libcitadel/lib/mime_parser.c

   1 /*
   2  * $Id$
   3  *
   4  * This is the MIME parser for Citadel.
   5  *
   6  * Copyright (c) 1998-2007 by the citadel.org development team.
   7  * This code is distributed under the GNU General Public License v3.
   8  *
   9  */
  10
  11 #include <stdlib.h>
  12 #include <unistd.h>
  13 #include <stdio.h>
  14 #include <signal.h>
  15 #include <sys/types.h>
  16 #include <ctype.h>
  17 #include <string.h>
  18 #include <sys/stat.h>
  19 #include <sys/types.h>
  20 #include <dirent.h>
  21 #include <errno.h>
  22
  23 #include "xdgmime/xdgmime.h"
  24 #include "libcitadel.h"
  25 #include "libcitadellocal.h"
  26
  27 void extract_key(char *target, char *source, long sourcelen, char *key, long keylen)
  28 {
  29         char *lookat;
  30         char *ptr;
  31         //char looking_for[256];
  32         int double_quotes = 0;
  33
  34         lookat = source;
  35         //snprintf(looking_for, sizeof looking_for, "%s=", key);
  36
  37         while ((lookat != NULL)) {
  38                 ptr = bmstrcasestr_len(lookat, sourcelen, key, keylen);
  39                 lookat = ptr;
  40                 if ((ptr != NULL) &&
  41                     (*(ptr + keylen) == '='))
  42                         lookat = NULL;
  43
  44
  45         }
  46         //ptr = bmstrcasestr(source, looking_for);
  47         if (ptr == NULL) {
  48                 strcpy(target, "");
  49                 return;
  50         }
  51         strcpy(target, (ptr + keylen + 1));
  52
  53         for (ptr=target; (*ptr != 0); ++ptr) {
  54
  55                 /* A semicolon means we've hit the end of the key, unless we're inside double quotes */
  56                 if ( (double_quotes != 1) && (*ptr == ';')) {
  57                         *ptr = 0;
  58                 }
  59
  60                 /* if we find double quotes, we've got a great set of string boundaries */
  61                 if (*ptr == '\"') {
  62                         ++double_quotes;
  63                         if (double_quotes == 1) {
  64                                 strcpy(ptr, ptr+1);
  65                         }
  66                         else {
  67                                 *ptr = 0;
  68                         }
  69                 }
  70         }
  71         *ptr = '\0';
  72 }
  73
  74
  75 /*
  76  * For non-multipart messages, we need to generate a quickie partnum of "1"
  77  * to return to callback functions.  Some callbacks demand it.
  78  */
  79 char *fixed_partnum(char *supplied_partnum) {
  80         if (supplied_partnum == NULL) return "1";
  81         if (strlen(supplied_partnum)==0) return "1";
  82         return supplied_partnum;
  83 }
  84
  85
  86 static inline unsigned int _decode_hex(char *Source)
  87 {
  88         int ret = 0;
  89         if      ((*Source >= '0') && (*Source <= '9')) {
  90                 ret += (*Source - '0');
  91         }
  92         else if ((*Source >= 'A') && (*Source <= 'F')) {
  93                 ret += (*Source - 'A' + 10);
  94         }
  95         else if ((*Source >= 'a') && (*Source <= 'f')) {
  96                 ret += (*Source - 'a' + 10);
  97         }
  98         else {
  99                 return '?';
 100         }
 101
 102         ret = ret << 4;
 103
 104         if      ((*(Source + 1) >= '0') && (*(Source + 1) <= '9')) {
 105                 ret |= (*(Source + 1) - '0');
 106         }
 107         else if ((*(Source + 1) >= 'A') && (*(Source + 1) <= 'F')) {
 108                 ret |= (*(Source + 1) - 'A' + 10);
 109         }
 110         else if ((*(Source + 1) >= 'a') && (*(Source + 1) <= 'f')) {
 111                 ret |= (*(Source + 1) - 'a' + 10);
 112         }
 113         else {
 114                 return '?';
 115         }
 116         return ret;
 117 }
 118
 119 unsigned int decode_hex(char *Source) {return _decode_hex(Source);}
 120
 121 /*
 122  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
 123  * according to RFC2045 section 6.7
 124  */
 125 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
 126         unsigned int ch;
 127         int decoded_length = 0;
 128         int pos = 0;
 129
 130         while (pos < sourcelen)
 131         {
 132                 if (!strncmp(&encoded[pos], "=\r\n", 3))
 133                 {
 134                         pos += 3;
 135                 }
 136                 else if (!strncmp(&encoded[pos], "=\n", 2))
 137                 {
 138                         pos += 2;
 139                 }
 140                 else if (encoded[pos] == '=')
 141                 {
 142                         ch = 0;
 143                         ch = _decode_hex(&encoded[pos+1]);
 144                         pos += 3;
 145                         decoded[decoded_length++] = ch;
 146                 }
 147                 else
 148                 {
 149                         decoded[decoded_length++] = encoded[pos];
 150                         pos += 1;
 151                 }
 152         }
 153         decoded[decoded_length] = 0;
 154         return(decoded_length);
 155 }
 156
 157
 158 /*
 159  * Given a message or message-part body and a length, handle any necessary
 160  * decoding and pass the request up the stack.
 161  */
 162 void mime_decode(char *partnum,
 163                  char *part_start, size_t length,
 164                  char *content_type, char *charset, char *encoding,
 165                  char *disposition,
 166                  char *id,
 167                  char *name, char *filename,
 168                  void (*CallBack)
 169                   (char *cbname,
 170                    char *cbfilename,
 171                    char *cbpartnum,
 172                    char *cbdisp,
 173                    void *cbcontent,
 174                    char *cbtype,
 175                    char *cbcharset,
 176                    size_t cblength,
 177                    char *cbencoding,
 178                    char *cbid,
 179                    void *cbuserdata),
 180                  void (*PreMultiPartCallBack)
 181                   (char *cbname,
 182                    char *cbfilename,
 183                    char *cbpartnum,
 184                    char *cbdisp,
 185                    void *cbcontent,
 186                    char *cbtype,
 187                    char *cbcharset,
 188                    size_t cblength,
 189                    char *cbencoding,
 190                    char *cbid,
 191                    void *cbuserdata),
 192                  void (*PostMultiPartCallBack)
 193                   (char *cbname,
 194                    char *cbfilename,
 195                    char *cbpartnum,
 196                    char *cbdisp,
 197                    void *cbcontent,
 198                    char *cbtype,
 199                    char *cbcharset,
 200                    size_t cblength,
 201                    char *cbencoding,
 202                    char *cbid,
 203                    void *cbuserdata),
 204                   void *userdata,
 205                   int dont_decode
 206 )
 207 {
 208
 209         char *decoded;
 210         size_t bytes_decoded = 0;
 211
 212         /* Some encodings aren't really encodings */
 213         if (!strcasecmp(encoding, "7bit"))
 214                 strcpy(encoding, "");
 215         if (!strcasecmp(encoding, "8bit"))
 216                 strcpy(encoding, "");
 217         if (!strcasecmp(encoding, "binary"))
 218                 strcpy(encoding, "");
 219
 220         /* If this part is not encoded, send as-is */
 221         if ( (strlen(encoding) == 0) || (dont_decode)) {
 222                 if (CallBack != NULL) {
 223                         CallBack(name, filename, fixed_partnum(partnum),
 224                                 disposition, part_start,
 225                                 content_type, charset, length, encoding, id, userdata);
 226                         }
 227                 return;
 228         }
 229
 230         /* Fail silently if we hit an unknown encoding. */
 231         if ((strcasecmp(encoding, "base64"))
 232             && (strcasecmp(encoding, "quoted-printable"))) {
 233                 return;
 234         }
 235
 236         /*
 237          * Allocate a buffer for the decoded data.  The output buffer is slightly
 238          * larger than the input buffer; this assumes that the decoded data
 239          * will never be significantly larger than the encoded data.  This is a
 240          * safe assumption with base64, uuencode, and quoted-printable.
 241          */
 242         decoded = malloc(length + 32768);
 243         if (decoded == NULL) {
 244                 return;
 245         }
 246
 247         if (!strcasecmp(encoding, "base64")) {
 248                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 249         }
 250         else if (!strcasecmp(encoding, "quoted-printable")) {
 251                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded, part_start, length);
 252         }
 253
 254         if (bytes_decoded > 0) if (CallBack != NULL) {
 255                 CallBack(name, filename, fixed_partnum(partnum),
 256                         disposition, decoded,
 257                         content_type, charset, bytes_decoded, "binary", id, userdata);
 258         }
 259
 260         free(decoded);
 261 }
 262
 263 /*
 264  * Break out the components of a multipart message
 265  * (This function expects to be fed HEADERS + CONTENT)
 266  * Note: NULL can be supplied as content_end; in this case, the message is
 267  * considered to have ended when the parser encounters a 0x00 byte.
 268  */
 269 void the_mime_parser(char *partnum,
 270                      char *content_start, char *content_end,
 271                      void (*CallBack)
 272                       (char *cbname,
 273                        char *cbfilename,
 274                        char *cbpartnum,
 275                        char *cbdisp,
 276                        void *cbcontent,
 277                        char *cbtype,
 278                        char *cbcharset,
 279                        size_t cblength,
 280                        char *cbencoding,
 281                        char *cbid,
 282                        void *cbuserdata),
 283                      void (*PreMultiPartCallBack)
 284                       (char *cbname,
 285                        char *cbfilename,
 286                        char *cbpartnum,
 287                        char *cbdisp,
 288                        void *cbcontent,
 289                        char *cbtype,
 290                        char *cbcharset,
 291                        size_t cblength,
 292                        char *cbencoding,
 293                        char *cbid,
 294                        void *cbuserdata),
 295                      void (*PostMultiPartCallBack)
 296                       (char *cbname,
 297                        char *cbfilename,
 298                        char *cbpartnum,
 299                        char *cbdisp,
 300                        void *cbcontent,
 301                        char *cbtype,
 302                        char *cbcharset,
 303                        size_t cblength,
 304                        char *cbencoding,
 305                        char *cbid,
 306                        void *cbuserdata),
 307                       void *userdata,
 308                       int dont_decode
 309 )
 310 {
 311
 312         char *ptr;
 313         char *srch = NULL;
 314         char *part_start, *part_end = NULL;
 315         char buf[SIZ];
 316         char *header;
 317         char *boundary;
 318         char *startary;
 319         size_t startary_len = 0;
 320         char *endary;
 321         char *next_boundary;
 322         char *content_type;
 323         char *charset;
 324         size_t content_type_len;
 325         size_t content_length;
 326         char *encoding;
 327         char *disposition;
 328         size_t disposition_len;
 329         char *id;
 330         char *name = NULL;
 331         char *content_type_name;
 332         char *content_disposition_name;
 333         char *filename;
 334         int is_multipart;
 335         int part_seq = 0;
 336         int i;
 337         size_t length;
 338         char nested_partnum[256];
 339         int crlf_in_use = 0;
 340         char *evaluate_crlf_ptr = NULL;
 341         int buflen = 0;
 342         int headerlen = 0;
 343
 344         ptr = content_start;
 345         content_length = 0;
 346
 347         boundary = malloc(SIZ * 12);
 348         *boundary = '\0';
 349
 350         startary = boundary + SIZ * 1;
 351         *startary = '\0';
 352
 353         endary = boundary + SIZ * 2;
 354         *endary = '\0';
 355
 356         header = boundary + SIZ * 3;
 357         *header = '\0';
 358
 359         content_type = boundary + SIZ * 4;
 360         *content_type = '\0';
 361
 362         charset = boundary + SIZ * 5;
 363         *charset = '\0';
 364
 365         encoding = boundary + SIZ * 6;
 366         *encoding = '\0';
 367
 368         content_type_name = boundary + SIZ * 7;
 369         *content_type_name = '\0';
 370
 371         content_disposition_name = boundary + SIZ * 8;
 372         *content_disposition_name = '\0';
 373
 374         filename = boundary + SIZ * 9;
 375         *filename = '\0';
 376
 377         disposition = boundary + SIZ * 10;
 378         *disposition = '\0';
 379
 380         id = boundary + SIZ * 11;
 381         *id = '\0';
 382
 383         /* If the caller didn't supply an endpointer, generate one by measure */
 384         if (content_end == NULL) {
 385                 content_end = &content_start[strlen(content_start)];
 386         }
 387
 388         /* Learn interesting things from the headers */
 389         strcpy(header, "");
 390         headerlen = 0;
 391         do {
 392                 ptr = memreadlinelen(ptr, buf, SIZ, &buflen);
 393                 if (ptr >= content_end) {
 394                         goto end_parser;
 395                 }
 396
 397                 for (i = 0; i < buflen; ++i) {
 398                         if (isspace(buf[i])) {
 399                                 buf[i] = ' ';
 400                         }
 401                 }
 402
 403                 if (!isspace(buf[0])) {
 404                         if (!strncasecmp(header, "Content-type:", 13)) {
 405                                 memcpy (content_type, &header[13], headerlen - 12);
 406                                 content_type_len = striplt (content_type);
 407
 408                                 extract_key(content_type_name, content_type, content_type_len, HKEY("name"));
 409                                 extract_key(charset,           content_type, content_type_len, HKEY("charset"));
 410                                 extract_key(boundary,          header,       headerlen,        HKEY("boundary"));
 411
 412                                 /* Deal with weird headers */
 413                                 if (strchr(content_type, ' '))
 414                                         *(strchr(content_type, ' ')) = '\0';
 415                                 if (strchr(content_type, ';'))
 416                                         *(strchr(content_type, ';')) = '\0';
 417                         }
 418                         if (!strncasecmp(header, "Content-Disposition:", 20)) {
 419                                 memcpy (disposition, &header[20], headerlen - 19);
 420                                 disposition_len = striplt(disposition);
 421                                 extract_key(content_disposition_name, disposition, disposition_len,  HKEY("name"));
 422                                 extract_key(filename,                 disposition, disposition_len, HKEY("filename"));
 423                         }
 424                         if (!strncasecmp(header, "Content-ID:", 11)) {
 425                                 strcpy(id, &header[11]);
 426                                 striplt(id);
 427                                 stripallbut(id, '<', '>');
 428                         }
 429                         if (!strncasecmp(header, "Content-length: ", 15)) {
 430                                 char clbuf[10];
 431                                 safestrncpy(clbuf, &header[15], sizeof clbuf);
 432                                 striplt(clbuf);
 433                                 content_length = (size_t) atol(clbuf);
 434                         }
 435                         if (!strncasecmp(header, "Content-transfer-encoding: ", 26)) {
 436                                 strcpy(encoding, &header[26]);
 437                                 striplt(encoding);
 438                         }
 439                         strcpy(header, "");
 440                         headerlen = 0;
 441                 }
 442                 if ((headerlen + buflen + 2) < SIZ) {
 443                         memcpy(&header[headerlen], buf, buflen);
 444                         headerlen += buflen;
 445                         header[headerlen] = '\0';
 446                 }
 447         } while ((!IsEmptyStr(buf)) && (*ptr != 0));
 448
 449         if (strchr(disposition, ';'))
 450                 *(strchr(disposition, ';')) = '\0';
 451         striplt(disposition);
 452         if (strchr(content_type, ';'))
 453                 *(strchr(content_type, ';')) = '\0';
 454         striplt(content_type);
 455
 456         if (!IsEmptyStr(boundary)) {
 457                 is_multipart = 1;
 458         } else {
 459                 is_multipart = 0;
 460         }
 461
 462         /* If this is a multipart message, then recursively process it */
 463         part_start = NULL;
 464         if (is_multipart) {
 465
 466                 /* Tell the client about this message's multipartedness */
 467                 if (PreMultiPartCallBack != NULL) {
 468                         PreMultiPartCallBack("", "", partnum, "",
 469                                 NULL, content_type, charset,
 470                                 0, encoding, id, userdata);
 471                 }
 472
 473                 /* Figure out where the boundaries are */
 474                 startary_len = snprintf(startary, SIZ, "--%s", boundary);
 475                 snprintf(endary, SIZ, "--%s--", boundary);
 476
 477                 part_start = NULL;
 478                 do {
 479                         next_boundary = NULL;
 480                         for (srch=ptr; srch<content_end; ++srch) {
 481                                 if (!memcmp(srch, startary, startary_len)) {
 482                                         next_boundary = srch;
 483                                         srch = content_end;
 484                                 }
 485                         }
 486
 487                         if ( (part_start != NULL) && (next_boundary != NULL) ) {
 488                                 part_end = next_boundary;
 489                                 --part_end;             /* omit the trailing LF */
 490                                 if (crlf_in_use) {
 491                                         --part_end;     /* omit the trailing CR */
 492                                 }
 493
 494                                 if (!IsEmptyStr(partnum)) {
 495                                         snprintf(nested_partnum,
 496                                                  sizeof nested_partnum,
 497                                                  "%s.%d", partnum,
 498                                                  ++part_seq);
 499                                 }
 500                                 else {
 501                                         snprintf(nested_partnum,
 502                                                  sizeof nested_partnum,
 503                                                  "%d", ++part_seq);
 504                                 }
 505                                 the_mime_parser(nested_partnum,
 506                                             part_start, part_end,
 507                                                 CallBack,
 508                                                 PreMultiPartCallBack,
 509                                                 PostMultiPartCallBack,
 510                                                 userdata,
 511                                                 dont_decode);
 512                         }
 513
 514                         if (next_boundary != NULL) {
 515                                 /* If we pass out of scope, don't attempt to
 516                                  * read past the end boundary. */
 517                                 if (!strcmp(next_boundary, endary)) {
 518                                         ptr = content_end;
 519                                 }
 520                                 else {
 521                                         /* Set up for the next part. */
 522                                         part_start = strstr(next_boundary, "\n");
 523
 524                                         /* Determine whether newlines are LF or CRLF */
 525                                         evaluate_crlf_ptr = part_start;
 526                                         --evaluate_crlf_ptr;
 527                                         if (!memcmp(evaluate_crlf_ptr, "\r\n", 2)) {
 528                                                 crlf_in_use = 1;
 529                                         }
 530                                         else {
 531                                                 crlf_in_use = 0;
 532                                         }
 533
 534                                         /* Advance past the LF ... now we're in the next part */
 535                                         ++part_start;
 536                                         ptr = part_start;
 537                                 }
 538                         }
 539                         else {
 540                                 /* Invalid end of multipart.  Bail out! */
 541                                 ptr = content_end;
 542                         }
 543                 } while ( (ptr < content_end) && (next_boundary != NULL) );
 544
 545                 if (PostMultiPartCallBack != NULL) {
 546                         PostMultiPartCallBack("", "", partnum, "", NULL,
 547                                 content_type, charset, 0, encoding, id, userdata);
 548                 }
 549                 goto end_parser;
 550         }
 551
 552         /* If it's not a multipart message, then do something with it */
 553         if (!is_multipart) {
 554                 part_start = ptr;
 555                 length = 0;
 556                 while (ptr < content_end) {
 557                         ++ptr;
 558                         ++length;
 559                 }
 560                 part_end = content_end;
 561
 562
 563                 /* The following code will truncate the MIME part to the size
 564                  * specified by the Content-length: header.   We have commented it
 565                  * out because these headers have a tendency to be wrong.
 566                  *
 567                  *      if ( (content_length > 0) && (length > content_length) ) {
 568                  *              length = content_length;
 569                  *      }
 570                  */
 571
 572                 /* Sometimes the "name" field is tacked on to Content-type,
 573                  * and sometimes it's tacked on to Content-disposition.  Use
 574                  * whichever one we have.
 575                  */
 576                 if (strlen(content_disposition_name) > strlen(content_type_name)) {
 577                         name = content_disposition_name;
 578                 }
 579                 else {
 580                         name = content_type_name;
 581                 }
 582
 583                 /* Ok, we've got a non-multipart part here, so do something with it.
 584                  */
 585                 mime_decode(partnum,
 586                         part_start, length,
 587                         content_type, charset, encoding, disposition, id,
 588                         name, filename,
 589                         CallBack, NULL, NULL,
 590                         userdata, dont_decode
 591                 );
 592
 593                 /*
 594                  * Now if it's an encapsulated message/rfc822 then we have to recurse into it
 595                  */
 596                 if (!strcasecmp(content_type, "message/rfc822")) {
 597
 598                         if (PreMultiPartCallBack != NULL) {
 599                                 PreMultiPartCallBack("", "", partnum, "",
 600                                         NULL, content_type, charset,
 601                                         0, encoding, id, userdata);
 602                         }
 603                         if (CallBack != NULL) {
 604                                 if (strlen(partnum) > 0) {
 605                                         snprintf(nested_partnum,
 606                                                  sizeof nested_partnum,
 607                                                  "%s.%d", partnum,
 608                                                  ++part_seq);
 609                                 }
 610                                 else {
 611                                         snprintf(nested_partnum,
 612                                                  sizeof nested_partnum,
 613                                                  "%d", ++part_seq);
 614                                 }
 615                                 the_mime_parser(nested_partnum,
 616                                         part_start, part_end,
 617                                         CallBack,
 618                                         PreMultiPartCallBack,
 619                                         PostMultiPartCallBack,
 620                                         userdata,
 621                                         dont_decode
 622                                 );
 623                         }
 624                         if (PostMultiPartCallBack != NULL) {
 625                                 PostMultiPartCallBack("", "", partnum, "", NULL,
 626                                         content_type, charset, 0, encoding, id, userdata);
 627                         }
 628
 629
 630                 }
 631
 632         }
 633
 634 end_parser:     /* free the buffers!  end the oppression!! */
 635         free(boundary);
 636 }
 637
 638
 639
 640 /*
 641  * Entry point for the MIME parser.
 642  * (This function expects to be fed HEADERS + CONTENT)
 643  * Note: NULL can be supplied as content_end; in this case, the message is
 644  * considered to have ended when the parser encounters a 0x00 byte.
 645  */
 646 void mime_parser(char *content_start,
 647                 char *content_end,
 648
 649                  void (*CallBack)
 650                   (char *cbname,
 651                    char *cbfilename,
 652                    char *cbpartnum,
 653                    char *cbdisp,
 654                    void *cbcontent,
 655                    char *cbtype,
 656                    char *cbcharset,
 657                    size_t cblength,
 658                    char *cbencoding,
 659                    char *cbid,
 660                    void *cbuserdata),
 661
 662                  void (*PreMultiPartCallBack)
 663                   (char *cbname,
 664                    char *cbfilename,
 665                    char *cbpartnum,
 666                    char *cbdisp,
 667                    void *cbcontent,
 668                    char *cbtype,
 669                    char *cbcharset,
 670                    size_t cblength,
 671                    char *cbencoding,
 672                    char *cbid,
 673                    void *cbuserdata),
 674
 675                  void (*PostMultiPartCallBack)
 676                   (char *cbname,
 677                    char *cbfilename,
 678                    char *cbpartnum,
 679                    char *cbdisp,
 680                    void *cbcontent,
 681                    char *cbtype,
 682                    char *cbcharset,
 683                    size_t cblength,
 684                    char *cbencoding,
 685                    char *cbid,
 686                    void *cbuserdata),
 687
 688                   void *userdata,
 689                   int dont_decode
 690 )
 691 {
 692
 693         the_mime_parser("", content_start, content_end,
 694                         CallBack,
 695                         PreMultiPartCallBack,
 696                         PostMultiPartCallBack,
 697                         userdata, dont_decode);
 698 }
 699
 700
 701
 702
 703
 704
 705 typedef struct _MimeGuess {
 706         const char *Pattern;
 707         size_t PatternLen;
 708         long PatternOffset;
 709         const char *MimeString;
 710 } MimeGuess;
 711
 712 MimeGuess MyMimes [] = {
 713         {
 714                 "GIF",
 715                 3,
 716                 0,
 717                 "image/gif"
 718         },
 719         {
 720                 "\xff\xd8",
 721                 2,
 722                 0,
 723                 "image/jpeg"
 724         },
 725         {
 726                 "\x89PNG",
 727                 4,
 728                 0,
 729                 "image/png"
 730         },
 731         { // last...
 732                 "",
 733                 0,
 734                 0,
 735                 ""
 736         }
 737 };
 738
 739
 740 const char *GuessMimeType(const char *data, size_t dlen)
 741 {
 742         int MimeIndex = 0;
 743
 744         while (MyMimes[MimeIndex].PatternLen != 0)
 745         {
 746                 if ((MyMimes[MimeIndex].PatternLen +
 747                      MyMimes[MimeIndex].PatternOffset < dlen) &&
 748                     strncmp(MyMimes[MimeIndex].Pattern,
 749                             &data[MyMimes[MimeIndex].PatternOffset],
 750                             MyMimes[MimeIndex].PatternLen) == 0)
 751                 {
 752                         return MyMimes[MimeIndex].MimeString;
 753                 }
 754                 MimeIndex ++;
 755         }
 756         /*
 757          * ok, our simple minded algorythm didn't find anything,
 758          * let the big chegger try it, he wil default to application/octet-stream
 759          */
 760         return (xdg_mime_get_mime_type_for_data(data, dlen));
 761 }
 762
 763
 764 const char* GuessMimeByFilename(const char *what, size_t len)
 765 {
 766         /* we know some hardcoded on our own, try them... */
 767         if ((len > 3) && !strncasecmp(&what[len - 4], ".gif", 4))
 768                 return "image/gif";
 769         else if ((len > 2) && !strncasecmp(&what[len - 3], ".js", 3))
 770                 return  "text/javascript";
 771         else if ((len > 3) && !strncasecmp(&what[len - 4], ".txt", 4))
 772                 return "text/plain";
 773         else if ((len > 3) && !strncasecmp(&what[len - 4], ".css", 4))
 774                 return "text/css";
 775         else if ((len > 3) && !strncasecmp(&what[len - 4], ".jpg", 4))
 776                 return "image/jpeg";
 777         else if ((len > 3) && !strncasecmp(&what[len - 4], ".png", 4))
 778                 return "image/png";
 779         else if ((len > 3) && !strncasecmp(&what[len - 4], ".ico", 4))
 780                 return "image/x-icon";
 781         else if ((len > 3) && !strncasecmp(&what[len - 4], ".vcf", 4))
 782                 return "text/x-vcard";
 783         else if ((len > 4) && !strncasecmp(&what[len - 5], ".html", 5))
 784                 return "text/html";
 785         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htm", 4))
 786                 return "text/html";
 787         else if ((len > 3) && !strncasecmp(&what[len - 4], ".wml", 4))
 788                 return "text/vnd.wap.wml";
 789         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmls", 5))
 790                 return "text/vnd.wap.wmlscript";
 791         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmlc", 5))
 792                 return "application/vnd.wap.wmlc";
 793         else if ((len > 5) && !strncasecmp(&what[len - 6], ".wmlsc", 6))
 794                 return "application/vnd.wap.wmlscriptc";
 795         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wbmp", 5))
 796                 return "image/vnd.wap.wbmp";
 797         else
 798                 /* and let xdgmime do the fallback. */
 799                 return xdg_mime_get_mime_type_from_file_name(what);
 800 }
 801
 802 static HashList *IconHash = NULL;
 803
 804 typedef struct IconName IconName;
 805
 806 struct IconName {
 807         char *FlatName;
 808         char *FileName;
 809 };
 810
 811 static void DeleteIcon(void *IconNamePtr)
 812 {
 813         IconName *Icon = (IconName*) IconNamePtr;
 814         free(Icon->FlatName);
 815         free(Icon->FileName);
 816         free(Icon);
 817 }
 818
 819 /*
 820 static const char *PrintFlat(void *IconNamePtr)
 821 {
 822         IconName *Icon = (IconName*) IconNamePtr;
 823         return Icon->FlatName;
 824 }
 825 static const char *PrintFile(void *IconNamePtr)
 826 {
 827         IconName *Icon = (IconName*) IconNamePtr;
 828         return Icon->FileName;
 829 }
 830 */
 831
 832 #define GENSTR "x-generic"
 833 #define IGNORE_PREFIX_1 "gnome-mime"
 834 int LoadIconDir(const char *DirName)
 835 {
 836         DIR *filedir = NULL;
 837         struct dirent *filedir_entry;
 838         int d_namelen;
 839         int d_without_ext;
 840         IconName *Icon;
 841
 842         filedir = opendir (DirName);
 843         IconHash = NewHash(1, NULL);
 844         if (filedir == NULL) {
 845                 return 0;
 846         }
 847
 848         while ((filedir_entry = readdir(filedir)))
 849         {
 850                 char *MinorPtr;
 851                 char *PStart;
 852 #ifdef _DIRENT_HAVE_D_NAMELEN
 853                 d_namelen = filedir_entry->d_namelen;
 854 #else
 855                 d_namelen = strlen(filedir_entry->d_name);
 856 #endif
 857                 d_without_ext = d_namelen;
 858                 while ((d_without_ext > 0) && (filedir_entry->d_name[d_without_ext] != '.'))
 859                         d_without_ext --;
 860                 if ((d_without_ext == 0) || (d_namelen < 3))
 861                         continue;
 862
 863                 if ((sizeof(IGNORE_PREFIX_1) < d_namelen) &&
 864                     (strncmp(IGNORE_PREFIX_1,
 865                              filedir_entry->d_name,
 866                              sizeof(IGNORE_PREFIX_1) - 1) == 0)) {
 867                         PStart = filedir_entry->d_name + sizeof(IGNORE_PREFIX_1);
 868                         d_without_ext -= sizeof(IGNORE_PREFIX_1);
 869                 }
 870                 else {
 871                         PStart = filedir_entry->d_name;
 872                 }
 873                 Icon = malloc(sizeof(IconName));
 874
 875                 Icon->FileName = malloc(d_namelen + 1);
 876                 memcpy(Icon->FileName, filedir_entry->d_name, d_namelen + 1);
 877
 878                 Icon->FlatName = malloc(d_without_ext + 1);
 879                 memcpy(Icon->FlatName, PStart, d_without_ext);
 880                 Icon->FlatName[d_without_ext] = '\0';
 881                 /* Try to find Minor type in image-jpeg */
 882                 MinorPtr = strchr(Icon->FlatName, '-');
 883                 if (MinorPtr != NULL) {
 884                         size_t MinorLen;
 885                         MinorLen = 1 + d_without_ext - (MinorPtr - Icon->FlatName + 1);
 886                         if ((MinorLen == sizeof(GENSTR)) &&
 887                             (strncmp(MinorPtr + 1, GENSTR, sizeof(GENSTR)) == 0)) {
 888                                 /* ok, we found a generic filename. cut the generic. */
 889                                 *MinorPtr = '\0';
 890                                 d_without_ext = d_without_ext - (MinorPtr - Icon->FlatName);
 891                         }
 892                         else { /* Map the major / minor separator to / */
 893                                 *MinorPtr = '/';
 894                         }
 895                 }
 896
 897 //              PrintHash(IconHash, PrintFlat, PrintFile);
 898 //              printf("%s - %s\n", Icon->FlatName, Icon->FileName);
 899                 Put(IconHash, Icon->FlatName, d_without_ext, Icon, DeleteIcon);
 900 //              PrintHash(IconHash, PrintFlat, PrintFile);
 901         }
 902         closedir(filedir);
 903         return 1;
 904 }
 905
 906 const char *GetIconFilename(char *MimeType, size_t len)
 907 {
 908         void *vIcon;
 909         IconName *Icon;
 910
 911         if(IconHash == NULL)
 912                 return NULL;
 913
 914         GetHash(IconHash, MimeType, len, &vIcon), Icon = (IconName*) vIcon;
 915         /* didn't find the exact mimetype? try major only. */
 916         if (Icon == NULL) {
 917                 char * pMinor;
 918                 pMinor = strchr(MimeType, '/');
 919                 if (pMinor != NULL) {
 920                         *pMinor = '\0';
 921                         GetHash(IconHash, MimeType, pMinor - MimeType, &vIcon),
 922                                 Icon = (IconName*) vIcon;
 923                 }
 924         }
 925         if (Icon == NULL) {
 926                 return NULL;
 927         }
 928
 929         /*printf("Getting: [%s] == [%s] -> [%s]\n", MimeType, Icon->FlatName, Icon->FileName);*/
 930         return Icon->FileName;
 931 }
 932
 933 void ShutDownLibCitadelMime(void)
 934 {
 935         DeleteHash(&IconHash);
 936 }