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  * Given a message or message-part body and a length, handle any necessary
  66  * decoding and pass the request up the stack.
  67  */
  68 void mime_decode(char *partnum,
  69                  char *part_start, size_t length,
  70                  char *content_type, char *encoding,
  71                  char *disposition,
  72                  char *name, char *filename,
  73                  void (*CallBack)
  74                   (char *cbname,
  75                    char *cbfilename,
  76                    char *cbpartnum,
  77                    char *cbdisp,
  78                    void *cbcontent,
  79                    char *cbtype,
  80                    size_t cblength,
  81                    char *cbencoding,
  82                    void *cbuserdata),
  83                  void (*PreMultiPartCallBack)
  84                   (char *cbname,
  85                    char *cbfilename,
  86                    char *cbpartnum,
  87                    char *cbdisp,
  88                    void *cbcontent,
  89                    char *cbtype,
  90                    size_t cblength,
  91                    char *cbencoding,
  92                    void *cbuserdata),
  93                  void (*PostMultiPartCallBack)
  94                   (char *cbname,
  95                    char *cbfilename,
  96                    char *cbpartnum,
  97                    char *cbdisp,
  98                    void *cbcontent,
  99                    char *cbtype,
 100                    size_t cblength,
 101                    char *cbencoding,
 102                    void *cbuserdata),
 103                   void *userdata,
 104                   int dont_decode
 105 )
 106 {
 107
 108         char *decoded;
 109         struct stat statbuf;
 110         int sendpipe[2];
 111         int recvpipe[2];
 112         int childpid;
 113         size_t bytes_sent = 0;
 114         size_t bytes_recv = 0;
 115         size_t blocksize;
 116         int write_error = 0;
 117
 118         lprintf(9, "mime_decode() called\n");
 119
 120         /* Some encodings aren't really encodings */
 121         if (!strcasecmp(encoding, "7bit"))
 122                 strcpy(encoding, "");
 123         if (!strcasecmp(encoding, "8bit"))
 124                 strcpy(encoding, "");
 125         if (!strcasecmp(encoding, "binary"))
 126                 strcpy(encoding, "");
 127
 128         /* If this part is not encoded, send as-is */
 129         if ( (strlen(encoding) == 0) || (dont_decode)) {
 130                 if (CallBack != NULL) {
 131                         CallBack(name, filename, fixed_partnum(partnum),
 132                                 disposition, part_start,
 133                                 content_type, length, encoding, userdata);
 134                         }
 135                 return;
 136         }
 137         if ((strcasecmp(encoding, "base64"))
 138             && (strcasecmp(encoding, "quoted-printable"))) {
 139                 lprintf(9, "ERROR: unknown MIME encoding '%s'\n", encoding);
 140                 return;
 141         }
 142         /*
 143          * Allocate a buffer for the decoded data.  The output buffer is the
 144          * same size as the input buffer; this assumes that the decoded data
 145          * will never be larger than the encoded data.  This is a safe
 146          * assumption with base64, uuencode, and quoted-printable.  Just to
 147          * be safe, we still pad the buffer a bit.
 148          */
 149         decoded = malloc(length + 1024);
 150         if (decoded == NULL) {
 151                 lprintf(9, "ERROR: cannot allocate memory.\n");
 152                 return;
 153         }
 154         if (pipe(sendpipe) != 0)
 155                 return;
 156         if (pipe(recvpipe) != 0)
 157                 return;
 158
 159         childpid = fork();
 160         if (childpid < 0) {
 161                 free(decoded);
 162                 return;
 163         }
 164         if (childpid == 0) {
 165                 close(2);
 166                 /* send stdio to the pipes */
 167                 if (dup2(sendpipe[0], 0) < 0)
 168                         lprintf(9, "ERROR dup2()\n");
 169                 if (dup2(recvpipe[1], 1) < 0)
 170                         lprintf(9, "ERROR dup2()\n");
 171                 close(sendpipe[1]);     /* Close the ends we're not using */
 172                 close(recvpipe[0]);
 173                 if (!strcasecmp(encoding, "base64"))
 174                         execlp("./base64", "base64", "-d", NULL);
 175                 else if (!strcasecmp(encoding, "quoted-printable"))
 176                         execlp("./qpdecode", "qpdecode", NULL);
 177                 lprintf(9, "ERROR: cannot exec decoder for %s\n", encoding);
 178                 exit(1);
 179         }
 180         close(sendpipe[0]);     /* Close the ends we're not using  */
 181         close(recvpipe[1]);
 182
 183         while ((bytes_sent < length) && (write_error == 0)) {
 184                 /* Empty the input pipe FIRST */
 185                 while (fstat(recvpipe[0], &statbuf), (statbuf.st_size > 0)) {
 186                         blocksize = read(recvpipe[0], &decoded[bytes_recv],
 187                                          statbuf.st_size);
 188                         if (blocksize < 0)
 189                                 lprintf(9, "ERROR: cannot read from pipe\n");
 190                         else
 191                                 bytes_recv = bytes_recv + blocksize;
 192                 }
 193                 /* Then put some data into the output pipe */
 194                 blocksize = length - bytes_sent;
 195                 if (blocksize > 2048)
 196                         blocksize = 2048;
 197                 if (write(sendpipe[1], &part_start[bytes_sent], blocksize) < 0) {
 198                         lprintf(9, "ERROR: cannot write to pipe: %s\n",
 199                                 strerror(errno));
 200                         write_error = 1;
 201                 }
 202                 bytes_sent = bytes_sent + blocksize;
 203         }
 204         close(sendpipe[1]);
 205         /* Empty the input pipe */
 206         while ((blocksize = read(recvpipe[0], &decoded[bytes_recv], 1)),
 207                (blocksize > 0)) {
 208                 bytes_recv = bytes_recv + blocksize;
 209         }
 210
 211         if (bytes_recv > 0) if (CallBack != NULL) {
 212                 CallBack(name, filename, fixed_partnum(partnum),
 213                         disposition, decoded,
 214                         content_type, bytes_recv, "binary", userdata);
 215         }
 216
 217         free(decoded);
 218 }
 219
 220 /*
 221  * Break out the components of a multipart message
 222  * (This function expects to be fed HEADERS + CONTENT)
 223  * Note: NULL can be supplied as content_end; in this case, the message is
 224  * considered to have ended when the parser encounters a 0x00 byte.
 225  */
 226 void the_mime_parser(char *partnum,
 227                      char *content_start, char *content_end,
 228                      void (*CallBack)
 229                       (char *cbname,
 230                        char *cbfilename,
 231                        char *cbpartnum,
 232                        char *cbdisp,
 233                        void *cbcontent,
 234                        char *cbtype,
 235                        size_t cblength,
 236                        char *cbencoding,
 237                        void *cbuserdata),
 238                      void (*PreMultiPartCallBack)
 239                       (char *cbname,
 240                        char *cbfilename,
 241                        char *cbpartnum,
 242                        char *cbdisp,
 243                        void *cbcontent,
 244                        char *cbtype,
 245                        size_t cblength,
 246                        char *cbencoding,
 247                        void *cbuserdata),
 248                      void (*PostMultiPartCallBack)
 249                       (char *cbname,
 250                        char *cbfilename,
 251                        char *cbpartnum,
 252                        char *cbdisp,
 253                        void *cbcontent,
 254                        char *cbtype,
 255                        size_t cblength,
 256                        char *cbencoding,
 257                        void *cbuserdata),
 258                       void *userdata,
 259                       int dont_decode
 260 )
 261 {
 262
 263         char *ptr;
 264         char *part_start, *part_end = NULL;
 265         char buf[SIZ];
 266         char header[SIZ];
 267         char boundary[SIZ];
 268         char startary[SIZ];
 269         char endary[SIZ];
 270         char content_type[SIZ];
 271         size_t content_length;
 272         char encoding[SIZ];
 273         char disposition[SIZ];
 274         char name[SIZ];
 275         char filename[SIZ];
 276         int is_multipart;
 277         int part_seq = 0;
 278         int i;
 279         size_t length;
 280         char nested_partnum[SIZ];
 281
 282         lprintf(9, "the_mime_parser() called\n");
 283         ptr = content_start;
 284         memset(boundary, 0, sizeof boundary);
 285         memset(content_type, 0, sizeof content_type);
 286         memset(encoding, 0, sizeof encoding);
 287         memset(name, 0, sizeof name);
 288         memset(filename, 0, sizeof filename);
 289         memset(disposition, 0, sizeof disposition);
 290         content_length = 0;
 291
 292         /* If the caller didn't supply an endpointer, generate one by measure */
 293         if (content_end == NULL) {
 294                 content_end = &content_start[strlen(content_start)];
 295         }
 296
 297         /* Learn interesting things from the headers */
 298         strcpy(header, "");
 299         do {
 300                 ptr = memreadline(ptr, buf, sizeof buf);
 301                 if (ptr >= content_end)
 302                         return;
 303
 304                 for (i = 0; i < strlen(buf); ++i)
 305                         if (isspace(buf[i]))
 306                                 buf[i] = ' ';
 307                 if (!isspace(buf[0])) {
 308                         if (!strncasecmp(header, "Content-type: ", 14)) {
 309                                 strcpy(content_type, &header[14]);
 310                                 extract_key(name, content_type, "name");
 311                                 lprintf(9, "Extracted content-type <%s>\n",
 312                                         content_type);
 313                         }
 314                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 315                                 strcpy(disposition, &header[21]);
 316                                 extract_key(filename, disposition, "filename");
 317                         }
 318                         if (!strncasecmp(header, "Content-length: ", 16)) {
 319                                 content_length = (size_t) atol(&header[16]);
 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         lprintf(9, "is_multipart=%d, boundary=<%s>\n",
 350                 is_multipart, boundary);
 351
 352         /* If this is a multipart message, then recursively process it */
 353         part_start = NULL;
 354         if (is_multipart) {
 355
 356                 /* Tell the client about this message's multipartedness */
 357                 if (PreMultiPartCallBack != NULL) {
 358                         PreMultiPartCallBack("", "", partnum, "",
 359                                 NULL, content_type,
 360                                 0, encoding, userdata);
 361                 }
 362
 363                 /* Figure out where the boundaries are */
 364                 sprintf(startary, "--%s", boundary);
 365                 sprintf(endary, "--%s--", boundary);
 366                 do {
 367                         if ( (!strncasecmp(ptr, startary, strlen(startary)))
 368                            || (!strncasecmp(ptr, endary, strlen(endary))) ) {
 369                                 lprintf(9, "hit boundary!\n");
 370                                 if (part_start != NULL) {
 371                                         if (strlen(partnum) > 0) {
 372                                                 sprintf(nested_partnum, "%s.%d",
 373                                                         partnum, ++part_seq);
 374                                         }
 375                                         else {
 376                                                 sprintf(nested_partnum, "%d",
 377                                                         ++part_seq);
 378                                         }
 379                                         the_mime_parser(nested_partnum,
 380                                                     part_start, part_end,
 381                                                         CallBack,
 382                                                         PreMultiPartCallBack,
 383                                                         PostMultiPartCallBack,
 384                                                         userdata,
 385                                                         dont_decode);
 386                                 }
 387                                 ptr = memreadline(ptr, buf, sizeof(buf));
 388                                 part_start = ptr;
 389                         }
 390                         else {
 391                                 part_end = ptr;
 392                                 ++ptr;
 393                         }
 394                 } while ( (strcasecmp(ptr, endary)) && (ptr <= content_end) );
 395                 if (PostMultiPartCallBack != NULL) {
 396                         PostMultiPartCallBack("", "", partnum, "", NULL,
 397                                 content_type, 0, encoding, userdata);
 398                 }
 399                 return;
 400         }
 401
 402         /* If it's not a multipart message, then do something with it */
 403         if (!is_multipart) {
 404                 lprintf(9, "doing non-multipart thing\n");
 405                 part_start = ptr;
 406                 length = 0;
 407                 while (ptr < content_end) {
 408                         ++ptr;
 409                         ++length;
 410                 }
 411                 part_end = content_end;
 412
 413                 /* Truncate if the header told us to */
 414                 if ( (content_length > 0) && (length > content_length) ) {
 415                         length = content_length;
 416                         lprintf(9, "truncated to %d\n", content_length);
 417                 }
 418
 419                 mime_decode(partnum,
 420                             part_start, length,
 421                             content_type, encoding, disposition,
 422                             name, filename,
 423                             CallBack, NULL, NULL,
 424                             userdata, dont_decode);
 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 }