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  * Given a message or message-part body and a length, handle any necessary
  80  * decoding and pass the request up the stack.
  81  */
  82 void mime_decode(char *partnum,
  83                  char *part_start, size_t length,
  84                  char *content_type, char *encoding,
  85                  char *disposition,
  86                  char *name, char *filename,
  87                  void (*CallBack)
  88                   (char *cbname,
  89                    char *cbfilename,
  90                    char *cbpartnum,
  91                    char *cbdisp,
  92                    void *cbcontent,
  93                    char *cbtype,
  94                    size_t cblength,
  95                    char *cbencoding,
  96                    void *cbuserdata),
  97                  void (*PreMultiPartCallBack)
  98                   (char *cbname,
  99                    char *cbfilename,
 100                    char *cbpartnum,
 101                    char *cbdisp,
 102                    void *cbcontent,
 103                    char *cbtype,
 104                    size_t cblength,
 105                    char *cbencoding,
 106                    void *cbuserdata),
 107                  void (*PostMultiPartCallBack)
 108                   (char *cbname,
 109                    char *cbfilename,
 110                    char *cbpartnum,
 111                    char *cbdisp,
 112                    void *cbcontent,
 113                    char *cbtype,
 114                    size_t cblength,
 115                    char *cbencoding,
 116                    void *cbuserdata),
 117                   void *userdata,
 118                   int dont_decode
 119 )
 120 {
 121
 122         char *decoded;
 123         struct stat statbuf;
 124         int sendpipe[2];
 125         int recvpipe[2];
 126         int childpid;
 127         size_t bytes_sent = 0;
 128         size_t bytes_recv = 0;
 129         size_t blocksize;
 130         int write_error = 0;
 131
 132         lprintf(9, "mime_decode() called\n");
 133
 134         /* Some encodings aren't really encodings */
 135         if (!strcasecmp(encoding, "7bit"))
 136                 strcpy(encoding, "");
 137         if (!strcasecmp(encoding, "8bit"))
 138                 strcpy(encoding, "");
 139         if (!strcasecmp(encoding, "binary"))
 140                 strcpy(encoding, "");
 141
 142         /* If this part is not encoded, send as-is */
 143         if ( (strlen(encoding) == 0) || (dont_decode)) {
 144                 CallBack(name, filename, partnum, disposition, part_start,
 145                          content_type, length, encoding, userdata);
 146                 return;
 147         }
 148         if ((strcasecmp(encoding, "base64"))
 149             && (strcasecmp(encoding, "quoted-printable"))) {
 150                 lprintf(5, "ERROR: unknown MIME encoding '%s'\n", encoding);
 151                 return;
 152         }
 153         /*
 154          * Allocate a buffer for the decoded data.  The output buffer is the
 155          * same size as the input buffer; this assumes that the decoded data
 156          * will never be larger than the encoded data.  This is a safe
 157          * assumption with base64, uuencode, and quoted-printable.  Just to
 158          * be safe, we still pad the buffer a bit.
 159          */
 160         decoded = mallok(length + 1024);
 161         if (decoded == NULL) {
 162                 lprintf(5, "ERROR: cannot allocate memory.\n");
 163                 return;
 164         }
 165         if (pipe(sendpipe) != 0)
 166                 return;
 167         if (pipe(recvpipe) != 0)
 168                 return;
 169
 170         childpid = fork();
 171         if (childpid < 0) {
 172                 phree(decoded);
 173                 return;
 174         }
 175         if (childpid == 0) {
 176                 close(2);
 177                 /* send stdio to the pipes */
 178                 if (dup2(sendpipe[0], 0) < 0)
 179                         lprintf(5, "ERROR dup2()\n");
 180                 if (dup2(recvpipe[1], 1) < 0)
 181                         lprintf(5, "ERROR dup2()\n");
 182                 close(sendpipe[1]);     /* Close the ends we're not using */
 183                 close(recvpipe[0]);
 184                 if (!strcasecmp(encoding, "base64"))
 185                         execlp("./base64", "base64", "-d", NULL);
 186                 else if (!strcasecmp(encoding, "quoted-printable"))
 187                         execlp("./qpdecode", "qpdecode", NULL);
 188                 lprintf(5, "ERROR: cannot exec decoder for %s\n", encoding);
 189                 exit(1);
 190         }
 191         close(sendpipe[0]);     /* Close the ends we're not using  */
 192         close(recvpipe[1]);
 193
 194         while ((bytes_sent < length) && (write_error == 0)) {
 195                 /* Empty the input pipe FIRST */
 196                 while (fstat(recvpipe[0], &statbuf), (statbuf.st_size > 0)) {
 197                         blocksize = read(recvpipe[0], &decoded[bytes_recv],
 198                                          statbuf.st_size);
 199                         if (blocksize < 0)
 200                                 lprintf(5, "ERROR: cannot read from pipe\n");
 201                         else
 202                                 bytes_recv = bytes_recv + blocksize;
 203                 }
 204                 /* Then put some data into the output pipe */
 205                 blocksize = length - bytes_sent;
 206                 if (blocksize > 2048)
 207                         blocksize = 2048;
 208                 if (write(sendpipe[1], &part_start[bytes_sent], blocksize) < 0) {
 209                         lprintf(5, "ERROR: cannot write to pipe: %s\n",
 210                                 strerror(errno));
 211                         write_error = 1;
 212                 }
 213                 bytes_sent = bytes_sent + blocksize;
 214         }
 215         close(sendpipe[1]);
 216         /* Empty the input pipe */
 217         while ((blocksize = read(recvpipe[0], &decoded[bytes_recv], 1)),
 218                (blocksize > 0)) {
 219                 bytes_recv = bytes_recv + blocksize;
 220         }
 221
 222         if (bytes_recv > 0)
 223                 CallBack(name, filename, partnum, disposition, decoded,
 224                          content_type, bytes_recv, "binary", userdata);
 225
 226         phree(decoded);
 227 }
 228
 229 /*
 230  * Break out the components of a multipart message
 231  * (This function expects to be fed HEADERS + CONTENT)
 232  * Note: NULL can be supplied as content_end; in this case, the message is
 233  * considered to have ended when the parser encounters a 0x00 byte.
 234  */
 235 void the_mime_parser(char *partnum,
 236                      char *content_start, char *content_end,
 237                      void (*CallBack)
 238                       (char *cbname,
 239                        char *cbfilename,
 240                        char *cbpartnum,
 241                        char *cbdisp,
 242                        void *cbcontent,
 243                        char *cbtype,
 244                        size_t cblength,
 245                        char *cbencoding,
 246                        void *cbuserdata),
 247                      void (*PreMultiPartCallBack)
 248                       (char *cbname,
 249                        char *cbfilename,
 250                        char *cbpartnum,
 251                        char *cbdisp,
 252                        void *cbcontent,
 253                        char *cbtype,
 254                        size_t cblength,
 255                        char *cbencoding,
 256                        void *cbuserdata),
 257                      void (*PostMultiPartCallBack)
 258                       (char *cbname,
 259                        char *cbfilename,
 260                        char *cbpartnum,
 261                        char *cbdisp,
 262                        void *cbcontent,
 263                        char *cbtype,
 264                        size_t cblength,
 265                        char *cbencoding,
 266                        void *cbuserdata),
 267                       void *userdata,
 268                       int dont_decode
 269 )
 270 {
 271
 272         char *ptr;
 273         char *part_start, *part_end;
 274         char buf[SIZ];
 275         char header[SIZ];
 276         char boundary[SIZ];
 277         char startary[SIZ];
 278         char endary[SIZ];
 279         char content_type[SIZ];
 280         char encoding[SIZ];
 281         char disposition[SIZ];
 282         char name[SIZ];
 283         char filename[SIZ];
 284         int is_multipart;
 285         int part_seq = 0;
 286         int i;
 287         size_t length;
 288         char nested_partnum[SIZ];
 289
 290         lprintf(9, "the_mime_parser() called\n");
 291         ptr = content_start;
 292         memset(boundary, 0, sizeof boundary);
 293         memset(content_type, 0, sizeof content_type);
 294         memset(encoding, 0, sizeof encoding);
 295         memset(name, 0, sizeof name);
 296         memset(filename, 0, sizeof filename);
 297         memset(disposition, 0, sizeof disposition);
 298
 299         /* Learn interesting things from the headers */
 300         strcpy(header, "");
 301         do {
 302                 ptr = memreadline(ptr, buf, sizeof buf);
 303                 if (*ptr == 0)
 304                         return; /* premature end of message */
 305                 if (content_end != NULL)
 306                         if (ptr >= content_end)
 307                                 return;
 308
 309                 for (i = 0; i < strlen(buf); ++i)
 310                         if (isspace(buf[i]))
 311                                 buf[i] = ' ';
 312                 if (!isspace(buf[0])) {
 313                         if (!strncasecmp(header, "Content-type: ", 14)) {
 314                                 strcpy(content_type, &header[14]);
 315                                 extract_key(name, content_type, "name");
 316                         }
 317                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 318                                 strcpy(disposition, &header[21]);
 319                                 extract_key(filename, disposition, "filename");
 320                         }
 321                         if (!strncasecmp(header,
 322                                       "Content-transfer-encoding: ", 27))
 323                                 strcpy(encoding, &header[27]);
 324                         if (strlen(boundary) == 0)
 325                                 extract_key(boundary, header, "boundary");
 326                         strcpy(header, "");
 327                 }
 328                 if ((strlen(header) + strlen(buf) + 2) < sizeof(header))
 329                         strcat(header, buf);
 330         } while ((strlen(buf) > 0) && (*ptr != 0));
 331
 332         for (i = 0; i < strlen(disposition); ++i)
 333                 if (disposition[i] == ';')
 334                         disposition[i] = 0;
 335         while (isspace(disposition[0]))
 336                 strcpy(disposition, &disposition[1]);
 337         for (i = 0; i < strlen(content_type); ++i)
 338                 if (content_type[i] == ';')
 339                         content_type[i] = 0;
 340         while (isspace(content_type[0]))
 341                 strcpy(content_type, &content_type[1]);
 342
 343         if (strlen(boundary) > 0) {
 344                 is_multipart = 1;
 345         } else {
 346                 is_multipart = 0;
 347         }
 348
 349         /* If this is a multipart message, then recursively process it */
 350         part_start = NULL;
 351         if (is_multipart) {
 352
 353                 /* Tell the client about this message's multipartedness */
 354                 PreMultiPartCallBack("", "", partnum, "", NULL, content_type,
 355                                 0, encoding, userdata);
 356                 /*
 357                 CallBack("", "", partnum, "", NULL, content_type,
 358                                 0, encoding, userdata);
 359                  */
 360
 361                 /* Figure out where the boundaries are */
 362                 sprintf(startary, "--%s", boundary);
 363                 sprintf(endary, "--%s--", boundary);
 364                 do {
 365                         part_end = ptr;
 366                         ptr = memreadline(ptr, buf, sizeof buf);
 367                         if (*ptr == 0) goto END_MULTI;  /* premature end */
 368                         if (content_end != NULL)
 369                                 if (ptr >= content_end) goto END_MULTI;
 370                         if ((!strcasecmp(buf, startary))
 371                             || (!strcasecmp(buf, endary))) {
 372                                 if (part_start != NULL) {
 373                                         if (strlen(partnum) > 0) {
 374                                                 sprintf(nested_partnum, "%s.%d",
 375                                                         partnum, ++part_seq);
 376                                         }
 377                                         else {
 378                                                 sprintf(nested_partnum, "%d",
 379                                                         ++part_seq);
 380                                         }
 381                                         the_mime_parser(nested_partnum,
 382                                                     part_start, part_end,
 383                                                         CallBack,
 384                                                         PreMultiPartCallBack,
 385                                                         PostMultiPartCallBack,
 386                                                         userdata,
 387                                                         dont_decode);
 388                                 }
 389                                 part_start = ptr;
 390                         }
 391                 } while (strcasecmp(buf, endary));
 392 END_MULTI:      PostMultiPartCallBack("", "", partnum, "", NULL, content_type,
 393                                 0, encoding, userdata);
 394                 return;
 395         }
 396
 397         /* If it's not a multipart message, then do something with it */
 398         if (!is_multipart) {
 399                 part_start = ptr;
 400                 length = 0;
 401                 while ((*ptr != 0)
 402                       && ((content_end == NULL) || (ptr < content_end))) {
 403                         ++length;
 404                         part_end = ptr++;
 405                 }
 406                 mime_decode(partnum,
 407                             part_start, length,
 408                             content_type, encoding, disposition,
 409                             name, filename,
 410                             CallBack, NULL, NULL,
 411                             userdata, dont_decode);
 412         }
 413
 414
 415 }
 416
 417
 418
 419 /*
 420  * Entry point for the MIME parser.
 421  * (This function expects to be fed HEADERS + CONTENT)
 422  * Note: NULL can be supplied as content_end; in this case, the message is
 423  * considered to have ended when the parser encounters a 0x00 byte.
 424  */
 425 void mime_parser(char *content_start,
 426                 char *content_end,
 427
 428                  void (*CallBack)
 429                   (char *cbname,
 430                    char *cbfilename,
 431                    char *cbpartnum,
 432                    char *cbdisp,
 433                    void *cbcontent,
 434                    char *cbtype,
 435                    size_t cblength,
 436                    char *cbencoding,
 437                    void *cbuserdata),
 438
 439                  void (*PreMultiPartCallBack)
 440                   (char *cbname,
 441                    char *cbfilename,
 442                    char *cbpartnum,
 443                    char *cbdisp,
 444                    void *cbcontent,
 445                    char *cbtype,
 446                    size_t cblength,
 447                    char *cbencoding,
 448                    void *cbuserdata),
 449
 450                  void (*PostMultiPartCallBack)
 451                   (char *cbname,
 452                    char *cbfilename,
 453                    char *cbpartnum,
 454                    char *cbdisp,
 455                    void *cbcontent,
 456                    char *cbtype,
 457                    size_t cblength,
 458                    char *cbencoding,
 459                    void *cbuserdata),
 460
 461                   void *userdata,
 462                   int dont_decode
 463 )
 464 {
 465
 466         lprintf(9, "mime_parser() called\n");
 467         the_mime_parser("", content_start, content_end,
 468                         CallBack,
 469                         PreMultiPartCallBack,
 470                         PostMultiPartCallBack,
 471                         userdata, dont_decode);
 472 }