webcit/mime_parser.c

   1 /*
   2  * $Id$
   3  *
   4  * This is the MIME parser for Citadel.  Sometimes it actually works.
   5  *
   6  * Copyright (c) 1998-2002 by Art Cancro
   7  * This code is distributed under the terms of the GNU General Public License.
   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 #include "webcit.h"
  21
  22
  23 void extract_key(char *target, char *source, char *key)
  24 {
  25         int a, b;
  26
  27         strcpy(target, source);
  28         for (a = 0; a < strlen(target); ++a) {
  29                 if ((!strncasecmp(&target[a], key, strlen(key)))
  30                     && (target[a + strlen(key)] == '=')) {
  31                         strcpy(target, &target[a + strlen(key) + 1]);
  32                         if (target[0] == 34)
  33                                 strcpy(target, &target[1]);
  34                         for (b = 0; b < strlen(target); ++b)
  35                                 if (target[b] == 34)
  36                                         target[b] = 0;
  37                         return;
  38                 }
  39         }
  40         strcpy(target, "");
  41 }
  42
  43
  44 /*
  45  * For non-multipart messages, we need to generate a quickie partnum of "1"
  46  * to return to callback functions.  Some callbacks demand it.
  47  */
  48 char *fixed_partnum(char *supplied_partnum) {
  49         if (supplied_partnum == NULL) return "1";
  50         if (strlen(supplied_partnum)==0) return "1";
  51         return supplied_partnum;
  52 }
  53
  54
  55
  56 /*
  57  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
  58  */
  59 int decode_quoted_printable(char *decoded, char *encoded, int sourcelen) {
  60         char buf[SIZ];
  61         int buf_length = 0;
  62         int soft_line_break = 0;
  63         int ch;
  64         int decoded_length = 0;
  65         int i;
  66
  67         decoded[0] = 0;
  68         decoded_length = 0;
  69         buf[0] = 0;
  70         buf_length = 0;
  71
  72         for (i = 0; i < sourcelen; ++i) {
  73
  74                 buf[buf_length++] = encoded[i];
  75
  76                 if ( (encoded[i] == '\n')
  77                    || (encoded[i] == 0)
  78                    || (i == (sourcelen-1)) ) {
  79                         buf[buf_length++] = 0;
  80
  81                         /*** begin -- process one line ***/
  82
  83                         if (buf[strlen(buf)-1] == '\n') {
  84                                 buf[strlen(buf)-1] = 0;
  85                         }
  86                         if (buf[strlen(buf)-1] == '\r') {
  87                                 buf[strlen(buf)-1] = 0;
  88                         }
  89                         while (isspace(buf[strlen(buf)-1])) {
  90                                 buf[strlen(buf)-1] = 0;
  91                         }
  92                         soft_line_break = 0;
  93
  94                         while (strlen(buf) > 0) {
  95                                 if (!strcmp(buf, "=")) {
  96                                         soft_line_break = 1;
  97                                         strcpy(buf, "");
  98                                 } else if ((strlen(buf)>=3) && (buf[0]=='=')) {
  99                                         sscanf(&buf[1], "%02x", &ch);
 100                                         decoded[decoded_length++] = ch;
 101                                         strcpy(buf, &buf[3]);
 102                                 } else {
 103                                         decoded[decoded_length++] = buf[0];
 104                                         strcpy(buf, &buf[1]);
 105                                 }
 106                         }
 107                         if (soft_line_break == 0) {
 108                                 decoded[decoded_length++] = '\r';
 109                                 decoded[decoded_length++] = '\n';
 110                         }
 111                         buf_length = 0;
 112                         /*** end -- process one line ***/
 113                 }
 114         }
 115
 116         decoded[decoded_length++] = 0;
 117         return(decoded_length);
 118 }
 119
 120 /*
 121  * Given a message or message-part body and a length, handle any necessary
 122  * decoding and pass the request up the stack.
 123  */
 124 void mime_decode(char *partnum,
 125                  char *part_start, size_t length,
 126                  char *content_type, char *encoding,
 127                  char *disposition,
 128                  char *name, char *filename,
 129                  void (*CallBack)
 130                   (char *cbname,
 131                    char *cbfilename,
 132                    char *cbpartnum,
 133                    char *cbdisp,
 134                    void *cbcontent,
 135                    char *cbtype,
 136                    size_t cblength,
 137                    char *cbencoding,
 138                    void *cbuserdata),
 139                  void (*PreMultiPartCallBack)
 140                   (char *cbname,
 141                    char *cbfilename,
 142                    char *cbpartnum,
 143                    char *cbdisp,
 144                    void *cbcontent,
 145                    char *cbtype,
 146                    size_t cblength,
 147                    char *cbencoding,
 148                    void *cbuserdata),
 149                  void (*PostMultiPartCallBack)
 150                   (char *cbname,
 151                    char *cbfilename,
 152                    char *cbpartnum,
 153                    char *cbdisp,
 154                    void *cbcontent,
 155                    char *cbtype,
 156                    size_t cblength,
 157                    char *cbencoding,
 158                    void *cbuserdata),
 159                   void *userdata,
 160                   int dont_decode
 161 )
 162 {
 163
 164         char *decoded;
 165         size_t bytes_decoded = 0;
 166
 167         fprintf(stderr, "mime_decode() called\n");
 168
 169         /* Some encodings aren't really encodings */
 170         if (!strcasecmp(encoding, "7bit"))
 171                 strcpy(encoding, "");
 172         if (!strcasecmp(encoding, "8bit"))
 173                 strcpy(encoding, "");
 174         if (!strcasecmp(encoding, "binary"))
 175                 strcpy(encoding, "");
 176
 177         /* If this part is not encoded, send as-is */
 178         if ( (strlen(encoding) == 0) || (dont_decode)) {
 179                 if (CallBack != NULL) {
 180                         CallBack(name, filename, fixed_partnum(partnum),
 181                                 disposition, part_start,
 182                                 content_type, length, encoding, userdata);
 183                         }
 184                 return;
 185         }
 186
 187         if ((strcasecmp(encoding, "base64"))
 188             && (strcasecmp(encoding, "quoted-printable"))) {
 189                 fprintf(stderr, "ERROR: unknown MIME encoding '%s'\n", encoding);
 190                 return;
 191         }
 192         /*
 193          * Allocate a buffer for the decoded data.  The output buffer is the
 194          * same size as the input buffer; this assumes that the decoded data
 195          * will never be larger than the encoded data.  This is a safe
 196          * assumption with base64, uuencode, and quoted-printable.
 197          */
 198         fprintf(stderr, "About to allocate %d bytes for decoded part\n",
 199                 length+2048);
 200         decoded = malloc(length+2048);
 201         if (decoded == NULL) {
 202                 fprintf(stderr, "ERROR: cannot allocate memory.\n");
 203                 return;
 204         }
 205         fprintf(stderr, "Got it!\n");
 206
 207         fprintf(stderr, "Decoding %s\n", encoding);
 208         if (!strcasecmp(encoding, "base64")) {
 209                 bytes_decoded = decode_base64(decoded, part_start);
 210         }
 211         else if (!strcasecmp(encoding, "quoted-printable")) {
 212                 bytes_decoded = decode_quoted_printable(decoded,
 213                                                         part_start, length);
 214         }
 215         fprintf(stderr, "Bytes decoded: %d\n", bytes_decoded);
 216
 217         if (bytes_decoded > 0) if (CallBack != NULL) {
 218                 CallBack(name, filename, fixed_partnum(partnum),
 219                         disposition, decoded,
 220                         content_type, bytes_decoded, "binary", userdata);
 221         }
 222
 223         free(decoded);
 224 }
 225
 226 /*
 227  * Break out the components of a multipart message
 228  * (This function expects to be fed HEADERS + CONTENT)
 229  * Note: NULL can be supplied as content_end; in this case, the message is
 230  * considered to have ended when the parser encounters a 0x00 byte.
 231  */
 232 void the_mime_parser(char *partnum,
 233                      char *content_start, char *content_end,
 234                      void (*CallBack)
 235                       (char *cbname,
 236                        char *cbfilename,
 237                        char *cbpartnum,
 238                        char *cbdisp,
 239                        void *cbcontent,
 240                        char *cbtype,
 241                        size_t cblength,
 242                        char *cbencoding,
 243                        void *cbuserdata),
 244                      void (*PreMultiPartCallBack)
 245                       (char *cbname,
 246                        char *cbfilename,
 247                        char *cbpartnum,
 248                        char *cbdisp,
 249                        void *cbcontent,
 250                        char *cbtype,
 251                        size_t cblength,
 252                        char *cbencoding,
 253                        void *cbuserdata),
 254                      void (*PostMultiPartCallBack)
 255                       (char *cbname,
 256                        char *cbfilename,
 257                        char *cbpartnum,
 258                        char *cbdisp,
 259                        void *cbcontent,
 260                        char *cbtype,
 261                        size_t cblength,
 262                        char *cbencoding,
 263                        void *cbuserdata),
 264                       void *userdata,
 265                       int dont_decode
 266 )
 267 {
 268
 269         char *ptr;
 270         char *part_start, *part_end = NULL;
 271         char buf[SIZ];
 272         char *header;
 273         char *boundary;
 274         char *startary;
 275         char *endary;
 276         char *content_type;
 277         size_t content_length;
 278         char *encoding;
 279         char *disposition;
 280         char *name;
 281         char *filename;
 282         int is_multipart;
 283         int part_seq = 0;
 284         int i;
 285         size_t length;
 286         char nested_partnum[SIZ];
 287
 288         fprintf(stderr, "the_mime_parser() called\n");
 289         ptr = content_start;
 290         content_length = 0;
 291
 292         boundary = malloc(SIZ);
 293         memset(boundary, 0, SIZ);
 294
 295         startary = malloc(SIZ);
 296         memset(startary, 0, SIZ);
 297
 298         endary = malloc(SIZ);
 299         memset(endary, 0, SIZ);
 300
 301         header = malloc(SIZ);
 302         memset(header, 0, SIZ);
 303
 304         content_type = malloc(SIZ);
 305         memset(content_type, 0, SIZ);
 306
 307         encoding = malloc(SIZ);
 308         memset(encoding, 0, SIZ);
 309
 310         name = malloc(SIZ);
 311         memset(name, 0, SIZ);
 312
 313         filename = malloc(SIZ);
 314         memset(filename, 0, SIZ);
 315
 316         disposition = malloc(SIZ);
 317         memset(disposition, 0, SIZ);
 318
 319         /* If the caller didn't supply an endpointer, generate one by measure */
 320         if (content_end == NULL) {
 321                 content_end = &content_start[strlen(content_start)];
 322         }
 323
 324         /* Learn interesting things from the headers */
 325         strcpy(header, "");
 326         do {
 327                 ptr = memreadline(ptr, buf, SIZ);
 328                 if (ptr >= content_end) {
 329                         goto end_parser;
 330                 }
 331
 332                 for (i = 0; i < strlen(buf); ++i)
 333                         if (isspace(buf[i]))
 334                                 buf[i] = ' ';
 335                 if (!isspace(buf[0])) {
 336                         if (!strncasecmp(header, "Content-type: ", 14)) {
 337                                 strcpy(content_type, &header[14]);
 338                                 extract_key(name, content_type, "name");
 339                                 fprintf(stderr, "Extracted content-type <%s>\n",
 340                                         content_type);
 341                         }
 342                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 343                                 strcpy(disposition, &header[21]);
 344                                 extract_key(filename, disposition, "filename");
 345                         }
 346                         if (!strncasecmp(header, "Content-length: ", 16)) {
 347                                 content_length = (size_t) atol(&header[16]);
 348                         }
 349                         if (!strncasecmp(header,
 350                                       "Content-transfer-encoding: ", 27))
 351                                 strcpy(encoding, &header[27]);
 352                         if (strlen(boundary) == 0)
 353                                 extract_key(boundary, header, "boundary");
 354                         strcpy(header, "");
 355                 }
 356                 if ((strlen(header) + strlen(buf) + 2) < SIZ)
 357                         strcat(header, buf);
 358         } while ((strlen(buf) > 0) && (*ptr != 0));
 359
 360         for (i = 0; i < strlen(disposition); ++i)
 361                 if (disposition[i] == ';')
 362                         disposition[i] = 0;
 363         while (isspace(disposition[0]))
 364                 strcpy(disposition, &disposition[1]);
 365         for (i = 0; i < strlen(content_type); ++i)
 366                 if (content_type[i] == ';')
 367                         content_type[i] = 0;
 368         while (isspace(content_type[0]))
 369                 strcpy(content_type, &content_type[1]);
 370
 371         if (strlen(boundary) > 0) {
 372                 is_multipart = 1;
 373         } else {
 374                 is_multipart = 0;
 375         }
 376
 377         fprintf(stderr, "is_multipart=%d, boundary=<%s>\n",
 378                 is_multipart, boundary);
 379
 380         /* If this is a multipart message, then recursively process it */
 381         part_start = NULL;
 382         if (is_multipart) {
 383
 384                 /* Tell the client about this message's multipartedness */
 385                 if (PreMultiPartCallBack != NULL) {
 386                         PreMultiPartCallBack("", "", partnum, "",
 387                                 NULL, content_type,
 388                                 0, encoding, userdata);
 389                 }
 390
 391                 /* Figure out where the boundaries are */
 392                 sprintf(startary, "--%s", boundary);
 393                 sprintf(endary, "--%s--", boundary);
 394                 do {
 395                         if ( (!strncasecmp(ptr, startary, strlen(startary)))
 396                            || (!strncasecmp(ptr, endary, strlen(endary))) ) {
 397                                 fprintf(stderr, "hit boundary!\n");
 398                                 if (part_start != NULL) {
 399                                         if (strlen(partnum) > 0) {
 400                                                 sprintf(nested_partnum, "%s.%d",
 401                                                         partnum, ++part_seq);
 402                                         }
 403                                         else {
 404                                                 sprintf(nested_partnum, "%d",
 405                                                         ++part_seq);
 406                                         }
 407                                         the_mime_parser(nested_partnum,
 408                                                     part_start, part_end,
 409                                                         CallBack,
 410                                                         PreMultiPartCallBack,
 411                                                         PostMultiPartCallBack,
 412                                                         userdata,
 413                                                         dont_decode);
 414                                 }
 415                                 ptr = memreadline(ptr, buf, SIZ);
 416                                 part_start = ptr;
 417                         }
 418                         else {
 419                                 part_end = ptr;
 420                                 ++ptr;
 421                         }
 422                 } while ( (strcasecmp(ptr, endary)) && (ptr <= content_end) );
 423                 if (PostMultiPartCallBack != NULL) {
 424                         PostMultiPartCallBack("", "", partnum, "", NULL,
 425                                 content_type, 0, encoding, userdata);
 426                 }
 427                 goto end_parser;
 428         }
 429
 430         /* If it's not a multipart message, then do something with it */
 431         if (!is_multipart) {
 432                 fprintf(stderr, "doing non-multipart thing\n");
 433                 part_start = ptr;
 434                 length = 0;
 435                 while (ptr < content_end) {
 436                         ++ptr;
 437                         ++length;
 438                 }
 439                 part_end = content_end;
 440
 441                 /* Truncate if the header told us to */
 442                 if ( (content_length > 0) && (length > content_length) ) {
 443                         length = content_length;
 444                         fprintf(stderr, "truncated to %ld\n", (long)content_length);
 445                 }
 446
 447                 mime_decode(partnum,
 448                             part_start, length,
 449                             content_type, encoding, disposition,
 450                             name, filename,
 451                             CallBack, NULL, NULL,
 452                             userdata, dont_decode);
 453         }
 454
 455 end_parser:     /* free the buffers!  end the oppression!! */
 456         free(boundary);
 457         free(startary);
 458         free(endary);
 459         free(header);
 460         free(content_type);
 461         free(encoding);
 462         free(name);
 463         free(filename);
 464         free(disposition);
 465 }
 466
 467
 468
 469 /*
 470  * Entry point for the MIME parser.
 471  * (This function expects to be fed HEADERS + CONTENT)
 472  * Note: NULL can be supplied as content_end; in this case, the message is
 473  * considered to have ended when the parser encounters a 0x00 byte.
 474  */
 475 void mime_parser(char *content_start,
 476                 char *content_end,
 477
 478                  void (*CallBack)
 479                   (char *cbname,
 480                    char *cbfilename,
 481                    char *cbpartnum,
 482                    char *cbdisp,
 483                    void *cbcontent,
 484                    char *cbtype,
 485                    size_t cblength,
 486                    char *cbencoding,
 487                    void *cbuserdata),
 488
 489                  void (*PreMultiPartCallBack)
 490                   (char *cbname,
 491                    char *cbfilename,
 492                    char *cbpartnum,
 493                    char *cbdisp,
 494                    void *cbcontent,
 495                    char *cbtype,
 496                    size_t cblength,
 497                    char *cbencoding,
 498                    void *cbuserdata),
 499
 500                  void (*PostMultiPartCallBack)
 501                   (char *cbname,
 502                    char *cbfilename,
 503                    char *cbpartnum,
 504                    char *cbdisp,
 505                    void *cbcontent,
 506                    char *cbtype,
 507                    size_t cblength,
 508                    char *cbencoding,
 509                    void *cbuserdata),
 510
 511                   void *userdata,
 512                   int dont_decode
 513 )
 514 {
 515
 516         fprintf(stderr, "mime_parser() called\n");
 517         the_mime_parser("", content_start, content_end,
 518                         CallBack,
 519                         PreMultiPartCallBack,
 520                         PostMultiPartCallBack,
 521                         userdata, dont_decode);
 522 }