);
 234
 235         // Allocate a buffer for the decoded data.  The output buffer is slightly
 236         // larger than the input buffer; this assumes that the decoded data
 237         // will never be significantly larger than the encoded data.  This is a
 238         // safe assumption with base64, uuencode, and quoted-printable.
 239         decoded = malloc(length + 32768);
 240         if (decoded == NULL) {
 241                 return;
 242         }
 243
 244         if (!strcasecmp(encoding, "base64")) {
 245                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 246         }
 247         else if (!strcasecmp(encoding, "quoted-printable")) {
 248                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded, part_start, length);
 249         }
 250         fprintf(stderr, "\033[33mTarget decoded length: %d\033[0m\n", bytes_decoded);
 251
 252         if (bytes_decoded > 0) if (CallBack != NULL) {
 253                         char encoding_buf[SIZ];
 254
 255                         strcpy(encoding_buf, "binary");
 256                         CallBack(name,
 257                                  filename,
 258                                  fixed_partnum(partnum),
 259                                  disposition,
 260                                  decoded,
 261                                  content_type,
 262                                  charset,
 263                                  bytes_decoded,
 264                                  encoding_buf,
 265                                  id,
 266                                  userdata);
 267         }
 268
 269         free(decoded);
 270 }
 271
 272 /*
 273  * this is the extract of mime_decode which can be called if 'dont_decode' was set;
 274  * to save the cpu intense process of decoding to the time when it realy wants the content.
 275  * returns:
 276  *   - > 0 we decoded something, its on *decoded, you need to free it.
 277  *   - = 0 no need to decode stuff. *decoded will be NULL.
 278  *   - < 0 an error occured, either an unknown encoding, or alloc failed. no need to free.
 279  */
 280 int mime_decode_now (char *part_start,
 281                      size_t length,
 282                      char *encoding,
 283                      char **decoded,
 284                      size_t *bytes_decoded)
 285 {
 286         *bytes_decoded = 0;
 287         *decoded = NULL;
 288         /* Some encodings aren't really encodings */
 289         if (!strcasecmp(encoding, "7bit"))
 290                 *encoding = '\0';
 291         if (!strcasecmp(encoding, "8bit"))
 292                 *encoding = '\0';
 293         if (!strcasecmp(encoding, "binary"))
 294                 *encoding = '\0';
 295
 296         /* If this part is not encoded, send as-is */
 297         if (strlen(encoding) == 0) {
 298                 return 0;
 299         }
 300
 301
 302         /* Fail if we hit an unknown encoding. */
 303         if ((strcasecmp(encoding, "base64"))
 304             && (strcasecmp(encoding, "quoted-printable"))) {
 305                 return -1;
 306         }
 307
 308         /*
 309          * Allocate a buffer for the decoded data.  The output buffer is slightly
 310          * larger than the input buffer; this assumes that the decoded data
 311          * will never be significantly larger than the encoded data.  This is a
 312          * safe assumption with base64, uuencode, and quoted-printable.
 313          */
 314         *decoded = malloc(length + 32768);
 315         if (decoded == NULL) {
 316                 return -1;
 317         }
 318
 319         if (!strcasecmp(encoding, "base64")) {
 320                 *bytes_decoded = CtdlDecodeBase64(*decoded, part_start, length);
 321                 return 1;
 322         }
 323         else if (!strcasecmp(encoding, "quoted-printable")) {
 324                 *bytes_decoded = CtdlDecodeQuotedPrintable(*decoded, part_start, length);
 325                 return 1;
 326         }
 327         return -1;
 328 }
 329
 330 typedef enum _eIntMimeHdrs {
 331         boundary,
 332         startary,
 333         endary,
 334         content_type,
 335         charset,
 336         encoding,
 337         content_type_name,
 338         content_disposition_name,
 339         filename,
 340         disposition,
 341         id,
 342         eMax /* don't move ! */
 343 } eIntMimeHdrs;
 344
 345 typedef struct _CBufStr {
 346         char Key[SIZ];
 347         long len;
 348 }CBufStr;
 349
 350 typedef struct _interesting_mime_headers {
 351         CBufStr b[eMax];
 352         long content_length;
 353         long is_multipart;
 354 } interesting_mime_headers;
 355
 356
 357 static void FlushInterestingMimes(interesting_mime_headers *m)
 358 {
 359         int i;
 360
 361         for (i = 0; i < eMax; i++) {
 362              m->b[i].Key[0] = '\0';
 363              m->b[i].len = 0;
 364         }
 365         m->content_length = -1;
 366 }
 367 static interesting_mime_headers *InitInterestingMimes(void)
 368 {
 369         interesting_mime_headers *m;
 370         m = (interesting_mime_headers*) malloc( sizeof(interesting_mime_headers));
 371
 372         FlushInterestingMimes(m);
 373
 374         return m;
 375 }
 376
 377
 378 static long parse_MimeHeaders(interesting_mime_headers *m,
 379                               char** pcontent_start,
 380                               char *content_end)
 381 {
 382         char buf[SIZ];
 383         char header[SIZ];
 384         long headerlen;
 385         char *ptr, *pch;
 386         int buflen = 0;
 387         int i;
 388
 389         /* Learn interesting things from the headers */
 390         ptr = *pcontent_start;
 391         *header = '\0';
 392         headerlen = 0;
 393         do {
 394                 ptr = memreadlinelen(ptr, buf, SIZ, &buflen);
 395
 396                 for (i = 0; i < buflen; ++i) {
 397                         if (isspace(buf[i])) {
 398                                 buf[i] = ' ';
 399                         }
 400                 }
 401
 402                 if (!isspace(buf[0]) && (headerlen > 0)) {
 403                         if (!strncasecmp(header, "Content-type:", 13)) {
 404                                 memcpy (m->b[content_type].Key, &header[13], headerlen - 12);
 405                                 m->b[content_type].Key[headerlen - 12] = '\0';
 406                                 m->b[content_type].len = striplt (m->b[content_type].Key);
 407
 408                                 m->b[content_type_name].len = extract_key(m->b[content_type_name].Key, CKEY(m->b[content_type]), HKEY("name"), '=');
 409                                 m->b[charset].len           = extract_key(m->b[charset].Key,           CKEY(m->b[content_type]), HKEY("charset"), '=');
 410                                 m->b[boundary].len          = extract_key(m->b[boundary].Key,          header,       headerlen,  HKEY("boundary"), '=');
 411
 412                                 /* Deal with weird headers */
 413                                 pch = strchr(m->b[content_type].Key, ' ');
 414                                 if (pch != NULL) {
 415                                         *pch = '\0';
 416                                         m->b[content_type].len = m->b[content_type].Key - pch;
 417                                 }
 418                                 pch = strchr(m->b[content_type].Key, ';');
 419                                 if (pch != NULL) {
 420                                         *pch = '\0';
 421                                         m->b[content_type].len = m->b[content_type].Key - pch;
 422                                 }
 423                         }
 424                         else if (!strncasecmp(header, "Content-Disposition:", 20)) {
 425                                 memcpy (m->b[disposition].Key, &header[20], headerlen - 19);
 426                                 m->b[disposition].Key[headerlen - 19] = '\0';
 427                                 m->b[disposition].len = striplt(m->b[disposition].Key);
 428
 429                                 m->b[content_disposition_name].len = extract_key(m->b[content_disposition_name].Key, CKEY(m->b[disposition]), HKEY("name"), '=');
 430                                 m->b[filename].len                 = extract_key(m->b[filename].Key,                 CKEY(m->b[disposition]), HKEY("filename"), '=');
 431                                 pch = strchr(m->b[disposition].Key, ';');
 432                                 if (pch != NULL) *pch = '\0';
 433                                 m->b[disposition].len = striplt(m->b[disposition].Key);
 434                         }
 435                         else if (!strncasecmp(header, "Content-ID:", 11)) {
 436                                 memcpy(m->b[id].Key, &header[11], headerlen - 11);
 437                                 m->b[id].Key[headerlen - 11] = '\0';
 438                                 striplt(m->b[id].Key);
 439                                 m->b[id].len = stripallbut(m->b[id].Key, '<', '>');
 440                         }
 441                         else if (!strncasecmp(header, "Content-length: ", 15)) {
 442                                 char *clbuf;
 443                                 clbuf = &header[15];
 444                                 while (isspace(*clbuf))
 445                                         clbuf ++;
 446                                 m->content_length = (size_t) atol(clbuf);
 447                         }
 448                         else if (!strncasecmp(header, "Content-transfer-encoding: ", 26)) {
 449                                 memcpy(m->b[encoding].Key, &header[26], headerlen - 26);
 450                                 m->b[encoding].Key[headerlen - 26] = '\0';
 451                                 m->b[encoding].len = striplt(m->b[encoding].Key);
 452                         }
 453                         *header = '\0';
 454                         headerlen = 0;
 455                 }
 456                 if ((headerlen + buflen + 2) < SIZ) {
 457                         memcpy(&header[headerlen], buf, buflen);
 458                         headerlen += buflen;
 459                         header[headerlen] = '\0';
 460                 }
 461                 if (ptr >= content_end) {
 462                         return -1;
 463                 }
 464         } while ((!IsEmptyStr(buf)) && (*ptr != 0));
 465
 466         m->is_multipart = m->b[boundary].len != 0;
 467         *pcontent_start = ptr;
 468
 469         return 0;
 470 }
 471
 472
 473 static int IsAsciiEncoding(interesting_mime_headers *m)
 474 {
 475
 476         if ((m->b[encoding].len != 0) &&
 477             (strcasecmp(m->b[encoding].Key, "base64") == 0))
 478                 return 1;
 479         if ((m->b[encoding].len != 0) &&
 480             (strcmp(m->b[encoding].Key, "quoted-printable") == 0))
 481                 return 1;
 482
 483         return 0;
 484 }
 485
 486 static char *FindNextContent(char *ptr,
 487                              char *content_end,
 488                              interesting_mime_headers *SubMimeHeaders,
 489                              interesting_mime_headers *m)
 490 {
 491         char *next_boundary;
 492         char  tmp;
 493
 494         if (IsAsciiEncoding(SubMimeHeaders)) {
 495                 tmp = *content_end;
 496                 *content_end = '\0';
 497
 498                 /**
 499                  * ok, if we have a content length of the mime part,
 500                  * try skipping the content on the search for the next
 501                  * boundary. since we don't trust the content_length
 502                  * to be all accurate, and suspect it to lose one digit
 503                  * per line with a line length of 80 chars, we need
 504                  * to start searching a little before..
 505                  */
 506
 507                 if ((SubMimeHeaders->content_length != -1) &&
 508                     (SubMimeHeaders->content_length > 10))
 509                 {
 510                         char *pptr;
 511                         long lines;
 512
 513                         lines = SubMimeHeaders->content_length / 80;
 514                         pptr = ptr + SubMimeHeaders->content_length - lines - 10;
 515                         if (pptr < content_end)
 516                                 ptr = pptr;
 517                 }
 518
 519                 next_boundary = strstr(ptr, m->b[startary].Key);
 520                 *content_end = tmp;
 521         }
 522         else {
 523                 char *srch;
 524                 /**
 525                  * ok, if we have a content length of the mime part,
 526                  * try skipping the content on the search for the next
 527                  * boundary. since we don't trust the content_length
 528                  * to be all accurate, start searching a little before..
 529                  */
 530
 531                 if ((SubMimeHeaders->content_length != -1) &&
 532                     (SubMimeHeaders->content_length > 10))
 533                 {
 534                         char *pptr;
 535                         pptr = ptr + SubMimeHeaders->content_length - 10;
 536                         if (pptr < content_end)
 537                                 ptr = pptr;
 538                 }
 539
 540
 541                 srch = next_boundary = NULL;
 542                 for (srch = memchr(ptr, '-',  content_end - ptr);
 543                      (srch != NULL) && (srch < content_end);
 544                      srch = memchr(srch, '-',  content_end - srch))
 545                 {
 546                         if (!memcmp(srch,
 547                                     m->b[startary].Key,
 548                                     m->b[startary].len))
 549                         {
 550                                 next_boundary = srch;
 551                                 srch = content_end;
 552                         }
 553                         else srch ++;
 554
 555                 }
 556
 557         }
 558         return next_boundary;
 559 }
 560
 561 /*
 562  * Break out the components of a multipart message
 563  * (This function expects to be fed HEADERS + CONTENT)
 564  * Note: NULL can be supplied as content_end; in this case, the message is
 565  * considered to have ended when the parser encounters a 0x00 byte.
 566  */
 567 static void recurseable_mime_parser(char *partnum,
 568                                     char *content_start, char *content_end,
 569                                     MimeParserCallBackType CallBack,
 570                                     MimeParserCallBackType PreMultiPartCallBack,
 571                                     MimeParserCallBackType PostMultiPartCallBack,
 572                                     void *userdata,
 573                                     int dont_decode,
 574                                     interesting_mime_headers *m)
 575 {
 576         interesting_mime_headers *SubMimeHeaders;
 577         char     *ptr;
 578         char     *part_start;
 579         char     *part_end = NULL;
 580         char     *evaluate_crlf_ptr = NULL;
 581         char     *next_boundary;
 582         char      nested_partnum[256];
 583         int       crlf_in_use = 0;
 584         int       part_seq = 0;
 585         CBufStr  *chosen_name;
 586
 587
 588         /* If this is a multipart message, then recursively process it */
 589         ptr = content_start;
 590         part_start = NULL;
 591         if (m->is_multipart) {
 592
 593                 /* Tell the client about this message's multipartedness */
 594                 if (PreMultiPartCallBack != NULL) {
 595                         PreMultiPartCallBack("",
 596                                              "",
 597                                              partnum,
 598                                              "",
 599                                              NULL,
 600                                              m->b[content_type].Key,
 601                                              m->b[charset].Key,
 602                                              0,
 603                                              m->b[encoding].Key,
 604                                              m->b[id].Key,
 605                                              userdata);
 606                 }
 607
 608                 /* Figure out where the boundaries are */
 609                 m->b[startary].len = snprintf(m->b[startary].Key, SIZ, "--%s", m->b[boundary].Key);
 610                 SubMimeHeaders = InitInterestingMimes ();
 611
 612                 while ((*ptr == '\r') || (*ptr == '\n')) ptr ++;
 613
 614                 if (strncmp(ptr, m->b[startary].Key, m->b[startary].len) == 0)
 615                         ptr += m->b[startary].len;
 616
 617                 while ((*ptr == '\r') || (*ptr == '\n')) ptr ++;
 618
 619                 part_start = NULL;
 620                 do {
 621                         char *optr;
 622
 623                         optr = ptr;
 624                         if (parse_MimeHeaders(SubMimeHeaders, &ptr, content_end) != 0)
 625                                 break;
 626                         if ((ptr - optr > 2) && (*(ptr - 2) == '\r')) {
 627                                 crlf_in_use = 1;
 628                         }
 629
 630                         part_start = ptr;
 631
 632                         next_boundary = FindNextContent(ptr, content_end, SubMimeHeaders, m);
 633                         if ((next_boundary != NULL) && (next_boundary - part_start < 3)) {
 634                                 FlushInterestingMimes(SubMimeHeaders);
 635                                 continue;
 636                         }
 637
 638                         if ( (part_start != NULL) && (next_boundary != NULL) ) {
 639                                 part_end = next_boundary;
 640                                 --part_end;             /* omit the trailing LF */
 641                                 if (crlf_in_use) {
 642                                         --part_end;     /* omit the trailing CR */
 643                                 }
 644
 645                                 if (!IsEmptyStr(partnum)) {
 646                                         snprintf(nested_partnum,
 647                                                  sizeof nested_partnum,
 648                                                  "%s.%d", partnum,
 649                                                  ++part_seq);
 650                                 }
 651                                 else {
 652                                         snprintf(nested_partnum,
 653                                                  sizeof nested_partnum,
 654                                                  "%d", ++part_seq);
 655                                 }
 656                                 recurseable_mime_parser(nested_partnum,
 657                                                         part_start,
 658                                                         part_end,
 659                                                         CallBack,
 660                                                         PreMultiPartCallBack,
 661                                                         PostMultiPartCallBack,
 662                                                         userdata,
 663                                                         dont_decode,
 664                                                         SubMimeHeaders);
 665                         }
 666
 667                         if (next_boundary != NULL) {
 668                                 /* If we pass out of scope, don't attempt to
 669                                  * read past the end boundary. */
 670                                 if ((*(next_boundary + m->b[startary].len) == '-') &&
 671                                     (*(next_boundary + m->b[startary].len + 1) == '-') ){
 672                                         ptr = content_end;
 673                                 }
 674                                 else {
 675                                         /* Set up for the next part. */
 676                                         part_start = strstr(next_boundary, "\n");
 677
 678                                         /* Determine whether newlines are LF or CRLF */
 679                                         evaluate_crlf_ptr = part_start;
 680                                         --evaluate_crlf_ptr;
 681                                         if ((*evaluate_crlf_ptr == '\r') && (*(evaluate_crlf_ptr + 1) == '\n')) {
 682                                                 crlf_in_use = 1;
 683                                         }
 684                                         else {
 685                                                 crlf_in_use = 0;
 686                                         }
 687
 688                                         /* Advance past the LF ... now we're in the next part */
 689                                         ++part_start;
 690                                         ptr = part_start;
 691                                 }
 692                         }
 693                         else {
 694                                 /* Invalid end of multipart.  Bail out! */
 695                                 ptr = content_end;
 696                         }
 697                         FlushInterestingMimes(SubMimeHeaders);
 698                 } while ( (ptr < content_end) && (next_boundary != NULL) );
 699
 700                 free(SubMimeHeaders);
 701
 702                 if (PostMultiPartCallBack != NULL) {
 703                         PostMultiPartCallBack("",
 704                                               "",
 705                                               partnum,
 706                                               "",
 707                                               NULL,
 708                                               m->b[content_type].Key,
 709                                               m->b[charset].Key,
 710                                               0,
 711                                               m->b[encoding].Key,
 712                                               m->b[id].Key,
 713                                               userdata);
 714                 }
 715         } /* If it's not a multipart message, then do something with it */
 716         else {
 717                 size_t length;
 718                 part_start = ptr;
 719                 length = content_end - part_start;
 720                 ptr = part_end = content_end;
 721
 722                 /* The following code will truncate the MIME part to the size
 723                  * specified by the Content-length: header.   We have commented it
 724                  * out because these headers have a tendency to be wrong.
 725                  *
 726                  *      if ( (content_length > 0) && (length > content_length) ) {
 727                  *              length = content_length;
 728                  *      }
 729                  */
 730
 731                 /* Sometimes the "name" field is tacked on to Content-type,
 732                  * and sometimes it's tacked on to Content-disposition.  Use
 733                  * whichever one we have.
 734                  */
 735                 if (m->b[content_disposition_name].len > m->b[content_type_name].len) {
 736                         chosen_name = &m->b[content_disposition_name];
 737                 }
 738                 else {
 739                         chosen_name = &m->b[content_type_name];
 740                 }
 741
 742                 // Ok, we've got a non-multipart part here, so do something with it.
 743                 mime_decode(partnum,
 744                             part_start,
 745                             length,
 746                             m->b[content_type].Key,
 747                             m->b[charset].Key,
 748                             m->b[encoding].Key,
 749                             m->b[disposition].Key,
 750                             m->b[id].Key,
 751                             chosen_name->Key,
 752                             m->b[filename].Key,
 753                             CallBack,
 754                             NULL, NULL,
 755                             userdata,
 756                             dont_decode
 757                         );
 758
 759                 /*
 760                  * Now if it's an encapsulated message/rfc822 then we have to recurse into it
 761                  */
 762                 if (!strcasecmp(&m->b[content_type].Key[0], "message/rfc822")) {
 763
 764                         if (PreMultiPartCallBack != NULL) {
 765                                 PreMultiPartCallBack("",
 766                                                      "",
 767                                                      partnum,
 768                                                      "",
 769                                                      NULL,
 770                                                      m->b[content_type].Key,
 771                                                      m->b[charset].Key,
 772                                                      0,
 773                                                      m->b[encoding].Key,
 774                                                      m->b[id].Key,
 775                                                      userdata);
 776                         }
 777                         if (CallBack != NULL) {
 778                                 if (strlen(partnum) > 0) {
 779                                         snprintf(nested_partnum,
 780                                                  sizeof nested_partnum,
 781                                                  "%s.%d", partnum,
 782                                                  ++part_seq);
 783                                 }
 784                                 else {
 785                                         snprintf(nested_partnum,
 786                                                  sizeof nested_partnum,
 787                                                  "%d", ++part_seq);
 788                                 }
 789                                 the_mime_parser(nested_partnum,
 790                                                 part_start,
 791                                                 part_end,
 792                                                 CallBack,
 793                                                 PreMultiPartCallBack,
 794                                                 PostMultiPartCallBack,
 795                                                 userdata,
 796                                                 dont_decode
 797                                         );
 798                         }
 799                         if (PostMultiPartCallBack != NULL) {
 800                                 PostMultiPartCallBack("",
 801                                                       "",
 802                                                       partnum,
 803                                                       "",
 804                                                       NULL,
 805                                                       m->b[content_type].Key,
 806                                                       m->b[charset].Key,
 807                                                       0,
 808                                                       m->b[encoding].Key,
 809                                                       m->b[id].Key,
 810                                                       userdata);
 811                         }
 812
 813
 814                 }
 815
 816         }
 817
 818 }
 819
 820 /*
 821  * Break out the components of a multipart message
 822  * (This function expects to be fed HEADERS + CONTENT)
 823  * Note: NULL can be supplied as content_end; in this case, the message is
 824  * considered to have ended when the parser encounters a 0x00 byte.
 825  */
 826 void the_mime_parser(char *partnum,
 827                      char *content_start, char *content_end,
 828                      MimeParserCallBackType CallBack,
 829                      MimeParserCallBackType PreMultiPartCallBack,
 830                      MimeParserCallBackType PostMultiPartCallBack,
 831                      void *userdata,
 832                      int dont_decode)
 833 {
 834         interesting_mime_headers *m;
 835
 836         /* If the caller didn't supply an endpointer, generate one by measure */
 837         if (content_end == NULL) {
 838                 content_end = &content_start[strlen(content_start)];
 839         }
 840
 841         m = InitInterestingMimes();
 842
 843         if (!parse_MimeHeaders(m, &content_start, content_end))
 844         {
 845
 846                 recurseable_mime_parser(partnum,
 847                                         content_start, content_end,
 848                                         CallBack,
 849                                         PreMultiPartCallBack,
 850                                         PostMultiPartCallBack,
 851                                         userdata,
 852                                         dont_decode,
 853                                         m);
 854         }
 855         free(m);
 856 }
 857
 858
 859 /*
 860  * Entry point for the MIME parser.
 861  * (This function expects to be fed HEADERS + CONTENT)
 862  * Note: NULL can be supplied as content_end; in this case, the message is
 863  * considered to have ended when the parser encounters a 0x00 byte.
 864  */
 865 void mime_parser(char *content_start,
 866                  char *content_end,
 867                  MimeParserCallBackType CallBack,
 868                  MimeParserCallBackType PreMultiPartCallBack,
 869                  MimeParserCallBackType PostMultiPartCallBack,
 870                  void *userdata,
 871                  int dont_decode)
 872 {
 873
 874         the_mime_parser("", content_start, content_end,
 875                         CallBack,
 876                         PreMultiPartCallBack,
 877                         PostMultiPartCallBack,
 878                         userdata, dont_decode);
 879 }
 880
 881
 882
 883
 884
 885
 886 typedef struct _MimeGuess {
 887         const char *Pattern;
 888         size_t PatternLen;
 889         long PatternOffset;
 890         const char *MimeString;
 891 } MimeGuess;
 892
 893 MimeGuess MyMimes [] = {
 894         {
 895                 "GIF",
 896                 3,
 897                 0,
 898                 "image/gif"
 899         },
 900         {
 901                 "\xff\xd8",
 902                 2,
 903                 0,
 904                 "image/jpeg"
 905         },
 906         {
 907                 "\x89PNG",
 908                 4,
 909                 0,
 910                 "image/png"
 911         },
 912         { // last...
 913                 "",
 914                 0,
 915                 0,
 916                 ""
 917         }
 918 };
 919
 920
 921 const char *GuessMimeType(const char *data, size_t dlen)
 922 {
 923         int MimeIndex = 0;
 924
 925         while (MyMimes[MimeIndex].PatternLen != 0)
 926         {
 927                 if ((MyMimes[MimeIndex].PatternLen +
 928                      MyMimes[MimeIndex].PatternOffset < dlen) &&
 929                     strncmp(MyMimes[MimeIndex].Pattern,
 930                             &data[MyMimes[MimeIndex].PatternOffset],
 931                             MyMimes[MimeIndex].PatternLen) == 0)
 932                 {
 933                         return MyMimes[MimeIndex].MimeString;
 934                 }
 935                 MimeIndex ++;
 936         }
 937         /*
 938          * ok, our simple minded algorythm didn't find anything,
 939          * let the big chegger try it, he wil default to application/octet-stream
 940          */
 941         return (xdg_mime_get_mime_type_for_data(data, dlen));
 942 }
 943
 944
 945 const char* GuessMimeByFilename(const char *what, size_t len)
 946 {
 947         /* we know some hardcoded on our own, try them... */
 948         if ((len > 3) && !strncasecmp(&what[len - 4], ".gif", 4))
 949                 return "image/gif";
 950         else if ((len > 2) && !strncasecmp(&what[len - 3], ".js", 3))
 951                 return  "text/javascript";
 952         else if ((len > 3) && !strncasecmp(&what[len - 4], ".txt", 4))
 953                 return "text/plain";
 954         else if ((len > 3) && !strncasecmp(&what[len - 4], ".css", 4))
 955                 return "text/css";
 956         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htc", 4))
 957                 return "text/x-component";
 958         else if ((len > 3) && !strncasecmp(&what[len - 4], ".jpg", 4))
 959                 return "image/jpeg";
 960         else if ((len > 4) && !strncasecmp(&what[len - 5], ".jpeg", 5))
 961                 return "image/jpeg";
 962         else if ((len > 3) && !strncasecmp(&what[len - 4], ".png", 4))
 963                 return "image/png";
 964         else if ((len > 3) && !strncasecmp(&what[len - 4], ".ico", 4))
 965                 return "image/x-icon";
 966         else if ((len > 3) && !strncasecmp(&what[len - 4], ".vcf", 4))
 967                 return "text/x-vcard";
 968         else if ((len > 4) && !strncasecmp(&what[len - 5], ".html", 5))
 969                 return "text/html";
 970         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htm", 4))
 971                 return "text/html";
 972         else if ((len > 3) && !strncasecmp(&what[len - 4], ".wml", 4))
 973                 return "text/vnd.wap.wml";
 974         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmls", 5))
 975                 return "text/vnd.wap.wmlscript";
 976         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmlc", 5))
 977                 return "application/vnd.wap.wmlc";
 978         else if ((len > 5) && !strncasecmp(&what[len - 6], ".wmlsc", 6))
 979                 return "application/vnd.wap.wmlscriptc";
 980         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wbmp", 5))
 981                 return "image/vnd.wap.wbmp";
 982         else
 983                 /* and let xdgmime do the fallback. */
 984                 return xdg_mime_get_mime_type_from_file_name(what);
 985 }
 986
 987 static HashList *IconHash = NULL;
 988
 989 typedef struct IconName IconName;
 990
 991 struct IconName {
 992         char *FlatName;
 993         char *FileName;
 994 };
 995
 996 static void DeleteIcon(void *IconNamePtr)
 997 {
 998         IconName *Icon = (IconName*) IconNamePtr;
 999         free(Icon->FlatName);
1000         free(Icon->FileName);
1001         free(Icon);
1002 }
1003
1004 /*
1005 static const char *PrintFlat(void *IconNamePtr)
1006 {
1007         IconName *Icon = (IconName*) IconNamePtr;
1008         return Icon->FlatName;
1009 }
1010 static const char *PrintFile(void *IconNamePtr)
1011 {
1012         IconName *Icon = (IconName*) IconNamePtr;
1013         return Icon->FileName;
1014 }
1015 */
1016
1017 #define GENSTR "x-generic"
1018 #define IGNORE_PREFIX_1 "gnome-mime"
1019 int LoadIconDir(const char *DirName)
1020 {
1021         DIR *filedir = NULL;
1022         struct dirent *filedir_entry;
1023         int d_namelen;
1024         int d_without_ext;
1025         IconName *Icon;
1026
1027         filedir = opendir (DirName);
1028         IconHash = NewHash(1, NULL);
1029         if (filedir == NULL) {
1030                 return 0;
1031         }
1032
1033         while ((filedir_entry = readdir(filedir)))
1034         {
1035                 char *MinorPtr;
1036                 char *PStart;
1037 #ifdef _DIRENT_HAVE_D_NAMLEN
1038                 d_namelen = filedir_entry->d_namlen;
1039 #else
1040                 d_namelen = strlen(filedir_entry->d_name);
1041 #endif
1042                 d_without_ext = d_namelen;
1043                 while ((d_without_ext > 0) && (filedir_entry->d_name[d_without_ext] != '.'))
1044                         d_without_ext --;
1045                 if ((d_without_ext == 0) || (d_namelen < 3))
1046                         continue;
1047
1048                 if ((sizeof(IGNORE_PREFIX_1) < d_namelen) &&
1049                     (strncmp(IGNORE_PREFIX_1,
1050                              filedir_entry->d_name,
1051                              sizeof(IGNORE_PREFIX_1) - 1) == 0)) {
1052                         PStart = filedir_entry->d_name + sizeof(IGNORE_PREFIX_1);
1053                         d_without_ext -= sizeof(IGNORE_PREFIX_1);
1054                 }
1055                 else {
1056                         PStart = filedir_entry->d_name;
1057                 }
1058                 Icon = malloc(sizeof(IconName));
1059
1060                 Icon->FileName = malloc(d_namelen + 1);
1061                 memcpy(Icon->FileName, filedir_entry->d_name, d_namelen + 1);
1062
1063                 Icon->FlatName = malloc(d_without_ext + 1);
1064                 memcpy(Icon->FlatName, PStart, d_without_ext);
1065                 Icon->FlatName[d_without_ext] = '\0';
1066                 /* Try to find Minor type in image-jpeg */
1067                 MinorPtr = strchr(Icon->FlatName, '-');
1068                 if (MinorPtr != NULL) {
1069                         size_t MinorLen;
1070                         MinorLen = 1 + d_without_ext - (MinorPtr - Icon->FlatName + 1);
1071                         if ((MinorLen == sizeof(GENSTR)) &&
1072                             (strncmp(MinorPtr + 1, GENSTR, sizeof(GENSTR)) == 0)) {
1073                                 /* ok, we found a generic filename. cut the generic. */
1074                                 *MinorPtr = '\0';
1075                                 d_without_ext = d_without_ext - (MinorPtr - Icon->FlatName);
1076                         }
1077                         else { /* Map the major / minor separator to / */
1078                                 *MinorPtr = '/';
1079                         }
1080                 }
1081
1082 //              PrintHash(IconHash, PrintFlat, PrintFile);
1083 //              printf("%s - %s\n", Icon->FlatName, Icon->FileName);
1084                 Put(IconHash, Icon->FlatName, d_without_ext, Icon, DeleteIcon);
1085 //              PrintHash(IconHash, PrintFlat, PrintFile);
1086         }
1087         closedir(filedir);
1088         return 1;
1089 }
1090
1091 const char *GetIconFilename(char *MimeType, size_t len)
1092 {
1093         void *vIcon;
1094         IconName *Icon;
1095
1096         if(IconHash == NULL)
1097                 return NULL;
1098
1099         GetHash(IconHash, MimeType, len, &vIcon), Icon = (IconName*) vIcon;
1100         /* didn't find the exact mimetype? try major only. */
1101         if (Icon == NULL) {
1102                 char * pMinor;
1103                 pMinor = strchr(MimeType, '/');
1104                 if (pMinor != NULL) {
1105                         *pMinor = '\0';
1106                         GetHash(IconHash, MimeType, pMinor - MimeType, &vIcon),
1107                                 Icon = (IconName*) vIcon;
1108                 }
1109         }
1110         if (Icon == NULL) {
1111                 return NULL;
1112         }
1113
1114         /*printf("Getting: [%s] == [%s] -> [%s]\n", MimeType, Icon->FlatName, Icon->FileName);*/
1115         return Icon->FileName;
1116 }
1117
1118 void ShutDownLibCitadelMime(void)
1119 {
1120         DeleteHash(&IconHash);
1121 }