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
 196         if ((strcasecmp(encoding, "base64"))
 197             && (strcasecmp(encoding, "quoted-printable"))) {
 198                 lprintf(9, "ERROR: unknown MIME encoding '%s'\n", encoding);
 199                 return;
 200         }
 201         /*
 202          * Allocate a buffer for the decoded data.  The output buffer is the
 203          * same size as the input buffer; this assumes that the decoded data
 204          * will never be larger than the encoded data.  This is a safe
 205          * assumption with base64, uuencode, and quoted-printable.
 206          */
 207         lprintf(9, "About to allocate %d bytes for decoded part\n",
 208                 length+2048);
 209         decoded = mallok(length+2048);
 210         if (decoded == NULL) {
 211                 lprintf(9, "ERROR: cannot allocate memory.\n");
 212                 return;
 213         }
 214         lprintf(9, "Got it!\n");
 215
 216         lprintf(9, "Decoding %s\n", encoding);
 217         if (!strcasecmp(encoding, "base64")) {
 218                 bytes_decoded = decode_base64(decoded, part_start, length);
 219         }
 220         else if (!strcasecmp(encoding, "quoted-printable")) {
 221                 bytes_decoded = decode_quoted_printable(decoded,
 222                                                         part_start, length);
 223         }
 224         lprintf(9, "Bytes decoded: %d\n", bytes_decoded);
 225
 226         if (bytes_decoded > 0) if (CallBack != NULL) {
 227                 CallBack(name, filename, fixed_partnum(partnum),
 228                         disposition, decoded,
 229                         content_type, bytes_decoded, "binary", userdata);
 230         }
 231
 232         phree(decoded);
 233 }
 234
 235 /*
 236  * Break out the components of a multipart message
 237  * (This function expects to be fed HEADERS + CONTENT)
 238  * Note: NULL can be supplied as content_end; in this case, the message is
 239  * considered to have ended when the parser encounters a 0x00 byte.
 240  */
 241 void the_mime_parser(char *partnum,
 242                      char *content_start, char *content_end,
 243                      void (*CallBack)
 244                       (char *cbname,
 245                        char *cbfilename,
 246                        char *cbpartnum,
 247                        char *cbdisp,
 248                        void *cbcontent,
 249                        char *cbtype,
 250                        size_t cblength,
 251                        char *cbencoding,
 252                        void *cbuserdata),
 253                      void (*PreMultiPartCallBack)
 254                       (char *cbname,
 255                        char *cbfilename,
 256                        char *cbpartnum,
 257                        char *cbdisp,
 258                        void *cbcontent,
 259                        char *cbtype,
 260                        size_t cblength,
 261                        char *cbencoding,
 262                        void *cbuserdata),
 263                      void (*PostMultiPartCallBack)
 264                       (char *cbname,
 265                        char *cbfilename,
 266                        char *cbpartnum,
 267                        char *cbdisp,
 268                        void *cbcontent,
 269                        char *cbtype,
 270                        size_t cblength,
 271                        char *cbencoding,
 272                        void *cbuserdata),
 273                       void *userdata,
 274                       int dont_decode
 275 )
 276 {
 277
 278         char *ptr;
 279         char *part_start, *part_end = NULL;
 280         char buf[SIZ];
 281         char *header;
 282         char *boundary;
 283         char *startary;
 284         char *endary;
 285         char *content_type;
 286         size_t content_length;
 287         char *encoding;
 288         char *disposition;
 289         char *name;
 290         char *filename;
 291         int is_multipart;
 292         int part_seq = 0;
 293         int i;
 294         size_t length;
 295         char nested_partnum[SIZ];
 296
 297         lprintf(9, "the_mime_parser() called\n");
 298         ptr = content_start;
 299         content_length = 0;
 300
 301         boundary = mallok(SIZ);
 302         memset(boundary, 0, SIZ);
 303
 304         startary = mallok(SIZ);
 305         memset(startary, 0, SIZ);
 306
 307         endary = mallok(SIZ);
 308         memset(endary, 0, SIZ);
 309
 310         header = mallok(SIZ);
 311         memset(header, 0, SIZ);
 312
 313         content_type = mallok(SIZ);
 314         memset(content_type, 0, SIZ);
 315
 316         encoding = mallok(SIZ);
 317         memset(encoding, 0, SIZ);
 318
 319         name = mallok(SIZ);
 320         memset(name, 0, SIZ);
 321
 322         filename = mallok(SIZ);
 323         memset(filename, 0, SIZ);
 324
 325         disposition = mallok(SIZ);
 326         memset(disposition, 0, SIZ);
 327
 328         /* If the caller didn't supply an endpointer, generate one by measure */
 329         if (content_end == NULL) {
 330                 content_end = &content_start[strlen(content_start)];
 331         }
 332
 333         /* Learn interesting things from the headers */
 334         strcpy(header, "");
 335         do {
 336                 ptr = memreadline(ptr, buf, SIZ);
 337                 if (ptr >= content_end) {
 338                         goto end_parser;
 339                 }
 340
 341                 for (i = 0; i < strlen(buf); ++i)
 342                         if (isspace(buf[i]))
 343                                 buf[i] = ' ';
 344                 if (!isspace(buf[0])) {
 345                         if (!strncasecmp(header, "Content-type: ", 14)) {
 346                                 strcpy(content_type, &header[14]);
 347                                 extract_key(name, content_type, "name");
 348                                 /* Deal with weird headers */
 349                                 if (strchr(content_type, ' '))
 350                                         *(strchr(content_type, ' ')) = '\0';
 351                                 if (strchr(content_type, ';'))
 352                                         *(strchr(content_type, ';')) = '\0';
 353                                 lprintf(9, "Extracted content-type <%s>\n",
 354                                         content_type);
 355                         }
 356                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 357                                 strcpy(disposition, &header[21]);
 358                                 extract_key(filename, disposition, "filename");
 359                         }
 360                         if (!strncasecmp(header, "Content-length: ", 16)) {
 361                                 content_length = (size_t) atol(&header[16]);
 362                         }
 363                         if (!strncasecmp(header,
 364                                       "Content-transfer-encoding: ", 27))
 365                                 strcpy(encoding, &header[27]);
 366                         if (strlen(boundary) == 0)
 367                                 extract_key(boundary, header, "boundary");
 368                         strcpy(header, "");
 369                 }
 370                 if ((strlen(header) + strlen(buf) + 2) < SIZ)
 371                         strcat(header, buf);
 372         } while ((strlen(buf) > 0) && (*ptr != 0));
 373
 374         for (i = 0; i < strlen(disposition); ++i)
 375                 if (disposition[i] == ';')
 376                         disposition[i] = 0;
 377         while (isspace(disposition[0]))
 378                 strcpy(disposition, &disposition[1]);
 379         for (i = 0; i < strlen(content_type); ++i)
 380                 if (content_type[i] == ';')
 381                         content_type[i] = 0;
 382         while (isspace(content_type[0]))
 383                 strcpy(content_type, &content_type[1]);
 384
 385         if (strlen(boundary) > 0) {
 386                 is_multipart = 1;
 387         } else {
 388                 is_multipart = 0;
 389         }
 390
 391         lprintf(9, "is_multipart=%d, boundary=<%s>\n",
 392                 is_multipart, boundary);
 393
 394         /* If this is a multipart message, then recursively process it */
 395         part_start = NULL;
 396         if (is_multipart) {
 397
 398                 /* Tell the client about this message's multipartedness */
 399                 if (PreMultiPartCallBack != NULL) {
 400                         PreMultiPartCallBack("", "", partnum, "",
 401                                 NULL, content_type,
 402                                 0, encoding, userdata);
 403                 }
 404
 405                 /* Figure out where the boundaries are */
 406                 snprintf(startary, SIZ, "--%s", boundary);
 407                 snprintf(endary, SIZ, "--%s--", boundary);
 408                 do {
 409                         if ( (!strncasecmp(ptr, startary, strlen(startary)))
 410                            || (!strncasecmp(ptr, endary, strlen(endary))) ) {
 411                                 lprintf(9, "hit boundary!\n");
 412                                 if (part_start != NULL) {
 413                                         if (strlen(partnum) > 0) {
 414                                                 snprintf(nested_partnum,
 415                                                          sizeof nested_partnum,
 416                                                          "%s.%d", partnum,
 417                                                          ++part_seq);
 418                                         }
 419                                         else {
 420                                                 snprintf(nested_partnum,
 421                                                          sizeof nested_partnum,
 422                                                          "%d", ++part_seq);
 423                                         }
 424                                         the_mime_parser(nested_partnum,
 425                                                     part_start, part_end,
 426                                                         CallBack,
 427                                                         PreMultiPartCallBack,
 428                                                         PostMultiPartCallBack,
 429                                                         userdata,
 430                                                         dont_decode);
 431                                 }
 432                                 ptr = memreadline(ptr, buf, SIZ);
 433                                 part_start = ptr;
 434                         }
 435                         else {
 436                                 part_end = ptr;
 437                                 ++ptr;
 438                         }
 439                 } while ( (strcasecmp(ptr, endary)) && (ptr <= content_end) );
 440                 if (PostMultiPartCallBack != NULL) {
 441                         PostMultiPartCallBack("", "", partnum, "", NULL,
 442                                 content_type, 0, encoding, userdata);
 443                 }
 444                 goto end_parser;
 445         }
 446
 447         /* If it's not a multipart message, then do something with it */
 448         if (!is_multipart) {
 449                 lprintf(9, "doing non-multipart thing\n");
 450                 part_start = ptr;
 451                 length = 0;
 452                 while (ptr < content_end) {
 453                         ++ptr;
 454                         ++length;
 455                 }
 456                 part_end = content_end;
 457
 458                 /* Truncate if the header told us to */
 459                 if ( (content_length > 0) && (length > content_length) ) {
 460                         length = content_length;
 461                         lprintf(9, "truncated to %ld\n", (long)content_length);
 462                 }
 463
 464                 mime_decode(partnum,
 465                             part_start, length,
 466                             content_type, encoding, disposition,
 467                             name, filename,
 468                             CallBack, NULL, NULL,
 469                             userdata, dont_decode);
 470         }
 471
 472 end_parser:     /* free the buffers!  end the oppression!! */
 473         phree(boundary);
 474         phree(startary);
 475         phree(endary);
 476         phree(header);
 477         phree(content_type);
 478         phree(encoding);
 479         phree(name);
 480         phree(filename);
 481         phree(disposition);
 482 }
 483
 484
 485
 486 /*
 487  * Entry point for the MIME parser.
 488  * (This function expects to be fed HEADERS + CONTENT)
 489  * Note: NULL can be supplied as content_end; in this case, the message is
 490  * considered to have ended when the parser encounters a 0x00 byte.
 491  */
 492 void mime_parser(char *content_start,
 493                 char *content_end,
 494
 495                  void (*CallBack)
 496                   (char *cbname,
 497                    char *cbfilename,
 498                    char *cbpartnum,
 499                    char *cbdisp,
 500                    void *cbcontent,
 501                    char *cbtype,
 502                    size_t cblength,
 503                    char *cbencoding,
 504                    void *cbuserdata),
 505
 506                  void (*PreMultiPartCallBack)
 507                   (char *cbname,
 508                    char *cbfilename,
 509                    char *cbpartnum,
 510                    char *cbdisp,
 511                    void *cbcontent,
 512                    char *cbtype,
 513                    size_t cblength,
 514                    char *cbencoding,
 515                    void *cbuserdata),
 516
 517                  void (*PostMultiPartCallBack)
 518                   (char *cbname,
 519                    char *cbfilename,
 520                    char *cbpartnum,
 521                    char *cbdisp,
 522                    void *cbcontent,
 523                    char *cbtype,
 524                    size_t cblength,
 525                    char *cbencoding,
 526                    void *cbuserdata),
 527
 528                   void *userdata,
 529                   int dont_decode
 530 )
 531 {
 532
 533         lprintf(9, "mime_parser() called\n");
 534         the_mime_parser("", content_start, content_end,
 535                         CallBack,
 536                         PreMultiPartCallBack,
 537                         PostMultiPartCallBack,
 538                         userdata, dont_decode);
 539 }