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** pcontent_start, char *content_end)
 366 {
 367         char buf[SIZ];
 368         char header[SIZ];
 369         long headerlen;
 370         char *ptr, *pch;
 371         int buflen = 0;
 372         int i;
 373
 374         /* Learn interesting things from the headers */
 375         ptr = *pcontent_start;
 376         *header = '\0';
 377         headerlen = 0;
 378         do {
 379                 ptr = memreadlinelen(ptr, buf, SIZ, &buflen);
 380
 381                 for (i = 0; i < buflen; ++i) {
 382                         if (isspace(buf[i])) {
 383                                 buf[i] = ' ';
 384                         }
 385                 }
 386
 387                 if (!isspace(buf[0]) && (headerlen > 0)) {
 388                         if (!strncasecmp(header, "Content-type:", 13)) {
 389                                 memcpy (m->b[content_type].Key, &header[13], headerlen - 12);
 390                                 m->b[content_type].len = striplt (m->b[content_type].Key);
 391
 392                                 m->b[content_type_name].len = extract_key(m->b[content_type_name].Key, CKEY(m->b[content_type]), HKEY("name"), '=');
 393                                 m->b[charset].len           = extract_key(m->b[charset].Key,           CKEY(m->b[content_type]), HKEY("charset"), '=');
 394                                 m->b[boundary].len          = extract_key(m->b[boundary].Key,          header,       headerlen,  HKEY("boundary"), '=');
 395
 396                                 /* Deal with weird headers */
 397                                 pch = strchr(m->b[content_type].Key, ' ');
 398                                 if (pch != NULL) {
 399                                         *pch = '\0';
 400                                         m->b[content_type].len = m->b[content_type].Key - pch;
 401                                 }
 402                                 pch = strchr(m->b[content_type].Key, ';');
 403                                 if (pch != NULL) {
 404                                         *pch = '\0';
 405                                         m->b[content_type].len = m->b[content_type].Key - pch;
 406                                 }
 407                         }
 408                         else if (!strncasecmp(header, "Content-Disposition:", 20)) {
 409                                 memcpy (m->b[disposition].Key, &header[20], headerlen - 19);
 410                                 m->b[disposition].len = striplt(m->b[disposition].Key);
 411
 412                                 m->b[content_disposition_name].len = extract_key(m->b[content_disposition_name].Key, CKEY(m->b[disposition]), HKEY("name"), '=');
 413                                 m->b[filename].len                 = extract_key(m->b[filename].Key,                 CKEY(m->b[disposition]), HKEY("filename"), '=');
 414                                 pch = strchr(m->b[disposition].Key, ';');
 415                                 if (pch != NULL) *ptr = '\0';
 416                                 m->b[disposition].len = striplt(m->b[disposition].Key);
 417                         }
 418                         else if (!strncasecmp(header, "Content-ID:", 11)) {
 419                                 memcpy(m->b[id].Key, &header[11], headerlen);
 420                                 striplt(m->b[id].Key);
 421                                 m->b[id].len = stripallbut(m->b[id].Key, '<', '>');
 422                         }
 423                         else if (!strncasecmp(header, "Content-length: ", 15)) {
 424                                 char *clbuf;
 425                                 clbuf = &header[15];
 426                                 while (isspace(*clbuf))
 427                                         clbuf ++;
 428                                 m->content_length = (size_t) atol(clbuf);
 429                         }
 430                         else if (!strncasecmp(header, "Content-transfer-encoding: ", 26)) {
 431                                 memcpy(m->b[encoding].Key, &header[26], headerlen - 26);
 432                                 m->b[encoding].len = striplt(m->b[encoding].Key);
 433                         }
 434                         *header = '\0';
 435                         headerlen = 0;
 436                 }
 437                 if ((headerlen + buflen + 2) < SIZ) {
 438                         memcpy(&header[headerlen], buf, buflen);
 439                         headerlen += buflen;
 440                         header[headerlen] = '\0';
 441                 }
 442                 if (ptr >= content_end) {
 443                         return -1;
 444                 }
 445         } while ((!IsEmptyStr(buf)) && (*ptr != 0));
 446
 447         m->is_multipart = m->b[boundary].len != 0;
 448         *pcontent_start = ptr;
 449
 450         return 0;
 451 }
 452
 453 /*
 454  * Break out the components of a multipart message
 455  * (This function expects to be fed HEADERS + CONTENT)
 456  * Note: NULL can be supplied as content_end; in this case, the message is
 457  * considered to have ended when the parser encounters a 0x00 byte.
 458  */
 459 void the_mime_parser(char *partnum,
 460                      char *content_start, char *content_end,
 461                      MimeParserCallBackType CallBack,
 462                      MimeParserCallBackType PreMultiPartCallBack,
 463                      MimeParserCallBackType PostMultiPartCallBack,
 464                      void *userdata,
 465                      int dont_decode)
 466 {
 467
 468         char *ptr;
 469         char *part_start, *part_end = NULL;
 470         char *next_boundary;
 471
 472         size_t content_length;
 473         int part_seq = 0;
 474         size_t length;
 475         char nested_partnum[256];
 476         int crlf_in_use = 0;
 477         char *evaluate_crlf_ptr = NULL;
 478
 479         interesting_mime_headers *m;
 480         CBufStr *chosen_name;
 481
 482         content_length = 0;
 483
 484         m = InitInterestingMimes();
 485
 486
 487         /* If the caller didn't supply an endpointer, generate one by measure */
 488         if (content_end == NULL) {
 489                 content_end = &content_start[strlen(content_start)];
 490         }
 491
 492
 493         if (parse_MimeHeaders(m, &content_start, content_end) != 0)
 494                 goto end_parser;
 495
 496         /* If this is a multipart message, then recursively process it */
 497         ptr = content_start;
 498         part_start = NULL;
 499         if (m->is_multipart) {
 500
 501                 /* Tell the client about this message's multipartedness */
 502                 if (PreMultiPartCallBack != NULL) {
 503                         PreMultiPartCallBack("",
 504                                              "",
 505                                              partnum,
 506                                              "",
 507                                              NULL,
 508                                              m->b[content_type].Key,
 509                                              m->b[charset].Key,
 510                                              0,
 511                                              m->b[encoding].Key,
 512                                              m->b[id].Key,
 513                                              userdata);
 514                 }
 515
 516                 /* Figure out where the boundaries are */
 517                 m->b[startary].len = snprintf(m->b[startary].Key, SIZ, "--%s", m->b[boundary].Key);
 518
 519                 part_start = NULL;
 520                 do {
 521                         char tmp;
 522
 523                         tmp = *content_end;
 524                         *content_end = '\0';
 525
 526                         next_boundary = strstr(ptr, m->b[startary].Key);
 527                         *content_end = tmp;
 528
 529                         if ( (part_start != NULL) && (next_boundary != NULL) ) {
 530                                 part_end = next_boundary;
 531                                 --part_end;             /* omit the trailing LF */
 532                                 if (crlf_in_use) {
 533                                         --part_end;     /* omit the trailing CR */
 534                                 }
 535
 536                                 if (!IsEmptyStr(partnum)) {
 537                                         snprintf(nested_partnum,
 538                                                  sizeof nested_partnum,
 539                                                  "%s.%d", partnum,
 540                                                  ++part_seq);
 541                                 }
 542                                 else {
 543                                         snprintf(nested_partnum,
 544                                                  sizeof nested_partnum,
 545                                                  "%d", ++part_seq);
 546                                 }
 547                                 the_mime_parser(nested_partnum,
 548                                                 part_start,
 549                                                 part_end,
 550                                                 CallBack,
 551                                                 PreMultiPartCallBack,
 552                                                 PostMultiPartCallBack,
 553                                                 userdata,
 554                                                 dont_decode);
 555                         }
 556
 557                         if (next_boundary != NULL) {
 558                                 /* If we pass out of scope, don't attempt to
 559                                  * read past the end boundary. */
 560                                 if ((*(next_boundary + m->b[startary].len + 1) == '-') &&
 561                                     (*(next_boundary + m->b[startary].len + 2) == '-') ){
 562                                         ptr = content_end;
 563                                 }
 564                                 else {
 565                                         /* Set up for the next part. */
 566                                         part_start = strstr(next_boundary, "\n");
 567
 568                                         /* Determine whether newlines are LF or CRLF */
 569                                         evaluate_crlf_ptr = part_start;
 570                                         --evaluate_crlf_ptr;
 571                                         if ((*evaluate_crlf_ptr == '\r') &&
 572                                             (*(evaluate_crlf_ptr + 1) == '\n'))
 573                                         {
 574                                                 crlf_in_use = 1;
 575                                         }
 576                                         else {
 577                                                 crlf_in_use = 0;
 578                                         }
 579
 580                                         /* Advance past the LF ... now we're in the next part */
 581                                         ++part_start;
 582                                         ptr = part_start;
 583                                 }
 584                         }
 585                         else {
 586                                 /* Invalid end of multipart.  Bail out! */
 587                                 ptr = content_end;
 588                         }
 589                 } while ( (ptr < content_end) && (next_boundary != NULL) );
 590
 591                 if (PostMultiPartCallBack != NULL) {
 592                         PostMultiPartCallBack("",
 593                                               "",
 594                                               partnum,
 595                                               "",
 596                                               NULL,
 597                                               m->b[content_type].Key,
 598                                               m->b[charset].Key,
 599                                               0,
 600                                               m->b[encoding].Key,
 601                                               m->b[id].Key,
 602                                               userdata);
 603                 }
 604         } /* If it's not a multipart message, then do something with it */
 605         else {
 606                 part_start = ptr;
 607                 length = content_end - part_start;
 608                 ptr = part_end = content_end;
 609
 610
 611                 /* The following code will truncate the MIME part to the size
 612                  * specified by the Content-length: header.   We have commented it
 613                  * out because these headers have a tendency to be wrong.
 614                  *
 615                  *      if ( (content_length > 0) && (length > content_length) ) {
 616                  *              length = content_length;
 617                  *      }
 618                  */
 619
 620                 /* Sometimes the "name" field is tacked on to Content-type,
 621                  * and sometimes it's tacked on to Content-disposition.  Use
 622                  * whichever one we have.
 623                  */
 624                 if (m->b[content_disposition_name].len > m->b[content_type_name].len) {
 625                         chosen_name = &m->b[content_disposition_name];
 626                 }
 627                 else {
 628                         chosen_name = &m->b[content_type_name];
 629                 }
 630
 631                 /* Ok, we've got a non-multipart part here, so do something with it.
 632                  */
 633                 mime_decode(partnum,
 634                             part_start,
 635                             length,
 636                             m->b[content_type].Key,
 637                             m->b[charset].Key,
 638                             m->b[encoding].Key,
 639                             m->b[disposition].Key,
 640                             m->b[id].Key,
 641                             chosen_name->Key,
 642                             m->b[filename].Key,
 643                             CallBack,
 644                             NULL, NULL,
 645                             userdata,
 646                             dont_decode
 647                         );
 648
 649                 /*
 650                  * Now if it's an encapsulated message/rfc822 then we have to recurse into it
 651                  */
 652                 if (!strcasecmp(&m->b[content_type].Key[0], "message/rfc822")) {
 653
 654                         if (PreMultiPartCallBack != NULL) {
 655                                 PreMultiPartCallBack("",
 656                                                      "",
 657                                                      partnum,
 658                                                      "",
 659                                                      NULL,
 660                                                      m->b[content_type].Key,
 661                                                      m->b[charset].Key,
 662                                                      0,
 663                                                      m->b[encoding].Key,
 664                                                      m->b[id].Key,
 665                                                      userdata);
 666                         }
 667                         if (CallBack != NULL) {
 668                                 if (strlen(partnum) > 0) {
 669                                         snprintf(nested_partnum,
 670                                                  sizeof nested_partnum,
 671                                                  "%s.%d", partnum,
 672                                                  ++part_seq);
 673                                 }
 674                                 else {
 675                                         snprintf(nested_partnum,
 676                                                  sizeof nested_partnum,
 677                                                  "%d", ++part_seq);
 678                                 }
 679                                 the_mime_parser(nested_partnum,
 680                                                 part_start,
 681                                                 part_end,
 682                                                 CallBack,
 683                                                 PreMultiPartCallBack,
 684                                                 PostMultiPartCallBack,
 685                                                 userdata,
 686                                                 dont_decode
 687                                         );
 688                         }
 689                         if (PostMultiPartCallBack != NULL) {
 690                                 PostMultiPartCallBack("",
 691                                                       "",
 692                                                       partnum,
 693                                                       "",
 694                                                       NULL,
 695                                                       m->b[content_type].Key,
 696                                                       m->b[charset].Key,
 697                                                       0,
 698                                                       m->b[encoding].Key,
 699                                                       m->b[id].Key,
 700                                                       userdata);
 701                         }
 702
 703
 704                 }
 705
 706         }
 707
 708 end_parser:     /* free the buffers!  end the oppression!! */
 709         free(m);
 710 }
 711
 712
 713
 714 /*
 715  * Entry point for the MIME parser.
 716  * (This function expects to be fed HEADERS + CONTENT)
 717  * Note: NULL can be supplied as content_end; in this case, the message is
 718  * considered to have ended when the parser encounters a 0x00 byte.
 719  */
 720 void mime_parser(char *content_start,
 721                  char *content_end,
 722                  MimeParserCallBackType CallBack,
 723                  MimeParserCallBackType PreMultiPartCallBack,
 724                  MimeParserCallBackType PostMultiPartCallBack,
 725                  void *userdata,
 726                  int dont_decode)
 727 {
 728
 729         the_mime_parser("", content_start, content_end,
 730                         CallBack,
 731                         PreMultiPartCallBack,
 732                         PostMultiPartCallBack,
 733                         userdata, dont_decode);
 734 }
 735
 736
 737
 738
 739
 740
 741 typedef struct _MimeGuess {
 742         const char *Pattern;
 743         size_t PatternLen;
 744         long PatternOffset;
 745         const char *MimeString;
 746 } MimeGuess;
 747
 748 MimeGuess MyMimes [] = {
 749         {
 750                 "GIF",
 751                 3,
 752                 0,
 753                 "image/gif"
 754         },
 755         {
 756                 "\xff\xd8",
 757                 2,
 758                 0,
 759                 "image/jpeg"
 760         },
 761         {
 762                 "\x89PNG",
 763                 4,
 764                 0,
 765                 "image/png"
 766         },
 767         { // last...
 768                 "",
 769                 0,
 770                 0,
 771                 ""
 772         }
 773 };
 774
 775
 776 const char *GuessMimeType(const char *data, size_t dlen)
 777 {
 778         int MimeIndex = 0;
 779
 780         while (MyMimes[MimeIndex].PatternLen != 0)
 781         {
 782                 if ((MyMimes[MimeIndex].PatternLen +
 783                      MyMimes[MimeIndex].PatternOffset < dlen) &&
 784                     strncmp(MyMimes[MimeIndex].Pattern,
 785                             &data[MyMimes[MimeIndex].PatternOffset],
 786                             MyMimes[MimeIndex].PatternLen) == 0)
 787                 {
 788                         return MyMimes[MimeIndex].MimeString;
 789                 }
 790                 MimeIndex ++;
 791         }
 792         /*
 793          * ok, our simple minded algorythm didn't find anything,
 794          * let the big chegger try it, he wil default to application/octet-stream
 795          */
 796         return (xdg_mime_get_mime_type_for_data(data, dlen));
 797 }
 798
 799
 800 const char* GuessMimeByFilename(const char *what, size_t len)
 801 {
 802         /* we know some hardcoded on our own, try them... */
 803         if ((len > 3) && !strncasecmp(&what[len - 4], ".gif", 4))
 804                 return "image/gif";
 805         else if ((len > 2) && !strncasecmp(&what[len - 3], ".js", 3))
 806                 return  "text/javascript";
 807         else if ((len > 3) && !strncasecmp(&what[len - 4], ".txt", 4))
 808                 return "text/plain";
 809         else if ((len > 3) && !strncasecmp(&what[len - 4], ".css", 4))
 810                 return "text/css";
 811         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htc", 4))
 812                 return "text/x-component";
 813         else if ((len > 3) && !strncasecmp(&what[len - 4], ".jpg", 4))
 814                 return "image/jpeg";
 815         else if ((len > 3) && !strncasecmp(&what[len - 4], ".png", 4))
 816                 return "image/png";
 817         else if ((len > 3) && !strncasecmp(&what[len - 4], ".ico", 4))
 818                 return "image/x-icon";
 819         else if ((len > 3) && !strncasecmp(&what[len - 4], ".vcf", 4))
 820                 return "text/x-vcard";
 821         else if ((len > 4) && !strncasecmp(&what[len - 5], ".html", 5))
 822                 return "text/html";
 823         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htm", 4))
 824                 return "text/html";
 825         else if ((len > 3) && !strncasecmp(&what[len - 4], ".wml", 4))
 826                 return "text/vnd.wap.wml";
 827         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmls", 5))
 828                 return "text/vnd.wap.wmlscript";
 829         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmlc", 5))
 830                 return "application/vnd.wap.wmlc";
 831         else if ((len > 5) && !strncasecmp(&what[len - 6], ".wmlsc", 6))
 832                 return "application/vnd.wap.wmlscriptc";
 833         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wbmp", 5))
 834                 return "image/vnd.wap.wbmp";
 835         else
 836                 /* and let xdgmime do the fallback. */
 837                 return xdg_mime_get_mime_type_from_file_name(what);
 838 }
 839
 840 static HashList *IconHash = NULL;
 841
 842 typedef struct IconName IconName;
 843
 844 struct IconName {
 845         char *FlatName;
 846         char *FileName;
 847 };
 848
 849 static void DeleteIcon(void *IconNamePtr)
 850 {
 851         IconName *Icon = (IconName*) IconNamePtr;
 852         free(Icon->FlatName);
 853         free(Icon->FileName);
 854         free(Icon);
 855 }
 856
 857 /*
 858 static const char *PrintFlat(void *IconNamePtr)
 859 {
 860         IconName *Icon = (IconName*) IconNamePtr;
 861         return Icon->FlatName;
 862 }
 863 static const char *PrintFile(void *IconNamePtr)
 864 {
 865         IconName *Icon = (IconName*) IconNamePtr;
 866         return Icon->FileName;
 867 }
 868 */
 869
 870 #define GENSTR "x-generic"
 871 #define IGNORE_PREFIX_1 "gnome-mime"
 872 int LoadIconDir(const char *DirName)
 873 {
 874         DIR *filedir = NULL;
 875         struct dirent *filedir_entry;
 876         int d_namelen;
 877         int d_without_ext;
 878         IconName *Icon;
 879
 880         filedir = opendir (DirName);
 881         IconHash = NewHash(1, NULL);
 882         if (filedir == NULL) {
 883                 return 0;
 884         }
 885
 886         while ((filedir_entry = readdir(filedir)))
 887         {
 888                 char *MinorPtr;
 889                 char *PStart;
 890 #ifdef _DIRENT_HAVE_D_NAMELEN
 891                 d_namelen = filedir_entry->d_namelen;
 892 #else
 893                 d_namelen = strlen(filedir_entry->d_name);
 894 #endif
 895                 d_without_ext = d_namelen;
 896                 while ((d_without_ext > 0) && (filedir_entry->d_name[d_without_ext] != '.'))
 897                         d_without_ext --;
 898                 if ((d_without_ext == 0) || (d_namelen < 3))
 899                         continue;
 900
 901                 if ((sizeof(IGNORE_PREFIX_1) < d_namelen) &&
 902                     (strncmp(IGNORE_PREFIX_1,
 903                              filedir_entry->d_name,
 904                              sizeof(IGNORE_PREFIX_1) - 1) == 0)) {
 905                         PStart = filedir_entry->d_name + sizeof(IGNORE_PREFIX_1);
 906                         d_without_ext -= sizeof(IGNORE_PREFIX_1);
 907                 }
 908                 else {
 909                         PStart = filedir_entry->d_name;
 910                 }
 911                 Icon = malloc(sizeof(IconName));
 912
 913                 Icon->FileName = malloc(d_namelen + 1);
 914                 memcpy(Icon->FileName, filedir_entry->d_name, d_namelen + 1);
 915
 916                 Icon->FlatName = malloc(d_without_ext + 1);
 917                 memcpy(Icon->FlatName, PStart, d_without_ext);
 918                 Icon->FlatName[d_without_ext] = '\0';
 919                 /* Try to find Minor type in image-jpeg */
 920                 MinorPtr = strchr(Icon->FlatName, '-');
 921                 if (MinorPtr != NULL) {
 922                         size_t MinorLen;
 923                         MinorLen = 1 + d_without_ext - (MinorPtr - Icon->FlatName + 1);
 924                         if ((MinorLen == sizeof(GENSTR)) &&
 925                             (strncmp(MinorPtr + 1, GENSTR, sizeof(GENSTR)) == 0)) {
 926                                 /* ok, we found a generic filename. cut the generic. */
 927                                 *MinorPtr = '\0';
 928                                 d_without_ext = d_without_ext - (MinorPtr - Icon->FlatName);
 929                         }
 930                         else { /* Map the major / minor separator to / */
 931                                 *MinorPtr = '/';
 932                         }
 933                 }
 934
 935 //              PrintHash(IconHash, PrintFlat, PrintFile);
 936 //              printf("%s - %s\n", Icon->FlatName, Icon->FileName);
 937                 Put(IconHash, Icon->FlatName, d_without_ext, Icon, DeleteIcon);
 938 //              PrintHash(IconHash, PrintFlat, PrintFile);
 939         }
 940         closedir(filedir);
 941         return 1;
 942 }
 943
 944 const char *GetIconFilename(char *MimeType, size_t len)
 945 {
 946         void *vIcon;
 947         IconName *Icon;
 948
 949         if(IconHash == NULL)
 950                 return NULL;
 951
 952         GetHash(IconHash, MimeType, len, &vIcon), Icon = (IconName*) vIcon;
 953         /* didn't find the exact mimetype? try major only. */
 954         if (Icon == NULL) {
 955                 char * pMinor;
 956                 pMinor = strchr(MimeType, '/');
 957                 if (pMinor != NULL) {
 958                         *pMinor = '\0';
 959                         GetHash(IconHash, MimeType, pMinor - MimeType, &vIcon),
 960                                 Icon = (IconName*) vIcon;
 961                 }
 962         }
 963         if (Icon == NULL) {
 964                 return NULL;
 965         }
 966
 967         /*printf("Getting: [%s] == [%s] -> [%s]\n", MimeType, Icon->FlatName, Icon->FileName);*/
 968         return Icon->FileName;
 969 }
 970
 971 void ShutDownLibCitadelMime(void)
 972 {
 973         DeleteHash(&IconHash);
 974 }