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 #include "webserver.h"
  22
  23
  24 void extract_key(char *target, char *source, char *key)
  25 {
  26         int a, b;
  27
  28         strcpy(target, source);
  29         for (a = 0; a < strlen(target); ++a) {
  30                 if ((!strncasecmp(&target[a], key, strlen(key)))
  31                     && (target[a + strlen(key)] == '=')) {
  32                         strcpy(target, &target[a + strlen(key) + 1]);
  33                         if (target[0] == 34)
  34                                 strcpy(target, &target[1]);
  35                         for (b = 0; b < strlen(target); ++b)
  36                                 if (target[b] == 34)
  37                                         target[b] = 0;
  38                         return;
  39                 }
  40         }
  41         strcpy(target, "");
  42 }
  43
  44
  45 /*
  46  * For non-multipart messages, we need to generate a quickie partnum of "1"
  47  * to return to callback functions.  Some callbacks demand it.
  48  */
  49 char *fixed_partnum(char *supplied_partnum) {
  50         if (supplied_partnum == NULL) return "1";
  51         if (strlen(supplied_partnum)==0) return "1";
  52         return supplied_partnum;
  53 }
  54
  55
  56
  57 /*
  58  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
  59  */
  60 int decode_quoted_printable(char *decoded, char *encoded, int sourcelen) {
  61         char buf[SIZ];
  62         int buf_length = 0;
  63         int soft_line_break = 0;
  64         int ch;
  65         int decoded_length = 0;
  66         int i;
  67
  68         decoded[0] = 0;
  69         decoded_length = 0;
  70         buf[0] = 0;
  71         buf_length = 0;
  72
  73         for (i = 0; i < sourcelen; ++i) {
  74
  75                 buf[buf_length++] = encoded[i];
  76
  77                 if ( (encoded[i] == '\n')
  78                    || (encoded[i] == 0)
  79                    || (i == (sourcelen-1)) ) {
  80                         buf[buf_length++] = 0;
  81
  82                         /*** begin -- process one line ***/
  83
  84                         if (buf[strlen(buf)-1] == '\n') {
  85                                 buf[strlen(buf)-1] = 0;
  86                         }
  87                         if (buf[strlen(buf)-1] == '\r') {
  88                                 buf[strlen(buf)-1] = 0;
  89                         }
  90                         while (isspace(buf[strlen(buf)-1])) {
  91                                 buf[strlen(buf)-1] = 0;
  92                         }
  93                         soft_line_break = 0;
  94
  95                         while (strlen(buf) > 0) {
  96                                 if (!strcmp(buf, "=")) {
  97                                         soft_line_break = 1;
  98                                         strcpy(buf, "");
  99                                 } else if ((strlen(buf)>=3) && (buf[0]=='=')) {
 100                                         sscanf(&buf[1], "%02x", &ch);
 101                                         decoded[decoded_length++] = ch;
 102                                         strcpy(buf, &buf[3]);
 103                                 } else {
 104                                         decoded[decoded_length++] = buf[0];
 105                                         strcpy(buf, &buf[1]);
 106                                 }
 107                         }
 108                         if (soft_line_break == 0) {
 109                                 decoded[decoded_length++] = '\r';
 110                                 decoded[decoded_length++] = '\n';
 111                         }
 112                         buf_length = 0;
 113                         /*** end -- process one line ***/
 114                 }
 115         }
 116
 117         decoded[decoded_length++] = 0;
 118         return(decoded_length);
 119 }
 120
 121 /*
 122  * Given a message or message-part body and a length, handle any necessary
 123  * decoding and pass the request up the stack.
 124  */
 125 void mime_decode(char *partnum,
 126                  char *part_start, size_t length,
 127                  char *content_type, char *encoding,
 128                  char *disposition,
 129                  char *name, char *filename,
 130                  void (*CallBack)
 131                   (char *cbname,
 132                    char *cbfilename,
 133                    char *cbpartnum,
 134                    char *cbdisp,
 135                    void *cbcontent,
 136                    char *cbtype,
 137                    size_t cblength,
 138                    char *cbencoding,
 139                    void *cbuserdata),
 140                  void (*PreMultiPartCallBack)
 141                   (char *cbname,
 142                    char *cbfilename,
 143                    char *cbpartnum,
 144                    char *cbdisp,
 145                    void *cbcontent,
 146                    char *cbtype,
 147                    size_t cblength,
 148                    char *cbencoding,
 149                    void *cbuserdata),
 150                  void (*PostMultiPartCallBack)
 151                   (char *cbname,
 152                    char *cbfilename,
 153                    char *cbpartnum,
 154                    char *cbdisp,
 155                    void *cbcontent,
 156                    char *cbtype,
 157                    size_t cblength,
 158                    char *cbencoding,
 159                    void *cbuserdata),
 160                   void *userdata,
 161                   int dont_decode
 162 )
 163 {
 164
 165         char *decoded;
 166         size_t bytes_decoded = 0;
 167
 168         lprintf(9, "mime_decode() called\n");
 169
 170         /* Some encodings aren't really encodings */
 171         if (!strcasecmp(encoding, "7bit"))
 172                 strcpy(encoding, "");
 173         if (!strcasecmp(encoding, "8bit"))
 174                 strcpy(encoding, "");
 175         if (!strcasecmp(encoding, "binary"))
 176                 strcpy(encoding, "");
 177
 178         /* If this part is not encoded, send as-is */
 179         if ( (strlen(encoding) == 0) || (dont_decode)) {
 180                 if (CallBack != NULL) {
 181                         CallBack(name, filename, fixed_partnum(partnum),
 182                                 disposition, part_start,
 183                                 content_type, length, encoding, userdata);
 184                         }
 185                 return;
 186         }
 187
 188         if ((strcasecmp(encoding, "base64"))
 189             && (strcasecmp(encoding, "quoted-printable"))) {
 190                 lprintf(1, "ERROR: unknown MIME encoding '%s'\n", encoding);
 191                 return;
 192         }
 193         /*
 194          * Allocate a buffer for the decoded data.  The output buffer is the
 195          * same size as the input buffer; this assumes that the decoded data
 196          * will never be larger than the encoded data.  This is a safe
 197          * assumption with base64, uuencode, and quoted-printable.
 198          */
 199         lprintf(9, "About to allocate %d bytes for decoded part\n",
 200                 length+2048);
 201         decoded = malloc(length+2048);
 202         if (decoded == NULL) {
 203                 lprintf(1, "ERROR: cannot allocate memory.\n");
 204                 return;
 205         }
 206         lprintf(9, "Got it!\n");
 207
 208         lprintf(9, "Decoding %s\n", encoding);
 209         if (!strcasecmp(encoding, "base64")) {
 210                 bytes_decoded = decode_base64(decoded, part_start);
 211         }
 212         else if (!strcasecmp(encoding, "quoted-printable")) {
 213                 bytes_decoded = decode_quoted_printable(decoded,
 214                                                         part_start, length);
 215         }
 216         lprintf(9, "Bytes decoded: %d\n", bytes_decoded);
 217
 218         if (bytes_decoded > 0) if (CallBack != NULL) {
 219                 CallBack(name, filename, fixed_partnum(partnum),
 220                         disposition, decoded,
 221                         content_type, bytes_decoded, "binary", userdata);
 222         }
 223
 224         free(decoded);
 225 }
 226
 227 /*
 228  * Break out the components of a multipart message
 229  * (This function expects to be fed HEADERS + CONTENT)
 230  * Note: NULL can be supplied as content_end; in this case, the message is
 231  * considered to have ended when the parser encounters a 0x00 byte.
 232  */
 233 void the_mime_parser(char *partnum,
 234                      char *content_start, char *content_end,
 235                      void (*CallBack)
 236                       (char *cbname,
 237                        char *cbfilename,
 238                        char *cbpartnum,
 239                        char *cbdisp,
 240                        void *cbcontent,
 241                        char *cbtype,
 242                        size_t cblength,
 243                        char *cbencoding,
 244                        void *cbuserdata),
 245                      void (*PreMultiPartCallBack)
 246                       (char *cbname,
 247                        char *cbfilename,
 248                        char *cbpartnum,
 249                        char *cbdisp,
 250                        void *cbcontent,
 251                        char *cbtype,
 252                        size_t cblength,
 253                        char *cbencoding,
 254                        void *cbuserdata),
 255                      void (*PostMultiPartCallBack)
 256                       (char *cbname,
 257                        char *cbfilename,
 258                        char *cbpartnum,
 259                        char *cbdisp,
 260                        void *cbcontent,
 261                        char *cbtype,
 262                        size_t cblength,
 263                        char *cbencoding,
 264                        void *cbuserdata),
 265                       void *userdata,
 266                       int dont_decode
 267 )
 268 {
 269
 270         char *ptr;
 271         char *part_start, *part_end = NULL;
 272         char buf[SIZ];
 273         char *header;
 274         char *boundary;
 275         char *startary;
 276         char *endary;
 277         char *content_type;
 278         size_t content_length;
 279         char *encoding;
 280         char *disposition;
 281         char *name;
 282         char *filename;
 283         int is_multipart;
 284         int part_seq = 0;
 285         int i;
 286         size_t length;
 287         char nested_partnum[SIZ];
 288
 289         lprintf(9, "the_mime_parser() called\n");
 290         ptr = content_start;
 291         content_length = 0;
 292
 293         boundary = malloc(SIZ);
 294         memset(boundary, 0, SIZ);
 295
 296         startary = malloc(SIZ);
 297         memset(startary, 0, SIZ);
 298
 299         endary = malloc(SIZ);
 300         memset(endary, 0, SIZ);
 301
 302         header = malloc(SIZ);
 303         memset(header, 0, SIZ);
 304
 305         content_type = malloc(SIZ);
 306         memset(content_type, 0, SIZ);
 307
 308         encoding = malloc(SIZ);
 309         memset(encoding, 0, SIZ);
 310
 311         name = malloc(SIZ);
 312         memset(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                 if (!isspace(buf[0])) {
 337                         if (!strncasecmp(header, "Content-type: ", 14)) {
 338                                 strcpy(content_type, &header[14]);
 339                                 extract_key(name, content_type, "name");
 340                                 lprintf(9, "Extracted content-type <%s>\n",
 341                                         content_type);
 342                         }
 343                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 344                                 strcpy(disposition, &header[21]);
 345                                 extract_key(filename, disposition, "filename");
 346                         }
 347                         if (!strncasecmp(header, "Content-length: ", 16)) {
 348                                 content_length = (size_t) atol(&header[16]);
 349                         }
 350                         if (!strncasecmp(header,
 351                                       "Content-transfer-encoding: ", 27))
 352                                 strcpy(encoding, &header[27]);
 353                         if (strlen(boundary) == 0)
 354                                 extract_key(boundary, header, "boundary");
 355                         strcpy(header, "");
 356                 }
 357                 if ((strlen(header) + strlen(buf) + 2) < SIZ)
 358                         strcat(header, buf);
 359         } while ((strlen(buf) > 0) && (*ptr != 0));
 360
 361         for (i = 0; i < strlen(disposition); ++i)
 362                 if (disposition[i] == ';')
 363                         disposition[i] = 0;
 364         while (isspace(disposition[0]))
 365                 strcpy(disposition, &disposition[1]);
 366         for (i = 0; i < strlen(content_type); ++i)
 367                 if (content_type[i] == ';')
 368                         content_type[i] = 0;
 369         while (isspace(content_type[0]))
 370                 strcpy(content_type, &content_type[1]);
 371
 372         if (strlen(boundary) > 0) {
 373                 is_multipart = 1;
 374         } else {
 375                 is_multipart = 0;
 376         }
 377
 378         lprintf(9, "is_multipart=%d, boundary=<%s>\n",
 379                 is_multipart, boundary);
 380
 381         /* If this is a multipart message, then recursively process it */
 382         part_start = NULL;
 383         if (is_multipart) {
 384
 385                 /* Tell the client about this message's multipartedness */
 386                 if (PreMultiPartCallBack != NULL) {
 387                         PreMultiPartCallBack("", "", partnum, "",
 388                                 NULL, content_type,
 389                                 0, encoding, userdata);
 390                 }
 391
 392                 /* Figure out where the boundaries are */
 393                 sprintf(startary, "--%s", boundary);
 394                 sprintf(endary, "--%s--", boundary);
 395                 do {
 396                         if ( (!strncasecmp(ptr, startary, strlen(startary)))
 397                            || (!strncasecmp(ptr, endary, strlen(endary))) ) {
 398                                 lprintf(9, "hit boundary!\n");
 399                                 if (part_start != NULL) {
 400                                         if (strlen(partnum) > 0) {
 401                                                 sprintf(nested_partnum, "%s.%d",
 402                                                         partnum, ++part_seq);
 403                                         }
 404                                         else {
 405                                                 sprintf(nested_partnum, "%d",
 406                                                         ++part_seq);
 407                                         }
 408                                         the_mime_parser(nested_partnum,
 409                                                     part_start, part_end,
 410                                                         CallBack,
 411                                                         PreMultiPartCallBack,
 412                                                         PostMultiPartCallBack,
 413                                                         userdata,
 414                                                         dont_decode);
 415                                 }
 416                                 ptr = memreadline(ptr, buf, SIZ);
 417                                 part_start = ptr;
 418                         }
 419                         else {
 420                                 part_end = ptr;
 421                                 ++ptr;
 422                         }
 423                 } while ( (strcasecmp(ptr, endary)) && (ptr <= content_end) );
 424                 if (PostMultiPartCallBack != NULL) {
 425                         PostMultiPartCallBack("", "", partnum, "", NULL,
 426                                 content_type, 0, encoding, userdata);
 427                 }
 428                 goto end_parser;
 429         }
 430
 431         /* If it's not a multipart message, then do something with it */
 432         if (!is_multipart) {
 433                 lprintf(9, "doing non-multipart thing\n");
 434                 part_start = ptr;
 435                 length = 0;
 436                 while (ptr < content_end) {
 437                         ++ptr;
 438                         ++length;
 439                 }
 440                 part_end = content_end;
 441
 442                 /* Truncate if the header told us to */
 443                 if ( (content_length > 0) && (length > content_length) ) {
 444                         length = content_length;
 445                         lprintf(9, "truncated to %ld\n", (long)content_length);
 446                 }
 447
 448                 mime_decode(partnum,
 449                             part_start, length,
 450                             content_type, encoding, disposition,
 451                             name, filename,
 452                             CallBack, NULL, NULL,
 453                             userdata, dont_decode);
 454         }
 455
 456 end_parser:     /* free the buffers!  end the oppression!! */
 457         free(boundary);
 458         free(startary);
 459         free(endary);
 460         free(header);
 461         free(content_type);
 462         free(encoding);
 463         free(name);
 464         free(filename);
 465         free(disposition);
 466 }
 467
 468
 469
 470 /*
 471  * Entry point for the MIME parser.
 472  * (This function expects to be fed HEADERS + CONTENT)
 473  * Note: NULL can be supplied as content_end; in this case, the message is
 474  * considered to have ended when the parser encounters a 0x00 byte.
 475  */
 476 void mime_parser(char *content_start,
 477                 char *content_end,
 478
 479                  void (*CallBack)
 480                   (char *cbname,
 481                    char *cbfilename,
 482                    char *cbpartnum,
 483                    char *cbdisp,
 484                    void *cbcontent,
 485                    char *cbtype,
 486                    size_t cblength,
 487                    char *cbencoding,
 488                    void *cbuserdata),
 489
 490                  void (*PreMultiPartCallBack)
 491                   (char *cbname,
 492                    char *cbfilename,
 493                    char *cbpartnum,
 494                    char *cbdisp,
 495                    void *cbcontent,
 496                    char *cbtype,
 497                    size_t cblength,
 498                    char *cbencoding,
 499                    void *cbuserdata),
 500
 501                  void (*PostMultiPartCallBack)
 502                   (char *cbname,
 503                    char *cbfilename,
 504                    char *cbpartnum,
 505                    char *cbdisp,
 506                    void *cbcontent,
 507                    char *cbtype,
 508                    size_t cblength,
 509                    char *cbencoding,
 510                    void *cbuserdata),
 511
 512                   void *userdata,
 513                   int dont_decode
 514 )
 515 {
 516
 517         lprintf(9, "mime_parser() called\n");
 518         the_mime_parser("", content_start, content_end,
 519                         CallBack,
 520                         PreMultiPartCallBack,
 521                         PostMultiPartCallBack,
 522                         userdata, dont_decode);
 523 }