libcitadel/lib/mime_parser.c

   1 // This is the MIME parser for Citadel.
   2 //
   3 // Copyright (c) 1998-2023 by the citadel.org development team.
   4 //
   5 // This program is open source software.  Use, duplication, or disclosure
   6 // is subject to the terms of the GNU General Public License, version 3.
   7
   8 #include <stdlib.h>
   9 #include <unistd.h>
  10 #include <stdio.h>
  11 #include <signal.h>
  12 #include <sys/types.h>
  13 #include <ctype.h>
  14 #include <string.h>
  15 #include <sys/stat.h>
  16 #include <sys/types.h>
  17 #include <dirent.h>
  18 #include <errno.h>
  19
  20 #include "xdgmime/xdgmime.h"
  21 #include "libcitadel.h"
  22 #include "libcitadellocal.h"
  23
  24 const unsigned char FromHexTable[256] = {
  25         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //  0
  26         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 10
  27         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 20
  28         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 30
  29         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x00, 0x01, // 40
  30         0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 0x09, 0xFF, 0xFF, // 50
  31         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0A, 0x0B, 0x0C, 0x0D, 0x0E, // 60
  32         0x0F, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 70
  33         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, // 80
  34         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0x0A, 0x0B, 0x0C, // 90
  35         0x0D, 0x0E, 0x0F, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //100
  36         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //110
  37         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //120
  38         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //130
  39         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //140
  40         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //150
  41         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //160
  42         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //170
  43         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //180
  44         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //190
  45         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //200
  46         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //210
  47         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //220
  48         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //230
  49         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF, //240
  50         0xFF, 0xFF, 0xFF, 0xFF, 0xFF, 0xFF                          //250
  51 };
  52
  53
  54 long extract_key(char *target, char *source, long sourcelen, char *key, long keylen, char KeyEnd) {
  55         char *sptr, *ptr = NULL;
  56         int double_quotes = 0;
  57         long RealKeyLen = keylen;
  58
  59         sptr = source;
  60
  61         while (sptr != NULL) {
  62                 ptr = bmstrcasestr_len(sptr, sourcelen - (sptr - source), key, keylen);
  63                 if (ptr != NULL) {
  64                         while (isspace(*(ptr + RealKeyLen)))
  65                                 RealKeyLen ++;
  66                         if (*(ptr + RealKeyLen) == KeyEnd) {
  67                                 sptr = NULL;
  68                                 RealKeyLen ++;
  69                         }
  70                         else {
  71                                 sptr = ptr + RealKeyLen + 1;
  72                         }
  73                 }
  74                 else
  75                         sptr = ptr;
  76         }
  77         if (ptr == NULL) {
  78                 *target = '\0';
  79                 return 0;
  80         }
  81         strcpy(target, (ptr + RealKeyLen));
  82
  83         for (ptr=target; (*ptr != 0); ptr++) {
  84
  85                 // A semicolon means we've hit the end of the key, unless we're inside double quotes
  86                 if ( (double_quotes != 1) && (*ptr == ';')) {
  87                         *ptr = 0;
  88                 }
  89
  90                 // if we find double quotes, we've got a great set of string boundaries
  91                 if (*ptr == '\"') {
  92                         ++double_quotes;
  93                         if (double_quotes == 1) {
  94                                 strcpy(ptr, ptr+1);
  95                         }
  96                         else {
  97                                 *ptr = 0;
  98                         }
  99                 }
 100         }
 101         *ptr = '\0';
 102         return ptr - target;
 103 }
 104
 105
 106 // For non-multipart messages, we need to generate a quickie partnum of "1"
 107 // to return to callback functions.  Some callbacks demand it.
 108 char *fixed_partnum(char *supplied_partnum) {
 109         if (supplied_partnum == NULL) return "1";
 110         if (strlen(supplied_partnum)==0) return "1";
 111         return supplied_partnum;
 112 }
 113
 114
 115 static inline unsigned int _decode_hex(const char *Source) {
 116         unsigned int ret = '?';
 117         unsigned char LO_NIBBLE;
 118         unsigned char HI_NIBBLE;
 119
 120         HI_NIBBLE = FromHexTable[(unsigned char) *Source];
 121         LO_NIBBLE = FromHexTable[(unsigned char) *(Source+1)];
 122
 123         if ((LO_NIBBLE == 0xFF) || (LO_NIBBLE == 0xFF))
 124                 return ret;
 125         ret = HI_NIBBLE;
 126         ret = ret << 4;
 127         ret = ret | LO_NIBBLE;
 128         return ret;
 129 }
 130
 131 unsigned int decode_hex(char *Source) {return _decode_hex(Source);}
 132
 133
 134 // Convert "quoted-printable" to binary.  Returns number of bytes decoded.
 135 // according to RFC2045 section 6.7
 136 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
 137         unsigned int ch;
 138         int decoded_length = 0;
 139         int pos = 0;
 140
 141         while (pos < sourcelen) {
 142                 if (*(encoded + pos) == '=') {
 143                         pos ++;
 144                         if (*(encoded + pos) == '\n') {
 145                                 pos ++;
 146                         }
 147                         else if (*(encoded + pos) == '\r') {
 148                                 pos ++;
 149                                 if (*(encoded + pos) == '\n')
 150                                         pos++;
 151                         }
 152                         else {
 153                                 ch = _decode_hex(&encoded[pos]);
 154                                 pos += 2;
 155                                 decoded[decoded_length++] = ch;
 156                         }
 157                 }
 158                 else {
 159                         decoded[decoded_length++] = encoded[pos];
 160                         pos += 1;
 161                 }
 162         }
 163         decoded[decoded_length] = 0;
 164         return(decoded_length);
 165 }
 166
 167
 168 // Given a message or message-part body and a length, handle any necessary
 169 // decoding and pass the request up the stack.
 170 void mime_decode(char *partnum,
 171                  char *part_start, size_t length,
 172                  char *content_type, char *charset, char *encoding,
 173                  char *disposition,
 174                  char *id,
 175                  char *name, char *filename,
 176                  MimeParserCallBackType CallBack,
 177                  MimeParserCallBackType PreMultiPartCallBack,
 178                  MimeParserCallBackType PostMultiPartCallBack,
 179                  void *userdata,
 180                  int dont_decode
 181 ) {
 182         char *decoded;
 183         size_t bytes_decoded = 0;
 184
 185         // Some encodings aren't really encodings
 186         if (!strcasecmp(encoding, "7bit"))
 187                 *encoding = '\0';
 188         if (!strcasecmp(encoding, "8bit"))
 189                 *encoding = '\0';
 190         if (!strcasecmp(encoding, "binary"))
 191                 *encoding = '\0';
 192         if (!strcasecmp(encoding, "ISO-8859-1"))
 193                 *encoding = '\0';
 194
 195         // If this part is not encoded, send as-is
 196         if ( (strlen(encoding) == 0) || (dont_decode)) {
 197                 if (CallBack != NULL) {
 198                         CallBack(name,
 199                                  filename,
 200                                  fixed_partnum(partnum),
 201                                  disposition,
 202                                  part_start,
 203                                  content_type,
 204                                  charset,
 205                                  length,
 206                                  encoding,
 207                                  id,
 208                                  userdata);
 209                         }
 210                 return;
 211         }
 212
 213         // Fail silently if we hit an unknown encoding.
 214         if ((strcasecmp(encoding, "base64")) && (strcasecmp(encoding, "quoted-printable"))) {
 215                 return;
 216         }
 217
 218         // Allocate a buffer for the decoded data.  The output buffer is slightly
 219         // larger than the input buffer; this assumes that the decoded data
 220         // will never be significantly larger than the encoded data.  This is a
 221         // safe assumption with base64, uuencode, and quoted-printable.
 222         decoded = malloc(length + 32768);
 223         if (decoded == NULL) {
 224                 return;
 225         }
 226
 227         if (!strcasecmp(encoding, "base64")) {
 228                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 229         }
 230         else if (!strcasecmp(encoding, "quoted-printable")) {
 231                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded, part_start, length);
 232         }
 233
 234         if (bytes_decoded > 0) if (CallBack != NULL) {
 235                         char encoding_buf[SIZ];
 236
 237                         strcpy(encoding_buf, "binary");
 238                         CallBack(name,
 239                                  filename,
 240                                  fixed_partnum(partnum),
 241                                  disposition,
 242                                  decoded,
 243                                  content_type,
 244                                  charset,
 245                                  bytes_decoded,
 246                                  encoding_buf,
 247                                  id,
 248                                  userdata);
 249         }
 250
 251         free(decoded);
 252 }
 253
 254
 255 // this is a bastardization of mime_decode() which can be called if 'dont_decode' was set;
 256 // to postpone the cpu intense process of decoding until the time when it realy wants the content.
 257 // returns:
 258 //   - > 0 we decoded something, its on *decoded, you need to free it.
 259 //   - = 0 no need to decode stuff. *decoded will be NULL.
 260 //   - < 0 an error occured, either an unknown encoding, or alloc failed. no need to free.
 261 int mime_decode_now (char *part_start,
 262                      size_t length,
 263                      char *encoding,
 264                      char **decoded,
 265                      size_t *bytes_decoded)
 266 {
 267         *bytes_decoded = 0;
 268         *decoded = NULL;
 269         // Some encodings aren't really encodings
 270         if (!strcasecmp(encoding, "7bit"))
 271                 *encoding = '\0';
 272         if (!strcasecmp(encoding, "8bit"))
 273                 *encoding = '\0';
 274         if (!strcasecmp(encoding, "binary"))
 275                 *encoding = '\0';
 276
 277         // If this part is not encoded, send as-is
 278         if (strlen(encoding) == 0) {
 279                 return 0;
 280         }
 281
 282         // Fail if we hit an unknown encoding.
 283         if ((strcasecmp(encoding, "base64")) && (strcasecmp(encoding, "quoted-printable"))) {
 284                 return -1;
 285         }
 286
 287         // Allocate a buffer for the decoded data.  The output buffer is slightly
 288         // larger than the input buffer; this assumes that the decoded data
 289         // will never be significantly larger than the encoded data.  This is a
 290         // safe assumption with base64, uuencode, and quoted-printable.
 291         *decoded = malloc(length + 32768);
 292         if (decoded == NULL) {
 293                 return -1;
 294         }
 295
 296         if (!strcasecmp(encoding, "base64")) {
 297                 *bytes_decoded = CtdlDecodeBase64(*decoded, part_start, length);
 298                 return 1;
 299         }
 300         else if (!strcasecmp(encoding, "quoted-printable")) {
 301                 *bytes_decoded = CtdlDecodeQuotedPrintable(*decoded, part_start, length);
 302                 return 1;
 303         }
 304         return -1;
 305 }
 306
 307 typedef enum _eIntMimeHdrs {
 308         boundary,
 309         startary,
 310         endary,
 311         content_type,
 312         charset,
 313         encoding,
 314         content_type_name,
 315         content_disposition_name,
 316         filename,
 317         disposition,
 318         id,
 319         eMax // don't move!
 320 } eIntMimeHdrs;
 321
 322 typedef struct _CBufStr {
 323         char Key[SIZ];
 324         long len;
 325 }CBufStr;
 326
 327 typedef struct _interesting_mime_headers {
 328         CBufStr b[eMax];
 329         long content_length;
 330         long is_multipart;
 331 } interesting_mime_headers;
 332
 333
 334 static void FlushInterestingMimes(interesting_mime_headers *m) {
 335         int i;
 336
 337         for (i = 0; i < eMax; i++) {
 338                 m->b[i].Key[0] = '\0';
 339                 m->b[i].len = 0;
 340         }
 341         m->content_length = -1;
 342 }
 343
 344
 345 static interesting_mime_headers *InitInterestingMimes(void) {
 346         interesting_mime_headers *m;
 347         m = (interesting_mime_headers*) malloc( sizeof(interesting_mime_headers));
 348
 349         FlushInterestingMimes(m);
 350
 351         return m;
 352 }
 353
 354
 355 static long parse_MimeHeaders(interesting_mime_headers *m, char** pcontent_start, char *content_end) {
 356         char buf[SIZ];
 357         char header[SIZ];
 358         long headerlen;
 359         char *ptr, *pch;
 360         int buflen = 0;
 361         int i;
 362
 363         // Learn interesting things from the headers
 364         ptr = *pcontent_start;
 365         *header = '\0';
 366         headerlen = 0;
 367         do {
 368                 ptr = memreadlinelen(ptr, buf, SIZ, &buflen);
 369
 370                 for (i = 0; i < buflen; ++i) {
 371                         if (isspace(buf[i])) {
 372                                 buf[i] = ' ';
 373                         }
 374                 }
 375
 376                 if (!isspace(buf[0]) && (headerlen > 0)) {
 377                         if (!strncasecmp(header, "Content-type:", 13)) {
 378                                 memcpy (m->b[content_type].Key, &header[13], headerlen - 12);
 379                                 m->b[content_type].Key[headerlen - 12] = '\0';
 380                                 m->b[content_type].len = string_trim (m->b[content_type].Key);
 381
 382                                 m->b[content_type_name].len = extract_key(m->b[content_type_name].Key, CKEY(m->b[content_type]), HKEY("name"), '=');
 383                                 m->b[charset].len           = extract_key(m->b[charset].Key,           CKEY(m->b[content_type]), HKEY("charset"), '=');
 384                                 m->b[boundary].len          = extract_key(m->b[boundary].Key,          header,       headerlen,  HKEY("boundary"), '=');
 385
 386                                 // Deal with weird headers
 387                                 pch = strchr(m->b[content_type].Key, ' ');
 388                                 if (pch != NULL) {
 389                                         *pch = '\0';
 390                                         m->b[content_type].len = m->b[content_type].Key - pch;
 391                                 }
 392                                 pch = strchr(m->b[content_type].Key, ';');
 393                                 if (pch != NULL) {
 394                                         *pch = '\0';
 395                                         m->b[content_type].len = m->b[content_type].Key - pch;
 396                                 }
 397                         }
 398                         else if (!strncasecmp(header, "Content-Disposition:", 20)) {
 399                                 memcpy (m->b[disposition].Key, &header[20], headerlen - 19);
 400                                 m->b[disposition].Key[headerlen - 19] = '\0';
 401                                 m->b[disposition].len = string_trim(m->b[disposition].Key);
 402
 403                                 m->b[content_disposition_name].len = extract_key(m->b[content_disposition_name].Key, CKEY(m->b[disposition]), HKEY("name"), '=');
 404                                 m->b[filename].len                 = extract_key(m->b[filename].Key,                 CKEY(m->b[disposition]), HKEY("filename"), '=');
 405                                 pch = strchr(m->b[disposition].Key, ';');
 406                                 if (pch != NULL) *pch = '\0';
 407                                 m->b[disposition].len = string_trim(m->b[disposition].Key);
 408                         }
 409                         else if (!strncasecmp(header, "Content-ID:", 11)) {
 410                                 memcpy(m->b[id].Key, &header[11], headerlen - 11);
 411                                 m->b[id].Key[headerlen - 11] = '\0';
 412                                 string_trim(m->b[id].Key);
 413                                 m->b[id].len = stripallbut(m->b[id].Key, '<', '>');
 414                         }
 415                         else if (!strncasecmp(header, "Content-length: ", 15)) {
 416                                 char *clbuf;
 417                                 clbuf = &header[15];
 418                                 while (isspace(*clbuf))
 419                                         clbuf ++;
 420                                 m->content_length = (size_t) atol(clbuf);
 421                         }
 422                         else if (!strncasecmp(header, "Content-transfer-encoding: ", 26)) {
 423                                 memcpy(m->b[encoding].Key, &header[26], headerlen - 26);
 424                                 m->b[encoding].Key[headerlen - 26] = '\0';
 425                                 m->b[encoding].len = string_trim(m->b[encoding].Key);
 426                         }
 427                         *header = '\0';
 428                         headerlen = 0;
 429                 }
 430                 if ((headerlen + buflen + 2) < SIZ) {
 431                         memcpy(&header[headerlen], buf, buflen);
 432                         headerlen += buflen;
 433                         header[headerlen] = '\0';
 434                 }
 435                 if (ptr >= content_end) {
 436                         return -1;
 437                 }
 438         } while ((!IsEmptyStr(buf)) && (*ptr != 0));
 439
 440         m->is_multipart = m->b[boundary].len != 0;
 441         *pcontent_start = ptr;
 442
 443         return 0;
 444 }
 445
 446
 447 static int IsAsciiEncoding(interesting_mime_headers *m) {
 448
 449         if ((m->b[encoding].len != 0) && (strcasecmp(m->b[encoding].Key, "base64") == 0)) {
 450                 return 1;
 451         }
 452         if ((m->b[encoding].len != 0) && (strcmp(m->b[encoding].Key, "quoted-printable") == 0)) {
 453                 return 1;
 454         }
 455
 456         return 0;
 457 }
 458
 459
 460 static char *FindNextContent(char *ptr,
 461                         char *content_end,
 462                         interesting_mime_headers *SubMimeHeaders,
 463                         interesting_mime_headers *m)
 464 {
 465         char *next_boundary;
 466         char  tmp;
 467
 468         if (IsAsciiEncoding(SubMimeHeaders)) {
 469                 tmp = *content_end;
 470                 *content_end = '\0';
 471
 472                 // ok, if we have a content length of the mime part,
 473                 // try skipping the content on the search for the next
 474                 // boundary. since we don't trust the content_length
 475                 // to be all accurate, and suspect it to lose one digit
 476                 // per line with a line length of 80 chars, we need
 477                 // to start searching a little before..
 478
 479                 if ((SubMimeHeaders->content_length != -1) && (SubMimeHeaders->content_length > 10)) {
 480                         char *pptr;
 481                         long lines;
 482
 483                         lines = SubMimeHeaders->content_length / 80;
 484                         pptr = ptr + SubMimeHeaders->content_length - lines - 10;
 485                         if (pptr < content_end)
 486                                 ptr = pptr;
 487                 }
 488
 489                 next_boundary = strstr(ptr, m->b[startary].Key);
 490                 *content_end = tmp;
 491         }
 492         else {
 493                 char *srch;
 494                 // ok, if we have a content length of the mime part,
 495                 // try skipping the content on the search for the next
 496                 // boundary. since we don't trust the content_length
 497                 // to be all accurate, start searching a little before..
 498
 499                 if ((SubMimeHeaders->content_length != -1) && (SubMimeHeaders->content_length > 10)) {
 500                         char *pptr;
 501                         pptr = ptr + SubMimeHeaders->content_length - 10;
 502                         if (pptr < content_end)
 503                                 ptr = pptr;
 504                 }
 505
 506                 srch = next_boundary = NULL;
 507                 for (srch = memchr(ptr, '-',  content_end - ptr);
 508                         (srch != NULL) && (srch < content_end);
 509                         srch = memchr(srch, '-',  content_end - srch))
 510                 {
 511                         if (!memcmp(srch, m->b[startary].Key, m->b[startary].len)) {
 512                                 next_boundary = srch;
 513                                 srch = content_end;
 514                         }
 515                         else srch ++;
 516
 517                 }
 518
 519         }
 520         return next_boundary;
 521 }
 522
 523
 524 // Break out the components of a multipart message
 525 // (This function expects to be fed HEADERS + CONTENT)
 526 // Note: NULL can be supplied as content_end; in this case, the message is
 527 // considered to have ended when the parser encounters a 0x00 byte.
 528 static void recurseable_mime_parser(char *partnum,
 529                 char *content_start, char *content_end,
 530                 MimeParserCallBackType CallBack,
 531                 MimeParserCallBackType PreMultiPartCallBack,
 532                 MimeParserCallBackType PostMultiPartCallBack,
 533                 void *userdata,
 534                 int dont_decode,
 535                 interesting_mime_headers *m
 536 ) {
 537         interesting_mime_headers *SubMimeHeaders;
 538         char     *ptr;
 539         char     *part_start;
 540         char     *part_end = NULL;
 541         char     *evaluate_crlf_ptr = NULL;
 542         char     *next_boundary;
 543         char      nested_partnum[256];
 544         int       crlf_in_use = 0;
 545         int       part_seq = 0;
 546         CBufStr  *chosen_name;
 547
 548         // If this is a multipart message, then recursively process it
 549         ptr = content_start;
 550         part_start = NULL;
 551         if (m->is_multipart) {
 552
 553                 // Tell the client about this message's multipartedness
 554                 if (PreMultiPartCallBack != NULL) {
 555                         PreMultiPartCallBack("",
 556                                 "",
 557                                 partnum,
 558                                 "",
 559                                 NULL,
 560                                 m->b[content_type].Key,
 561                                 m->b[charset].Key,
 562                                 0,
 563                                 m->b[encoding].Key,
 564                                 m->b[id].Key,
 565                                 userdata
 566                         );
 567                 }
 568
 569                 // Figure out where the boundaries are
 570                 m->b[startary].len = snprintf(m->b[startary].Key, SIZ, "--%s", m->b[boundary].Key);
 571                 SubMimeHeaders = InitInterestingMimes ();
 572
 573                 while ((*ptr == '\r') || (*ptr == '\n')) {
 574                         ptr++;
 575                 }
 576
 577                 if (strncmp(ptr, m->b[startary].Key, m->b[startary].len) == 0) {
 578                         ptr += m->b[startary].len;
 579                 }
 580
 581                 while ((*ptr == '\r') || (*ptr == '\n')) {
 582                         ptr ++;
 583                 }
 584
 585                 part_start = NULL;
 586                 do {
 587                         char *optr;
 588
 589                         optr = ptr;
 590                         if (parse_MimeHeaders(SubMimeHeaders, &ptr, content_end) != 0)
 591                                 break;
 592                         if ((ptr - optr > 2) && (*(ptr - 2) == '\r')) {
 593                                 crlf_in_use = 1;
 594                         }
 595
 596                         part_start = ptr;
 597
 598                         next_boundary = FindNextContent(ptr, content_end, SubMimeHeaders, m);
 599                         if ((next_boundary != NULL) && (next_boundary - part_start < 3)) {
 600                                 FlushInterestingMimes(SubMimeHeaders);
 601                                 continue;
 602                         }
 603
 604                         if ( (part_start != NULL) && (next_boundary != NULL) ) {
 605                                 part_end = next_boundary;
 606                                 --part_end;             // omit the trailing LF
 607                                 if (crlf_in_use) {
 608                                         --part_end;     // omit the trailing CR
 609                                 }
 610
 611                                 if (!IsEmptyStr(partnum)) {
 612                                         snprintf(nested_partnum,
 613                                                  sizeof nested_partnum,
 614                                                  "%s.%d", partnum,
 615                                                  ++part_seq);
 616                                 }
 617                                 else {
 618                                         snprintf(nested_partnum,
 619                                                  sizeof nested_partnum,
 620                                                  "%d", ++part_seq);
 621                                 }
 622                                 recurseable_mime_parser(nested_partnum,
 623                                                         part_start,
 624                                                         part_end,
 625                                                         CallBack,
 626                                                         PreMultiPartCallBack,
 627                                                         PostMultiPartCallBack,
 628                                                         userdata,
 629                                                         dont_decode,
 630                                                         SubMimeHeaders);
 631                         }
 632
 633                         if (next_boundary != NULL) {
 634                                 // If we pass out of scope, don't attempt to read past the end boundary.
 635                                 if ((*(next_boundary + m->b[startary].len) == '-') &&
 636                                     (*(next_boundary + m->b[startary].len + 1) == '-') ){
 637                                         ptr = content_end;
 638                                 }
 639                                 else {
 640                                         // Set up for the next part.
 641                                         part_start = strstr(next_boundary, "\n");
 642
 643                                         // Determine whether newlines are LF or CRLF
 644                                         evaluate_crlf_ptr = part_start;
 645                                         --evaluate_crlf_ptr;
 646                                         if ((*evaluate_crlf_ptr == '\r') && (*(evaluate_crlf_ptr + 1) == '\n')) {
 647                                                 crlf_in_use = 1;
 648                                         }
 649                                         else {
 650                                                 crlf_in_use = 0;
 651                                         }
 652
 653                                         // Advance past the LF ... now we're in the next part
 654                                         ++part_start;
 655                                         ptr = part_start;
 656                                 }
 657                         }
 658                         else {
 659                                 // Invalid end of multipart.  Bail out!
 660                                 ptr = content_end;
 661                         }
 662                         FlushInterestingMimes(SubMimeHeaders);
 663                 } while ( (ptr < content_end) && (next_boundary != NULL) );
 664
 665                 free(SubMimeHeaders);
 666
 667                 if (PostMultiPartCallBack != NULL) {
 668                         PostMultiPartCallBack("",
 669                                 "",
 670                                 partnum,
 671                                 "",
 672                                 NULL,
 673                                 m->b[content_type].Key,
 674                                 m->b[charset].Key,
 675                                 0,
 676                                 m->b[encoding].Key,
 677                                 m->b[id].Key,
 678                                 userdata
 679                         );
 680                 }
 681         }
 682         // If it's not a multipart message, then do something with it
 683         else {
 684                 size_t length;
 685                 part_start = ptr;
 686                 length = content_end - part_start;
 687                 ptr = part_end = content_end;
 688
 689                 // The following code will truncate the MIME part to the size
 690                 // specified by the Content-length: header.   We have commented it
 691                 // out because these headers have a tendency to be wrong.
 692                 //
 693                 //      if ( (content_length > 0) && (length > content_length) ) {
 694                 //              length = content_length;
 695                 //      }
 696
 697                 // Sometimes the "name" field is tacked on to Content-type,
 698                 // and sometimes it's tacked on to Content-disposition.  Use
 699                 // whichever one we have.
 700                 if (m->b[content_disposition_name].len > m->b[content_type_name].len) {
 701                         chosen_name = &m->b[content_disposition_name];
 702                 }
 703                 else {
 704                         chosen_name = &m->b[content_type_name];
 705                 }
 706
 707                 // Ok, we've got a non-multipart part here, so do something with it.
 708                 mime_decode(partnum,
 709                                 part_start,
 710                                 length,
 711                                 m->b[content_type].Key,
 712                                 m->b[charset].Key,
 713                                 m->b[encoding].Key,
 714                                 m->b[disposition].Key,
 715                                 m->b[id].Key,
 716                                 chosen_name->Key,
 717                                 m->b[filename].Key,
 718                                 CallBack,
 719                                 NULL,
 720                                 NULL,
 721                                 userdata,
 722                                 dont_decode
 723                         );
 724
 725                 // Now if it's an encapsulated message/rfc822 then we have to recurse into it
 726                 if (!strcasecmp(&m->b[content_type].Key[0], "message/rfc822")) {
 727
 728                         if (PreMultiPartCallBack != NULL) {
 729                                 PreMultiPartCallBack("",
 730                                                          "",
 731                                                          partnum,
 732                                                          "",
 733                                                          NULL,
 734                                                          m->b[content_type].Key,
 735                                                          m->b[charset].Key,
 736                                                          0,
 737                                                          m->b[encoding].Key,
 738                                                          m->b[id].Key,
 739                                                          userdata);
 740                         }
 741                         if (CallBack != NULL) {
 742                                 if (strlen(partnum) > 0) {
 743                                         snprintf(nested_partnum,
 744                                                  sizeof nested_partnum,
 745                                                  "%s.%d", partnum,
 746                                                  ++part_seq);
 747                                 }
 748                                 else {
 749                                         snprintf(nested_partnum,
 750                                                  sizeof nested_partnum,
 751                                                  "%d", ++part_seq);
 752                                 }
 753                                 the_mime_parser(nested_partnum,
 754                                                 part_start,
 755                                                 part_end,
 756                                                 CallBack,
 757                                                 PreMultiPartCallBack,
 758                                                 PostMultiPartCallBack,
 759                                                 userdata,
 760                                                 dont_decode
 761                                         );
 762                         }
 763                         if (PostMultiPartCallBack != NULL) {
 764                                 PostMultiPartCallBack("",
 765                                         "",
 766                                         partnum,
 767                                         "",
 768                                         NULL,
 769                                         m->b[content_type].Key,
 770                                         m->b[charset].Key,
 771                                         0,
 772                                         m->b[encoding].Key,
 773                                         m->b[id].Key,
 774                                         userdata
 775                                 );
 776                         }
 777
 778
 779                 }
 780
 781         }
 782
 783 }
 784
 785
 786 // Break out the components of a multipart message
 787 // (This function expects to be fed HEADERS + CONTENT)
 788 // Note: NULL can be supplied as content_end; in this case, the message is
 789 // considered to have ended when the parser encounters a 0x00 byte.
 790 void the_mime_parser(char *partnum,
 791                      char *content_start, char *content_end,
 792                      MimeParserCallBackType CallBack,
 793                      MimeParserCallBackType PreMultiPartCallBack,
 794                      MimeParserCallBackType PostMultiPartCallBack,
 795                      void *userdata,
 796                      int dont_decode)
 797 {
 798         interesting_mime_headers *m;
 799
 800         // If the caller didn't supply an endpointer, generate one by measure
 801         if (content_end == NULL) {
 802                 content_end = &content_start[strlen(content_start)];
 803         }
 804
 805         m = InitInterestingMimes();
 806
 807         if (!parse_MimeHeaders(m, &content_start, content_end)) {
 808
 809                 recurseable_mime_parser(partnum,
 810                                         content_start, content_end,
 811                                         CallBack,
 812                                         PreMultiPartCallBack,
 813                                         PostMultiPartCallBack,
 814                                         userdata,
 815                                         dont_decode,
 816                                         m);
 817         }
 818         free(m);
 819 }
 820
 821
 822 // Entry point for the MIME parser.
 823 // (This function expects to be fed HEADERS + CONTENT)
 824 // Note: NULL can be supplied as content_end; in this case, the message is
 825 // considered to have ended when the parser encounters a 0x00 byte.
 826 void mime_parser(char *content_start,
 827                  char *content_end,
 828                  MimeParserCallBackType CallBack,
 829                  MimeParserCallBackType PreMultiPartCallBack,
 830                  MimeParserCallBackType PostMultiPartCallBack,
 831                  void *userdata,
 832                  int dont_decode)
 833 {
 834         the_mime_parser("", content_start, content_end,
 835                         CallBack,
 836                         PreMultiPartCallBack,
 837                         PostMultiPartCallBack,
 838                         userdata, dont_decode);
 839 }
 840
 841
 842 typedef struct _MimeGuess {
 843         const char *Pattern;
 844         size_t PatternLen;
 845         long PatternOffset;
 846         const char *MimeString;
 847 } MimeGuess;
 848
 849 MimeGuess MyMimes [] = {
 850         {
 851                 "GIF",
 852                 3,
 853                 0,
 854                 "image/gif"
 855         },
 856         {
 857                 "\xff\xd8",
 858                 2,
 859                 0,
 860                 "image/jpeg"
 861         },
 862         {
 863                 "\x89PNG",
 864                 4,
 865                 0,
 866                 "image/png"
 867         },
 868         { // last...
 869                 "",
 870                 0,
 871                 0,
 872                 ""
 873         }
 874 };
 875
 876
 877 const char *GuessMimeType(const char *data, size_t dlen) {
 878         int MimeIndex = 0;
 879
 880         while (MyMimes[MimeIndex].PatternLen != 0) {
 881                 if ((MyMimes[MimeIndex].PatternLen +
 882                      MyMimes[MimeIndex].PatternOffset < dlen) &&
 883                     strncmp(MyMimes[MimeIndex].Pattern,
 884                             &data[MyMimes[MimeIndex].PatternOffset],
 885                             MyMimes[MimeIndex].PatternLen) == 0)
 886                 {
 887                         return MyMimes[MimeIndex].MimeString;
 888                 }
 889                 MimeIndex ++;
 890         }
 891         // ok, our simple minded algorythm didn't find anything,
 892         // let the big chegger try it, he wil default to application/octet-stream
 893         return (xdg_mime_get_mime_type_for_data(data, dlen));
 894 }
 895
 896
 897 const char* GuessMimeByFilename(const char *what, size_t len) {
 898         // we know some hardcoded on our own, try them...
 899         if ((len > 3) && !strncasecmp(&what[len - 4], ".gif", 4))
 900                 return "image/gif";
 901         else if ((len > 2) && !strncasecmp(&what[len - 3], ".js", 3))
 902                 return  "text/javascript";
 903         else if ((len > 3) && !strncasecmp(&what[len - 4], ".txt", 4))
 904                 return "text/plain";
 905         else if ((len > 3) && !strncasecmp(&what[len - 4], ".css", 4))
 906                 return "text/css";
 907         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htc", 4))
 908                 return "text/x-component";
 909         else if ((len > 3) && !strncasecmp(&what[len - 4], ".jpg", 4))
 910                 return "image/jpeg";
 911         else if ((len > 4) && !strncasecmp(&what[len - 5], ".jpeg", 5))
 912                 return "image/jpeg";
 913         else if ((len > 3) && !strncasecmp(&what[len - 4], ".png", 4))
 914                 return "image/png";
 915         else if ((len > 3) && !strncasecmp(&what[len - 4], ".ico", 4))
 916                 return "image/x-icon";
 917         else if ((len > 3) && !strncasecmp(&what[len - 4], ".vcf", 4))
 918                 return "text/x-vcard";
 919         else if ((len > 4) && !strncasecmp(&what[len - 5], ".html", 5))
 920                 return "text/html";
 921         else if ((len > 3) && !strncasecmp(&what[len - 4], ".htm", 4))
 922                 return "text/html";
 923         else if ((len > 3) && !strncasecmp(&what[len - 4], ".wml", 4))
 924                 return "text/vnd.wap.wml";
 925         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmls", 5))
 926                 return "text/vnd.wap.wmlscript";
 927         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wmlc", 5))
 928                 return "application/vnd.wap.wmlc";
 929         else if ((len > 5) && !strncasecmp(&what[len - 6], ".wmlsc", 6))
 930                 return "application/vnd.wap.wmlscriptc";
 931         else if ((len > 4) && !strncasecmp(&what[len - 5], ".wbmp", 5))
 932                 return "image/vnd.wap.wbmp";
 933         else
 934                 // and let xdgmime do the fallback.
 935                 return xdg_mime_get_mime_type_from_file_name(what);
 936 }
 937
 938 static HashList *IconHash = NULL;
 939
 940 typedef struct IconName IconName;
 941
 942 struct IconName {
 943         char *FlatName;
 944         char *FileName;
 945 };
 946
 947
 948 static void DeleteIcon(void *IconNamePtr) {
 949         IconName *Icon = (IconName*) IconNamePtr;
 950         free(Icon->FlatName);
 951         free(Icon->FileName);
 952         free(Icon);
 953 }
 954
 955
 956 #define GENSTR "x-generic"
 957 #define IGNORE_PREFIX_1 "gnome-mime"
 958 int LoadIconDir(const char *DirName) {
 959         DIR *filedir = NULL;
 960         struct dirent *filedir_entry;
 961         int d_namelen;
 962         int d_without_ext;
 963         IconName *Icon;
 964
 965         filedir = opendir (DirName);
 966         IconHash = NewHash(1, NULL);
 967         if (filedir == NULL) {
 968                 return 0;
 969         }
 970
 971         while ((filedir_entry = readdir(filedir))) {
 972                 char *MinorPtr;
 973                 char *PStart;
 974 #ifdef _DIRENT_HAVE_D_NAMLEN
 975                 d_namelen = filedir_entry->d_namlen;
 976 #else
 977                 d_namelen = strlen(filedir_entry->d_name);
 978 #endif
 979                 d_without_ext = d_namelen;
 980                 while ((d_without_ext > 0) && (filedir_entry->d_name[d_without_ext] != '.'))
 981                         d_without_ext --;
 982                 if ((d_without_ext == 0) || (d_namelen < 3))
 983                         continue;
 984
 985                 if ((sizeof(IGNORE_PREFIX_1) < d_namelen) &&
 986                     (strncmp(IGNORE_PREFIX_1,
 987                              filedir_entry->d_name,
 988                              sizeof(IGNORE_PREFIX_1) - 1) == 0)) {
 989                         PStart = filedir_entry->d_name + sizeof(IGNORE_PREFIX_1);
 990                         d_without_ext -= sizeof(IGNORE_PREFIX_1);
 991                 }
 992                 else {
 993                         PStart = filedir_entry->d_name;
 994                 }
 995                 Icon = malloc(sizeof(IconName));
 996
 997                 Icon->FileName = malloc(d_namelen + 1);
 998                 memcpy(Icon->FileName, filedir_entry->d_name, d_namelen + 1);
 999
1000                 Icon->FlatName = malloc(d_without_ext + 1);
1001                 memcpy(Icon->FlatName, PStart, d_without_ext);
1002                 Icon->FlatName[d_without_ext] = '\0';
1003                 // Try to find Minor type in image-jpeg
1004                 MinorPtr = strchr(Icon->FlatName, '-');
1005                 if (MinorPtr != NULL) {
1006                         size_t MinorLen;
1007                         MinorLen = 1 + d_without_ext - (MinorPtr - Icon->FlatName + 1);
1008                         if ((MinorLen == sizeof(GENSTR)) &&
1009                             (strncmp(MinorPtr + 1, GENSTR, sizeof(GENSTR)) == 0)) {
1010                                 // ok, we found a generic filename. cut the generic.
1011                                 *MinorPtr = '\0';
1012                                 d_without_ext = d_without_ext - (MinorPtr - Icon->FlatName);
1013                         }
1014                         else { // Map the major / minor separator to /
1015                                 *MinorPtr = '/';
1016                         }
1017                 }
1018
1019                 Put(IconHash, Icon->FlatName, d_without_ext, Icon, DeleteIcon);
1020         }
1021         closedir(filedir);
1022         return 1;
1023 }
1024
1025
1026 const char *GetIconFilename(char *MimeType, size_t len) {
1027         void *vIcon;
1028         IconName *Icon;
1029
1030         if (IconHash == NULL) {
1031                 return NULL;
1032         }
1033
1034         GetHash(IconHash, MimeType, len, &vIcon), Icon = (IconName*) vIcon;
1035         // didn't find the exact mimetype? try major only.
1036         if (Icon == NULL) {
1037                 char * pMinor;
1038                 pMinor = strchr(MimeType, '/');
1039                 if (pMinor != NULL) {
1040                         *pMinor = '\0';
1041                         GetHash(IconHash, MimeType, pMinor - MimeType, &vIcon),
1042                                 Icon = (IconName*) vIcon;
1043                 }
1044         }
1045         if (Icon == NULL) {
1046                 return NULL;
1047         }
1048
1049         return Icon->FileName;
1050 }
1051
1052
1053 void ShutDownLibCitadelMime(void) {
1054         DeleteHash(&IconHash);
1055 }