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