citadel/mime_parser.c

   1 /*
   2  * $Id$
   3  *
   4  * This is the MIME parser for Citadel.
   5  *
   6  * Copyright (c) 1998-2006 by Art Cancro
   7  * This code is distributed under the GNU General Public License v2.
   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 <errno.h>
  20
  21 #include "citadel.h"
  22 #include "server.h"
  23 #include "sysdep_decls.h"
  24 #include "tools.h"
  25
  26 #include "mime_parser.h"
  27
  28
  29 void extract_key(char *target, char *source, char *key)
  30 {
  31         int a, b;
  32
  33         strcpy(target, source);
  34         for (a = 0; a < strlen(target); ++a) {
  35                 if ((!strncasecmp(&target[a], key, strlen(key)))
  36                     && (target[a + strlen(key)] == '=')) {
  37                         strcpy(target, &target[a + strlen(key) + 1]);
  38                         if (target[0] == 34)
  39                                 strcpy(target, &target[1]);
  40                         for (b = 0; b < strlen(target); ++b)
  41                                 if (target[b] == 34)
  42                                         target[b] = 0;
  43                         return;
  44                 }
  45         }
  46         strcpy(target, "");
  47 }
  48
  49
  50 /*
  51  * For non-multipart messages, we need to generate a quickie partnum of "1"
  52  * to return to callback functions.  Some callbacks demand it.
  53  */
  54 char *fixed_partnum(char *supplied_partnum) {
  55         if (supplied_partnum == NULL) return "1";
  56         if (strlen(supplied_partnum)==0) return "1";
  57         return supplied_partnum;
  58 }
  59
  60
  61
  62 /*
  63  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
  64  * according to RFC2045 section 6.7
  65  */
  66 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
  67         unsigned int ch;
  68         int decoded_length = 0;
  69         int pos = 0;
  70
  71         while (pos < sourcelen)
  72         {
  73                 if (!strncmp(&encoded[pos], "=\r\n", 3))
  74                 {
  75                         pos += 3;
  76                 }
  77                 else if (!strncmp(&encoded[pos], "=\n", 2))
  78                 {
  79                         pos += 2;
  80                 }
  81                 else if (encoded[pos] == '=')
  82                 {
  83                         ch = 0;
  84                         sscanf(&encoded[pos+1], "%02x", &ch);
  85                         pos += 3;
  86                         decoded[decoded_length++] = ch;
  87                 }
  88                 else
  89                 {
  90                         decoded[decoded_length++] = encoded[pos];
  91                         pos += 1;
  92                 }
  93         }
  94         decoded[decoded_length] = 0;
  95         return(decoded_length);
  96 }
  97
  98
  99 /*
 100  * Given a message or message-part body and a length, handle any necessary
 101  * decoding and pass the request up the stack.
 102  */
 103 void mime_decode(char *partnum,
 104                  char *part_start, size_t length,
 105                  char *content_type, char *charset, char *encoding,
 106                  char *disposition,
 107                  char *name, char *filename,
 108                  void (*CallBack)
 109                   (char *cbname,
 110                    char *cbfilename,
 111                    char *cbpartnum,
 112                    char *cbdisp,
 113                    void *cbcontent,
 114                    char *cbtype,
 115                    char *cbcharset,
 116                    size_t cblength,
 117                    char *cbencoding,
 118                    void *cbuserdata),
 119                  void (*PreMultiPartCallBack)
 120                   (char *cbname,
 121                    char *cbfilename,
 122                    char *cbpartnum,
 123                    char *cbdisp,
 124                    void *cbcontent,
 125                    char *cbtype,
 126                    char *cbcharset,
 127                    size_t cblength,
 128                    char *cbencoding,
 129                    void *cbuserdata),
 130                  void (*PostMultiPartCallBack)
 131                   (char *cbname,
 132                    char *cbfilename,
 133                    char *cbpartnum,
 134                    char *cbdisp,
 135                    void *cbcontent,
 136                    char *cbtype,
 137                    char *cbcharset,
 138                    size_t cblength,
 139                    char *cbencoding,
 140                    void *cbuserdata),
 141                   void *userdata,
 142                   int dont_decode
 143 )
 144 {
 145
 146         char *decoded;
 147         size_t bytes_decoded = 0;
 148
 149         /* Some encodings aren't really encodings */
 150         if (!strcasecmp(encoding, "7bit"))
 151                 strcpy(encoding, "");
 152         if (!strcasecmp(encoding, "8bit"))
 153                 strcpy(encoding, "");
 154         if (!strcasecmp(encoding, "binary"))
 155                 strcpy(encoding, "");
 156
 157         /* If this part is not encoded, send as-is */
 158         if ( (strlen(encoding) == 0) || (dont_decode)) {
 159                 if (CallBack != NULL) {
 160                         CallBack(name, filename, fixed_partnum(partnum),
 161                                 disposition, part_start,
 162                                 content_type, charset, length, encoding, userdata);
 163                         }
 164                 return;
 165         }
 166
 167         /* Fail silently if we hit an unknown encoding. */
 168         if ((strcasecmp(encoding, "base64"))
 169             && (strcasecmp(encoding, "quoted-printable"))) {
 170                 return;
 171         }
 172
 173         /*
 174          * Allocate a buffer for the decoded data.  The output buffer is slightly
 175          * larger than the input buffer; this assumes that the decoded data
 176          * will never be significantly larger than the encoded data.  This is a
 177          * safe assumption with base64, uuencode, and quoted-printable.
 178          */
 179         decoded = malloc(length + 32768);
 180         if (decoded == NULL) {
 181                 return;
 182         }
 183
 184         if (!strcasecmp(encoding, "base64")) {
 185                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 186         }
 187         else if (!strcasecmp(encoding, "quoted-printable")) {
 188                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded, part_start, length);
 189         }
 190
 191         if (bytes_decoded > 0) if (CallBack != NULL) {
 192                 CallBack(name, filename, fixed_partnum(partnum),
 193                         disposition, decoded,
 194                         content_type, charset, bytes_decoded, "binary", userdata);
 195         }
 196
 197         free(decoded);
 198 }
 199
 200 /*
 201  * Break out the components of a multipart message
 202  * (This function expects to be fed HEADERS + CONTENT)
 203  * Note: NULL can be supplied as content_end; in this case, the message is
 204  * considered to have ended when the parser encounters a 0x00 byte.
 205  */
 206 void the_mime_parser(char *partnum,
 207                      char *content_start, char *content_end,
 208                      void (*CallBack)
 209                       (char *cbname,
 210                        char *cbfilename,
 211                        char *cbpartnum,
 212                        char *cbdisp,
 213                        void *cbcontent,
 214                        char *cbtype,
 215                        char *cbcharset,
 216                        size_t cblength,
 217                        char *cbencoding,
 218                        void *cbuserdata),
 219                      void (*PreMultiPartCallBack)
 220                       (char *cbname,
 221                        char *cbfilename,
 222                        char *cbpartnum,
 223                        char *cbdisp,
 224                        void *cbcontent,
 225                        char *cbtype,
 226                        char *cbcharset,
 227                        size_t cblength,
 228                        char *cbencoding,
 229                        void *cbuserdata),
 230                      void (*PostMultiPartCallBack)
 231                       (char *cbname,
 232                        char *cbfilename,
 233                        char *cbpartnum,
 234                        char *cbdisp,
 235                        void *cbcontent,
 236                        char *cbtype,
 237                        char *cbcharset,
 238                        size_t cblength,
 239                        char *cbencoding,
 240                        void *cbuserdata),
 241                       void *userdata,
 242                       int dont_decode
 243 )
 244 {
 245
 246         char *ptr;
 247         char *srch = NULL;
 248         char *part_start, *part_end = NULL;
 249         char buf[SIZ];
 250         char *header;
 251         char *boundary;
 252         char *startary;
 253         size_t startary_len = 0;
 254         char *endary;
 255         char *next_boundary;
 256         char *content_type;
 257         char *charset;
 258         size_t content_length;
 259         char *encoding;
 260         char *disposition;
 261         char *name = NULL;
 262         char *content_type_name;
 263         char *content_disposition_name;
 264         char *filename;
 265         int is_multipart;
 266         int part_seq = 0;
 267         int i;
 268         size_t length;
 269         char nested_partnum[256];
 270         int crlf_in_use = 0;
 271         char *evaluate_crlf_ptr = NULL;
 272         int buflen = 0;
 273         int headerlen = 0;
 274
 275         ptr = content_start;
 276         content_length = 0;
 277
 278         boundary = malloc(SIZ);
 279         memset(boundary, 0, SIZ);
 280
 281         startary = malloc(SIZ);
 282         memset(startary, 0, SIZ);
 283
 284         endary = malloc(SIZ);
 285         memset(endary, 0, SIZ);
 286
 287         header = malloc(SIZ);
 288         memset(header, 0, SIZ);
 289
 290         content_type = malloc(SIZ);
 291         memset(content_type, 0, SIZ);
 292
 293         charset = malloc(SIZ);
 294         memset(charset, 0, SIZ);
 295
 296         encoding = malloc(SIZ);
 297         memset(encoding, 0, SIZ);
 298
 299         content_type_name = malloc(SIZ);
 300         memset(content_type_name, 0, SIZ);
 301
 302         content_disposition_name = malloc(SIZ);
 303         memset(content_disposition_name, 0, SIZ);
 304
 305         filename = malloc(SIZ);
 306         memset(filename, 0, SIZ);
 307
 308         disposition = malloc(SIZ);
 309         memset(disposition, 0, SIZ);
 310
 311         /* If the caller didn't supply an endpointer, generate one by measure */
 312         if (content_end == NULL) {
 313                 content_end = &content_start[strlen(content_start)];
 314         }
 315
 316         /* Learn interesting things from the headers */
 317         strcpy(header, "");
 318         headerlen = 0;
 319         do {
 320                 ptr = memreadlinelen(ptr, buf, SIZ, &buflen);
 321                 if (ptr >= content_end) {
 322                         goto end_parser;
 323                 }
 324
 325                 for (i = 0; i < buflen; ++i) {
 326                         if (isspace(buf[i])) {
 327                                 buf[i] = ' ';
 328                         }
 329                 }
 330
 331                 if (!isspace(buf[0])) {
 332                         if (!strncasecmp(header, "Content-type:", 13)) {
 333                                 strcpy(content_type, &header[13]);
 334                                 striplt(content_type);
 335                                 extract_key(content_type_name, content_type, "name");
 336                                 extract_key(charset, content_type, "charset");
 337                                 /* Deal with weird headers */
 338                                 if (strchr(content_type, ' '))
 339                                         *(strchr(content_type, ' ')) = '\0';
 340                                 if (strchr(content_type, ';'))
 341                                         *(strchr(content_type, ';')) = '\0';
 342                         }
 343                         if (!strncasecmp(header, "Content-Disposition:", 20)) {
 344                                 strcpy(disposition, &header[20]);
 345                                 striplt(disposition);
 346                                 extract_key(content_disposition_name, disposition, "name");
 347                                 extract_key(filename, disposition, "filename");
 348                         }
 349                         if (!strncasecmp(header, "Content-length: ", 15)) {
 350                                 char clbuf[10];
 351                                 safestrncpy(clbuf, &header[15], sizeof clbuf);
 352                                 striplt(clbuf);
 353                                 content_length = (size_t) atol(clbuf);
 354                         }
 355                         if (!strncasecmp(header, "Content-transfer-encoding: ", 26)) {
 356                                 strcpy(encoding, &header[26]);
 357                                 striplt(encoding);
 358                         }
 359                         if (strlen(boundary) == 0)
 360                                 extract_key(boundary, header, "boundary");
 361                         strcpy(header, "");
 362                         headerlen = 0;
 363                 }
 364                 if ((headerlen + buflen + 2) < SIZ) {
 365                         memcpy(&header[headerlen], buf, buflen);
 366                         headerlen += buflen;
 367                         header[headerlen] = '\0';
 368                 }
 369         } while ((!IsEmptyStr(buf)) && (*ptr != 0));
 370
 371         if (strchr(disposition, ';'))
 372                 *(strchr(disposition, ';')) = '\0';
 373         striplt(disposition);
 374         if (strchr(content_type, ';'))
 375                 *(strchr(content_type, ';')) = '\0';
 376         striplt(content_type);
 377
 378         if (!IsEmptyStr(boundary)) {
 379                 is_multipart = 1;
 380         } else {
 381                 is_multipart = 0;
 382         }
 383
 384         /* If this is a multipart message, then recursively process it */
 385         part_start = NULL;
 386         if (is_multipart) {
 387
 388                 /* Tell the client about this message's multipartedness */
 389                 if (PreMultiPartCallBack != NULL) {
 390                         PreMultiPartCallBack("", "", partnum, "",
 391                                 NULL, content_type, charset,
 392                                 0, encoding, userdata);
 393                 }
 394
 395                 /* Figure out where the boundaries are */
 396                 snprintf(startary, SIZ, "--%s", boundary);
 397                 snprintf(endary, SIZ, "--%s--", boundary);
 398                 startary_len = strlen(startary);
 399
 400                 part_start = NULL;
 401                 do {
 402                         next_boundary = NULL;
 403                         for (srch=ptr; srch<content_end; ++srch) {
 404                                 if (!memcmp(srch, startary, startary_len)) {
 405                                         next_boundary = srch;
 406                                         srch = content_end;
 407                                 }
 408                         }
 409
 410                         if ( (part_start != NULL) && (next_boundary != NULL) ) {
 411                                 part_end = next_boundary;
 412                                 --part_end;             /* omit the trailing LF */
 413                                 if (crlf_in_use) {
 414                                         --part_end;     /* omit the trailing CR */
 415                                 }
 416
 417                                 if (!IsEmptyStr(partnum)) {
 418                                         snprintf(nested_partnum,
 419                                                  sizeof nested_partnum,
 420                                                  "%s.%d", partnum,
 421                                                  ++part_seq);
 422                                 }
 423                                 else {
 424                                         snprintf(nested_partnum,
 425                                                  sizeof nested_partnum,
 426                                                  "%d", ++part_seq);
 427                                 }
 428                                 the_mime_parser(nested_partnum,
 429                                             part_start, part_end,
 430                                                 CallBack,
 431                                                 PreMultiPartCallBack,
 432                                                 PostMultiPartCallBack,
 433                                                 userdata,
 434                                                 dont_decode);
 435                         }
 436
 437                         if (next_boundary != NULL) {
 438                                 /* If we pass out of scope, don't attempt to
 439                                  * read past the end boundary. */
 440                                 if (!strcmp(next_boundary, endary)) {
 441                                         ptr = content_end;
 442                                 }
 443                                 else {
 444                                         /* Set up for the next part. */
 445                                         part_start = strstr(next_boundary, "\n");
 446
 447                                         /* Determine whether newlines are LF or CRLF */
 448                                         evaluate_crlf_ptr = part_start;
 449                                         --evaluate_crlf_ptr;
 450                                         if (!memcmp(evaluate_crlf_ptr, "\r\n", 2)) {
 451                                                 crlf_in_use = 1;
 452                                         }
 453                                         else {
 454                                                 crlf_in_use = 0;
 455                                         }
 456
 457                                         /* Advance past the LF ... now we're in the next part */
 458                                         ++part_start;
 459                                         ptr = part_start;
 460                                 }
 461                         }
 462                         else {
 463                                 /* Invalid end of multipart.  Bail out! */
 464                                 ptr = content_end;
 465                         }
 466                 } while ( (ptr < content_end) && (next_boundary != NULL) );
 467
 468                 if (PostMultiPartCallBack != NULL) {
 469                         PostMultiPartCallBack("", "", partnum, "", NULL,
 470                                 content_type, charset, 0, encoding, userdata);
 471                 }
 472                 goto end_parser;
 473         }
 474
 475         /* If it's not a multipart message, then do something with it */
 476         if (!is_multipart) {
 477                 part_start = ptr;
 478                 length = 0;
 479                 while (ptr < content_end) {
 480                         ++ptr;
 481                         ++length;
 482                 }
 483                 part_end = content_end;
 484
 485                 /******
 486                  * I thought there was an off-by-one error here, but there isn't.
 487                  * This probably means that there's an off-by-one error somewhere
 488                  * else ... or maybe only in certain messages?
 489                 --part_end;
 490                 --length;
 491                 ******/
 492
 493                 /* Truncate if the header told us to */
 494                 if ( (content_length > 0) && (length > content_length) ) {
 495                         length = content_length;
 496                 }
 497
 498                 /* Sometimes the "name" field is tacked on to Content-type,
 499                  * and sometimes it's tacked on to Content-disposition.  Use
 500                  * whichever one we have.
 501                  */
 502                 if (strlen(content_disposition_name) > strlen(content_type_name)) {
 503                         name = content_disposition_name;
 504                 }
 505                 else {
 506                         name = content_type_name;
 507                 }
 508
 509                 /* lprintf(CTDL_DEBUG, "mime_decode part=%s, len=%d, type=%s, charset=%s, encoding=%s\n",
 510                         partnum, length, content_type, charset, encoding); */
 511
 512                 /* Ok, we've got a non-multipart part here, so do something with it.
 513                  */
 514                 mime_decode(partnum,
 515                         part_start, length,
 516                         content_type, charset, encoding, disposition,
 517                         name, filename,
 518                         CallBack, NULL, NULL,
 519                         userdata, dont_decode
 520                 );
 521
 522                 /*
 523                  * Now if it's an encapsulated message/rfc822 then we have to recurse into it
 524                  */
 525                 if (!strcasecmp(content_type, "message/rfc822")) {
 526
 527                         if (PreMultiPartCallBack != NULL) {
 528                                 PreMultiPartCallBack("", "", partnum, "",
 529                                         NULL, content_type, charset,
 530                                         0, encoding, userdata);
 531                         }
 532                         if (CallBack != NULL) {
 533                                 if (strlen(partnum) > 0) {
 534                                         snprintf(nested_partnum,
 535                                                  sizeof nested_partnum,
 536                                                  "%s.%d", partnum,
 537                                                  ++part_seq);
 538                                 }
 539                                 else {
 540                                         snprintf(nested_partnum,
 541                                                  sizeof nested_partnum,
 542                                                  "%d", ++part_seq);
 543                                 }
 544                                 the_mime_parser(nested_partnum,
 545                                         part_start, part_end,
 546                                         CallBack,
 547                                         PreMultiPartCallBack,
 548                                         PostMultiPartCallBack,
 549                                         userdata,
 550                                         dont_decode
 551                                 );
 552                         }
 553                         if (PostMultiPartCallBack != NULL) {
 554                                 PostMultiPartCallBack("", "", partnum, "", NULL,
 555                                         content_type, charset, 0, encoding, userdata);
 556                         }
 557
 558
 559                 }
 560
 561         }
 562
 563 end_parser:     /* free the buffers!  end the oppression!! */
 564         free(boundary);
 565         free(startary);
 566         free(endary);
 567         free(header);
 568         free(content_type);
 569         free(charset);
 570         free(encoding);
 571         free(content_type_name);
 572         free(content_disposition_name);
 573         free(filename);
 574         free(disposition);
 575 }
 576
 577
 578
 579 /*
 580  * Entry point for the MIME parser.
 581  * (This function expects to be fed HEADERS + CONTENT)
 582  * Note: NULL can be supplied as content_end; in this case, the message is
 583  * considered to have ended when the parser encounters a 0x00 byte.
 584  */
 585 void mime_parser(char *content_start,
 586                 char *content_end,
 587
 588                  void (*CallBack)
 589                   (char *cbname,
 590                    char *cbfilename,
 591                    char *cbpartnum,
 592                    char *cbdisp,
 593                    void *cbcontent,
 594                    char *cbtype,
 595                    char *cbcharset,
 596                    size_t cblength,
 597                    char *cbencoding,
 598                    void *cbuserdata),
 599
 600                  void (*PreMultiPartCallBack)
 601                   (char *cbname,
 602                    char *cbfilename,
 603                    char *cbpartnum,
 604                    char *cbdisp,
 605                    void *cbcontent,
 606                    char *cbtype,
 607                    char *cbcharset,
 608                    size_t cblength,
 609                    char *cbencoding,
 610                    void *cbuserdata),
 611
 612                  void (*PostMultiPartCallBack)
 613                   (char *cbname,
 614                    char *cbfilename,
 615                    char *cbpartnum,
 616                    char *cbdisp,
 617                    void *cbcontent,
 618                    char *cbtype,
 619                    char *cbcharset,
 620                    size_t cblength,
 621                    char *cbencoding,
 622                    void *cbuserdata),
 623
 624                   void *userdata,
 625                   int dont_decode
 626 )
 627 {
 628
 629         the_mime_parser("", content_start, content_end,
 630                         CallBack,
 631                         PreMultiPartCallBack,
 632                         PostMultiPartCallBack,
 633                         userdata, dont_decode);
 634 }