webcit/mime_parser.c

   1 /*
   2  * $Id$
   3  */
   4 /**
   5  * \defgroup MIME This is the MIME parser for Citadel.
   6  *
   7  * Copyright (c) 1998-2005 by Art Cancro
   8  * This code is distributed under the terms of the GNU General Public License.
   9  * \ingroup WebcitHttpServer
  10  */
  11 /*@{*/
  12 #include "webcit.h"
  13 #include "webserver.h"
  14 #include "mime_parser.h"
  15
  16 void extract_key(char *target, char *source, char *key)
  17 {
  18         int a, b;
  19
  20         strcpy(target, source);
  21         for (a = 0; a < strlen(target); ++a) {
  22                 if ((!strncasecmp(&target[a], key, strlen(key)))
  23                     && (target[a + strlen(key)] == '=')) {
  24                         strcpy(target, &target[a + strlen(key) + 1]);
  25                         if (target[0] == 34)
  26                                 strcpy(target, &target[1]);
  27                         for (b = 0; b < strlen(target); ++b)
  28                                 if (target[b] == 34)
  29                                         target[b] = 0;
  30                         return;
  31                 }
  32         }
  33         strcpy(target, "");
  34 }
  35
  36
  37 /*
  38  * For non-multipart messages, we need to generate a quickie partnum of "1"
  39  * to return to callback functions.  Some callbacks demand it.
  40  */
  41 char *fixed_partnum(char *supplied_partnum) {
  42         if (supplied_partnum == NULL) return "1";
  43         if (IsEmptyStr(supplied_partnum)) return "1";
  44         return supplied_partnum;
  45 }
  46
  47
  48
  49 /*
  50  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
  51  * according to RFC2045 section 6.7
  52  */
  53 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
  54         unsigned int ch;
  55         int decoded_length = 0;
  56         int pos = 0;
  57
  58         while (pos < sourcelen)
  59         {
  60                 if (!strncmp(&encoded[pos], "=\r\n", 3))
  61                 {
  62                         pos += 3;
  63                 }
  64                 else if (!strncmp(&encoded[pos], "=\n", 2))
  65                 {
  66                         pos += 2;
  67                 }
  68                 else if (encoded[pos] == '=')
  69                 {
  70                         ch = 0;
  71                         sscanf(&encoded[pos+1], "%02x", &ch);
  72                         pos += 3;
  73                         decoded[decoded_length++] = ch;
  74                 }
  75                 else
  76                 {
  77                         decoded[decoded_length++] = encoded[pos];
  78                         pos += 1;
  79                 }
  80         }
  81         decoded[decoded_length] = 0;
  82         return(decoded_length);
  83 }
  84
  85
  86 /*
  87  * Given a message or message-part body and a length, handle any necessary
  88  * decoding and pass the request up the stack.
  89  */
  90 void mime_decode(char *partnum,
  91                  char *part_start, size_t length,
  92                  char *content_type, char *charset, char *encoding,
  93                  char *disposition,
  94                  char *name, char *filename,
  95                  void (*CallBack)
  96                   (char *cbname,
  97                    char *cbfilename,
  98                    char *cbpartnum,
  99                    char *cbdisp,
 100                    void *cbcontent,
 101                    char *cbtype,
 102                    char *cbcharset,
 103                    size_t cblength,
 104                    char *cbencoding,
 105                    void *cbuserdata),
 106                  void (*PreMultiPartCallBack)
 107                   (char *cbname,
 108                    char *cbfilename,
 109                    char *cbpartnum,
 110                    char *cbdisp,
 111                    void *cbcontent,
 112                    char *cbtype,
 113                    char *cbcharset,
 114                    size_t cblength,
 115                    char *cbencoding,
 116                    void *cbuserdata),
 117                  void (*PostMultiPartCallBack)
 118                   (char *cbname,
 119                    char *cbfilename,
 120                    char *cbpartnum,
 121                    char *cbdisp,
 122                    void *cbcontent,
 123                    char *cbtype,
 124                    char *cbcharset,
 125                    size_t cblength,
 126                    char *cbencoding,
 127                    void *cbuserdata),
 128                   void *userdata,
 129                   int dont_decode
 130 )
 131 {
 132
 133         char *decoded;
 134         size_t bytes_decoded = 0;
 135
 136         /* Some encodings aren't really encodings */
 137         if (!strcasecmp(encoding, "7bit"))
 138                 strcpy(encoding, "");
 139         if (!strcasecmp(encoding, "8bit"))
 140                 strcpy(encoding, "");
 141         if (!strcasecmp(encoding, "binary"))
 142                 strcpy(encoding, "");
 143
 144         /* If this part is not encoded, send as-is */
 145         if ( (IsEmptyStr(encoding)) || (dont_decode)) {
 146                 if (CallBack != NULL) {
 147                         CallBack(name, filename, fixed_partnum(partnum),
 148                                 disposition, part_start,
 149                                 content_type, charset, length, encoding, userdata);
 150                         }
 151                 return;
 152         }
 153
 154         /* Fail silently if we hit an unknown encoding. */
 155         if ((strcasecmp(encoding, "base64"))
 156             && (strcasecmp(encoding, "quoted-printable"))) {
 157                 return;
 158         }
 159
 160         /*
 161          * Allocate a buffer for the decoded data.  The output buffer is slightly
 162          * larger than the input buffer; this assumes that the decoded data
 163          * will never be significantly larger than the encoded data.  This is a
 164          * safe assumption with base64, uuencode, and quoted-printable.
 165          */
 166         decoded = malloc(length + 32768);
 167         if (decoded == NULL) {
 168                 return;
 169         }
 170
 171         if (!strcasecmp(encoding, "base64")) {
 172                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 173         }
 174         else if (!strcasecmp(encoding, "quoted-printable")) {
 175                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded, part_start, length);
 176         }
 177
 178         if (bytes_decoded > 0) if (CallBack != NULL) {
 179                 CallBack(name, filename, fixed_partnum(partnum),
 180                         disposition, decoded,
 181                         content_type, charset, bytes_decoded, "binary", userdata);
 182         }
 183
 184         free(decoded);
 185 }
 186
 187 /*
 188  * Break out the components of a multipart message
 189  * (This function expects to be fed HEADERS + CONTENT)
 190  * Note: NULL can be supplied as content_end; in this case, the message is
 191  * considered to have ended when the parser encounters a 0x00 byte.
 192  */
 193 void the_mime_parser(char *partnum,
 194                      char *content_start, char *content_end,
 195                      void (*CallBack)
 196                       (char *cbname,
 197                        char *cbfilename,
 198                        char *cbpartnum,
 199                        char *cbdisp,
 200                        void *cbcontent,
 201                        char *cbtype,
 202                        char *cbcharset,
 203                        size_t cblength,
 204                        char *cbencoding,
 205                        void *cbuserdata),
 206                      void (*PreMultiPartCallBack)
 207                       (char *cbname,
 208                        char *cbfilename,
 209                        char *cbpartnum,
 210                        char *cbdisp,
 211                        void *cbcontent,
 212                        char *cbtype,
 213                        char *cbcharset,
 214                        size_t cblength,
 215                        char *cbencoding,
 216                        void *cbuserdata),
 217                      void (*PostMultiPartCallBack)
 218                       (char *cbname,
 219                        char *cbfilename,
 220                        char *cbpartnum,
 221                        char *cbdisp,
 222                        void *cbcontent,
 223                        char *cbtype,
 224                        char *cbcharset,
 225                        size_t cblength,
 226                        char *cbencoding,
 227                        void *cbuserdata),
 228                       void *userdata,
 229                       int dont_decode
 230 )
 231 {
 232
 233         char *ptr;
 234         char *srch = NULL;
 235         char *part_start, *part_end = NULL;
 236         char buf[SIZ];
 237         char *header;
 238         char *boundary;
 239         char *startary;
 240         size_t startary_len = 0;
 241         char *endary;
 242         char *next_boundary;
 243         char *content_type;
 244         char *charset;
 245         size_t content_length;
 246         char *encoding;
 247         char *disposition;
 248         char *name = NULL;
 249         char *content_type_name;
 250         char *content_disposition_name;
 251         char *filename;
 252         int is_multipart;
 253         int part_seq = 0;
 254         int i;
 255         size_t length;
 256         char nested_partnum[256];
 257         int crlf_in_use = 0;
 258         char *evaluate_crlf_ptr = NULL;
 259
 260         ptr = content_start;
 261         content_length = 0;
 262
 263         boundary = malloc(SIZ);
 264         memset(boundary, 0, SIZ);
 265
 266         startary = malloc(SIZ);
 267         memset(startary, 0, SIZ);
 268
 269         endary = malloc(SIZ);
 270         memset(endary, 0, SIZ);
 271
 272         header = malloc(SIZ);
 273         memset(header, 0, SIZ);
 274
 275         content_type = malloc(SIZ);
 276         memset(content_type, 0, SIZ);
 277
 278         charset = malloc(SIZ);
 279         memset(charset, 0, SIZ);
 280
 281         encoding = malloc(SIZ);
 282         memset(encoding, 0, SIZ);
 283
 284         content_type_name = malloc(SIZ);
 285         memset(content_type_name, 0, SIZ);
 286
 287         content_disposition_name = malloc(SIZ);
 288         memset(content_disposition_name, 0, SIZ);
 289
 290         filename = malloc(SIZ);
 291         memset(filename, 0, SIZ);
 292
 293         disposition = malloc(SIZ);
 294         memset(disposition, 0, SIZ);
 295
 296         /* If the caller didn't supply an endpointer, generate one by measure */
 297         if (content_end == NULL) {
 298                 content_end = &content_start[strlen(content_start)];
 299         }
 300
 301         /* Learn interesting things from the headers */
 302         strcpy(header, "");
 303         do {
 304                 int len;
 305                 ptr = memreadline(ptr, buf, SIZ);
 306                 if (ptr >= content_end) {
 307                         goto end_parser;
 308                 }
 309
 310                 len = strlen (buf);
 311                 for (i = 0; i < len; ++i) {
 312                         if (isspace(buf[i])) {
 313                                 buf[i] = ' ';
 314                         }
 315                 }
 316
 317                 if (!isspace(buf[0])) {
 318                         if (!strncasecmp(header, "Content-type:", 13)) {
 319                                 strcpy(content_type, &header[13]);
 320                                 striplt(content_type);
 321                                 extract_key(content_type_name, content_type, "name");
 322                                 extract_key(charset, content_type, "charset");
 323                                 /* Deal with weird headers */
 324                                 if (strchr(content_type, ' '))
 325                                         *(strchr(content_type, ' ')) = '\0';
 326                                 if (strchr(content_type, ';'))
 327                                         *(strchr(content_type, ';')) = '\0';
 328                         }
 329                         if (!strncasecmp(header, "Content-Disposition:", 20)) {
 330                                 strcpy(disposition, &header[20]);
 331                                 striplt(disposition);
 332                                 extract_key(content_disposition_name, disposition, "name");
 333                                 extract_key(filename, disposition, "filename");
 334                         }
 335                         if (!strncasecmp(header, "Content-length: ", 15)) {
 336                                 char clbuf[10];
 337                                 safestrncpy(clbuf, &header[15], sizeof clbuf);
 338                                 striplt(clbuf);
 339                                 content_length = (size_t) atol(clbuf);
 340                         }
 341                         if (!strncasecmp(header, "Content-transfer-encoding: ", 26)) {
 342                                 strcpy(encoding, &header[26]);
 343                                 striplt(encoding);
 344                         }
 345                         if (IsEmptyStr(boundary))
 346                                 extract_key(boundary, header, "boundary");
 347                         strcpy(header, "");
 348                 }
 349                 if ((strlen(header) + strlen(buf) + 2) < SIZ) {
 350                         strcat(header, buf);
 351                 }
 352         } while ((!IsEmptyStr(buf)) && (*ptr != 0));
 353
 354         if (strchr(disposition, ';'))
 355                 *(strchr(disposition, ';')) = '\0';
 356         striplt(disposition);
 357         if (strchr(content_type, ';'))
 358                 *(strchr(content_type, ';')) = '\0';
 359         striplt(content_type);
 360
 361         if (!IsEmptyStr(boundary)) {
 362                 is_multipart = 1;
 363         } else {
 364                 is_multipart = 0;
 365         }
 366
 367         /* If this is a multipart message, then recursively process it */
 368         part_start = NULL;
 369         if (is_multipart) {
 370
 371                 /* Tell the client about this message's multipartedness */
 372                 if (PreMultiPartCallBack != NULL) {
 373                         PreMultiPartCallBack("", "", partnum, "",
 374                                 NULL, content_type, charset,
 375                                 0, encoding, userdata);
 376                 }
 377
 378                 /* Figure out where the boundaries are */
 379                 snprintf(startary, SIZ, "--%s", boundary);
 380                 snprintf(endary, SIZ, "--%s--", boundary);
 381                 startary_len = strlen(startary);
 382
 383                 part_start = NULL;
 384                 do {
 385                         next_boundary = NULL;
 386                         for (srch=ptr; srch<content_end; ++srch) {
 387                                 if (!memcmp(srch, startary, startary_len)) {
 388                                         next_boundary = srch;
 389                                         srch = content_end;
 390                                 }
 391                         }
 392
 393                         if ( (part_start != NULL) && (next_boundary != NULL) ) {
 394                                 part_end = next_boundary;
 395                                 --part_end;             /* omit the trailing LF */
 396                                 if (crlf_in_use) {
 397                                         --part_end;     /* omit the trailing CR */
 398                                 }
 399
 400                                 if (!IsEmptyStr(partnum)) {
 401                                         snprintf(nested_partnum,
 402                                                  sizeof nested_partnum,
 403                                                  "%s.%d", partnum,
 404                                                  ++part_seq);
 405                                 }
 406                                 else {
 407                                         snprintf(nested_partnum,
 408                                                  sizeof nested_partnum,
 409                                                  "%d", ++part_seq);
 410                                 }
 411                                 the_mime_parser(nested_partnum,
 412                                             part_start, part_end,
 413                                                 CallBack,
 414                                                 PreMultiPartCallBack,
 415                                                 PostMultiPartCallBack,
 416                                                 userdata,
 417                                                 dont_decode);
 418                         }
 419
 420                         if (next_boundary != NULL) {
 421                                 /* If we pass out of scope, don't attempt to
 422                                  * read past the end boundary. */
 423                                 if (!strcmp(next_boundary, endary)) {
 424                                         ptr = content_end;
 425                                 }
 426                                 else {
 427                                         /* Set up for the next part. */
 428                                         part_start = strstr(next_boundary, "\n");
 429
 430                                         /* Determine whether newlines are LF or CRLF */
 431                                         evaluate_crlf_ptr = part_start;
 432                                         --evaluate_crlf_ptr;
 433                                         if (!memcmp(evaluate_crlf_ptr, "\r\n", 2)) {
 434                                                 crlf_in_use = 1;
 435                                         }
 436                                         else {
 437                                                 crlf_in_use = 0;
 438                                         }
 439
 440                                         /* Advance past the LF ... now we're in the next part */
 441                                         ++part_start;
 442                                         ptr = part_start;
 443                                 }
 444                         }
 445                         else {
 446                                 /* Invalid end of multipart.  Bail out! */
 447                                 ptr = content_end;
 448                         }
 449                 } while ( (ptr < content_end) && (next_boundary != NULL) );
 450
 451                 if (PostMultiPartCallBack != NULL) {
 452                         PostMultiPartCallBack("", "", partnum, "", NULL,
 453                                 content_type, charset, 0, encoding, userdata);
 454                 }
 455                 goto end_parser;
 456         }
 457
 458         /* If it's not a multipart message, then do something with it */
 459         if (!is_multipart) {
 460                 part_start = ptr;
 461                 length = 0;
 462                 while (ptr < content_end) {
 463                         ++ptr;
 464                         ++length;
 465                 }
 466                 part_end = content_end;
 467
 468                 /******
 469                  * I thought there was an off-by-one error here, but there isn't.
 470                  * This probably means that there's an off-by-one error somewhere
 471                  * else ... or maybe only in certain messages?
 472                 --part_end;
 473                 --length;
 474                 ******/
 475
 476                 /* Truncate if the header told us to */
 477                 if ( (content_length > 0) && (length > content_length) ) {
 478                         length = content_length;
 479                 }
 480
 481                 /* Sometimes the "name" field is tacked on to Content-type,
 482                  * and sometimes it's tacked on to Content-disposition.  Use
 483                  * whichever one we have.
 484                  */
 485                 if (strlen(content_disposition_name) > strlen(content_type_name)) {
 486                         name = content_disposition_name;
 487                 }
 488                 else {
 489                         name = content_type_name;
 490                 }
 491
 492                 /* lprintf(CTDL_DEBUG, "mime_decode part=%s, len=%d, type=%s, charset=%s, encoding=%s\n",
 493                         partnum, length, content_type, charset, encoding); */
 494
 495                 /* Ok, we've got a non-multipart part here, so do something with it.
 496                  */
 497                 mime_decode(partnum,
 498                         part_start, length,
 499                         content_type, charset, encoding, disposition,
 500                         name, filename,
 501                         CallBack, NULL, NULL,
 502                         userdata, dont_decode
 503                 );
 504
 505                 /*
 506                  * Now if it's an encapsulated message/rfc822 then we have to recurse into it
 507                  */
 508                 if (!strcasecmp(content_type, "message/rfc822")) {
 509
 510                         if (PreMultiPartCallBack != NULL) {
 511                                 PreMultiPartCallBack("", "", partnum, "",
 512                                         NULL, content_type, charset,
 513                                         0, encoding, userdata);
 514                         }
 515                         if (CallBack != NULL) {
 516                                 if (!IsEmptyStr(partnum)) {
 517                                         snprintf(nested_partnum,
 518                                                  sizeof nested_partnum,
 519                                                  "%s.%d", partnum,
 520                                                  ++part_seq);
 521                                 }
 522                                 else {
 523                                         snprintf(nested_partnum,
 524                                                  sizeof nested_partnum,
 525                                                  "%d", ++part_seq);
 526                                 }
 527                                 the_mime_parser(nested_partnum,
 528                                         part_start, part_end,
 529                                         CallBack,
 530                                         PreMultiPartCallBack,
 531                                         PostMultiPartCallBack,
 532                                         userdata,
 533                                         dont_decode
 534                                 );
 535                         }
 536                         if (PostMultiPartCallBack != NULL) {
 537                                 PostMultiPartCallBack("", "", partnum, "", NULL,
 538                                         content_type, charset, 0, encoding, userdata);
 539                         }
 540
 541
 542                 }
 543
 544         }
 545
 546 end_parser:     /* free the buffers!  end the oppression!! */
 547         free(boundary);
 548         free(startary);
 549         free(endary);
 550         free(header);
 551         free(content_type);
 552         free(charset);
 553         free(encoding);
 554         free(content_type_name);
 555         free(content_disposition_name);
 556         free(filename);
 557         free(disposition);
 558 }
 559
 560
 561
 562 /*
 563  * Entry point for the MIME parser.
 564  * (This function expects to be fed HEADERS + CONTENT)
 565  * Note: NULL can be supplied as content_end; in this case, the message is
 566  * considered to have ended when the parser encounters a 0x00 byte.
 567  */
 568 void mime_parser(char *content_start,
 569                 char *content_end,
 570
 571                  void (*CallBack)
 572                   (char *cbname,
 573                    char *cbfilename,
 574                    char *cbpartnum,
 575                    char *cbdisp,
 576                    void *cbcontent,
 577                    char *cbtype,
 578                    char *cbcharset,
 579                    size_t cblength,
 580                    char *cbencoding,
 581                    void *cbuserdata),
 582
 583                  void (*PreMultiPartCallBack)
 584                   (char *cbname,
 585                    char *cbfilename,
 586                    char *cbpartnum,
 587                    char *cbdisp,
 588                    void *cbcontent,
 589                    char *cbtype,
 590                    char *cbcharset,
 591                    size_t cblength,
 592                    char *cbencoding,
 593                    void *cbuserdata),
 594
 595                  void (*PostMultiPartCallBack)
 596                   (char *cbname,
 597                    char *cbfilename,
 598                    char *cbpartnum,
 599                    char *cbdisp,
 600                    void *cbcontent,
 601                    char *cbtype,
 602                    char *cbcharset,
 603                    size_t cblength,
 604                    char *cbencoding,
 605                    void *cbuserdata),
 606
 607                   void *userdata,
 608                   int dont_decode
 609 )
 610 {
 611
 612         the_mime_parser("", content_start, content_end,
 613                         CallBack,
 614                         PreMultiPartCallBack,
 615                         PostMultiPartCallBack,
 616                         userdata, dont_decode);
 617 }
 618
 619 /*@}*/