citadel/mime_parser.c

   1 /*
   2  * $Id$
   3  *
   4  * This is a really bad attempt at writing a parser to handle MIME-encoded
   5  * messages.
   6  *
   7  * Copyright (c) 1998-1999 by Art Cancro
   8  * This code is distributed under the terms of the GNU General Public License.
   9  *
  10  */
  11
  12 #include "sysdep.h"
  13 #include <stdlib.h>
  14 #include <unistd.h>
  15 #include <stdio.h>
  16 #include <signal.h>
  17 #include <sys/types.h>
  18 #include <ctype.h>
  19 #include <string.h>
  20 #include <sys/stat.h>
  21 #include <errno.h>
  22 #include "citadel.h"
  23 #include "mime_parser.h"
  24 #include "sysdep_decls.h"
  25 #include "server.h"
  26
  27
  28
  29 void extract_key(char *target, char *source, char *key)
  30 {
  31         int a, b;
  32
  33         strcpy(target, source);
  34         for (a = 0; a < strlen(target); ++a) {
  35                 if ((!strncasecmp(&target[a], key, strlen(key)))
  36                     && (target[a + strlen(key)] == '=')) {
  37                         strcpy(target, &target[a + strlen(key) + 1]);
  38                         if (target[0] == 34)
  39                                 strcpy(target, &target[1]);
  40                         for (b = 0; b < strlen(target); ++b)
  41                                 if (target[b] == 34)
  42                                         target[b] = 0;
  43                         return;
  44                 }
  45         }
  46         strcpy(target, "");
  47 }
  48
  49
  50
  51 /*
  52  * Utility function to "readline" from memory
  53  * (returns new pointer)
  54  */
  55 char *memreadline(char *start, char *buf, int maxlen)
  56 {
  57         char ch;
  58         char *ptr;
  59
  60         ptr = start;
  61         memset(buf, 0, maxlen);
  62
  63         while (1) {
  64                 ch = *ptr++;
  65                 if ((ch == 10) || (ch == 0)) {
  66                         if (strlen(buf) > 0)
  67                                 if (buf[strlen(buf) - 1] == 13)
  68                                         buf[strlen(buf) - 1] = 0;
  69                         return ptr;
  70                 }
  71                 if (strlen(buf) < (maxlen - 1)) {
  72                         buf[strlen(buf) + 1] = 0;
  73                         buf[strlen(buf)] = ch;
  74                 }
  75         }
  76 }
  77
  78
  79 /*
  80  * For non-multipart messages, we need to generate a quickie partnum of "1"
  81  * to return to callback functions.  Some callbacks demand it.
  82  */
  83 char *fixed_partnum(char *supplied_partnum) {
  84         if (supplied_partnum == NULL) return "1";
  85         if (strlen(supplied_partnum)==0) return "1";
  86         return supplied_partnum;
  87 }
  88
  89
  90 /*
  91  * Given a message or message-part body and a length, handle any necessary
  92  * decoding and pass the request up the stack.
  93  */
  94 void mime_decode(char *partnum,
  95                  char *part_start, size_t length,
  96                  char *content_type, char *encoding,
  97                  char *disposition,
  98                  char *name, char *filename,
  99                  void (*CallBack)
 100                   (char *cbname,
 101                    char *cbfilename,
 102                    char *cbpartnum,
 103                    char *cbdisp,
 104                    void *cbcontent,
 105                    char *cbtype,
 106                    size_t cblength,
 107                    char *cbencoding,
 108                    void *cbuserdata),
 109                  void (*PreMultiPartCallBack)
 110                   (char *cbname,
 111                    char *cbfilename,
 112                    char *cbpartnum,
 113                    char *cbdisp,
 114                    void *cbcontent,
 115                    char *cbtype,
 116                    size_t cblength,
 117                    char *cbencoding,
 118                    void *cbuserdata),
 119                  void (*PostMultiPartCallBack)
 120                   (char *cbname,
 121                    char *cbfilename,
 122                    char *cbpartnum,
 123                    char *cbdisp,
 124                    void *cbcontent,
 125                    char *cbtype,
 126                    size_t cblength,
 127                    char *cbencoding,
 128                    void *cbuserdata),
 129                   void *userdata,
 130                   int dont_decode
 131 )
 132 {
 133
 134         char *decoded;
 135         struct stat statbuf;
 136         int sendpipe[2];
 137         int recvpipe[2];
 138         int childpid;
 139         size_t bytes_sent = 0;
 140         size_t bytes_recv = 0;
 141         size_t blocksize;
 142         int write_error = 0;
 143
 144         lprintf(9, "mime_decode() called\n");
 145
 146         /* Some encodings aren't really encodings */
 147         if (!strcasecmp(encoding, "7bit"))
 148                 strcpy(encoding, "");
 149         if (!strcasecmp(encoding, "8bit"))
 150                 strcpy(encoding, "");
 151         if (!strcasecmp(encoding, "binary"))
 152                 strcpy(encoding, "");
 153
 154         /* If this part is not encoded, send as-is */
 155         if ( (strlen(encoding) == 0) || (dont_decode)) {
 156                 if (CallBack != NULL) {
 157                         CallBack(name, filename, fixed_partnum(partnum),
 158                                 disposition, part_start,
 159                                 content_type, length, encoding, userdata);
 160                         }
 161                 return;
 162         }
 163         if ((strcasecmp(encoding, "base64"))
 164             && (strcasecmp(encoding, "quoted-printable"))) {
 165                 lprintf(5, "ERROR: unknown MIME encoding '%s'\n", encoding);
 166                 return;
 167         }
 168         /*
 169          * Allocate a buffer for the decoded data.  The output buffer is the
 170          * same size as the input buffer; this assumes that the decoded data
 171          * will never be larger than the encoded data.  This is a safe
 172          * assumption with base64, uuencode, and quoted-printable.  Just to
 173          * be safe, we still pad the buffer a bit.
 174          */
 175         decoded = mallok(length + 1024);
 176         if (decoded == NULL) {
 177                 lprintf(5, "ERROR: cannot allocate memory.\n");
 178                 return;
 179         }
 180         if (pipe(sendpipe) != 0)
 181                 return;
 182         if (pipe(recvpipe) != 0)
 183                 return;
 184
 185         childpid = fork();
 186         if (childpid < 0) {
 187                 phree(decoded);
 188                 return;
 189         }
 190         if (childpid == 0) {
 191                 close(2);
 192                 /* send stdio to the pipes */
 193                 if (dup2(sendpipe[0], 0) < 0)
 194                         lprintf(5, "ERROR dup2()\n");
 195                 if (dup2(recvpipe[1], 1) < 0)
 196                         lprintf(5, "ERROR dup2()\n");
 197                 close(sendpipe[1]);     /* Close the ends we're not using */
 198                 close(recvpipe[0]);
 199                 if (!strcasecmp(encoding, "base64"))
 200                         execlp("./base64", "base64", "-d", NULL);
 201                 else if (!strcasecmp(encoding, "quoted-printable"))
 202                         execlp("./qpdecode", "qpdecode", NULL);
 203                 lprintf(5, "ERROR: cannot exec decoder for %s\n", encoding);
 204                 exit(1);
 205         }
 206         close(sendpipe[0]);     /* Close the ends we're not using  */
 207         close(recvpipe[1]);
 208
 209         while ((bytes_sent < length) && (write_error == 0)) {
 210                 /* Empty the input pipe FIRST */
 211                 while (fstat(recvpipe[0], &statbuf), (statbuf.st_size > 0)) {
 212                         blocksize = read(recvpipe[0], &decoded[bytes_recv],
 213                                          statbuf.st_size);
 214                         if (blocksize < 0)
 215                                 lprintf(5, "ERROR: cannot read from pipe\n");
 216                         else
 217                                 bytes_recv = bytes_recv + blocksize;
 218                 }
 219                 /* Then put some data into the output pipe */
 220                 blocksize = length - bytes_sent;
 221                 if (blocksize > 2048)
 222                         blocksize = 2048;
 223                 if (write(sendpipe[1], &part_start[bytes_sent], blocksize) < 0) {
 224                         lprintf(5, "ERROR: cannot write to pipe: %s\n",
 225                                 strerror(errno));
 226                         write_error = 1;
 227                 }
 228                 bytes_sent = bytes_sent + blocksize;
 229         }
 230         close(sendpipe[1]);
 231         /* Empty the input pipe */
 232         while ((blocksize = read(recvpipe[0], &decoded[bytes_recv], 1)),
 233                (blocksize > 0)) {
 234                 bytes_recv = bytes_recv + blocksize;
 235         }
 236
 237         if (bytes_recv > 0) if (CallBack != NULL) {
 238                 CallBack(name, filename, fixed_partnum(partnum),
 239                         disposition, decoded,
 240                         content_type, bytes_recv, "binary", userdata);
 241         }
 242
 243         phree(decoded);
 244 }
 245
 246 /*
 247  * Break out the components of a multipart message
 248  * (This function expects to be fed HEADERS + CONTENT)
 249  * Note: NULL can be supplied as content_end; in this case, the message is
 250  * considered to have ended when the parser encounters a 0x00 byte.
 251  */
 252 void the_mime_parser(char *partnum,
 253                      char *content_start, char *content_end,
 254                      void (*CallBack)
 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 (*PreMultiPartCallBack)
 265                       (char *cbname,
 266                        char *cbfilename,
 267                        char *cbpartnum,
 268                        char *cbdisp,
 269                        void *cbcontent,
 270                        char *cbtype,
 271                        size_t cblength,
 272                        char *cbencoding,
 273                        void *cbuserdata),
 274                      void (*PostMultiPartCallBack)
 275                       (char *cbname,
 276                        char *cbfilename,
 277                        char *cbpartnum,
 278                        char *cbdisp,
 279                        void *cbcontent,
 280                        char *cbtype,
 281                        size_t cblength,
 282                        char *cbencoding,
 283                        void *cbuserdata),
 284                       void *userdata,
 285                       int dont_decode
 286 )
 287 {
 288
 289         char *ptr;
 290         char *part_start, *part_end;
 291         char buf[SIZ];
 292         char header[SIZ];
 293         char boundary[SIZ];
 294         char startary[SIZ];
 295         char endary[SIZ];
 296         char content_type[SIZ];
 297         char encoding[SIZ];
 298         char disposition[SIZ];
 299         char name[SIZ];
 300         char filename[SIZ];
 301         int is_multipart;
 302         int part_seq = 0;
 303         int i;
 304         size_t length;
 305         char nested_partnum[SIZ];
 306
 307         lprintf(9, "the_mime_parser() called\n");
 308         ptr = content_start;
 309         memset(boundary, 0, sizeof boundary);
 310         memset(content_type, 0, sizeof content_type);
 311         memset(encoding, 0, sizeof encoding);
 312         memset(name, 0, sizeof name);
 313         memset(filename, 0, sizeof filename);
 314         memset(disposition, 0, sizeof disposition);
 315
 316         /* Learn interesting things from the headers */
 317         strcpy(header, "");
 318         do {
 319                 ptr = memreadline(ptr, buf, sizeof buf);
 320                 if (*ptr == 0)
 321                         return; /* premature end of message */
 322                 if (content_end != NULL)
 323                         if (ptr >= content_end)
 324                                 return;
 325
 326                 for (i = 0; i < strlen(buf); ++i)
 327                         if (isspace(buf[i]))
 328                                 buf[i] = ' ';
 329                 if (!isspace(buf[0])) {
 330                         if (!strncasecmp(header, "Content-type: ", 14)) {
 331                                 strcpy(content_type, &header[14]);
 332                                 extract_key(name, content_type, "name");
 333                         }
 334                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 335                                 strcpy(disposition, &header[21]);
 336                                 extract_key(filename, disposition, "filename");
 337                         }
 338                         if (!strncasecmp(header,
 339                                       "Content-transfer-encoding: ", 27))
 340                                 strcpy(encoding, &header[27]);
 341                         if (strlen(boundary) == 0)
 342                                 extract_key(boundary, header, "boundary");
 343                         strcpy(header, "");
 344                 }
 345                 if ((strlen(header) + strlen(buf) + 2) < sizeof(header))
 346                         strcat(header, buf);
 347         } while ((strlen(buf) > 0) && (*ptr != 0));
 348
 349         for (i = 0; i < strlen(disposition); ++i)
 350                 if (disposition[i] == ';')
 351                         disposition[i] = 0;
 352         while (isspace(disposition[0]))
 353                 strcpy(disposition, &disposition[1]);
 354         for (i = 0; i < strlen(content_type); ++i)
 355                 if (content_type[i] == ';')
 356                         content_type[i] = 0;
 357         while (isspace(content_type[0]))
 358                 strcpy(content_type, &content_type[1]);
 359
 360         if (strlen(boundary) > 0) {
 361                 is_multipart = 1;
 362         } else {
 363                 is_multipart = 0;
 364         }
 365
 366         /* If this is a multipart message, then recursively process it */
 367         part_start = NULL;
 368         if (is_multipart) {
 369
 370                 /* Tell the client about this message's multipartedness */
 371                 if (PreMultiPartCallBack != NULL) {
 372                         PreMultiPartCallBack("", "", partnum, "",
 373                                 NULL, content_type,
 374                                 0, encoding, userdata);
 375                 }
 376                 /*
 377                 if (CallBack != NULL) {
 378                         CallBack("", "", fixed_partnum(partnum),
 379                                 "", NULL, content_type,
 380                                 0, encoding, userdata);
 381                 }
 382                  */
 383
 384                 /* Figure out where the boundaries are */
 385                 sprintf(startary, "--%s", boundary);
 386                 sprintf(endary, "--%s--", boundary);
 387                 do {
 388                         part_end = ptr;
 389                         ptr = memreadline(ptr, buf, sizeof buf);
 390                         if (*ptr == 0) goto END_MULTI;  /* premature end */
 391                         if (content_end != NULL)
 392                                 if (ptr >= content_end) goto END_MULTI;
 393                         if ((!strcasecmp(buf, startary))
 394                             || (!strcasecmp(buf, endary))) {
 395                                 if (part_start != NULL) {
 396                                         if (strlen(partnum) > 0) {
 397                                                 sprintf(nested_partnum, "%s.%d",
 398                                                         partnum, ++part_seq);
 399                                         }
 400                                         else {
 401                                                 sprintf(nested_partnum, "%d",
 402                                                         ++part_seq);
 403                                         }
 404                                         the_mime_parser(nested_partnum,
 405                                                     part_start, part_end,
 406                                                         CallBack,
 407                                                         PreMultiPartCallBack,
 408                                                         PostMultiPartCallBack,
 409                                                         userdata,
 410                                                         dont_decode);
 411                                 }
 412                                 part_start = ptr;
 413                         }
 414                 } while (strcasecmp(buf, endary));
 415 END_MULTI:      if (PostMultiPartCallBack != NULL) {
 416                         PostMultiPartCallBack("", "", partnum, "", NULL,
 417                                 content_type, 0, encoding, userdata);
 418                 }
 419                 return;
 420         }
 421
 422         /* If it's not a multipart message, then do something with it */
 423         if (!is_multipart) {
 424                 part_start = ptr;
 425                 length = 0;
 426                 while ((*ptr != 0)
 427                       && ((content_end == NULL) || (ptr < content_end))) {
 428                         ++length;
 429                         part_end = ptr++;
 430                 }
 431                 mime_decode(partnum,
 432                             part_start, length,
 433                             content_type, encoding, disposition,
 434                             name, filename,
 435                             CallBack, NULL, NULL,
 436                             userdata, dont_decode);
 437         }
 438
 439
 440 }
 441
 442
 443
 444 /*
 445  * Entry point for the MIME parser.
 446  * (This function expects to be fed HEADERS + CONTENT)
 447  * Note: NULL can be supplied as content_end; in this case, the message is
 448  * considered to have ended when the parser encounters a 0x00 byte.
 449  */
 450 void mime_parser(char *content_start,
 451                 char *content_end,
 452
 453                  void (*CallBack)
 454                   (char *cbname,
 455                    char *cbfilename,
 456                    char *cbpartnum,
 457                    char *cbdisp,
 458                    void *cbcontent,
 459                    char *cbtype,
 460                    size_t cblength,
 461                    char *cbencoding,
 462                    void *cbuserdata),
 463
 464                  void (*PreMultiPartCallBack)
 465                   (char *cbname,
 466                    char *cbfilename,
 467                    char *cbpartnum,
 468                    char *cbdisp,
 469                    void *cbcontent,
 470                    char *cbtype,
 471                    size_t cblength,
 472                    char *cbencoding,
 473                    void *cbuserdata),
 474
 475                  void (*PostMultiPartCallBack)
 476                   (char *cbname,
 477                    char *cbfilename,
 478                    char *cbpartnum,
 479                    char *cbdisp,
 480                    void *cbcontent,
 481                    char *cbtype,
 482                    size_t cblength,
 483                    char *cbencoding,
 484                    void *cbuserdata),
 485
 486                   void *userdata,
 487                   int dont_decode
 488 )
 489 {
 490
 491         lprintf(9, "mime_parser() called\n");
 492         the_mime_parser("", content_start, content_end,
 493                         CallBack,
 494                         PreMultiPartCallBack,
 495                         PostMultiPartCallBack,
 496                         userdata, dont_decode);
 497 }