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         memset(buf, '\0', SIZ);
  77         buf_length = 0;
  78
  79         for (i = 0; i < sourcelen; ++i) {
  80                 size_t buflen = strlen(buf);
  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[buflen-1] == '\n') {
  91                                 buf[buflen-1] = 0;
  92                                 buflen --;
  93                         }
  94                         if (buf[buflen-1] == '\r') {
  95                                 buf[buflen-1] = 0;
  96                                 buflen --;
  97                         }
  98                         while (isspace(buf[buflen-1])) {
  99                                 buf[buflen-1] = 0;
 100                                 buflen --;
 101                         }
 102                         soft_line_break = 0;
 103
 104                         while (buflen > 0) {
 105                                 if ((buf[0] ==  '=') && (buf[1] == '\0')) {
 106                                         soft_line_break = 1;
 107                                         buf[0] = '\0';
 108                                         buflen = 0;
 109                                 } else if ((buflen>=3) && (buf[0]=='=')) {
 110                                         sscanf(&buf[1], "%02x", &ch);
 111                                         decoded[decoded_length++] = ch;
 112                                         strcpy(buf, &buf[3]);
 113                                         buflen -=3;
 114                                 } else {
 115                                         decoded[decoded_length++] = buf[0];
 116                                         strcpy(buf, &buf[1]);
 117                                         buflen --;
 118                                 }
 119                         }
 120                         if (soft_line_break == 0) {
 121                                 decoded[decoded_length++] = '\r';
 122                                 decoded[decoded_length++] = '\n';
 123                         }
 124                         buf_length = 0;
 125                         /*** end -- process one line ***/
 126                 }
 127         }
 128
 129         decoded[decoded_length++] = 0;
 130         return(decoded_length);
 131 }
 132
 133 /*
 134  * Given a message or message-part body and a length, handle any necessary
 135  * decoding and pass the request up the stack.
 136  */
 137 void mime_decode(char *partnum,
 138                  char *part_start, size_t length,
 139                  char *content_type, char *charset, char *encoding,
 140                  char *disposition,
 141                  char *name, char *filename,
 142                  void (*CallBack)
 143                   (char *cbname,
 144                    char *cbfilename,
 145                    char *cbpartnum,
 146                    char *cbdisp,
 147                    void *cbcontent,
 148                    char *cbtype,
 149                    char *cbcharset,
 150                    size_t cblength,
 151                    char *cbencoding,
 152                    void *cbuserdata),
 153                  void (*PreMultiPartCallBack)
 154                   (char *cbname,
 155                    char *cbfilename,
 156                    char *cbpartnum,
 157                    char *cbdisp,
 158                    void *cbcontent,
 159                    char *cbtype,
 160                    char *cbcharset,
 161                    size_t cblength,
 162                    char *cbencoding,
 163                    void *cbuserdata),
 164                  void (*PostMultiPartCallBack)
 165                   (char *cbname,
 166                    char *cbfilename,
 167                    char *cbpartnum,
 168                    char *cbdisp,
 169                    void *cbcontent,
 170                    char *cbtype,
 171                    char *cbcharset,
 172                    size_t cblength,
 173                    char *cbencoding,
 174                    void *cbuserdata),
 175                   void *userdata,
 176                   int dont_decode
 177 )
 178 {
 179
 180         char *decoded;
 181         size_t bytes_decoded = 0;
 182
 183         /* Some encodings aren't really encodings */
 184         if (!strcasecmp(encoding, "7bit"))
 185                 strcpy(encoding, "");
 186         if (!strcasecmp(encoding, "8bit"))
 187                 strcpy(encoding, "");
 188         if (!strcasecmp(encoding, "binary"))
 189                 strcpy(encoding, "");
 190
 191         /* If this part is not encoded, send as-is */
 192         if ( (strlen(encoding) == 0) || (dont_decode)) {
 193                 if (CallBack != NULL) {
 194                         CallBack(name, filename, fixed_partnum(partnum),
 195                                 disposition, part_start,
 196                                 content_type, charset, length, encoding, userdata);
 197                         }
 198                 return;
 199         }
 200
 201         /* Fail silently if we hit an unknown encoding. */
 202         if ((strcasecmp(encoding, "base64"))
 203             && (strcasecmp(encoding, "quoted-printable"))) {
 204                 return;
 205         }
 206
 207         /*
 208          * Allocate a buffer for the decoded data.  The output buffer is slightly
 209          * larger than the input buffer; this assumes that the decoded data
 210          * will never be significantly larger than the encoded data.  This is a
 211          * safe assumption with base64, uuencode, and quoted-printable.
 212          */
 213         decoded = malloc(length + 32768);
 214         if (decoded == NULL) {
 215                 return;
 216         }
 217
 218         if (!strcasecmp(encoding, "base64")) {
 219                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 220         }
 221         else if (!strcasecmp(encoding, "quoted-printable")) {
 222                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded, part_start, length);
 223         }
 224
 225         if (bytes_decoded > 0) if (CallBack != NULL) {
 226                 CallBack(name, filename, fixed_partnum(partnum),
 227                         disposition, decoded,
 228                         content_type, charset, bytes_decoded, "binary", userdata);
 229         }
 230
 231         free(decoded);
 232 }
 233
 234 /*
 235  * Break out the components of a multipart message
 236  * (This function expects to be fed HEADERS + CONTENT)
 237  * Note: NULL can be supplied as content_end; in this case, the message is
 238  * considered to have ended when the parser encounters a 0x00 byte.
 239  */
 240 void the_mime_parser(char *partnum,
 241                      char *content_start, char *content_end,
 242                      void (*CallBack)
 243                       (char *cbname,
 244                        char *cbfilename,
 245                        char *cbpartnum,
 246                        char *cbdisp,
 247                        void *cbcontent,
 248                        char *cbtype,
 249                        char *cbcharset,
 250                        size_t cblength,
 251                        char *cbencoding,
 252                        void *cbuserdata),
 253                      void (*PreMultiPartCallBack)
 254                       (char *cbname,
 255                        char *cbfilename,
 256                        char *cbpartnum,
 257                        char *cbdisp,
 258                        void *cbcontent,
 259                        char *cbtype,
 260                        char *cbcharset,
 261                        size_t cblength,
 262                        char *cbencoding,
 263                        void *cbuserdata),
 264                      void (*PostMultiPartCallBack)
 265                       (char *cbname,
 266                        char *cbfilename,
 267                        char *cbpartnum,
 268                        char *cbdisp,
 269                        void *cbcontent,
 270                        char *cbtype,
 271                        char *cbcharset,
 272                        size_t cblength,
 273                        char *cbencoding,
 274                        void *cbuserdata),
 275                       void *userdata,
 276                       int dont_decode
 277 )
 278 {
 279
 280         char *ptr;
 281         char *srch = NULL;
 282         char *part_start, *part_end = NULL;
 283         char buf[SIZ];
 284         char *header;
 285         char *boundary;
 286         char *startary;
 287         size_t startary_len = 0;
 288         char *endary;
 289         char *next_boundary;
 290         char *content_type;
 291         char *charset;
 292         size_t content_length;
 293         char *encoding;
 294         char *disposition;
 295         char *name = NULL;
 296         char *content_type_name;
 297         char *content_disposition_name;
 298         char *filename;
 299         int is_multipart;
 300         int part_seq = 0;
 301         int i;
 302         size_t length;
 303         char nested_partnum[SIZ];
 304
 305         ptr = content_start;
 306         content_length = 0;
 307
 308         boundary = malloc(SIZ);
 309         memset(boundary, 0, SIZ);
 310
 311         startary = malloc(SIZ);
 312         memset(startary, 0, SIZ);
 313
 314         endary = malloc(SIZ);
 315         memset(endary, 0, SIZ);
 316
 317         header = malloc(SIZ);
 318         memset(header, 0, SIZ);
 319
 320         content_type = malloc(SIZ);
 321         memset(content_type, 0, SIZ);
 322
 323         charset = malloc(SIZ);
 324         memset(charset, 0, SIZ);
 325
 326         encoding = malloc(SIZ);
 327         memset(encoding, 0, SIZ);
 328
 329         content_type_name = malloc(SIZ);
 330         memset(content_type_name, 0, SIZ);
 331
 332         content_disposition_name = malloc(SIZ);
 333         memset(content_disposition_name, 0, SIZ);
 334
 335         filename = malloc(SIZ);
 336         memset(filename, 0, SIZ);
 337
 338         disposition = malloc(SIZ);
 339         memset(disposition, 0, SIZ);
 340
 341         /* If the caller didn't supply an endpointer, generate one by measure */
 342         if (content_end == NULL) {
 343                 content_end = &content_start[strlen(content_start)];
 344         }
 345
 346         /* Learn interesting things from the headers */
 347         strcpy(header, "");
 348         do {
 349                 ptr = memreadline(ptr, buf, SIZ);
 350                 if (ptr >= content_end) {
 351                         goto end_parser;
 352                 }
 353
 354                 for (i = 0; i < strlen(buf); ++i) {
 355                         if (isspace(buf[i])) {
 356                                 buf[i] = ' ';
 357                         }
 358                 }
 359
 360                 if (!isspace(buf[0])) {
 361                         if (!strncasecmp(header, "Content-type: ", 14)) {
 362                                 strcpy(content_type, &header[14]);
 363                                 extract_key(content_type_name, content_type, "name");
 364                                 extract_key(charset, content_type, "charset");
 365                                 /* Deal with weird headers */
 366                                 if (strchr(content_type, ' '))
 367                                         *(strchr(content_type, ' ')) = '\0';
 368                                 if (strchr(content_type, ';'))
 369                                         *(strchr(content_type, ';')) = '\0';
 370                         }
 371                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 372                                 strcpy(disposition, &header[21]);
 373                                 extract_key(content_disposition_name, disposition, "name");
 374                                 extract_key(filename, disposition, "filename");
 375                         }
 376                         if (!strncasecmp(header, "Content-length: ", 16)) {
 377                                 content_length = (size_t) atol(&header[16]);
 378                         }
 379                         if (!strncasecmp(header,
 380                                       "Content-transfer-encoding: ", 27))
 381                                 strcpy(encoding, &header[27]);
 382                         if (strlen(boundary) == 0)
 383                                 extract_key(boundary, header, "boundary");
 384                         strcpy(header, "");
 385                 }
 386                 if ((strlen(header) + strlen(buf) + 2) < SIZ)
 387                         strcat(header, buf);
 388         } while ((strlen(buf) > 0) && (*ptr != 0));
 389
 390         if (strchr(disposition, ';'))
 391                 *(strchr(disposition, ';')) = '\0';
 392         striplt(disposition);
 393         if (strchr(content_type, ';'))
 394                 *(strchr(content_type, ';')) = '\0';
 395         striplt(content_type);
 396
 397         if (strlen(boundary) > 0) {
 398                 is_multipart = 1;
 399         } else {
 400                 is_multipart = 0;
 401         }
 402
 403         /* If this is a multipart message, then recursively process it */
 404         part_start = NULL;
 405         if (is_multipart) {
 406
 407                 /* Tell the client about this message's multipartedness */
 408                 if (PreMultiPartCallBack != NULL) {
 409                         PreMultiPartCallBack("", "", partnum, "",
 410                                 NULL, content_type, charset,
 411                                 0, encoding, userdata);
 412                 }
 413
 414                 /* Figure out where the boundaries are */
 415                 snprintf(startary, SIZ, "--%s", boundary);
 416                 snprintf(endary, SIZ, "--%s--", boundary);
 417                 startary_len = strlen(startary);
 418
 419                 part_start = NULL;
 420                 do {
 421                         next_boundary = NULL;
 422                         for (srch=ptr; srch<content_end; ++srch) {
 423                                 if (!memcmp(srch, startary, startary_len)) {
 424                                         next_boundary = srch;
 425                                         srch = content_end;
 426                                 }
 427                         }
 428
 429                         if ( (part_start != NULL) && (next_boundary != NULL) ) {
 430                                 part_end = next_boundary;
 431                                 --part_end;
 432
 433                                 if (strlen(partnum) > 0) {
 434                                         snprintf(nested_partnum,
 435                                                  sizeof nested_partnum,
 436                                                  "%s.%d", partnum,
 437                                                  ++part_seq);
 438                                 }
 439                                 else {
 440                                         snprintf(nested_partnum,
 441                                                  sizeof nested_partnum,
 442                                                  "%d", ++part_seq);
 443                                 }
 444                                 the_mime_parser(nested_partnum,
 445                                             part_start, part_end,
 446                                                 CallBack,
 447                                                 PreMultiPartCallBack,
 448                                                 PostMultiPartCallBack,
 449                                                 userdata,
 450                                                 dont_decode);
 451                         }
 452
 453                         if (next_boundary != NULL) {
 454                                 /* If we pass out of scope, don't attempt to
 455                                  * read past the end boundary. */
 456                                 if (!strcmp(next_boundary, endary)) {
 457                                         ptr = content_end;
 458                                 }
 459                                 else {
 460                                         /* Set up for the next part. */
 461                                         part_start = strstr(next_boundary, "\n");
 462                                         ++part_start;
 463                                         ptr = part_start;
 464                                 }
 465                         }
 466                         else {
 467                                 /* Invalid end of multipart.  Bail out! */
 468                                 ptr = content_end;
 469                         }
 470                 } while ( (ptr < content_end) && (next_boundary != NULL) );
 471
 472                 if (PostMultiPartCallBack != NULL) {
 473                         PostMultiPartCallBack("", "", partnum, "", NULL,
 474                                 content_type, charset, 0, encoding, userdata);
 475                 }
 476                 goto end_parser;
 477         }
 478
 479         /* If it's not a multipart message, then do something with it */
 480         if (!is_multipart) {
 481                 part_start = ptr;
 482                 length = 0;
 483                 while (ptr < content_end) {
 484                         ++ptr;
 485                         ++length;
 486                 }
 487                 part_end = content_end;
 488                 /* fix an off-by-one error */
 489                 --part_end;
 490                 --length;
 491
 492                 /* Truncate if the header told us to */
 493                 if ( (content_length > 0) && (length > content_length) ) {
 494                         length = content_length;
 495                 }
 496
 497                 /* Sometimes the "name" field is tacked on to Content-type,
 498                  * and sometimes it's tacked on to Content-disposition.  Use
 499                  * whichever one we have.
 500                  */
 501                 if (strlen(content_disposition_name) > strlen(content_type_name)) {
 502                         name = content_disposition_name;
 503                 }
 504                 else {
 505                         name = content_type_name;
 506                 }
 507
 508                 /* lprintf(CTDL_DEBUG, "mime_decode part=%s, len=%d, type=%s, charset=%s, encoding=%s\n",
 509                         partnum, length, content_type, charset, encoding); */
 510
 511                 /* Ok, we've got a non-multipart part here, so do something with it.
 512                  */
 513                 mime_decode(partnum,
 514                         part_start, length,
 515                         content_type, charset, encoding, disposition,
 516                         name, filename,
 517                         CallBack, NULL, NULL,
 518                         userdata, dont_decode
 519                 );
 520
 521                 /*
 522                  * Now if it's an encapsulated message/rfc822 then we have to recurse into it
 523                  */
 524                 if (!strcasecmp(content_type, "message/rfc822")) {
 525
 526                         if (PreMultiPartCallBack != NULL) {
 527                                 PreMultiPartCallBack("", "", partnum, "",
 528                                         NULL, content_type, charset,
 529                                         0, encoding, userdata);
 530                         }
 531                         if (CallBack != NULL) {
 532                                 if (strlen(partnum) > 0) {
 533                                         snprintf(nested_partnum,
 534                                                  sizeof nested_partnum,
 535                                                  "%s.%d", partnum,
 536                                                  ++part_seq);
 537                                 }
 538                                 else {
 539                                         snprintf(nested_partnum,
 540                                                  sizeof nested_partnum,
 541                                                  "%d", ++part_seq);
 542                                 }
 543                                 the_mime_parser(nested_partnum,
 544                                         part_start, part_end,
 545                                         CallBack,
 546                                         PreMultiPartCallBack,
 547                                         PostMultiPartCallBack,
 548                                         userdata,
 549                                         dont_decode
 550                                 );
 551                         }
 552                         if (PostMultiPartCallBack != NULL) {
 553                                 PostMultiPartCallBack("", "", partnum, "", NULL,
 554                                         content_type, charset, 0, encoding, userdata);
 555                         }
 556
 557
 558                 }
 559
 560         }
 561
 562 end_parser:     /* free the buffers!  end the oppression!! */
 563         free(boundary);
 564         free(startary);
 565         free(endary);
 566         free(header);
 567         free(content_type);
 568         free(charset);
 569         free(encoding);
 570         free(content_type_name);
 571         free(content_disposition_name);
 572         free(filename);
 573         free(disposition);
 574 }
 575
 576
 577
 578 /*
 579  * Entry point for the MIME parser.
 580  * (This function expects to be fed HEADERS + CONTENT)
 581  * Note: NULL can be supplied as content_end; in this case, the message is
 582  * considered to have ended when the parser encounters a 0x00 byte.
 583  */
 584 void mime_parser(char *content_start,
 585                 char *content_end,
 586
 587                  void (*CallBack)
 588                   (char *cbname,
 589                    char *cbfilename,
 590                    char *cbpartnum,
 591                    char *cbdisp,
 592                    void *cbcontent,
 593                    char *cbtype,
 594                    char *cbcharset,
 595                    size_t cblength,
 596                    char *cbencoding,
 597                    void *cbuserdata),
 598
 599                  void (*PreMultiPartCallBack)
 600                   (char *cbname,
 601                    char *cbfilename,
 602                    char *cbpartnum,
 603                    char *cbdisp,
 604                    void *cbcontent,
 605                    char *cbtype,
 606                    char *cbcharset,
 607                    size_t cblength,
 608                    char *cbencoding,
 609                    void *cbuserdata),
 610
 611                  void (*PostMultiPartCallBack)
 612                   (char *cbname,
 613                    char *cbfilename,
 614                    char *cbpartnum,
 615                    char *cbdisp,
 616                    void *cbcontent,
 617                    char *cbtype,
 618                    char *cbcharset,
 619                    size_t cblength,
 620                    char *cbencoding,
 621                    void *cbuserdata),
 622
 623                   void *userdata,
 624                   int dont_decode
 625 )
 626 {
 627
 628         the_mime_parser("", content_start, content_end,
 629                         CallBack,
 630                         PreMultiPartCallBack,
 631                         PostMultiPartCallBack,
 632                         userdata, dont_decode);
 633 }