webcit/mime_parser.c

   1 /*
   2  * $Id$
   3  *
   4  * This is the MIME parser for Citadel.
   5  *
   6  * Copyright (c) 1998-2005 by Art Cancro
   7  * This code is distributed under the terms of the GNU General Public License.
   8  *
   9  */
  10
  11 #include "webcit.h"
  12 #include "webserver.h"
  13 #include "mime_parser.h"
  14
  15
  16 void extract_key(char *target, char *source, char *key)
  17 {
  18         int a, b;
  19
  20         strcpy(target, source);
  21         for (a = 0; a < strlen(target); ++a) {
  22                 if ((!strncasecmp(&target[a], key, strlen(key)))
  23                     && (target[a + strlen(key)] == '=')) {
  24                         strcpy(target, &target[a + strlen(key) + 1]);
  25                         if (target[0] == 34)
  26                                 strcpy(target, &target[1]);
  27                         for (b = 0; b < strlen(target); ++b)
  28                                 if (target[b] == 34)
  29                                         target[b] = 0;
  30                         return;
  31                 }
  32         }
  33         strcpy(target, "");
  34 }
  35
  36
  37 /*
  38  * For non-multipart messages, we need to generate a quickie partnum of "1"
  39  * to return to callback functions.  Some callbacks demand it.
  40  */
  41 char *fixed_partnum(char *supplied_partnum) {
  42         if (supplied_partnum == NULL) return "1";
  43         if (strlen(supplied_partnum)==0) return "1";
  44         return supplied_partnum;
  45 }
  46
  47
  48
  49 /*
  50  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
  51  */
  52 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
  53         char buf[SIZ];
  54         int buf_length = 0;
  55         int soft_line_break = 0;
  56         unsigned int ch;
  57         int decoded_length = 0;
  58         int i;
  59
  60         decoded[0] = 0;
  61         decoded_length = 0;
  62         buf[0] = 0;
  63         buf_length = 0;
  64
  65         for (i = 0; i < sourcelen; ++i) {
  66
  67                 buf[buf_length++] = encoded[i];
  68
  69                 if ( (encoded[i] == '\n')
  70                    || (encoded[i] == 0)
  71                    || (i == (sourcelen-1)) ) {
  72                         buf[buf_length++] = 0;
  73
  74                         /*** begin -- process one line ***/
  75
  76                         if (buf[strlen(buf)-1] == '\n') {
  77                                 buf[strlen(buf)-1] = 0;
  78                         }
  79                         if (buf[strlen(buf)-1] == '\r') {
  80                                 buf[strlen(buf)-1] = 0;
  81                         }
  82                         while (isspace(buf[strlen(buf)-1])) {
  83                                 buf[strlen(buf)-1] = 0;
  84                         }
  85                         soft_line_break = 0;
  86
  87                         while (strlen(buf) > 0) {
  88                                 if (!strcmp(buf, "=")) {
  89                                         soft_line_break = 1;
  90                                         strcpy(buf, "");
  91                                 } else if ((strlen(buf)>=3) && (buf[0]=='=')) {
  92                                         sscanf(&buf[1], "%02x", &ch);
  93                                         decoded[decoded_length++] = ch;
  94                                         strcpy(buf, &buf[3]);
  95                                 } else {
  96                                         decoded[decoded_length++] = buf[0];
  97                                         strcpy(buf, &buf[1]);
  98                                 }
  99                         }
 100                         if (soft_line_break == 0) {
 101                                 decoded[decoded_length++] = '\r';
 102                                 decoded[decoded_length++] = '\n';
 103                         }
 104                         buf_length = 0;
 105                         /*** end -- process one line ***/
 106                 }
 107         }
 108
 109         decoded[decoded_length++] = 0;
 110         return(decoded_length);
 111 }
 112
 113 /*
 114  * Given a message or message-part body and a length, handle any necessary
 115  * decoding and pass the request up the stack.
 116  */
 117 void mime_decode(char *partnum,
 118                  char *part_start, size_t length,
 119                  char *content_type, char *charset, char *encoding,
 120                  char *disposition,
 121                  char *name, char *filename,
 122                  void (*CallBack)
 123                   (char *cbname,
 124                    char *cbfilename,
 125                    char *cbpartnum,
 126                    char *cbdisp,
 127                    void *cbcontent,
 128                    char *cbtype,
 129                    char *cbcharset,
 130                    size_t cblength,
 131                    char *cbencoding,
 132                    void *cbuserdata),
 133                  void (*PreMultiPartCallBack)
 134                   (char *cbname,
 135                    char *cbfilename,
 136                    char *cbpartnum,
 137                    char *cbdisp,
 138                    void *cbcontent,
 139                    char *cbtype,
 140                    char *cbcharset,
 141                    size_t cblength,
 142                    char *cbencoding,
 143                    void *cbuserdata),
 144                  void (*PostMultiPartCallBack)
 145                   (char *cbname,
 146                    char *cbfilename,
 147                    char *cbpartnum,
 148                    char *cbdisp,
 149                    void *cbcontent,
 150                    char *cbtype,
 151                    char *cbcharset,
 152                    size_t cblength,
 153                    char *cbencoding,
 154                    void *cbuserdata),
 155                   void *userdata,
 156                   int dont_decode
 157 )
 158 {
 159
 160         char *decoded;
 161         size_t bytes_decoded = 0;
 162
 163         /* Some encodings aren't really encodings */
 164         if (!strcasecmp(encoding, "7bit"))
 165                 strcpy(encoding, "");
 166         if (!strcasecmp(encoding, "8bit"))
 167                 strcpy(encoding, "");
 168         if (!strcasecmp(encoding, "binary"))
 169                 strcpy(encoding, "");
 170
 171         /* If this part is not encoded, send as-is */
 172         if ( (strlen(encoding) == 0) || (dont_decode)) {
 173                 if (CallBack != NULL) {
 174                         CallBack(name, filename, fixed_partnum(partnum),
 175                                 disposition, part_start,
 176                                 content_type, charset, length, encoding, userdata);
 177                         }
 178                 return;
 179         }
 180
 181         if ((strcasecmp(encoding, "base64"))
 182             && (strcasecmp(encoding, "quoted-printable"))) {
 183                 return;
 184         }
 185         /*
 186          * Allocate a buffer for the decoded data.  The output buffer is the
 187          * same size as the input buffer; this assumes that the decoded data
 188          * will never be larger than the encoded data.  This is a safe
 189          * assumption with base64, uuencode, and quoted-printable.
 190          */
 191         decoded = malloc(length+2048);
 192         if (decoded == NULL) {
 193                 return;
 194         }
 195
 196         if (!strcasecmp(encoding, "base64")) {
 197                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 198         }
 199         else if (!strcasecmp(encoding, "quoted-printable")) {
 200                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded,
 201                                                         part_start, length);
 202         }
 203
 204         if (bytes_decoded > 0) if (CallBack != NULL) {
 205                 CallBack(name, filename, fixed_partnum(partnum),
 206                         disposition, decoded,
 207                         content_type, charset, bytes_decoded, "binary", userdata);
 208         }
 209
 210         free(decoded);
 211 }
 212
 213 /*
 214  * Break out the components of a multipart message
 215  * (This function expects to be fed HEADERS + CONTENT)
 216  * Note: NULL can be supplied as content_end; in this case, the message is
 217  * considered to have ended when the parser encounters a 0x00 byte.
 218  */
 219 void the_mime_parser(char *partnum,
 220                      char *content_start, char *content_end,
 221                      void (*CallBack)
 222                       (char *cbname,
 223                        char *cbfilename,
 224                        char *cbpartnum,
 225                        char *cbdisp,
 226                        void *cbcontent,
 227                        char *cbtype,
 228                        char *cbcharset,
 229                        size_t cblength,
 230                        char *cbencoding,
 231                        void *cbuserdata),
 232                      void (*PreMultiPartCallBack)
 233                       (char *cbname,
 234                        char *cbfilename,
 235                        char *cbpartnum,
 236                        char *cbdisp,
 237                        void *cbcontent,
 238                        char *cbtype,
 239                        char *cbcharset,
 240                        size_t cblength,
 241                        char *cbencoding,
 242                        void *cbuserdata),
 243                      void (*PostMultiPartCallBack)
 244                       (char *cbname,
 245                        char *cbfilename,
 246                        char *cbpartnum,
 247                        char *cbdisp,
 248                        void *cbcontent,
 249                        char *cbtype,
 250                        char *cbcharset,
 251                        size_t cblength,
 252                        char *cbencoding,
 253                        void *cbuserdata),
 254                       void *userdata,
 255                       int dont_decode
 256 )
 257 {
 258
 259         char *ptr;
 260         char *srch = NULL;
 261         char *part_start, *part_end = NULL;
 262         char buf[SIZ];
 263         char *header;
 264         char *boundary;
 265         char *startary;
 266         size_t startary_len = 0;
 267         char *endary;
 268         char *next_boundary;
 269         char *content_type;
 270         char *charset;
 271         size_t content_length;
 272         char *encoding;
 273         char *disposition;
 274         char *name = NULL;
 275         char *content_type_name;
 276         char *content_disposition_name;
 277         char *filename;
 278         int is_multipart;
 279         int part_seq = 0;
 280         int i;
 281         size_t length;
 282         char nested_partnum[SIZ];
 283
 284         ptr = content_start;
 285         content_length = 0;
 286
 287         boundary = malloc(SIZ);
 288         memset(boundary, 0, SIZ);
 289
 290         startary = malloc(SIZ);
 291         memset(startary, 0, SIZ);
 292
 293         endary = malloc(SIZ);
 294         memset(endary, 0, SIZ);
 295
 296         header = malloc(SIZ);
 297         memset(header, 0, SIZ);
 298
 299         content_type = malloc(SIZ);
 300         memset(content_type, 0, SIZ);
 301
 302         charset = malloc(SIZ);
 303         memset(charset, 0, SIZ);
 304
 305         encoding = malloc(SIZ);
 306         memset(encoding, 0, SIZ);
 307
 308         content_type_name = malloc(SIZ);
 309         memset(content_type_name, 0, SIZ);
 310
 311         content_disposition_name = malloc(SIZ);
 312         memset(content_disposition_name, 0, SIZ);
 313
 314         filename = malloc(SIZ);
 315         memset(filename, 0, SIZ);
 316
 317         disposition = malloc(SIZ);
 318         memset(disposition, 0, SIZ);
 319
 320         /* If the caller didn't supply an endpointer, generate one by measure */
 321         if (content_end == NULL) {
 322                 content_end = &content_start[strlen(content_start)];
 323         }
 324
 325         /* Learn interesting things from the headers */
 326         strcpy(header, "");
 327         do {
 328                 ptr = memreadline(ptr, buf, SIZ);
 329                 if (ptr >= content_end) {
 330                         goto end_parser;
 331                 }
 332
 333                 for (i = 0; i < strlen(buf); ++i) {
 334                         if (isspace(buf[i])) {
 335                                 buf[i] = ' ';
 336                         }
 337                 }
 338
 339                 if (!isspace(buf[0])) {
 340                         if (!strncasecmp(header, "Content-type: ", 14)) {
 341                                 strcpy(content_type, &header[14]);
 342                                 extract_key(content_type_name, content_type, "name");
 343                                 extract_key(charset, content_type, "charset");
 344                                 /* Deal with weird headers */
 345                                 if (strchr(content_type, ' '))
 346                                         *(strchr(content_type, ' ')) = '\0';
 347                                 if (strchr(content_type, ';'))
 348                                         *(strchr(content_type, ';')) = '\0';
 349                         }
 350                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 351                                 strcpy(disposition, &header[21]);
 352                                 extract_key(content_disposition_name, disposition, "name");
 353                                 extract_key(filename, disposition, "filename");
 354                         }
 355                         if (!strncasecmp(header, "Content-length: ", 16)) {
 356                                 content_length = (size_t) atol(&header[16]);
 357                         }
 358                         if (!strncasecmp(header,
 359                                       "Content-transfer-encoding: ", 27))
 360                                 strcpy(encoding, &header[27]);
 361                         if (strlen(boundary) == 0)
 362                                 extract_key(boundary, header, "boundary");
 363                         strcpy(header, "");
 364                 }
 365                 if ((strlen(header) + strlen(buf) + 2) < SIZ)
 366                         strcat(header, buf);
 367         } while ((strlen(buf) > 0) && (*ptr != 0));
 368
 369         if (strchr(disposition, ';'))
 370                 *(strchr(disposition, ';')) = '\0';
 371         striplt(disposition);
 372         if (strchr(content_type, ';'))
 373                 *(strchr(content_type, ';')) = '\0';
 374         striplt(content_type);
 375
 376         if (strlen(boundary) > 0) {
 377                 is_multipart = 1;
 378         } else {
 379                 is_multipart = 0;
 380         }
 381
 382         /* If this is a multipart message, then recursively process it */
 383         part_start = NULL;
 384         if (is_multipart) {
 385
 386                 /* Tell the client about this message's multipartedness */
 387                 if (PreMultiPartCallBack != NULL) {
 388                         PreMultiPartCallBack("", "", partnum, "",
 389                                 NULL, content_type, charset,
 390                                 0, encoding, userdata);
 391                 }
 392
 393                 /* Figure out where the boundaries are */
 394                 snprintf(startary, SIZ, "--%s", boundary);
 395                 snprintf(endary, SIZ, "--%s--", boundary);
 396                 startary_len = strlen(startary);
 397
 398                 part_start = NULL;
 399                 do {
 400                         next_boundary = NULL;
 401                         for (srch=ptr; srch<content_end; ++srch) {
 402                                 if (!memcmp(srch, startary, startary_len)) {
 403                                         next_boundary = srch;
 404                                         srch = content_end;
 405                                 }
 406                         }
 407
 408                         if ( (part_start != NULL) && (next_boundary != NULL) ) {
 409                                 part_end = next_boundary;
 410                                 --part_end;
 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                                         ++part_start;
 442                                         ptr = part_start;
 443                                 }
 444                         }
 445                         else {
 446                                 /* Invalid end of multipart.  Bail out! */
 447                                 ptr = content_end;
 448                         }
 449                 } while ( (ptr < content_end) && (next_boundary != NULL) );
 450
 451                 if (PostMultiPartCallBack != NULL) {
 452                         PostMultiPartCallBack("", "", partnum, "", NULL,
 453                                 content_type, charset, 0, encoding, userdata);
 454                 }
 455                 goto end_parser;
 456         }
 457
 458         /* If it's not a multipart message, then do something with it */
 459         if (!is_multipart) {
 460                 part_start = ptr;
 461                 length = 0;
 462                 while (ptr < content_end) {
 463                         ++ptr;
 464                         ++length;
 465                 }
 466                 part_end = content_end;
 467                 /* fix an off-by-one error */
 468                 --part_end;
 469                 --length;
 470
 471                 /* Truncate if the header told us to */
 472                 if ( (content_length > 0) && (length > content_length) ) {
 473                         length = content_length;
 474                 }
 475
 476                 /* Sometimes the "name" field is tacked on to Content-type,
 477                  * and sometimes it's tacked on to Content-disposition.  Use
 478                  * whichever one we have.
 479                  */
 480                 if (strlen(content_disposition_name) > strlen(content_type_name)) {
 481                         name = content_disposition_name;
 482                 }
 483                 else {
 484                         name = content_type_name;
 485                 }
 486
 487                 /*
 488                 lprintf(9, "mime_decode part=%s, len=%d, type=%s, charset=%s, encoding=%s\n",
 489                         partnum, length, content_type, charset, encoding);
 490                 */
 491
 492                 /* Ok, we've got a non-multipart part here, so do something with it.
 493                  */
 494                 mime_decode(partnum,
 495                         part_start, length,
 496                         content_type, charset, encoding, disposition,
 497                         name, filename,
 498                         CallBack, NULL, NULL,
 499                         userdata, dont_decode
 500                 );
 501
 502                 /*
 503                  * Now if it's an encapsulated message/rfc822 then we have to recurse into it
 504                  */
 505                 if (!strcasecmp(content_type, "message/rfc822")) {
 506
 507                         if (PreMultiPartCallBack != NULL) {
 508                                 PreMultiPartCallBack("", "", partnum, "",
 509                                         NULL, content_type, charset,
 510                                         0, encoding, userdata);
 511                         }
 512                         if (CallBack != NULL) {
 513                                 if (strlen(partnum) > 0) {
 514                                         snprintf(nested_partnum,
 515                                                  sizeof nested_partnum,
 516                                                  "%s.%d", partnum,
 517                                                  ++part_seq);
 518                                 }
 519                                 else {
 520                                         snprintf(nested_partnum,
 521                                                  sizeof nested_partnum,
 522                                                  "%d", ++part_seq);
 523                                 }
 524                                 the_mime_parser(nested_partnum,
 525                                         part_start, part_end,
 526                                         CallBack,
 527                                         PreMultiPartCallBack,
 528                                         PostMultiPartCallBack,
 529                                         userdata,
 530                                         dont_decode
 531                                 );
 532                         }
 533                         if (PostMultiPartCallBack != NULL) {
 534                                 PostMultiPartCallBack("", "", partnum, "", NULL,
 535                                         content_type, charset, 0, encoding, userdata);
 536                         }
 537
 538
 539                 }
 540
 541         }
 542
 543 end_parser:     /* free the buffers!  end the oppression!! */
 544         free(boundary);
 545         free(startary);
 546         free(endary);
 547         free(header);
 548         free(content_type);
 549         free(charset);
 550         free(encoding);
 551         free(content_type_name);
 552         free(content_disposition_name);
 553         free(filename);
 554         free(disposition);
 555 }
 556
 557
 558
 559 /*
 560  * Entry point for the MIME parser.
 561  * (This function expects to be fed HEADERS + CONTENT)
 562  * Note: NULL can be supplied as content_end; in this case, the message is
 563  * considered to have ended when the parser encounters a 0x00 byte.
 564  */
 565 void mime_parser(char *content_start,
 566                 char *content_end,
 567
 568                  void (*CallBack)
 569                   (char *cbname,
 570                    char *cbfilename,
 571                    char *cbpartnum,
 572                    char *cbdisp,
 573                    void *cbcontent,
 574                    char *cbtype,
 575                    char *cbcharset,
 576                    size_t cblength,
 577                    char *cbencoding,
 578                    void *cbuserdata),
 579
 580                  void (*PreMultiPartCallBack)
 581                   (char *cbname,
 582                    char *cbfilename,
 583                    char *cbpartnum,
 584                    char *cbdisp,
 585                    void *cbcontent,
 586                    char *cbtype,
 587                    char *cbcharset,
 588                    size_t cblength,
 589                    char *cbencoding,
 590                    void *cbuserdata),
 591
 592                  void (*PostMultiPartCallBack)
 593                   (char *cbname,
 594                    char *cbfilename,
 595                    char *cbpartnum,
 596                    char *cbdisp,
 597                    void *cbcontent,
 598                    char *cbtype,
 599                    char *cbcharset,
 600                    size_t cblength,
 601                    char *cbencoding,
 602                    void *cbuserdata),
 603
 604                   void *userdata,
 605                   int dont_decode
 606 )
 607 {
 608
 609         the_mime_parser("", content_start, content_end,
 610                         CallBack,
 611                         PreMultiPartCallBack,
 612                         PostMultiPartCallBack,
 613                         userdata, dont_decode);
 614 }