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  *
   6  * This program is open source software; you can redistribute it and/or modify
   7  * it under the terms of the GNU General Public License as published by
   8  * the Free Software Foundation; either version 3 of the License, or
   9  * (at your option) any later version.
  10  *
  11  * This program is distributed in the hope that it will be useful,
  12  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  13  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  14  * GNU General Public License for more details.
  15  *
  16  * You should have received a copy of the GNU General Public License
  17  * along with this program; if not, write to the Free Software
  18  * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
  19  */
  20
  21 #include <stdlib.h>
  22 #include <unistd.h>
  23 #include <stdio.h>
  24 #include <signal.h>
  25 #include <sys/types.h>
  26 #include <ctype.h>
  27 #include <string.h>
  28 #include <sys/stat.h>
  29 #include <sys/types.h>
  30 #include <dirent.h>
  31 #include <errno.h>
  32
  33 #include "xdgmime/xdgmime.h"
  34 #include "libcitadel.h"
  35 #include "libcitadellocal.h"
  36
  37 const unsigned char FromHexTable [256] = {
  38         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //  0
  39         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 10
  40         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 20
  41         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 30
  42         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x00, 0x01, // 40
  43         0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0xFF, 0xFF, // 50
  44         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, // 60
  45         0x0F, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 70
  46         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 80
  47         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0A, 0x0B, 0x0C, // 90
  48         0x0D, 0x0E, 0x0F, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //100
  49         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //110
  50         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //120
  51         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //130
  52         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //140
  53         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //150
  54         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //160
  55         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //170
  56         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //180
  57         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //190
  58         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //200
  59         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //210
  60         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //220
  61         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //230
  62         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //240
  63         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF                          //250
  64 };
  65
  66
  67 long extract_key(char *target, char *source, long sourcelen, char *key, long keylen, char KeyEnd)
  68 {
  69         char *sptr, *ptr = NULL;
  70         int double_quotes = 0;
  71         long RealKeyLen = keylen;
  72
  73         sptr = source;
  74
  75         while (sptr != NULL)
  76         {
  77                 ptr = bmstrcasestr_len(sptr, sourcelen - (sptr - source),
  78                                        key, keylen);
  79                 if(ptr != NULL)
  80                 {
  81                         while (isspace(*(ptr + RealKeyLen)))
  82                                 RealKeyLen ++;
  83                         if (*(ptr + RealKeyLen) == KeyEnd)
  84                         {
  85                                 sptr = NULL;
  86                                 RealKeyLen ++;
  87                         }
  88                         else
  89                         {
  90                                 sptr = ptr + RealKeyLen + 1;
  91                         }
  92                 }
  93                 else
  94                         sptr = ptr;
  95         }
  96         if (ptr == NULL) {
  97                 *target = '\0';
  98                 return 0;
  99         }
 100         strcpy(target, (ptr + RealKeyLen));
 101
 102         for (ptr=target; (*ptr != 0); ptr++) {
 103
 104                 /* A semicolon means we've hit the end of the key, unless we're inside double quotes */
 105                 if ( (double_quotes != 1) && (*ptr == ';')) {
 106                         *ptr = 0;
 107                 }
 108
 109                 /* if we find double quotes, we've got a great set of string boundaries */
 110                 if (*ptr == '\"') {
 111                         ++double_quotes;
 112                         if (double_quotes == 1) {
 113                                 strcpy(ptr, ptr+1);
 114                         }
 115                         else {
 116                                 *ptr = 0;
 117                         }
 118                 }
 119         }
 120         *ptr = '\0';
 121         return ptr - target;
 122 }
 123
 124
 125 /*
 126  * For non-multipart messages, we need to generate a quickie partnum of "1"
 127  * to return to callback functions.  Some callbacks demand it.
 128  */
 129 char *fixed_partnum(char *supplied_partnum) {
 130         if (supplied_partnum == NULL) return "1";
 131         if (strlen(supplied_partnum)==0) return "1";
 132         return supplied_partnum;
 133 }
 134
 135
 136 static inline unsigned int _decode_hex(const char *Source)
 137 {
 138         int ret = '?';
 139         unsigned char LO_NIBBLE;
 140         unsigned char HI_NIBBLE;
 141
 142         HI_NIBBLE = FromHexTable[(unsigned char) *Source];
 143         LO_NIBBLE = FromHexTable[(unsigned char) *(Source+1)];
 144
 145         if ((LO_NIBBLE == 0xFF) || (LO_NIBBLE == 0xFF))
 146                 return ret;
 147         ret = HI_NIBBLE;
 148         ret = ret << 4;
 149         ret = ret | LO_NIBBLE;
 150         return ret;
 151 }
 152
 153 unsigned int decode_hex(char *Source) {return _decode_hex(Source);}
 154
 155 /*
 156  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
 157  * according to RFC2045 section 6.7
 158  */
 159 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
 160         unsigned int ch;
 161         int decoded_length = 0;
 162         int pos = 0;
 163
 164         while (pos < sourcelen)
 165         {
 166                 if (*(encoded + pos) == '=')
 167                 {
 168                         pos ++;
 169                         if (*(encoded + pos) == '\n')
 170                         {
 171                                 pos ++;
 172                         }
 173                         else if (*(encoded + pos) == '\r')
 174                         {
 175                                 pos ++;
 176                                 if (*(encoded + pos) == '\n')
 177                                         pos++;
 178                         }
 179                         else
 180                         {
 181                                 ch = 0;
 182                                 ch = _decode_hex(&encoded[pos]);
 183                                 pos += 2;
 184                                 decoded[decoded_length++] = ch;
 185                         }
 186                 }
 187                 else
 188                 {
 189                         decoded[decoded_length++] = encoded[pos];
 190                         pos += 1;
 191                 }
 192         }
 193         decoded[decoded_length] = 0;
 194         return(decoded_length);
 195 }
 196
 197
 198 /*
 199  * Given a message or message-part body and a length, handle any necessary
 200  * decoding and pass the request up the stack.
 201  */
 202 void mime_decode(char *partnum,
 203                  char *part_start, size_t length,
 204                  char *content_type, char *charset, char *encoding,
 205                  char *disposition,
 206                  char *id,
 207                  char *name, char *filename,
 208                  MimeParserCallBackType CallBack,
 209                  MimeParserCallBackType PreMultiPartCallBack,
 210                  MimeParserCallBackType PostMultiPartCallBack,
 211                  void *userdata,
 212                  int dont_decode)
 213 {
 214
 215         char *decoded;
 216         size_t bytes_decoded = 0;
 217
 218         /* Some encodings aren't really encodings */
 219         if (!strcasecmp(encoding, "7bit"))
 220                 strcpy(encoding, "");
 221         if (!strcasecmp(encoding, "8bit"))
 222                 strcpy(encoding, "");
 223         if (!strcasecmp(encoding, "binary"))
 224                 strcpy(encoding, "");
 225         if (!strcasecmp(encoding, "ISO-8859-1"))
 226                 strcpy(encoding, "");
 227
 228         /* If this part is not encoded, send as-is */
 229         if ( (strlen(encoding) == 0) || (dont_decode)) {
 230                 if (CallBack != NULL) {
 231                         CallBack(name,
 232                                  filename,
 233                                  fixed_partnum(partnum),
 234                                  disposition,
 235                                  part_start,
 236                                  content_type,
 237                                  charset,
 238                                  length,
 239                                  encoding,
 240                                  id,
 241                                  userdata);
 242                         }
 243                 return;
 244         }
 245
 246         /* Fail silently if we hit an unknown encoding. */
 247         if ((strcasecmp(encoding, "base64"))
 248             && (strcasecmp(encoding, "quoted-printable"))) {
 249                 return;
 250         }
 251
 252         /*
 253          * Allocate a buffer for the decoded data.  The output buffer is slightly
 254          * larger than the input buffer; this assumes that the decoded data
 255          * will never be significantly larger than the encoded data.  This is a
 256          * safe assumption with base64, uuencode, and quoted-printable.
 257          */
 258         decoded = malloc(length + 32768);
 259         if (decoded == NULL) {
 260                 return;
 261         }
 262
 263         if (!strcasecmp(encoding, "base64")) {
 264                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 265         }
 266         else if (!strcasecmp(encoding, "quoted-printable")) {
 267                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded, part_start, length);
 268         }
 269
 270         if (bytes_decoded > 0) if (CallBack != NULL) {
 271                         char encoding_buf[SIZ];
 272
 273                         strcpy(encoding_buf, "binary");
 274                         CallBack(name,
 275                                  filename,
 276                                  fixed_partnum(partnum),
 277                                  disposition,
 278                                  decoded,
 279                                  content_type,
 280                                  charset,
 281                                  bytes_decoded,
 282                                  encoding_buf,
 283                                  id,
 284                                  userdata);
 285         }
 286
 287         free(decoded);
 288 }
 289
 290 /*
 291  * this is the extract of mime_decode which can be called if 'dont_decode' was set;
 292  * to save the cpu intense process of decoding to the time when it realy wants the content.
 293  * returns:
 294  *   - > 0 we decoded something, its on *decoded, you need to free it.
 295  *   - = 0 no need to decode stuff. *decoded will be NULL.
 296  *   - < 0 an error occured, either an unknown encoding, or alloc failed. no need to free.
 297  */
 298 int mime_decode_now (char *part_start,
 299                      size_t length,
 300                      char *encoding,
 301                      char **decoded,
 302                      size_t *bytes_decoded)
 303 {
 304         *bytes_decoded = 0;
 305         *decoded = NULL;
 306         /* Some encodings aren't really encodings */
 307         if (!strcasecmp(encoding, "7bit"))
 308                 strcpy(encoding, "");
 309         if (!strcasecmp(encoding, "8bit"))
 310                 strcpy(encoding, "");
 311         if (!strcasecmp(encoding, "binary"))
 312                 strcpy(encoding, "");
 313
 314         /* If this part is not encoded, send as-is */
 315         if (strlen(encoding) == 0) {
 316                 return 0;
 317         }
 318
 319
 320         /* Fail if we hit an unknown encoding. */
 321         if ((strcasecmp(encoding, "base64"))
 322             && (strcasecmp(encoding, "quoted-printable"))) {
 323                 return -1;
 324         }
 325
 326         /*
 327          * Allocate a buffer for the decoded data.  The output buffer is slightly
 328          * larger than the input buffer; this assumes that the decoded data
 329          * will never be significantly larger than the encoded data.  This is a
 330          * safe assumption with base64, uuencode, and quoted-printable.
 331          */
 332         *decoded = malloc(length + 32768);
 333         if (decoded == NULL) {
 334                 return -1;
 335         }
 336
 337         if (!strcasecmp(encoding, "base64")) {
 338                 *bytes_decoded = CtdlDecodeBase64(*decoded, part_start, length);
 339                 return 1;
 340         }
 341         else if (!strcasecmp(encoding, "quoted-printable")) {
 342                 *bytes_decoded = CtdlDecodeQuotedPrintable(*decoded, part_start, length);
 343                 return 1;
 344         }
 345         return -1;
 346 }
 347
 348 typedef enum _eIntMimeHdrs {
 349         boundary,
 350         startary,
 351         endary,
 352         content_type,
 353         charset,
 354         encoding,
 355         content_type_name,
 356         content_disposition_name,
 357         filename,
 358         disposition,
 359         id,
 360         eMax /* don't move ! */
 361 } eIntMimeHdrs;
 362
 363 typedef struct _CBufStr {
 364         char Key[SIZ];
 365         long len;
 366 }CBufStr;
 367
 368 typedef struct _interesting_mime_headers {
 369         CBufStr b[eMax];
 370         long content_length;
 371         long is_multipart;
 372 } interesting_mime_headers;
 373
 374
 375 static void FlushInterestingMimes(interesting_mime_headers *m)
 376 {
 377         int i;
 378
 379         for (i = 0; i < eMax; i++) {
 380              m->b[i].Key[0] = '\0';
 381              m->b[i].len = 0;
 382         }
 383         m->content_length = -1;
 384 }
 385 static interesting_mime_headers *InitInterestingMimes(void)
 386 {
 387         interesting_mime_headers *m;
 388         m = (interesting_mime_headers*) malloc( sizeof(interesting_mime_headers));
 389
 390         FlushInterestingMimes(m);
 391
 392         return m;
 393 }
 394
 395
 396 static long parse_MimeHeaders(interesting_mime_headers *m,
 397                               char** pcontent_start,
 398                               char *content_end)
 399 {
 400         char buf[SIZ];
 401         char header[SIZ];
 402         long headerlen;
 403         char *ptr, *pch;
 404         int buflen = 0;
 405         int i;
 406
 407         /* Learn interesting things from the headers */
 408         ptr = *pcontent_start;
 409         *header = '\0';
 410         headerlen = 0;
 411         do {
 412                 ptr = memreadlinelen(ptr, buf, SIZ, &buflen);
 413
 414                 for (i = 0; i < buflen; ++i) {
 415                         if (isspace(buf[i])) {
 416                                 buf[i] = ' ';
 417                         }
 418                 }
 419
 420                 if (!isspace(buf[0]) && (headerlen > 0)) {
 421                         if (!strncasecmp(header, "Content-type:", 13)) {
 422                                 memcpy (m->b[content_type].Key, &header[13], headerlen - 12);
 423                                 m->b[content_type].Key[headerlen - 12] = '\0';
 424                                 m->b[content_type].len = striplt (m->b[content_type].Key);
 425
 426                                 m->b[content_type_name].len = extract_key(m->b[content_type_name].Key, CKEY(m->b[content_type]), HKEY("name"), '=');
 427                                 m->b[charset].len           = extract_key(m->b[charset].Key,           CKEY(m->b[content_type]), HKEY("charset"), '=');
 428                                 m->b[boundary].len          = extract_key(m->b[boundary].Key,          header,       headerlen,  HKEY("boundary"), '=');
 429
 430                                 /* Deal with weird headers */
 431                                 pch = strchr(m->b[content_type].Key, ' ');
 432                                 if (pch != NULL) {
 433                                         *pch = '\0';
 434                                         m->b[content_type].len = m->b[content_type].Key - pch;
 435                                 }
 436                                 pch = strchr(m->b[content_type].Key, ';');
 437                                 if (pch != NULL) {
 438                                         *pch = '\0';
 439                                         m->b[content_type].len = m->b[content_type].Key - pch;
 440                                 }
 441                         }
 442                         else if (!strncasecmp(header, "Content-Disposition:", 20)) {
 443                                 memcpy (m->b[disposition].Key, &header[20], headerlen - 19);
 444                                 m->b[disposition].Key[headerlen - 19] = '\0';
 445                                 m->b[disposition].len = striplt(m->b[disposition].Key);
 446
 447                                 m->b[content_disposition_name].len = extract_key(m->b[content_disposition_name].Key, CKEY(m->b[disposition]), HKEY("name"), '=');
 448                                 m->b[filename].len                 = extract_key(m->b[filename].Key,                 CKEY(m->b[disposition]), HKEY("filename"), '=');
 449                                 pch = strchr(m->b[disposition].Key, ';');
 450                                 if (pch != NULL) *pch = '\0';
 451                                 m->b[disposition].len = striplt(m->b[disposition].Key);
 452                         }
 453                         else if (!strncasecmp(header, "Content-ID:", 11)) {
 454                                 memcpy(m->b[id].Key, &header[11], headerlen - 11);
 455                                 m->b[id].Key[headerlen - 11] = '\0';
 456                                 striplt(m->b[id].Key);
 457                                 m->b[id].len = stripallbut(m->b[id].Key, '<', '>');
 458                         }
 459                         else if (!strncasecmp(header, "Content-length: ", 15)) {
 460                                 char *clbuf;
 461                                 clbuf = &header[15];
 462                                 while (isspace(*clbuf))
 463                                         clbuf ++;
 464                                 m->content_length = (size_t) atol(clbuf);
 465                         }
 466                         else if (!strncasecmp(header, "Content-transfer-encoding: ", 26)) {
 467                                 memcpy(m->b[encoding].Key, &header[26], headerlen - 26);
 468                                 m->b[encoding].Key[headerlen - 26] = '\0';
 469                                 m->b[encoding].len = striplt(m->b[encoding].Key);
 470                         }
 471                         *header = '\0';
 472                         headerlen = 0;
 473                 }
 474                 if ((headerlen + buflen + 2) < SIZ) {
 475                         memcpy(&header[headerlen], buf, buflen);
 476                         headerlen += buflen;
 477                         header[headerlen] = '\0';
 478                 }
 479                 if (ptr >= content_end) {
 480                         return -1;
 481                 }
 482         } while ((!IsEmptyStr(buf)) && (*ptr != 0));
 483
 484         m->is_multipart = m->b[boundary].len != 0;
 485         *pcontent_start = ptr;
 486
 487         return 0;
 488 }
 489
 490
 491 static int IsAsciiEncoding(interesting_mime_headers *m)
 492 {
 493
 494         if ((m->b[encoding].len != 0) &&
 495             (strcasecmp(m->b[encoding].Key, "base64") == 0))
 496                 return 1;
 497         if ((m->b[encoding].len != 0) &&
 498             (strcmp(m->b[encoding].Key, "quoted-printable") == 0))
 499                 return 1;
 500
 501         return 0;
 502 }
 503
 504 static char *FindNextContent(char *ptr,
 505                              char *content_end,
 506                              interesting_mime_headers *SubMimeHeaders,
 507                              interesting_mime_headers *m)
 508 {
 509         char *next_boundary;
 510         char  tmp;
 511
 512         if (IsAsciiEncoding(SubMimeHeaders)) {
 513                 tmp = *content_end;
 514                 *content_end = '\0';
 515
 516                 /**
 517                  * ok, if we have a content length of the mime part,
 518                  * try skipping the content on the search for the next
 519                  * boundary. since we don't trust the content_length
 520                  * to be all accurate, and suspect it to lose one digit
 521                  * per line with a line length of 80 chars, we need
 522                  * to start searching a little before..
 523                  */
 524
 525                 if ((SubMimeHeaders->content_length != -1) &&
 526                     (SubMimeHeaders->content_length > 10))
 527                 {
 528                         char *pptr;
 529                         long lines;
 530
 531                         lines = SubMimeHeaders->content_length / 80;
 532                         pptr = ptr + SubMimeHeaders->content_length - lines - 10;
 533                         if (pptr < content_end)
 534                                 ptr = pptr;
 535                 }
 536
 537                 next_boundary = strstr(ptr, m->b[startary].Key);
 538                 *content_end = tmp;
 539         }
 540         else {
 541                 char *srch;
 542                 /**
 543                  * ok, if we have a content length of the mime part,
 544                  * try skipping the content on the search for the next
 545                  * boundary. since we don't trust the content_length
 546                  * to be all accurate, start searching a little before..
 547                  */
 548
 549                 if ((SubMimeHeaders->content_length != -1) &&
 550                     (SubMimeHeaders->content_length > 10))
 551                 {
 552                         char *pptr;
 553                         pptr = ptr + SubMimeHeaders->content_length - 10;
 554                         if (pptr < content_end)
 555                                 ptr = pptr;
 556                 }
 557
 558
 559                 srch = next_boundary = NULL;
 560                 for (srch = memchr(ptr, '-',  content_end - ptr);
 561                      (srch != NULL) && (srch < content_end);
 562                      srch = memchr(srch, '-',  content_end - srch))
 563                 {
 564                         if (!memcmp(srch,
 565                                     m->b[startary].Key,
 566                                     m->b[startary].len))
 567                         {
 568                                 next_boundary = srch;
 569                                 srch = content_end;
 570                         }
 571                         else srch ++;
 572
 573                 }
 574
 575         }
 576         return next_boundary;
 577 }
 578
 579 /*
 580  * Break out the components of a multipart message
 581  * (This function expects to be fed HEADERS + CONTENT)
 582  * Note: NULL can be supplied as content_end; in this case, the message is
 583  * considered to have ended when the parser encounters a 0x00 byte.
 584  */
 585 static void recurseable_mime_parser(char *partnum,
 586                                     char *content_start, char *content_end,
 587                                     MimeParserCallBackType CallBack,
 588                                     MimeParserCallBackType PreMultiPartCallBack,
 589                                     MimeParserCallBackType PostMultiPartCallBack,
 590                                     void *userdata,
 591                                     int dont_decode,
 592                                     interesting_mime_headers *m)
 593 {
 594         interesting_mime_headers *SubMimeHeaders;
 595         char     *ptr;
 596         char     *part_start;
 597         char     *part_end = NULL;
 598         char     *evaluate_crlf_ptr = NULL;
 599         char     *next_boundary;
 600         char      nested_partnum[256];
 601         int       crlf_in_use = 0;
 602         int       part_seq = 0;
 603         CBufStr  *chosen_name;
 604
 605
 606         /* If this is a multipart message, then recursively process it */
 607         ptr = content_start;
 608         part_start = NULL;
 609         if (m->is_multipart) {
 610
 611                 /* Tell the client about this message's multipartedness */
 612                 if (PreMultiPartCallBack != NULL) {
 613                         PreMultiPartCallBack("",
 614                                              "",
 615                                              partnum,
 616                                              "",
 617                                              NULL,
 618                                              m->b[content_type].Key,
 619                                              m->b[charset].Key,
 620                                              0,
 621                                              m->b[encoding].Key,
 622                                              m->b[id].Key,
 623                                              userdata);
 624                 }
 625
 626                 /* Figure out where the boundaries are */
 627                 m->b[startary].len = snprintf(m->b[startary].Key, SIZ, "--%s", m->b[boundary].Key);
 628                 SubMimeHeaders = InitInterestingMimes ();
 629                 if (*ptr == '\r')
 630                         ptr ++;
 631                 if (*ptr == '\n')
 632                         ptr ++;
 633                 if (strncmp(ptr, m->b[startary].Key, m->b[startary].len) == 0)
 634                         ptr += m->b[startary].len;
 635                 if (*ptr == '\r')
 636                         ptr ++;
 637                 if (*ptr == '\n')
 638                         ptr ++;
 639                 part_start = NULL;
 640                 do {
 641                         char *optr;
 642
 643                         optr = ptr;
 644                         if (parse_MimeHeaders(SubMimeHeaders, &ptr, content_end) != 0)
 645                                 break;
 646                         if ((ptr - optr > 2) &&
 647                             (*(ptr - 2) == '\r'))
 648                                 crlf_in_use = 1;
 649
 650                         part_start = ptr;
 651
 652                         next_boundary = FindNextContent(ptr,
 653                                                         content_end,
 654                                                         SubMimeHeaders,
 655                                                         m);
 656                         if ((next_boundary != NULL) &&
 657                             (next_boundary - part_start < 3)) {
 658                                 FlushInterestingMimes(SubMimeHeaders);
 659
 660                                 continue;
 661                         }
 662
 663                         if ( (part_start != NULL) && (next_boundary != NULL) ) {
 664                                 part_end = next_boundary;
 665                                 --part_end;             /* omit the trailing LF */
 666                                 if (crlf_in_use) {
 667                                         --part_end;     /* omit the trailing CR */
 668                                 }
 669
 670                                 if (!IsEmptyStr(partnum)) {
 671                                         snprintf(nested_partnum,
 672                                                  sizeof nested_partnum,
 673                                                  "%s.%d", partnum,
 674                                                  ++part_seq);
 675                                 }
 676                                 else {
 677                                         snprintf(nested_partnum,
 678                                                  sizeof nested_partnum,
 679                                                  "%d", ++part_seq);
 680                                 }
 681                                 recurseable_mime_parser(nested_partnum,
 682                                                         part_start,
 683                                                         part_end,
 684                                                         CallBack,
 685                                                         PreMultiPartCallBack,
 686                                                         PostMultiPartCallBack,
 687                                                         userdata,
 688                                                         dont_decode,
 689                                                         SubMimeHeaders);
 690                         }
 691
 692                         if (next_boundary != NULL) {
 693                                 /* If we pass out of scope, don't attempt to
 694                                  * read past the end boundary. */
 695                                 if ((*(next_boundary + m->b[startary].len) == '-') &&
 696                                     (*(next_boundary + m->b[startary].len + 1) == '-') ){
 697                                         ptr = content_end;
 698                                 }
 699                                 else {
 700                                         /* Set up for the next part. */
 701                                         part_start = strstr(next_boundary, "\n");
 702
 703                                         /* Determine whether newlines are LF or CRLF */
 704                                         evaluate_crlf_ptr = part_start;
 705                                         --evaluate_crlf_ptr;
 706                                         if ((*evaluate_crlf_ptr == '\r') &&
 707                                             (*(evaluate_crlf_ptr + 1) == '\n'))
 708                                         {
 709                                                 crlf_in_use = 1;
 710                                         }
 711                                         else {
 712                                                 crlf_in_use = 0;
 713                                         }
 714
 715                                         /* Advance past the LF ... now we're in the next part */
 716                                         ++part_start;
 717                                         ptr = part_start;
 718                                 }
 719                         }
 720                         else {
 721                                 /* Invalid end of multipart.  Bail out! */
 722                                 ptr = content_end;
 723                         }
 724                         FlushInterestingMimes(SubMimeHeaders);
 725                 } while ( (ptr < content_end) && (next_boundary != NULL) );
 726
 727                 free(SubMimeHeaders);
 728
 729                 if (PostMultiPartCallBack != NULL) {
 730                         PostMultiPartCallBack("",
 731                                               "",
 732                                               partnum,
 733                                               "",
 734                                               NULL,
 735                                               m->b[content_type].Key,
 736                                               m->b[charset].Key,
 737                                               0,
 738                                               m->b[encoding].Key,
 739                                               m->b[id].Key,
 740                                               userdata);
 741                 }
 742         } /* If it's not a multipart message, then do something with it */
 743         else {
 744                 size_t length;
 745                 part_start = ptr;
 746                 length = content_end - part_start;
 747                 ptr = part_end = content_end;
 748
 749
 750                 /* The following code will truncate the MIME part to the size
 751                  * specified by the Content-length: header.   We have commented it
 752                  * out because these headers have a tendency to be wrong.
 753                  *
 754                  *      if ( (content_length > 0) && (length > content_length) ) {
 755                  *              length = content_length;
 756                  *      }
 757                  */
 758
 759                 /* Sometimes the "name" field is tacked on to Content-type,
 760                  * and sometimes it's tacked on to Content-disposition.  Use
 761                  * whichever one we have.
 762                  */
 763                 if (m->b[content_disposition_name].len > m->b[content_type_name].len) {
 764                         chosen_name = &m->b[content_disposition_name];
 765                 }
 766                 else {
 767                         chosen_name = &m->b[content_type_name];
 768                 }
 769
 770                 /* Ok, we've got a non-multipart part here, so do something with it.
 771                  */
 772                 mime_decode(partnum,
 773                             part_start,
 774                             length,
 775                             m->b[content_type].Key,
 776                             m->b[charset].Key,
 777                             m->b[encoding].Key,
 778                             m->b[disposition].Key,
 779                             m->b[id].Key,
 780                             chosen_name->Key,
 781                             m->b[filename].Key,
 782                             CallBack,
 783                             NULL, NULL,
 784                             userdata,
 785                             dont_decode
 786                         );
 787
 788                 /*
 789                  * Now if it's an encapsulated message/rfc822 then we have to recurse into it
 790                  */
 791                 if (!strcasecmp(&m->b[content_type].Key[0], "message/rfc822")) {
 792
 793                         if (PreMultiPartCallBack != NULL) {
 794                                 PreMultiPartCallBack("",
 795                                                      "",
 796                                                      partnum,
 797                                                      "",
 798                                                      NULL,
 799                                                      m->b[content_type].Key,
 800                                                      m->b[charset].Key,
 801                                                      0,
 802                                                      m->b[encoding].Key,
 803                                                      m->b[id].Key,
 804                                                      userdata);
 805                         }
 806                         if (CallBack != NULL) {
 807                                 if (strlen(partnum) > 0) {
 808                                         snprintf(nested_partnum,
 809                                                  sizeof nested_partnum,
 810                                                  "%s.%d", partnum,
 811                                                  ++part_seq);
 812                                 }
 813                                 else {
 814                                         snprintf(nested_partnum,
 815                                                  sizeof nested_partnum,
 816                                                  "%d", ++part_seq);
 817                                 }
 818                                 the_mime_parser(nested_partnum,
 819                                                 part_start,
 820                                                 part_end,
 821                                                 CallBack,
 822                                                 PreMultiPartCallBack,
 823                                                 PostMultiPartCallBack,
 824                                                 userdata,
 825                                                 dont_decode
 826                                         );
 827                         }
 828                         if (PostMultiPartCallBack != NULL) {
 829                                 PostMultiPartCallBack("",
 830                                                       "",
 831                                                       partnum,
 832                                                       "",
 833                                                       NULL,
 834                                                       m->b[content_type].Key,
 835                                                       m->b[charset].Key,
 836                                                       0,
 837                                                       m->b[encoding].Key,
 838                                                       m->b[id].Key,
 839                                                       userdata);
 840                         }
 841
 842
 843                 }
 844
 845         }
 846
 847 }
 848
 849 /*
 850  * Break out the components of a multipart message
 851  * (This function expects to be fed HEADERS + CONTENT)
 852  * Note: NULL can be supplied as content_end; in this case, the message is
 853  * considered to have ended when the parser encounters a 0x00 byte.
 854  */
 855 void the_mime_parser(char *partnum,
 856                      char *content_start, char *content_end,
 857                      MimeParserCallBackType CallBack,
 858                      MimeParserCallBackType PreMultiPartCallBack,
 859                      MimeParserCallBackType PostMultiPartCallBack,
 860                      void *userdata,
 861                      int dont_decode)
 862 {
 863         interesting_mime_headers *m;
 864
 865         /* If the caller didn't supply an endpointer, generate one by measure */
 866         if (content_end == NULL) {
 867                 content_end = &content_start[strlen(content_start)];
 868         }
 869
 870         m = InitInterestingMimes();
 871
 872         if (!parse_MimeHeaders(m, &content_start, content_end))
 873         {
 874
 875                 recurseable_mime_parser(partnum,
 876                                         content_start, content_end,
 877                                         CallBack,
 878                                         PreMultiPartCallBack,
 879                                         PostMultiPartCallBack,
 880                                         userdata,
 881                                         dont_decode,
 882                                         m);
 883         }
 884         free(m);
 885 }
 886
 887 /*
 888  * Entry point for the MIME parser.
 889  * (This function expects to be fed HEADERS + CONTENT)
 890  * Note: NULL can be supplied as content_end; in this case, the message is
 891  * considered to have ended when the parser encounters a 0x00 byte.
 892  */
 893 void mime_parser(char *content_start,
 894                  char *content_end,
 895                  MimeParserCallBackType CallBack,
 896                  MimeParserCallBackType PreMultiPartCallBack,
 897                  MimeParserCallBackType PostMultiPartCallBack,
 898                  void *userdata,
 899                  int dont_decode)
 900 {
 901
 902         the_mime_parser("", content_start, content_end,
 903                         CallBack,
 904                         PreMultiPartCallBack,
 905                         PostMultiPartCallBack,
 906                         userdata, dont_decode);
 907 }
 908
 909
 910
 911
 912
 913
 914 typedef struct _MimeGuess {
 915         const char *Pattern;
 916         size_t PatternLen;
 917         long PatternOffset;
 918         const char *MimeString;
 919 } MimeGuess;
 920
 921 MimeGuess MyMimes [] = {
 922         {
 923                 "GIF",
 924                 3,
 925                 0,
 926                 "image/gif"
 927         },
 928         {
 929                 "\xff\xd8",
 930                 2,
 931                 0,
 932                 "image/jpeg"
 933         },
 934         {
 935                 "\x89PNG",
 936                 4,
 937                 0,
 938                 "image/png"
 939         },
 940         { // last...
 941                 "",
 942                 0,
 943                 0,
 944                 ""
 945         }
 946 };
 947
 948
 949 const char *GuessMimeType(const char *data, size_t dlen)
 950 {
 951         int MimeIndex = 0;
 952
 953         while (MyMimes[MimeIndex].PatternLen != 0)
 954         {
 955                 if ((MyMimes[MimeIndex].PatternLen +
 956                      MyMimes[MimeIndex].PatternOffset < dlen) &&
 957                     strncmp(MyMimes[MimeIndex].Pattern,
 958                             &data[MyMimes[MimeIndex].PatternOffset],
 959                             MyMimes[MimeIndex].PatternLen) == 0)
 960                 {
 961                         return MyMimes[MimeIndex].MimeString;
 962                 }
 963                 MimeIndex ++;
 964         }
 965         /*
 966          * ok, our simple minded algorythm didn't find anything,
 967          * let the big chegger try it, he wil default to application/octet-stream
 968          */
 969         return (xdg_mime_get_mime_type_for_data(data, dlen));
 970 }
 971
 972
 973 const char* GuessMimeByFilename(const char *what, size_t len)
 974 {
 975         /* we know some hardcoded on our own, try them... */
 976         if ((len > 3) && !strncasecmp(&what[len - 4], ".gif", 4))
 977                 return "image/gif";
 978         else if ((len > 2) && !strncasecmp(&what[len - 3], ".js", 3))
 979                 return  "text/javascript";
 980         else if ((len > 3) && !strncasecmp(&what[len - 4], ".txt", 4))
 981                 return "text/plain";
 982         else if ((len > 3) && !strncasecmp(&what[len - 4], ".css", 4))
 983                 return "text/css";
 984         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htc", 4))
 985                 return "text/x-component";
 986         else if ((len > 3) && !strncasecmp(&what[len - 4], ".jpg", 4))
 987                 return "image/jpeg";
 988         else if ((len > 3) && !strncasecmp(&what[len - 4], ".png", 4))
 989                 return "image/png";
 990         else if ((len > 3) && !strncasecmp(&what[len - 4], ".ico", 4))
 991                 return "image/x-icon";
 992         else if ((len > 3) && !strncasecmp(&what[len - 4], ".vcf", 4))
 993                 return "text/x-vcard";
 994         else if ((len > 4) && !strncasecmp(&what[len - 5], ".html", 5))
 995                 return "text/html";
 996         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htm", 4))
 997                 return "text/html";
 998         else if ((len > 3) && !strncasecmp(&what[len - 4], ".wml", 4))
 999                 return "text/vnd.wap.wml";
1000         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmls", 5))
1001                 return "text/vnd.wap.wmlscript";
1002         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmlc", 5))
1003                 return "application/vnd.wap.wmlc";
1004         else if ((len > 5) && !strncasecmp(&what[len - 6], ".wmlsc", 6))
1005                 return "application/vnd.wap.wmlscriptc";
1006         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wbmp", 5))
1007                 return "image/vnd.wap.wbmp";
1008         else
1009                 /* and let xdgmime do the fallback. */
1010                 return xdg_mime_get_mime_type_from_file_name(what);
1011 }
1012
1013 static HashList *IconHash = NULL;
1014
1015 typedef struct IconName IconName;
1016
1017 struct IconName {
1018         char *FlatName;
1019         char *FileName;
1020 };
1021
1022 static void DeleteIcon(void *IconNamePtr)
1023 {
1024         IconName *Icon = (IconName*) IconNamePtr;
1025         free(Icon->FlatName);
1026         free(Icon->FileName);
1027         free(Icon);
1028 }
1029
1030 /*
1031 static const char *PrintFlat(void *IconNamePtr)
1032 {
1033         IconName *Icon = (IconName*) IconNamePtr;
1034         return Icon->FlatName;
1035 }
1036 static const char *PrintFile(void *IconNamePtr)
1037 {
1038         IconName *Icon = (IconName*) IconNamePtr;
1039         return Icon->FileName;
1040 }
1041 */
1042
1043 #define GENSTR "x-generic"
1044 #define IGNORE_PREFIX_1 "gnome-mime"
1045 int LoadIconDir(const char *DirName)
1046 {
1047         DIR *filedir = NULL;
1048         struct dirent *filedir_entry;
1049         int d_namelen;
1050         int d_without_ext;
1051         IconName *Icon;
1052
1053         filedir = opendir (DirName);
1054         IconHash = NewHash(1, NULL);
1055         if (filedir == NULL) {
1056                 return 0;
1057         }
1058
1059         while ((filedir_entry = readdir(filedir)))
1060         {
1061                 char *MinorPtr;
1062                 char *PStart;
1063 #ifdef _DIRENT_HAVE_D_NAMELEN
1064                 d_namelen = filedir_entry->d_namelen;
1065 #else
1066                 d_namelen = strlen(filedir_entry->d_name);
1067 #endif
1068                 d_without_ext = d_namelen;
1069                 while ((d_without_ext > 0) && (filedir_entry->d_name[d_without_ext] != '.'))
1070                         d_without_ext --;
1071                 if ((d_without_ext == 0) || (d_namelen < 3))
1072                         continue;
1073
1074                 if ((sizeof(IGNORE_PREFIX_1) < d_namelen) &&
1075                     (strncmp(IGNORE_PREFIX_1,
1076                              filedir_entry->d_name,
1077                              sizeof(IGNORE_PREFIX_1) - 1) == 0)) {
1078                         PStart = filedir_entry->d_name + sizeof(IGNORE_PREFIX_1);
1079                         d_without_ext -= sizeof(IGNORE_PREFIX_1);
1080                 }
1081                 else {
1082                         PStart = filedir_entry->d_name;
1083                 }
1084                 Icon = malloc(sizeof(IconName));
1085
1086                 Icon->FileName = malloc(d_namelen + 1);
1087                 memcpy(Icon->FileName, filedir_entry->d_name, d_namelen + 1);
1088
1089                 Icon->FlatName = malloc(d_without_ext + 1);
1090                 memcpy(Icon->FlatName, PStart, d_without_ext);
1091                 Icon->FlatName[d_without_ext] = '\0';
1092                 /* Try to find Minor type in image-jpeg */
1093                 MinorPtr = strchr(Icon->FlatName, '-');
1094                 if (MinorPtr != NULL) {
1095                         size_t MinorLen;
1096                         MinorLen = 1 + d_without_ext - (MinorPtr - Icon->FlatName + 1);
1097                         if ((MinorLen == sizeof(GENSTR)) &&
1098                             (strncmp(MinorPtr + 1, GENSTR, sizeof(GENSTR)) == 0)) {
1099                                 /* ok, we found a generic filename. cut the generic. */
1100                                 *MinorPtr = '\0';
1101                                 d_without_ext = d_without_ext - (MinorPtr - Icon->FlatName);
1102                         }
1103                         else { /* Map the major / minor separator to / */
1104                                 *MinorPtr = '/';
1105                         }
1106                 }
1107
1108 //              PrintHash(IconHash, PrintFlat, PrintFile);
1109 //              printf("%s - %s\n", Icon->FlatName, Icon->FileName);
1110                 Put(IconHash, Icon->FlatName, d_without_ext, Icon, DeleteIcon);
1111 //              PrintHash(IconHash, PrintFlat, PrintFile);
1112         }
1113         closedir(filedir);
1114         return 1;
1115 }
1116
1117 const char *GetIconFilename(char *MimeType, size_t len)
1118 {
1119         void *vIcon;
1120         IconName *Icon;
1121
1122         if(IconHash == NULL)
1123                 return NULL;
1124
1125         GetHash(IconHash, MimeType, len, &vIcon), Icon = (IconName*) vIcon;
1126         /* didn't find the exact mimetype? try major only. */
1127         if (Icon == NULL) {
1128                 char * pMinor;
1129                 pMinor = strchr(MimeType, '/');
1130                 if (pMinor != NULL) {
1131                         *pMinor = '\0';
1132                         GetHash(IconHash, MimeType, pMinor - MimeType, &vIcon),
1133                                 Icon = (IconName*) vIcon;
1134                 }
1135         }
1136         if (Icon == NULL) {
1137                 return NULL;
1138         }
1139
1140         /*printf("Getting: [%s] == [%s] -> [%s]\n", MimeType, Icon->FlatName, Icon->FileName);*/
1141         return Icon->FileName;
1142 }
1143
1144 void ShutDownLibCitadelMime(void)
1145 {
1146         DeleteHash(&IconHash);
1147 }