citadel/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 #ifdef DLL_EXPORT
  12 #define IN_LIBCIT
  13 #endif
  14
  15 #include <stdlib.h>
  16 #include <unistd.h>
  17 #include <stdio.h>
  18 #include <signal.h>
  19 #include <sys/types.h>
  20 #include <ctype.h>
  21 #include <string.h>
  22 #include <sys/stat.h>
  23 #include <errno.h>
  24 #include "citadel.h"
  25 #include "server.h"
  26 #include "dynloader.h"
  27 #include "sysdep_decls.h"
  28 #include "mime_parser.h"
  29 #include "tools.h"
  30
  31
  32 void extract_key(char *target, char *source, char *key)
  33 {
  34         int a, b;
  35
  36         strcpy(target, source);
  37         for (a = 0; a < strlen(target); ++a) {
  38                 if ((!strncasecmp(&target[a], key, strlen(key)))
  39                     && (target[a + strlen(key)] == '=')) {
  40                         strcpy(target, &target[a + strlen(key) + 1]);
  41                         if (target[0] == 34)
  42                                 strcpy(target, &target[1]);
  43                         for (b = 0; b < strlen(target); ++b)
  44                                 if (target[b] == 34)
  45                                         target[b] = 0;
  46                         return;
  47                 }
  48         }
  49         strcpy(target, "");
  50 }
  51
  52
  53 /*
  54  * For non-multipart messages, we need to generate a quickie partnum of "1"
  55  * to return to callback functions.  Some callbacks demand it.
  56  */
  57 char *fixed_partnum(char *supplied_partnum) {
  58         if (supplied_partnum == NULL) return "1";
  59         if (strlen(supplied_partnum)==0) return "1";
  60         return supplied_partnum;
  61 }
  62
  63
  64
  65 /*
  66  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
  67  */
  68 int decode_quoted_printable(char *decoded, char *encoded, int sourcelen) {
  69         char buf[SIZ];
  70         int buf_length = 0;
  71         int soft_line_break = 0;
  72         int ch;
  73         int decoded_length = 0;
  74         int i;
  75
  76         decoded[0] = 0;
  77         decoded_length = 0;
  78         buf[0] = 0;
  79         buf_length = 0;
  80
  81         for (i = 0; i < sourcelen; ++i) {
  82
  83                 buf[buf_length++] = encoded[i];
  84
  85                 if ( (encoded[i] == '\n')
  86                    || (encoded[i] == 0)
  87                    || (i == (sourcelen-1)) ) {
  88                         buf[buf_length++] = 0;
  89
  90                         /*** begin -- process one line ***/
  91
  92                         if (buf[strlen(buf)-1] == '\n') {
  93                                 buf[strlen(buf)-1] = 0;
  94                         }
  95                         if (buf[strlen(buf)-1] == '\r') {
  96                                 buf[strlen(buf)-1] = 0;
  97                         }
  98                         while (isspace(buf[strlen(buf)-1])) {
  99                                 buf[strlen(buf)-1] = 0;
 100                         }
 101                         soft_line_break = 0;
 102
 103                         while (strlen(buf) > 0) {
 104                                 if (!strcmp(buf, "=")) {
 105                                         soft_line_break = 1;
 106                                         strcpy(buf, "");
 107                                 } else if ((strlen(buf)>=3) && (buf[0]=='=')) {
 108                                         sscanf(&buf[1], "%02x", &ch);
 109                                         decoded[decoded_length++] = ch;
 110                                         strcpy(buf, &buf[3]);
 111                                 } else {
 112                                         decoded[decoded_length++] = buf[0];
 113                                         strcpy(buf, &buf[1]);
 114                                 }
 115                         }
 116                         if (soft_line_break == 0) {
 117                                 decoded[decoded_length++] = '\r';
 118                                 decoded[decoded_length++] = '\n';
 119                         }
 120                         buf_length = 0;
 121                         /*** end -- process one line ***/
 122                 }
 123         }
 124
 125         decoded[decoded_length++] = 0;
 126         return(decoded_length);
 127 }
 128
 129 /*
 130  * Given a message or message-part body and a length, handle any necessary
 131  * decoding and pass the request up the stack.
 132  */
 133 void mime_decode(char *partnum,
 134                  char *part_start, size_t length,
 135                  char *content_type, char *encoding,
 136                  char *disposition,
 137                  char *name, char *filename,
 138                  void (*CallBack)
 139                   (char *cbname,
 140                    char *cbfilename,
 141                    char *cbpartnum,
 142                    char *cbdisp,
 143                    void *cbcontent,
 144                    char *cbtype,
 145                    size_t cblength,
 146                    char *cbencoding,
 147                    void *cbuserdata),
 148                  void (*PreMultiPartCallBack)
 149                   (char *cbname,
 150                    char *cbfilename,
 151                    char *cbpartnum,
 152                    char *cbdisp,
 153                    void *cbcontent,
 154                    char *cbtype,
 155                    size_t cblength,
 156                    char *cbencoding,
 157                    void *cbuserdata),
 158                  void (*PostMultiPartCallBack)
 159                   (char *cbname,
 160                    char *cbfilename,
 161                    char *cbpartnum,
 162                    char *cbdisp,
 163                    void *cbcontent,
 164                    char *cbtype,
 165                    size_t cblength,
 166                    char *cbencoding,
 167                    void *cbuserdata),
 168                   void *userdata,
 169                   int dont_decode
 170 )
 171 {
 172
 173         char *decoded;
 174         size_t bytes_decoded = 0;
 175
 176         lprintf(9, "mime_decode() called\n");
 177
 178         /* Some encodings aren't really encodings */
 179         if (!strcasecmp(encoding, "7bit"))
 180                 strcpy(encoding, "");
 181         if (!strcasecmp(encoding, "8bit"))
 182                 strcpy(encoding, "");
 183         if (!strcasecmp(encoding, "binary"))
 184                 strcpy(encoding, "");
 185
 186         /* If this part is not encoded, send as-is */
 187         if ( (strlen(encoding) == 0) || (dont_decode)) {
 188                 if (CallBack != NULL) {
 189                         CallBack(name, filename, fixed_partnum(partnum),
 190                                 disposition, part_start,
 191                                 content_type, length, encoding, userdata);
 192                         }
 193                 return;
 194         }
 195         if ((strcasecmp(encoding, "base64"))
 196             && (strcasecmp(encoding, "quoted-printable"))) {
 197                 lprintf(9, "ERROR: unknown MIME encoding '%s'\n", encoding);
 198                 return;
 199         }
 200         /*
 201          * Allocate a buffer for the decoded data.  The output buffer is the
 202          * same size as the input buffer; this assumes that the decoded data
 203          * will never be larger than the encoded data.  This is a safe
 204          * assumption with base64, uuencode, and quoted-printable.  Just to
 205          * be safe, we still pad the buffer a bit.
 206          */
 207         decoded = mallok(length + 1024);
 208         if (decoded == NULL) {
 209                 lprintf(9, "ERROR: cannot allocate memory.\n");
 210                 return;
 211         }
 212
 213         if (!strcasecmp(encoding, "base64")) {
 214                 bytes_decoded = decode_base64(decoded, part_start);
 215         }
 216         else if (!strcasecmp(encoding, "quoted-printable")) {
 217                 bytes_decoded = decode_quoted_printable(decoded,
 218                                                         part_start, length);
 219         }
 220
 221         if (bytes_decoded > 0) if (CallBack != NULL) {
 222                 CallBack(name, filename, fixed_partnum(partnum),
 223                         disposition, decoded,
 224                         content_type, bytes_decoded, "binary", userdata);
 225         }
 226
 227         phree(decoded);
 228 }
 229
 230 /*
 231  * Break out the components of a multipart message
 232  * (This function expects to be fed HEADERS + CONTENT)
 233  * Note: NULL can be supplied as content_end; in this case, the message is
 234  * considered to have ended when the parser encounters a 0x00 byte.
 235  */
 236 void the_mime_parser(char *partnum,
 237                      char *content_start, char *content_end,
 238                      void (*CallBack)
 239                       (char *cbname,
 240                        char *cbfilename,
 241                        char *cbpartnum,
 242                        char *cbdisp,
 243                        void *cbcontent,
 244                        char *cbtype,
 245                        size_t cblength,
 246                        char *cbencoding,
 247                        void *cbuserdata),
 248                      void (*PreMultiPartCallBack)
 249                       (char *cbname,
 250                        char *cbfilename,
 251                        char *cbpartnum,
 252                        char *cbdisp,
 253                        void *cbcontent,
 254                        char *cbtype,
 255                        size_t cblength,
 256                        char *cbencoding,
 257                        void *cbuserdata),
 258                      void (*PostMultiPartCallBack)
 259                       (char *cbname,
 260                        char *cbfilename,
 261                        char *cbpartnum,
 262                        char *cbdisp,
 263                        void *cbcontent,
 264                        char *cbtype,
 265                        size_t cblength,
 266                        char *cbencoding,
 267                        void *cbuserdata),
 268                       void *userdata,
 269                       int dont_decode
 270 )
 271 {
 272
 273         char *ptr;
 274         char *part_start, *part_end = NULL;
 275         char buf[SIZ];
 276         char header[SIZ];
 277         char boundary[SIZ];
 278         char startary[SIZ];
 279         char endary[SIZ];
 280         char content_type[SIZ];
 281         size_t content_length;
 282         char encoding[SIZ];
 283         char disposition[SIZ];
 284         char name[SIZ];
 285         char filename[SIZ];
 286         int is_multipart;
 287         int part_seq = 0;
 288         int i;
 289         size_t length;
 290         char nested_partnum[SIZ];
 291
 292         lprintf(9, "the_mime_parser() called\n");
 293         ptr = content_start;
 294         memset(boundary, 0, sizeof boundary);
 295         memset(content_type, 0, sizeof content_type);
 296         memset(encoding, 0, sizeof encoding);
 297         memset(name, 0, sizeof name);
 298         memset(filename, 0, sizeof filename);
 299         memset(disposition, 0, sizeof disposition);
 300         content_length = 0;
 301
 302         /* If the caller didn't supply an endpointer, generate one by measure */
 303         if (content_end == NULL) {
 304                 content_end = &content_start[strlen(content_start)];
 305         }
 306
 307         /* Learn interesting things from the headers */
 308         strcpy(header, "");
 309         do {
 310                 ptr = memreadline(ptr, buf, sizeof buf);
 311                 if (ptr >= content_end)
 312                         return;
 313
 314                 for (i = 0; i < strlen(buf); ++i)
 315                         if (isspace(buf[i]))
 316                                 buf[i] = ' ';
 317                 if (!isspace(buf[0])) {
 318                         if (!strncasecmp(header, "Content-type: ", 14)) {
 319                                 strcpy(content_type, &header[14]);
 320                                 extract_key(name, content_type, "name");
 321                                 lprintf(9, "Extracted content-type <%s>\n",
 322                                         content_type);
 323                         }
 324                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 325                                 strcpy(disposition, &header[21]);
 326                                 extract_key(filename, disposition, "filename");
 327                         }
 328                         if (!strncasecmp(header, "Content-length: ", 16)) {
 329                                 content_length = (size_t) atol(&header[16]);
 330                         }
 331                         if (!strncasecmp(header,
 332                                       "Content-transfer-encoding: ", 27))
 333                                 strcpy(encoding, &header[27]);
 334                         if (strlen(boundary) == 0)
 335                                 extract_key(boundary, header, "boundary");
 336                         strcpy(header, "");
 337                 }
 338                 if ((strlen(header) + strlen(buf) + 2) < sizeof(header))
 339                         strcat(header, buf);
 340         } while ((strlen(buf) > 0) && (*ptr != 0));
 341
 342         for (i = 0; i < strlen(disposition); ++i)
 343                 if (disposition[i] == ';')
 344                         disposition[i] = 0;
 345         while (isspace(disposition[0]))
 346                 strcpy(disposition, &disposition[1]);
 347         for (i = 0; i < strlen(content_type); ++i)
 348                 if (content_type[i] == ';')
 349                         content_type[i] = 0;
 350         while (isspace(content_type[0]))
 351                 strcpy(content_type, &content_type[1]);
 352
 353         if (strlen(boundary) > 0) {
 354                 is_multipart = 1;
 355         } else {
 356                 is_multipart = 0;
 357         }
 358
 359         lprintf(9, "is_multipart=%d, boundary=<%s>\n",
 360                 is_multipart, boundary);
 361
 362         /* If this is a multipart message, then recursively process it */
 363         part_start = NULL;
 364         if (is_multipart) {
 365
 366                 /* Tell the client about this message's multipartedness */
 367                 if (PreMultiPartCallBack != NULL) {
 368                         PreMultiPartCallBack("", "", partnum, "",
 369                                 NULL, content_type,
 370                                 0, encoding, userdata);
 371                 }
 372
 373                 /* Figure out where the boundaries are */
 374                 sprintf(startary, "--%s", boundary);
 375                 sprintf(endary, "--%s--", boundary);
 376                 do {
 377                         if ( (!strncasecmp(ptr, startary, strlen(startary)))
 378                            || (!strncasecmp(ptr, endary, strlen(endary))) ) {
 379                                 lprintf(9, "hit boundary!\n");
 380                                 if (part_start != NULL) {
 381                                         if (strlen(partnum) > 0) {
 382                                                 sprintf(nested_partnum, "%s.%d",
 383                                                         partnum, ++part_seq);
 384                                         }
 385                                         else {
 386                                                 sprintf(nested_partnum, "%d",
 387                                                         ++part_seq);
 388                                         }
 389                                         the_mime_parser(nested_partnum,
 390                                                     part_start, part_end,
 391                                                         CallBack,
 392                                                         PreMultiPartCallBack,
 393                                                         PostMultiPartCallBack,
 394                                                         userdata,
 395                                                         dont_decode);
 396                                 }
 397                                 ptr = memreadline(ptr, buf, sizeof(buf));
 398                                 part_start = ptr;
 399                         }
 400                         else {
 401                                 part_end = ptr;
 402                                 ++ptr;
 403                         }
 404                 } while ( (strcasecmp(ptr, endary)) && (ptr <= content_end) );
 405                 if (PostMultiPartCallBack != NULL) {
 406                         PostMultiPartCallBack("", "", partnum, "", NULL,
 407                                 content_type, 0, encoding, userdata);
 408                 }
 409                 return;
 410         }
 411
 412         /* If it's not a multipart message, then do something with it */
 413         if (!is_multipart) {
 414                 lprintf(9, "doing non-multipart thing\n");
 415                 part_start = ptr;
 416                 length = 0;
 417                 while (ptr < content_end) {
 418                         ++ptr;
 419                         ++length;
 420                 }
 421                 part_end = content_end;
 422
 423                 /* Truncate if the header told us to */
 424                 if ( (content_length > 0) && (length > content_length) ) {
 425                         length = content_length;
 426                         lprintf(9, "truncated to %ld\n", (long)content_length);
 427                 }
 428
 429                 mime_decode(partnum,
 430                             part_start, length,
 431                             content_type, encoding, disposition,
 432                             name, filename,
 433                             CallBack, NULL, NULL,
 434                             userdata, dont_decode);
 435         }
 436 }
 437
 438
 439
 440 /*
 441  * Entry point for the MIME parser.
 442  * (This function expects to be fed HEADERS + CONTENT)
 443  * Note: NULL can be supplied as content_end; in this case, the message is
 444  * considered to have ended when the parser encounters a 0x00 byte.
 445  */
 446 void mime_parser(char *content_start,
 447                 char *content_end,
 448
 449                  void (*CallBack)
 450                   (char *cbname,
 451                    char *cbfilename,
 452                    char *cbpartnum,
 453                    char *cbdisp,
 454                    void *cbcontent,
 455                    char *cbtype,
 456                    size_t cblength,
 457                    char *cbencoding,
 458                    void *cbuserdata),
 459
 460                  void (*PreMultiPartCallBack)
 461                   (char *cbname,
 462                    char *cbfilename,
 463                    char *cbpartnum,
 464                    char *cbdisp,
 465                    void *cbcontent,
 466                    char *cbtype,
 467                    size_t cblength,
 468                    char *cbencoding,
 469                    void *cbuserdata),
 470
 471                  void (*PostMultiPartCallBack)
 472                   (char *cbname,
 473                    char *cbfilename,
 474                    char *cbpartnum,
 475                    char *cbdisp,
 476                    void *cbcontent,
 477                    char *cbtype,
 478                    size_t cblength,
 479                    char *cbencoding,
 480                    void *cbuserdata),
 481
 482                   void *userdata,
 483                   int dont_decode
 484 )
 485 {
 486
 487         lprintf(9, "mime_parser() called\n");
 488         the_mime_parser("", content_start, content_end,
 489                         CallBack,
 490                         PreMultiPartCallBack,
 491                         PostMultiPartCallBack,
 492                         userdata, dont_decode);
 493 }