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