libcitadel/lib/mime_parser.c

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