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  */
  66 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
  67         char buf[SIZ];
  68         int buf_length = 0;
  69         int soft_line_break = 0;
  70         unsigned int ch;
  71         int decoded_length = 0;
  72         int i;
  73
  74         decoded[0] = 0;
  75         decoded_length = 0;
  76         buf[0] = 0;
  77         buf_length = 0;
  78
  79         for (i = 0; i < sourcelen; ++i) {
  80
  81                 buf[buf_length++] = encoded[i];
  82
  83                 if ( (encoded[i] == '\n')
  84                    || (encoded[i] == 0)
  85                    || (i == (sourcelen-1)) ) {
  86                         buf[buf_length++] = 0;
  87
  88                         /*** begin -- process one line ***/
  89
  90                         if (buf[strlen(buf)-1] == '\n') {
  91                                 buf[strlen(buf)-1] = 0;
  92                         }
  93                         if (buf[strlen(buf)-1] == '\r') {
  94                                 buf[strlen(buf)-1] = 0;
  95                         }
  96                         while (isspace(buf[strlen(buf)-1])) {
  97                                 buf[strlen(buf)-1] = 0;
  98                         }
  99                         soft_line_break = 0;
 100
 101                         while (strlen(buf) > 0) {
 102                                 if (!strcmp(buf, "=")) {
 103                                         soft_line_break = 1;
 104                                         strcpy(buf, "");
 105                                 } else if ((strlen(buf)>=3) && (buf[0]=='=')) {
 106                                         sscanf(&buf[1], "%02x", &ch);
 107                                         decoded[decoded_length++] = ch;
 108                                         strcpy(buf, &buf[3]);
 109                                 } else {
 110                                         decoded[decoded_length++] = buf[0];
 111                                         strcpy(buf, &buf[1]);
 112                                 }
 113                         }
 114                         if (soft_line_break == 0) {
 115                                 decoded[decoded_length++] = '\r';
 116                                 decoded[decoded_length++] = '\n';
 117                         }
 118                         buf_length = 0;
 119                         /*** end -- process one line ***/
 120                 }
 121         }
 122
 123         decoded[decoded_length++] = 0;
 124         return(decoded_length);
 125 }
 126
 127 /*
 128  * Given a message or message-part body and a length, handle any necessary
 129  * decoding and pass the request up the stack.
 130  */
 131 void mime_decode(char *partnum,
 132                  char *part_start, size_t length,
 133                  char *content_type, char *charset, char *encoding,
 134                  char *disposition,
 135                  char *name, char *filename,
 136                  void (*CallBack)
 137                   (char *cbname,
 138                    char *cbfilename,
 139                    char *cbpartnum,
 140                    char *cbdisp,
 141                    void *cbcontent,
 142                    char *cbtype,
 143                    char *cbcharset,
 144                    size_t cblength,
 145                    char *cbencoding,
 146                    void *cbuserdata),
 147                  void (*PreMultiPartCallBack)
 148                   (char *cbname,
 149                    char *cbfilename,
 150                    char *cbpartnum,
 151                    char *cbdisp,
 152                    void *cbcontent,
 153                    char *cbtype,
 154                    char *cbcharset,
 155                    size_t cblength,
 156                    char *cbencoding,
 157                    void *cbuserdata),
 158                  void (*PostMultiPartCallBack)
 159                   (char *cbname,
 160                    char *cbfilename,
 161                    char *cbpartnum,
 162                    char *cbdisp,
 163                    void *cbcontent,
 164                    char *cbtype,
 165                    char *cbcharset,
 166                    size_t cblength,
 167                    char *cbencoding,
 168                    void *cbuserdata),
 169                   void *userdata,
 170                   int dont_decode
 171 )
 172 {
 173
 174         char *decoded;
 175         size_t bytes_decoded = 0;
 176
 177         /* Some encodings aren't really encodings */
 178         if (!strcasecmp(encoding, "7bit"))
 179                 strcpy(encoding, "");
 180         if (!strcasecmp(encoding, "8bit"))
 181                 strcpy(encoding, "");
 182         if (!strcasecmp(encoding, "binary"))
 183                 strcpy(encoding, "");
 184
 185         /* If this part is not encoded, send as-is */
 186         if ( (strlen(encoding) == 0) || (dont_decode)) {
 187                 if (CallBack != NULL) {
 188                         CallBack(name, filename, fixed_partnum(partnum),
 189                                 disposition, part_start,
 190                                 content_type, charset, length, encoding, userdata);
 191                         }
 192                 return;
 193         }
 194
 195         /* Fail silently if we hit an unknown encoding. */
 196         if ((strcasecmp(encoding, "base64"))
 197             && (strcasecmp(encoding, "quoted-printable"))) {
 198                 return;
 199         }
 200
 201         /*
 202          * Allocate a buffer for the decoded data.  The output buffer is slightly
 203          * larger than the input buffer; this assumes that the decoded data
 204          * will never be significantly larger than the encoded data.  This is a
 205          * safe assumption with base64, uuencode, and quoted-printable.
 206          */
 207         decoded = malloc(length + 32768);
 208         if (decoded == NULL) {
 209                 return;
 210         }
 211
 212         if (!strcasecmp(encoding, "base64")) {
 213                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 214         }
 215         else if (!strcasecmp(encoding, "quoted-printable")) {
 216                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded, part_start, length);
 217         }
 218
 219         if (bytes_decoded > 0) if (CallBack != NULL) {
 220                 CallBack(name, filename, fixed_partnum(partnum),
 221                         disposition, decoded,
 222                         content_type, charset, bytes_decoded, "binary", userdata);
 223         }
 224
 225         free(decoded);
 226 }
 227
 228 /*
 229  * Break out the components of a multipart message
 230  * (This function expects to be fed HEADERS + CONTENT)
 231  * Note: NULL can be supplied as content_end; in this case, the message is
 232  * considered to have ended when the parser encounters a 0x00 byte.
 233  */
 234 void the_mime_parser(char *partnum,
 235                      char *content_start, char *content_end,
 236                      void (*CallBack)
 237                       (char *cbname,
 238                        char *cbfilename,
 239                        char *cbpartnum,
 240                        char *cbdisp,
 241                        void *cbcontent,
 242                        char *cbtype,
 243                        char *cbcharset,
 244                        size_t cblength,
 245                        char *cbencoding,
 246                        void *cbuserdata),
 247                      void (*PreMultiPartCallBack)
 248                       (char *cbname,
 249                        char *cbfilename,
 250                        char *cbpartnum,
 251                        char *cbdisp,
 252                        void *cbcontent,
 253                        char *cbtype,
 254                        char *cbcharset,
 255                        size_t cblength,
 256                        char *cbencoding,
 257                        void *cbuserdata),
 258                      void (*PostMultiPartCallBack)
 259                       (char *cbname,
 260                        char *cbfilename,
 261                        char *cbpartnum,
 262                        char *cbdisp,
 263                        void *cbcontent,
 264                        char *cbtype,
 265                        char *cbcharset,
 266                        size_t cblength,
 267                        char *cbencoding,
 268                        void *cbuserdata),
 269                       void *userdata,
 270                       int dont_decode
 271 )
 272 {
 273
 274         char *ptr;
 275         char *srch = NULL;
 276         char *part_start, *part_end = NULL;
 277         char buf[SIZ];
 278         char *header;
 279         char *boundary;
 280         char *startary;
 281         size_t startary_len = 0;
 282         char *endary;
 283         char *next_boundary;
 284         char *content_type;
 285         char *charset;
 286         size_t content_length;
 287         char *encoding;
 288         char *disposition;
 289         char *name = NULL;
 290         char *content_type_name;
 291         char *content_disposition_name;
 292         char *filename;
 293         int is_multipart;
 294         int part_seq = 0;
 295         int i;
 296         size_t length;
 297         char nested_partnum[SIZ];
 298
 299         ptr = content_start;
 300         content_length = 0;
 301
 302         boundary = malloc(SIZ);
 303         memset(boundary, 0, SIZ);
 304
 305         startary = malloc(SIZ);
 306         memset(startary, 0, SIZ);
 307
 308         endary = malloc(SIZ);
 309         memset(endary, 0, SIZ);
 310
 311         header = malloc(SIZ);
 312         memset(header, 0, SIZ);
 313
 314         content_type = malloc(SIZ);
 315         memset(content_type, 0, SIZ);
 316
 317         charset = malloc(SIZ);
 318         memset(charset, 0, SIZ);
 319
 320         encoding = malloc(SIZ);
 321         memset(encoding, 0, SIZ);
 322
 323         content_type_name = malloc(SIZ);
 324         memset(content_type_name, 0, SIZ);
 325
 326         content_disposition_name = malloc(SIZ);
 327         memset(content_disposition_name, 0, SIZ);
 328
 329         filename = malloc(SIZ);
 330         memset(filename, 0, SIZ);
 331
 332         disposition = malloc(SIZ);
 333         memset(disposition, 0, SIZ);
 334
 335         /* If the caller didn't supply an endpointer, generate one by measure */
 336         if (content_end == NULL) {
 337                 content_end = &content_start[strlen(content_start)];
 338         }
 339
 340         /* Learn interesting things from the headers */
 341         strcpy(header, "");
 342         do {
 343                 ptr = memreadline(ptr, buf, SIZ);
 344                 if (ptr >= content_end) {
 345                         goto end_parser;
 346                 }
 347
 348                 for (i = 0; i < strlen(buf); ++i) {
 349                         if (isspace(buf[i])) {
 350                                 buf[i] = ' ';
 351                         }
 352                 }
 353
 354                 if (!isspace(buf[0])) {
 355                         if (!strncasecmp(header, "Content-type: ", 14)) {
 356                                 strcpy(content_type, &header[14]);
 357                                 extract_key(content_type_name, content_type, "name");
 358                                 extract_key(charset, content_type, "charset");
 359                                 /* Deal with weird headers */
 360                                 if (strchr(content_type, ' '))
 361                                         *(strchr(content_type, ' ')) = '\0';
 362                                 if (strchr(content_type, ';'))
 363                                         *(strchr(content_type, ';')) = '\0';
 364                         }
 365                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 366                                 strcpy(disposition, &header[21]);
 367                                 extract_key(content_disposition_name, disposition, "name");
 368                                 extract_key(filename, disposition, "filename");
 369                         }
 370                         if (!strncasecmp(header, "Content-length: ", 16)) {
 371                                 content_length = (size_t) atol(&header[16]);
 372                         }
 373                         if (!strncasecmp(header,
 374                                       "Content-transfer-encoding: ", 27))
 375                                 strcpy(encoding, &header[27]);
 376                         if (strlen(boundary) == 0)
 377                                 extract_key(boundary, header, "boundary");
 378                         strcpy(header, "");
 379                 }
 380                 if ((strlen(header) + strlen(buf) + 2) < SIZ)
 381                         strcat(header, buf);
 382         } while ((strlen(buf) > 0) && (*ptr != 0));
 383
 384         if (strchr(disposition, ';'))
 385                 *(strchr(disposition, ';')) = '\0';
 386         striplt(disposition);
 387         if (strchr(content_type, ';'))
 388                 *(strchr(content_type, ';')) = '\0';
 389         striplt(content_type);
 390
 391         if (strlen(boundary) > 0) {
 392                 is_multipart = 1;
 393         } else {
 394                 is_multipart = 0;
 395         }
 396
 397         /* If this is a multipart message, then recursively process it */
 398         part_start = NULL;
 399         if (is_multipart) {
 400
 401                 /* Tell the client about this message's multipartedness */
 402                 if (PreMultiPartCallBack != NULL) {
 403                         PreMultiPartCallBack("", "", partnum, "",
 404                                 NULL, content_type, charset,
 405                                 0, encoding, userdata);
 406                 }
 407
 408                 /* Figure out where the boundaries are */
 409                 snprintf(startary, SIZ, "--%s", boundary);
 410                 snprintf(endary, SIZ, "--%s--", boundary);
 411                 startary_len = strlen(startary);
 412
 413                 part_start = NULL;
 414                 do {
 415                         next_boundary = NULL;
 416                         for (srch=ptr; srch<content_end; ++srch) {
 417                                 if (!memcmp(srch, startary, startary_len)) {
 418                                         next_boundary = srch;
 419                                         srch = content_end;
 420                                 }
 421                         }
 422
 423                         if ( (part_start != NULL) && (next_boundary != NULL) ) {
 424                                 part_end = next_boundary;
 425                                 --part_end;
 426
 427                                 if (strlen(partnum) > 0) {
 428                                         snprintf(nested_partnum,
 429                                                  sizeof nested_partnum,
 430                                                  "%s.%d", partnum,
 431                                                  ++part_seq);
 432                                 }
 433                                 else {
 434                                         snprintf(nested_partnum,
 435                                                  sizeof nested_partnum,
 436                                                  "%d", ++part_seq);
 437                                 }
 438                                 the_mime_parser(nested_partnum,
 439                                             part_start, part_end,
 440                                                 CallBack,
 441                                                 PreMultiPartCallBack,
 442                                                 PostMultiPartCallBack,
 443                                                 userdata,
 444                                                 dont_decode);
 445                         }
 446
 447                         if (next_boundary != NULL) {
 448                                 /* If we pass out of scope, don't attempt to
 449                                  * read past the end boundary. */
 450                                 if (!strcmp(next_boundary, endary)) {
 451                                         ptr = content_end;
 452                                 }
 453                                 else {
 454                                         /* Set up for the next part. */
 455                                         part_start = strstr(next_boundary, "\n");
 456                                         ++part_start;
 457                                         ptr = part_start;
 458                                 }
 459                         }
 460                         else {
 461                                 /* Invalid end of multipart.  Bail out! */
 462                                 ptr = content_end;
 463                         }
 464                 } while ( (ptr < content_end) && (next_boundary != NULL) );
 465
 466                 if (PostMultiPartCallBack != NULL) {
 467                         PostMultiPartCallBack("", "", partnum, "", NULL,
 468                                 content_type, charset, 0, encoding, userdata);
 469                 }
 470                 goto end_parser;
 471         }
 472
 473         /* If it's not a multipart message, then do something with it */
 474         if (!is_multipart) {
 475                 part_start = ptr;
 476                 length = 0;
 477                 while (ptr < content_end) {
 478                         ++ptr;
 479                         ++length;
 480                 }
 481                 part_end = content_end;
 482                 /* fix an off-by-one error */
 483                 --part_end;
 484                 --length;
 485
 486                 /* Truncate if the header told us to */
 487                 if ( (content_length > 0) && (length > content_length) ) {
 488                         length = content_length;
 489                 }
 490
 491                 /* Sometimes the "name" field is tacked on to Content-type,
 492                  * and sometimes it's tacked on to Content-disposition.  Use
 493                  * whichever one we have.
 494                  */
 495                 if (strlen(content_disposition_name) > strlen(content_type_name)) {
 496                         name = content_disposition_name;
 497                 }
 498                 else {
 499                         name = content_type_name;
 500                 }
 501
 502                 /* lprintf(CTDL_DEBUG, "mime_decode part=%s, len=%d, type=%s, charset=%s, encoding=%s\n",
 503                         partnum, length, content_type, charset, encoding); */
 504
 505                 /* Ok, we've got a non-multipart part here, so do something with it.
 506                  */
 507                 mime_decode(partnum,
 508                         part_start, length,
 509                         content_type, charset, encoding, disposition,
 510                         name, filename,
 511                         CallBack, NULL, NULL,
 512                         userdata, dont_decode
 513                 );
 514
 515                 /*
 516                  * Now if it's an encapsulated message/rfc822 then we have to recurse into it
 517                  */
 518                 if (!strcasecmp(content_type, "message/rfc822")) {
 519
 520                         if (PreMultiPartCallBack != NULL) {
 521                                 PreMultiPartCallBack("", "", partnum, "",
 522                                         NULL, content_type, charset,
 523                                         0, encoding, userdata);
 524                         }
 525                         if (CallBack != NULL) {
 526                                 if (strlen(partnum) > 0) {
 527                                         snprintf(nested_partnum,
 528                                                  sizeof nested_partnum,
 529                                                  "%s.%d", partnum,
 530                                                  ++part_seq);
 531                                 }
 532                                 else {
 533                                         snprintf(nested_partnum,
 534                                                  sizeof nested_partnum,
 535                                                  "%d", ++part_seq);
 536                                 }
 537                                 the_mime_parser(nested_partnum,
 538                                         part_start, part_end,
 539                                         CallBack,
 540                                         PreMultiPartCallBack,
 541                                         PostMultiPartCallBack,
 542                                         userdata,
 543                                         dont_decode
 544                                 );
 545                         }
 546                         if (PostMultiPartCallBack != NULL) {
 547                                 PostMultiPartCallBack("", "", partnum, "", NULL,
 548                                         content_type, charset, 0, encoding, userdata);
 549                         }
 550
 551
 552                 }
 553
 554         }
 555
 556 end_parser:     /* free the buffers!  end the oppression!! */
 557         free(boundary);
 558         free(startary);
 559         free(endary);
 560         free(header);
 561         free(content_type);
 562         free(charset);
 563         free(encoding);
 564         free(content_type_name);
 565         free(content_disposition_name);
 566         free(filename);
 567         free(disposition);
 568 }
 569
 570
 571
 572 /*
 573  * Entry point for the MIME parser.
 574  * (This function expects to be fed HEADERS + CONTENT)
 575  * Note: NULL can be supplied as content_end; in this case, the message is
 576  * considered to have ended when the parser encounters a 0x00 byte.
 577  */
 578 void mime_parser(char *content_start,
 579                 char *content_end,
 580
 581                  void (*CallBack)
 582                   (char *cbname,
 583                    char *cbfilename,
 584                    char *cbpartnum,
 585                    char *cbdisp,
 586                    void *cbcontent,
 587                    char *cbtype,
 588                    char *cbcharset,
 589                    size_t cblength,
 590                    char *cbencoding,
 591                    void *cbuserdata),
 592
 593                  void (*PreMultiPartCallBack)
 594                   (char *cbname,
 595                    char *cbfilename,
 596                    char *cbpartnum,
 597                    char *cbdisp,
 598                    void *cbcontent,
 599                    char *cbtype,
 600                    char *cbcharset,
 601                    size_t cblength,
 602                    char *cbencoding,
 603                    void *cbuserdata),
 604
 605                  void (*PostMultiPartCallBack)
 606                   (char *cbname,
 607                    char *cbfilename,
 608                    char *cbpartnum,
 609                    char *cbdisp,
 610                    void *cbcontent,
 611                    char *cbtype,
 612                    char *cbcharset,
 613                    size_t cblength,
 614                    char *cbencoding,
 615                    void *cbuserdata),
 616
 617                   void *userdata,
 618                   int dont_decode
 619 )
 620 {
 621
 622         the_mime_parser("", content_start, content_end,
 623                         CallBack,
 624                         PreMultiPartCallBack,
 625                         PostMultiPartCallBack,
 626                         userdata, dont_decode);
 627 }