webcit/mime_parser.c

   1 /*
   2  * $Id$
   3  *
   4  * This is the MIME parser for Citadel.  Sometimes it actually works.
   5  *
   6  * Copyright (c) 1998-2001 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 <pthread.h>
  21 #include "webcit.h"
  22 #include "mime_parser.h"
  23
  24
  25 void extract_key(char *target, char *source, char *key)
  26 {
  27         int a, b;
  28
  29         strcpy(target, source);
  30         for (a = 0; a < strlen(target); ++a) {
  31                 if ((!strncasecmp(&target[a], key, strlen(key)))
  32                     && (target[a + strlen(key)] == '=')) {
  33                         strcpy(target, &target[a + strlen(key) + 1]);
  34                         if (target[0] == 34)
  35                                 strcpy(target, &target[1]);
  36                         for (b = 0; b < strlen(target); ++b)
  37                                 if (target[b] == 34)
  38                                         target[b] = 0;
  39                         return;
  40                 }
  41         }
  42         strcpy(target, "");
  43 }
  44
  45
  46 /*
  47  * For non-multipart messages, we need to generate a quickie partnum of "1"
  48  * to return to callback functions.  Some callbacks demand it.
  49  */
  50 char *fixed_partnum(char *supplied_partnum) {
  51         if (supplied_partnum == NULL) return "1";
  52         if (strlen(supplied_partnum)==0) return "1";
  53         return supplied_partnum;
  54 }
  55
  56
  57
  58 /*
  59  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
  60  */
  61 int decode_quoted_printable(char *decoded, char *encoded, int sourcelen) {
  62         char buf[SIZ];
  63         int buf_length = 0;
  64         int soft_line_break = 0;
  65         int ch;
  66         int decoded_length = 0;
  67         int i;
  68
  69         decoded[0] = 0;
  70         decoded_length = 0;
  71         buf[0] = 0;
  72         buf_length = 0;
  73
  74         for (i = 0; i < sourcelen; ++i) {
  75
  76                 buf[buf_length++] = encoded[i];
  77
  78                 if ( (encoded[i] == '\n')
  79                    || (encoded[i] == 0)
  80                    || (i == (sourcelen-1)) ) {
  81                         buf[buf_length++] = 0;
  82
  83                         /*** begin -- process one line ***/
  84
  85                         if (buf[strlen(buf)-1] == '\n') {
  86                                 buf[strlen(buf)-1] = 0;
  87                         }
  88                         if (buf[strlen(buf)-1] == '\r') {
  89                                 buf[strlen(buf)-1] = 0;
  90                         }
  91                         while (isspace(buf[strlen(buf)-1])) {
  92                                 buf[strlen(buf)-1] = 0;
  93                         }
  94                         soft_line_break = 0;
  95
  96                         while (strlen(buf) > 0) {
  97                                 if (!strcmp(buf, "=")) {
  98                                         soft_line_break = 1;
  99                                         strcpy(buf, "");
 100                                 } else if ((strlen(buf)>=3) && (buf[0]=='=')) {
 101                                         sscanf(&buf[1], "%02x", &ch);
 102                                         decoded[decoded_length++] = ch;
 103                                         strcpy(buf, &buf[3]);
 104                                 } else {
 105                                         decoded[decoded_length++] = buf[0];
 106                                         strcpy(buf, &buf[1]);
 107                                 }
 108                         }
 109                         if (soft_line_break == 0) {
 110                                 decoded[decoded_length++] = '\r';
 111                                 decoded[decoded_length++] = '\n';
 112                         }
 113                         buf_length = 0;
 114                         /*** end -- process one line ***/
 115                 }
 116         }
 117
 118         decoded[decoded_length++] = 0;
 119         return(decoded_length);
 120 }
 121
 122 /*
 123  * Given a message or message-part body and a length, handle any necessary
 124  * decoding and pass the request up the stack.
 125  */
 126 void mime_decode(char *partnum,
 127                  char *part_start, size_t length,
 128                  char *content_type, char *encoding,
 129                  char *disposition,
 130                  char *name, char *filename,
 131                  void (*CallBack)
 132                   (char *cbname,
 133                    char *cbfilename,
 134                    char *cbpartnum,
 135                    char *cbdisp,
 136                    void *cbcontent,
 137                    char *cbtype,
 138                    size_t cblength,
 139                    char *cbencoding,
 140                    void *cbuserdata),
 141                  void (*PreMultiPartCallBack)
 142                   (char *cbname,
 143                    char *cbfilename,
 144                    char *cbpartnum,
 145                    char *cbdisp,
 146                    void *cbcontent,
 147                    char *cbtype,
 148                    size_t cblength,
 149                    char *cbencoding,
 150                    void *cbuserdata),
 151                  void (*PostMultiPartCallBack)
 152                   (char *cbname,
 153                    char *cbfilename,
 154                    char *cbpartnum,
 155                    char *cbdisp,
 156                    void *cbcontent,
 157                    char *cbtype,
 158                    size_t cblength,
 159                    char *cbencoding,
 160                    void *cbuserdata),
 161                   void *userdata,
 162                   int dont_decode
 163 )
 164 {
 165
 166         char *decoded;
 167         size_t bytes_decoded = 0;
 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         if ((strcasecmp(encoding, "base64"))
 187             && (strcasecmp(encoding, "quoted-printable"))) {
 188                 fprintf(stderr, "ERROR: unknown MIME encoding '%s'\n", encoding);
 189                 return;
 190         }
 191         /*
 192          * Allocate a buffer for the decoded data.  The output buffer is the
 193          * same size as the input buffer; this assumes that the decoded data
 194          * will never be larger than the encoded data.  This is a safe
 195          * assumption with base64, uuencode, and quoted-printable.  Just to
 196          * be safe, we still pad the buffer a bit.
 197          */
 198         decoded = malloc(length + 1024);
 199         if (decoded == NULL) {
 200                 fprintf(stderr, "ERROR: cannot allocate memory.\n");
 201                 return;
 202         }
 203
 204         if (!strcasecmp(encoding, "base64")) {
 205                 bytes_decoded = decode_base64(decoded, part_start);
 206         }
 207         else if (!strcasecmp(encoding, "quoted-printable")) {
 208                 bytes_decoded = decode_quoted_printable(decoded,
 209                                                         part_start, length);
 210         }
 211
 212         if (bytes_decoded > 0) if (CallBack != NULL) {
 213                 CallBack(name, filename, fixed_partnum(partnum),
 214                         disposition, decoded,
 215                         content_type, bytes_decoded, "binary", userdata);
 216         }
 217
 218         free(decoded);
 219 }
 220
 221
 222
 223
 224 /*
 225  * Break out the components of a multipart message
 226  * (This function expects to be fed HEADERS + CONTENT)
 227  * Note: NULL can be supplied as content_end; in this case, the message is
 228  * considered to have ended when the parser encounters a 0x00 byte.
 229  */
 230 void the_mime_parser(char *partnum,
 231                      char *content_start, char *content_end,
 232                      void (*CallBack)
 233                       (char *cbname,
 234                        char *cbfilename,
 235                        char *cbpartnum,
 236                        char *cbdisp,
 237                        void *cbcontent,
 238                        char *cbtype,
 239                        size_t cblength,
 240                        char *cbencoding,
 241                        void *cbuserdata),
 242                      void (*PreMultiPartCallBack)
 243                       (char *cbname,
 244                        char *cbfilename,
 245                        char *cbpartnum,
 246                        char *cbdisp,
 247                        void *cbcontent,
 248                        char *cbtype,
 249                        size_t cblength,
 250                        char *cbencoding,
 251                        void *cbuserdata),
 252                      void (*PostMultiPartCallBack)
 253                       (char *cbname,
 254                        char *cbfilename,
 255                        char *cbpartnum,
 256                        char *cbdisp,
 257                        void *cbcontent,
 258                        char *cbtype,
 259                        size_t cblength,
 260                        char *cbencoding,
 261                        void *cbuserdata),
 262                       void *userdata,
 263                       int dont_decode
 264 )
 265 {
 266
 267         char *ptr;
 268         char *part_start, *part_end = NULL;
 269         char buf[SIZ];
 270         char *header;
 271         char *boundary;
 272         char *startary;
 273         char *endary;
 274         char *content_type;
 275         size_t content_length;
 276         char *encoding;
 277         char *disposition;
 278         char *name;
 279         char *filename;
 280         int is_multipart;
 281         int part_seq = 0;
 282         int i;
 283         size_t length;
 284         char nested_partnum[SIZ];
 285
 286         fprintf(stderr, "the_mime_parser() called\n");
 287         ptr = content_start;
 288         content_length = 0;
 289
 290         boundary = malloc(SIZ);
 291         memset(boundary, 0, SIZ);
 292
 293         startary = malloc(SIZ);
 294         memset(startary, 0, SIZ);
 295
 296         endary = malloc(SIZ);
 297         memset(endary, 0, SIZ);
 298
 299         header = malloc(SIZ);
 300         memset(header, 0, SIZ);
 301
 302         content_type = malloc(SIZ);
 303         memset(content_type, 0, SIZ);
 304
 305         encoding = malloc(SIZ);
 306         memset(encoding, 0, SIZ);
 307
 308         name = malloc(SIZ);
 309         memset(name, 0, SIZ);
 310
 311         filename = malloc(SIZ);
 312         memset(filename, 0, SIZ);
 313
 314         disposition = malloc(SIZ);
 315         memset(disposition, 0, SIZ);
 316
 317         /* If the caller didn't supply an endpointer, generate one by measure */
 318         if (content_end == NULL) {
 319                 content_end = &content_start[strlen(content_start)];
 320         }
 321
 322         /* Learn interesting things from the headers */
 323         strcpy(header, "");
 324         do {
 325                 ptr = memreadline(ptr, buf, sizeof buf);
 326                 if (ptr >= content_end) {
 327                         goto end_parser;
 328                 }
 329
 330                 for (i = 0; i < strlen(buf); ++i)
 331                         if (isspace(buf[i]))
 332                                 buf[i] = ' ';
 333                 if (!isspace(buf[0])) {
 334                         if (!strncasecmp(header, "Content-type: ", 14)) {
 335                                 strcpy(content_type, &header[14]);
 336                                 extract_key(name, content_type, "name");
 337                                 fprintf(stderr, "Extracted content-type <%s>\n",
 338                                         content_type);
 339                         }
 340                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 341                                 strcpy(disposition, &header[21]);
 342                                 extract_key(filename, disposition, "filename");
 343                         }
 344                         if (!strncasecmp(header, "Content-length: ", 16)) {
 345                                 content_length = (size_t) atol(&header[16]);
 346                         }
 347                         if (!strncasecmp(header,
 348                                       "Content-transfer-encoding: ", 27))
 349                                 strcpy(encoding, &header[27]);
 350                         if (strlen(boundary) == 0)
 351                                 extract_key(boundary, header, "boundary");
 352                         strcpy(header, "");
 353                 }
 354                 if ((strlen(header) + strlen(buf) + 2) < sizeof(header))
 355                         strcat(header, buf);
 356         } while ((strlen(buf) > 0) && (*ptr != 0));
 357
 358         for (i = 0; i < strlen(disposition); ++i)
 359                 if (disposition[i] == ';')
 360                         disposition[i] = 0;
 361         while (isspace(disposition[0]))
 362                 strcpy(disposition, &disposition[1]);
 363         for (i = 0; i < strlen(content_type); ++i)
 364                 if (content_type[i] == ';')
 365                         content_type[i] = 0;
 366         while (isspace(content_type[0]))
 367                 strcpy(content_type, &content_type[1]);
 368
 369         if (strlen(boundary) > 0) {
 370                 is_multipart = 1;
 371         } else {
 372                 is_multipart = 0;
 373         }
 374
 375         fprintf(stderr, "is_multipart=%d, boundary=<%s>\n",
 376                 is_multipart, boundary);
 377
 378         /* If this is a multipart message, then recursively process it */
 379         part_start = NULL;
 380         if (is_multipart) {
 381
 382                 /* Tell the client about this message's multipartedness */
 383                 if (PreMultiPartCallBack != NULL) {
 384                         PreMultiPartCallBack("", "", partnum, "",
 385                                 NULL, content_type,
 386                                 0, encoding, userdata);
 387                 }
 388
 389                 /* Figure out where the boundaries are */
 390                 sprintf(startary, "--%s", boundary);
 391                 sprintf(endary, "--%s--", boundary);
 392                 do {
 393                         if ( (!strncasecmp(ptr, startary, strlen(startary)))
 394                            || (!strncasecmp(ptr, endary, strlen(endary))) ) {
 395                                 fprintf(stderr, "hit boundary!\n");
 396                                 if (part_start != NULL) {
 397                                         if (strlen(partnum) > 0) {
 398                                                 sprintf(nested_partnum, "%s.%d",
 399                                                         partnum, ++part_seq);
 400                                         }
 401                                         else {
 402                                                 sprintf(nested_partnum, "%d",
 403                                                         ++part_seq);
 404                                         }
 405                                         the_mime_parser(nested_partnum,
 406                                                     part_start, part_end,
 407                                                         CallBack,
 408                                                         PreMultiPartCallBack,
 409                                                         PostMultiPartCallBack,
 410                                                         userdata,
 411                                                         dont_decode);
 412                                 }
 413                                 ptr = memreadline(ptr, buf, sizeof(buf));
 414                                 part_start = ptr;
 415                         }
 416                         else {
 417                                 part_end = ptr;
 418                                 ++ptr;
 419                         }
 420                 } while ( (strcasecmp(ptr, endary)) && (ptr <= content_end) );
 421                 if (PostMultiPartCallBack != NULL) {
 422                         PostMultiPartCallBack("", "", partnum, "", NULL,
 423                                 content_type, 0, encoding, userdata);
 424                 }
 425                 goto end_parser;
 426         }
 427
 428         /* If it's not a multipart message, then do something with it */
 429         if (!is_multipart) {
 430                 fprintf(stderr, "doing non-multipart thing\n");
 431                 part_start = ptr;
 432                 length = 0;
 433                 while (ptr < content_end) {
 434                         ++ptr;
 435                         ++length;
 436                 }
 437                 part_end = content_end;
 438
 439                 /* Truncate if the header told us to */
 440                 if ( (content_length > 0) && (length > content_length) ) {
 441                         length = content_length;
 442                         fprintf(stderr, "truncated to %ld\n", (long)content_length);
 443                 }
 444
 445                 mime_decode(partnum,
 446                             part_start, length,
 447                             content_type, encoding, disposition,
 448                             name, filename,
 449                             CallBack, NULL, NULL,
 450                             userdata, dont_decode);
 451         }
 452
 453 end_parser:     /* free the buffers!  end the oppression!! */
 454         free(boundary);
 455         free(startary);
 456         free(endary);
 457         free(header);
 458         free(content_type);
 459         free(encoding);
 460         free(name);
 461         free(filename);
 462         free(disposition);
 463 }
 464
 465
 466
 467
 468
 469
 470
 471 /*
 472  * Entry point for the MIME parser.
 473  * (This function expects to be fed HEADERS + CONTENT)
 474  * Note: NULL can be supplied as content_end; in this case, the message is
 475  * considered to have ended when the parser encounters a 0x00 byte.
 476  */
 477 void mime_parser(char *content_start,
 478                 char *content_end,
 479
 480                  void (*CallBack)
 481                   (char *cbname,
 482                    char *cbfilename,
 483                    char *cbpartnum,
 484                    char *cbdisp,
 485                    void *cbcontent,
 486                    char *cbtype,
 487                    size_t cblength,
 488                    char *cbencoding,
 489                    void *cbuserdata),
 490
 491                  void (*PreMultiPartCallBack)
 492                   (char *cbname,
 493                    char *cbfilename,
 494                    char *cbpartnum,
 495                    char *cbdisp,
 496                    void *cbcontent,
 497                    char *cbtype,
 498                    size_t cblength,
 499                    char *cbencoding,
 500                    void *cbuserdata),
 501
 502                  void (*PostMultiPartCallBack)
 503                   (char *cbname,
 504                    char *cbfilename,
 505                    char *cbpartnum,
 506                    char *cbdisp,
 507                    void *cbcontent,
 508                    char *cbtype,
 509                    size_t cblength,
 510                    char *cbencoding,
 511                    void *cbuserdata),
 512
 513                   void *userdata,
 514                   int dont_decode
 515 )
 516 {
 517
 518         fprintf(stderr, "mime_parser() called\n");
 519         the_mime_parser("", content_start, content_end,
 520                         CallBack,
 521                         PreMultiPartCallBack,
 522                         PostMultiPartCallBack,
 523                         userdata, dont_decode);
 524 }