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