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                 if (CallBack != NULL) {
 145                         CallBack(name, filename, partnum,
 146                                 disposition, part_start,
 147                                 content_type, length, encoding, userdata);
 148                         }
 149                 return;
 150         }
 151         if ((strcasecmp(encoding, "base64"))
 152             && (strcasecmp(encoding, "quoted-printable"))) {
 153                 lprintf(5, "ERROR: unknown MIME encoding '%s'\n", encoding);
 154                 return;
 155         }
 156         /*
 157          * Allocate a buffer for the decoded data.  The output buffer is the
 158          * same size as the input buffer; this assumes that the decoded data
 159          * will never be larger than the encoded data.  This is a safe
 160          * assumption with base64, uuencode, and quoted-printable.  Just to
 161          * be safe, we still pad the buffer a bit.
 162          */
 163         decoded = mallok(length + 1024);
 164         if (decoded == NULL) {
 165                 lprintf(5, "ERROR: cannot allocate memory.\n");
 166                 return;
 167         }
 168         if (pipe(sendpipe) != 0)
 169                 return;
 170         if (pipe(recvpipe) != 0)
 171                 return;
 172
 173         childpid = fork();
 174         if (childpid < 0) {
 175                 phree(decoded);
 176                 return;
 177         }
 178         if (childpid == 0) {
 179                 close(2);
 180                 /* send stdio to the pipes */
 181                 if (dup2(sendpipe[0], 0) < 0)
 182                         lprintf(5, "ERROR dup2()\n");
 183                 if (dup2(recvpipe[1], 1) < 0)
 184                         lprintf(5, "ERROR dup2()\n");
 185                 close(sendpipe[1]);     /* Close the ends we're not using */
 186                 close(recvpipe[0]);
 187                 if (!strcasecmp(encoding, "base64"))
 188                         execlp("./base64", "base64", "-d", NULL);
 189                 else if (!strcasecmp(encoding, "quoted-printable"))
 190                         execlp("./qpdecode", "qpdecode", NULL);
 191                 lprintf(5, "ERROR: cannot exec decoder for %s\n", encoding);
 192                 exit(1);
 193         }
 194         close(sendpipe[0]);     /* Close the ends we're not using  */
 195         close(recvpipe[1]);
 196
 197         while ((bytes_sent < length) && (write_error == 0)) {
 198                 /* Empty the input pipe FIRST */
 199                 while (fstat(recvpipe[0], &statbuf), (statbuf.st_size > 0)) {
 200                         blocksize = read(recvpipe[0], &decoded[bytes_recv],
 201                                          statbuf.st_size);
 202                         if (blocksize < 0)
 203                                 lprintf(5, "ERROR: cannot read from pipe\n");
 204                         else
 205                                 bytes_recv = bytes_recv + blocksize;
 206                 }
 207                 /* Then put some data into the output pipe */
 208                 blocksize = length - bytes_sent;
 209                 if (blocksize > 2048)
 210                         blocksize = 2048;
 211                 if (write(sendpipe[1], &part_start[bytes_sent], blocksize) < 0) {
 212                         lprintf(5, "ERROR: cannot write to pipe: %s\n",
 213                                 strerror(errno));
 214                         write_error = 1;
 215                 }
 216                 bytes_sent = bytes_sent + blocksize;
 217         }
 218         close(sendpipe[1]);
 219         /* Empty the input pipe */
 220         while ((blocksize = read(recvpipe[0], &decoded[bytes_recv], 1)),
 221                (blocksize > 0)) {
 222                 bytes_recv = bytes_recv + blocksize;
 223         }
 224
 225         if (bytes_recv > 0) if (CallBack != NULL) {
 226                 CallBack(name, filename, partnum, disposition, decoded,
 227                          content_type, bytes_recv, "binary", userdata);
 228         }
 229
 230         phree(decoded);
 231 }
 232
 233 /*
 234  * Break out the components of a multipart message
 235  * (This function expects to be fed HEADERS + CONTENT)
 236  * Note: NULL can be supplied as content_end; in this case, the message is
 237  * considered to have ended when the parser encounters a 0x00 byte.
 238  */
 239 void the_mime_parser(char *partnum,
 240                      char *content_start, char *content_end,
 241                      void (*CallBack)
 242                       (char *cbname,
 243                        char *cbfilename,
 244                        char *cbpartnum,
 245                        char *cbdisp,
 246                        void *cbcontent,
 247                        char *cbtype,
 248                        size_t cblength,
 249                        char *cbencoding,
 250                        void *cbuserdata),
 251                      void (*PreMultiPartCallBack)
 252                       (char *cbname,
 253                        char *cbfilename,
 254                        char *cbpartnum,
 255                        char *cbdisp,
 256                        void *cbcontent,
 257                        char *cbtype,
 258                        size_t cblength,
 259                        char *cbencoding,
 260                        void *cbuserdata),
 261                      void (*PostMultiPartCallBack)
 262                       (char *cbname,
 263                        char *cbfilename,
 264                        char *cbpartnum,
 265                        char *cbdisp,
 266                        void *cbcontent,
 267                        char *cbtype,
 268                        size_t cblength,
 269                        char *cbencoding,
 270                        void *cbuserdata),
 271                       void *userdata,
 272                       int dont_decode
 273 )
 274 {
 275
 276         char *ptr;
 277         char *part_start, *part_end;
 278         char buf[SIZ];
 279         char header[SIZ];
 280         char boundary[SIZ];
 281         char startary[SIZ];
 282         char endary[SIZ];
 283         char content_type[SIZ];
 284         char encoding[SIZ];
 285         char disposition[SIZ];
 286         char name[SIZ];
 287         char filename[SIZ];
 288         int is_multipart;
 289         int part_seq = 0;
 290         int i;
 291         size_t length;
 292         char nested_partnum[SIZ];
 293
 294         lprintf(9, "the_mime_parser() called\n");
 295         ptr = content_start;
 296         memset(boundary, 0, sizeof boundary);
 297         memset(content_type, 0, sizeof content_type);
 298         memset(encoding, 0, sizeof encoding);
 299         memset(name, 0, sizeof name);
 300         memset(filename, 0, sizeof filename);
 301         memset(disposition, 0, sizeof disposition);
 302
 303         /* Learn interesting things from the headers */
 304         strcpy(header, "");
 305         do {
 306                 ptr = memreadline(ptr, buf, sizeof buf);
 307                 if (*ptr == 0)
 308                         return; /* premature end of message */
 309                 if (content_end != NULL)
 310                         if (ptr >= content_end)
 311                                 return;
 312
 313                 for (i = 0; i < strlen(buf); ++i)
 314                         if (isspace(buf[i]))
 315                                 buf[i] = ' ';
 316                 if (!isspace(buf[0])) {
 317                         if (!strncasecmp(header, "Content-type: ", 14)) {
 318                                 strcpy(content_type, &header[14]);
 319                                 extract_key(name, content_type, "name");
 320                         }
 321                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 322                                 strcpy(disposition, &header[21]);
 323                                 extract_key(filename, disposition, "filename");
 324                         }
 325                         if (!strncasecmp(header,
 326                                       "Content-transfer-encoding: ", 27))
 327                                 strcpy(encoding, &header[27]);
 328                         if (strlen(boundary) == 0)
 329                                 extract_key(boundary, header, "boundary");
 330                         strcpy(header, "");
 331                 }
 332                 if ((strlen(header) + strlen(buf) + 2) < sizeof(header))
 333                         strcat(header, buf);
 334         } while ((strlen(buf) > 0) && (*ptr != 0));
 335
 336         for (i = 0; i < strlen(disposition); ++i)
 337                 if (disposition[i] == ';')
 338                         disposition[i] = 0;
 339         while (isspace(disposition[0]))
 340                 strcpy(disposition, &disposition[1]);
 341         for (i = 0; i < strlen(content_type); ++i)
 342                 if (content_type[i] == ';')
 343                         content_type[i] = 0;
 344         while (isspace(content_type[0]))
 345                 strcpy(content_type, &content_type[1]);
 346
 347         if (strlen(boundary) > 0) {
 348                 is_multipart = 1;
 349         } else {
 350                 is_multipart = 0;
 351         }
 352
 353         /* If this is a multipart message, then recursively process it */
 354         part_start = NULL;
 355         if (is_multipart) {
 356
 357                 /* Tell the client about this message's multipartedness */
 358                 if (PreMultiPartCallBack != NULL) {
 359                         PreMultiPartCallBack("", "", partnum, "",
 360                                 NULL, content_type,
 361                                 0, encoding, userdata);
 362                 }
 363                 /*
 364                 if (CallBack != NULL) {
 365                         CallBack("", "", partnum, "", NULL, content_type,
 366                                 0, encoding, userdata);
 367                 }
 368                  */
 369
 370                 /* Figure out where the boundaries are */
 371                 sprintf(startary, "--%s", boundary);
 372                 sprintf(endary, "--%s--", boundary);
 373                 do {
 374                         part_end = ptr;
 375                         ptr = memreadline(ptr, buf, sizeof buf);
 376                         if (*ptr == 0) goto END_MULTI;  /* premature end */
 377                         if (content_end != NULL)
 378                                 if (ptr >= content_end) goto END_MULTI;
 379                         if ((!strcasecmp(buf, startary))
 380                             || (!strcasecmp(buf, endary))) {
 381                                 if (part_start != NULL) {
 382                                         if (strlen(partnum) > 0) {
 383                                                 sprintf(nested_partnum, "%s.%d",
 384                                                         partnum, ++part_seq);
 385                                         }
 386                                         else {
 387                                                 sprintf(nested_partnum, "%d",
 388                                                         ++part_seq);
 389                                         }
 390                                         the_mime_parser(nested_partnum,
 391                                                     part_start, part_end,
 392                                                         CallBack,
 393                                                         PreMultiPartCallBack,
 394                                                         PostMultiPartCallBack,
 395                                                         userdata,
 396                                                         dont_decode);
 397                                 }
 398                                 part_start = ptr;
 399                         }
 400                 } while (strcasecmp(buf, endary));
 401 END_MULTI:      if (PostMultiPartCallBack != NULL) {
 402                         PostMultiPartCallBack("", "", partnum, "", NULL,
 403                                 content_type, 0, encoding, userdata);
 404                 }
 405                 return;
 406         }
 407
 408         /* If it's not a multipart message, then do something with it */
 409         if (!is_multipart) {
 410                 part_start = ptr;
 411                 length = 0;
 412                 while ((*ptr != 0)
 413                       && ((content_end == NULL) || (ptr < content_end))) {
 414                         ++length;
 415                         part_end = ptr++;
 416                 }
 417                 mime_decode(partnum,
 418                             part_start, length,
 419                             content_type, encoding, disposition,
 420                             name, filename,
 421                             CallBack, NULL, NULL,
 422                             userdata, dont_decode);
 423         }
 424
 425
 426 }
 427
 428
 429
 430 /*
 431  * Entry point for the MIME parser.
 432  * (This function expects to be fed HEADERS + CONTENT)
 433  * Note: NULL can be supplied as content_end; in this case, the message is
 434  * considered to have ended when the parser encounters a 0x00 byte.
 435  */
 436 void mime_parser(char *content_start,
 437                 char *content_end,
 438
 439                  void (*CallBack)
 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 (*PreMultiPartCallBack)
 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 (*PostMultiPartCallBack)
 462                   (char *cbname,
 463                    char *cbfilename,
 464                    char *cbpartnum,
 465                    char *cbdisp,
 466                    void *cbcontent,
 467                    char *cbtype,
 468                    size_t cblength,
 469                    char *cbencoding,
 470                    void *cbuserdata),
 471
 472                   void *userdata,
 473                   int dont_decode
 474 )
 475 {
 476
 477         lprintf(9, "mime_parser() called\n");
 478         the_mime_parser("", content_start, content_end,
 479                         CallBack,
 480                         PreMultiPartCallBack,
 481                         PostMultiPartCallBack,
 482                         userdata, dont_decode);
 483 }