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 #include <stdlib.h>
  12 #include <unistd.h>
  13 #include <stdio.h>
  14 #include <signal.h>
  15 #include <sys/types.h>
  16 #include <ctype.h>
  17 #include <string.h>
  18 #include <sys/stat.h>
  19 #include <errno.h>
  20 #include "citadel.h"
  21 #include "sysdep_decls.h"
  22 #include "mime_parser.h"
  23 #include "tools.h"
  24
  25
  26 void extract_key(char *target, char *source, char *key)
  27 {
  28         int a, b;
  29
  30         strcpy(target, source);
  31         for (a = 0; a < strlen(target); ++a) {
  32                 if ((!strncasecmp(&target[a], key, strlen(key)))
  33                     && (target[a + strlen(key)] == '=')) {
  34                         strcpy(target, &target[a + strlen(key) + 1]);
  35                         if (target[0] == 34)
  36                                 strcpy(target, &target[1]);
  37                         for (b = 0; b < strlen(target); ++b)
  38                                 if (target[b] == 34)
  39                                         target[b] = 0;
  40                         return;
  41                 }
  42         }
  43         strcpy(target, "");
  44 }
  45
  46
  47 /*
  48  * For non-multipart messages, we need to generate a quickie partnum of "1"
  49  * to return to callback functions.  Some callbacks demand it.
  50  */
  51 char *fixed_partnum(char *supplied_partnum) {
  52         if (supplied_partnum == NULL) return "1";
  53         if (strlen(supplied_partnum)==0) return "1";
  54         return supplied_partnum;
  55 }
  56
  57
  58 /*
  59  * Given a message or message-part body and a length, handle any necessary
  60  * decoding and pass the request up the stack.
  61  */
  62 void mime_decode(char *partnum,
  63                  char *part_start, size_t length,
  64                  char *content_type, char *encoding,
  65                  char *disposition,
  66                  char *name, char *filename,
  67                  void (*CallBack)
  68                   (char *cbname,
  69                    char *cbfilename,
  70                    char *cbpartnum,
  71                    char *cbdisp,
  72                    void *cbcontent,
  73                    char *cbtype,
  74                    size_t cblength,
  75                    char *cbencoding,
  76                    void *cbuserdata),
  77                  void (*PreMultiPartCallBack)
  78                   (char *cbname,
  79                    char *cbfilename,
  80                    char *cbpartnum,
  81                    char *cbdisp,
  82                    void *cbcontent,
  83                    char *cbtype,
  84                    size_t cblength,
  85                    char *cbencoding,
  86                    void *cbuserdata),
  87                  void (*PostMultiPartCallBack)
  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 *userdata,
  98                   int dont_decode
  99 )
 100 {
 101
 102         char *decoded;
 103         struct stat statbuf;
 104         int sendpipe[2];
 105         int recvpipe[2];
 106         int childpid;
 107         size_t bytes_sent = 0;
 108         size_t bytes_recv = 0;
 109         size_t blocksize;
 110         int write_error = 0;
 111
 112         lprintf(9, "mime_decode() called\n");
 113
 114         /* Some encodings aren't really encodings */
 115         if (!strcasecmp(encoding, "7bit"))
 116                 strcpy(encoding, "");
 117         if (!strcasecmp(encoding, "8bit"))
 118                 strcpy(encoding, "");
 119         if (!strcasecmp(encoding, "binary"))
 120                 strcpy(encoding, "");
 121
 122         /* If this part is not encoded, send as-is */
 123         if ( (strlen(encoding) == 0) || (dont_decode)) {
 124                 if (CallBack != NULL) {
 125                         CallBack(name, filename, fixed_partnum(partnum),
 126                                 disposition, part_start,
 127                                 content_type, length, encoding, userdata);
 128                         }
 129                 return;
 130         }
 131         if ((strcasecmp(encoding, "base64"))
 132             && (strcasecmp(encoding, "quoted-printable"))) {
 133                 lprintf(9, "ERROR: unknown MIME encoding '%s'\n", encoding);
 134                 return;
 135         }
 136         /*
 137          * Allocate a buffer for the decoded data.  The output buffer is the
 138          * same size as the input buffer; this assumes that the decoded data
 139          * will never be larger than the encoded data.  This is a safe
 140          * assumption with base64, uuencode, and quoted-printable.  Just to
 141          * be safe, we still pad the buffer a bit.
 142          */
 143         decoded = malloc(length + 1024);
 144         if (decoded == NULL) {
 145                 lprintf(9, "ERROR: cannot allocate memory.\n");
 146                 return;
 147         }
 148         if (pipe(sendpipe) != 0)
 149                 return;
 150         if (pipe(recvpipe) != 0)
 151                 return;
 152
 153         childpid = fork();
 154         if (childpid < 0) {
 155                 free(decoded);
 156                 return;
 157         }
 158         if (childpid == 0) {
 159                 close(2);
 160                 /* send stdio to the pipes */
 161                 if (dup2(sendpipe[0], 0) < 0)
 162                         lprintf(9, "ERROR dup2()\n");
 163                 if (dup2(recvpipe[1], 1) < 0)
 164                         lprintf(9, "ERROR dup2()\n");
 165                 close(sendpipe[1]);     /* Close the ends we're not using */
 166                 close(recvpipe[0]);
 167                 if (!strcasecmp(encoding, "base64"))
 168                         execlp("./base64", "base64", "-d", NULL);
 169                 else if (!strcasecmp(encoding, "quoted-printable"))
 170                         execlp("./qpdecode", "qpdecode", NULL);
 171                 lprintf(9, "ERROR: cannot exec decoder for %s\n", encoding);
 172                 exit(1);
 173         }
 174         close(sendpipe[0]);     /* Close the ends we're not using  */
 175         close(recvpipe[1]);
 176
 177         while ((bytes_sent < length) && (write_error == 0)) {
 178                 /* Empty the input pipe FIRST */
 179                 while (fstat(recvpipe[0], &statbuf), (statbuf.st_size > 0)) {
 180                         blocksize = read(recvpipe[0], &decoded[bytes_recv],
 181                                          statbuf.st_size);
 182                         if (blocksize < 0)
 183                                 lprintf(9, "ERROR: cannot read from pipe\n");
 184                         else
 185                                 bytes_recv = bytes_recv + blocksize;
 186                 }
 187                 /* Then put some data into the output pipe */
 188                 blocksize = length - bytes_sent;
 189                 if (blocksize > 2048)
 190                         blocksize = 2048;
 191                 if (write(sendpipe[1], &part_start[bytes_sent], blocksize) < 0) {
 192                         lprintf(9, "ERROR: cannot write to pipe: %s\n",
 193                                 strerror(errno));
 194                         write_error = 1;
 195                 }
 196                 bytes_sent = bytes_sent + blocksize;
 197         }
 198         close(sendpipe[1]);
 199         /* Empty the input pipe */
 200         while ((blocksize = read(recvpipe[0], &decoded[bytes_recv], 1)),
 201                (blocksize > 0)) {
 202                 bytes_recv = bytes_recv + blocksize;
 203         }
 204
 205         if (bytes_recv > 0) if (CallBack != NULL) {
 206                 CallBack(name, filename, fixed_partnum(partnum),
 207                         disposition, decoded,
 208                         content_type, bytes_recv, "binary", userdata);
 209         }
 210
 211         free(decoded);
 212 }
 213
 214 /*
 215  * Break out the components of a multipart message
 216  * (This function expects to be fed HEADERS + CONTENT)
 217  * Note: NULL can be supplied as content_end; in this case, the message is
 218  * considered to have ended when the parser encounters a 0x00 byte.
 219  */
 220 void the_mime_parser(char *partnum,
 221                      char *content_start, char *content_end,
 222                      void (*CallBack)
 223                       (char *cbname,
 224                        char *cbfilename,
 225                        char *cbpartnum,
 226                        char *cbdisp,
 227                        void *cbcontent,
 228                        char *cbtype,
 229                        size_t cblength,
 230                        char *cbencoding,
 231                        void *cbuserdata),
 232                      void (*PreMultiPartCallBack)
 233                       (char *cbname,
 234                        char *cbfilename,
 235                        char *cbpartnum,
 236                        char *cbdisp,
 237                        void *cbcontent,
 238                        char *cbtype,
 239                        size_t cblength,
 240                        char *cbencoding,
 241                        void *cbuserdata),
 242                      void (*PostMultiPartCallBack)
 243                       (char *cbname,
 244                        char *cbfilename,
 245                        char *cbpartnum,
 246                        char *cbdisp,
 247                        void *cbcontent,
 248                        char *cbtype,
 249                        size_t cblength,
 250                        char *cbencoding,
 251                        void *cbuserdata),
 252                       void *userdata,
 253                       int dont_decode
 254 )
 255 {
 256
 257         char *ptr;
 258         char *part_start, *part_end = NULL;
 259         char buf[SIZ];
 260         char header[SIZ];
 261         char boundary[SIZ];
 262         char startary[SIZ];
 263         char endary[SIZ];
 264         char content_type[SIZ];
 265         size_t content_length;
 266         char encoding[SIZ];
 267         char disposition[SIZ];
 268         char name[SIZ];
 269         char filename[SIZ];
 270         int is_multipart;
 271         int part_seq = 0;
 272         int i;
 273         size_t length;
 274         char nested_partnum[SIZ];
 275
 276         lprintf(9, "the_mime_parser() called\n");
 277         ptr = content_start;
 278         memset(boundary, 0, sizeof boundary);
 279         memset(content_type, 0, sizeof content_type);
 280         memset(encoding, 0, sizeof encoding);
 281         memset(name, 0, sizeof name);
 282         memset(filename, 0, sizeof filename);
 283         memset(disposition, 0, sizeof disposition);
 284         content_length = 0;
 285
 286         /* If the caller didn't supply an endpointer, generate one by measure */
 287         if (content_end == NULL) {
 288                 content_end = &content_start[strlen(content_start)];
 289         }
 290
 291         /* Learn interesting things from the headers */
 292         strcpy(header, "");
 293         do {
 294                 ptr = memreadline(ptr, buf, sizeof buf);
 295                 if (ptr >= content_end)
 296                         return;
 297
 298                 for (i = 0; i < strlen(buf); ++i)
 299                         if (isspace(buf[i]))
 300                                 buf[i] = ' ';
 301                 if (!isspace(buf[0])) {
 302                         if (!strncasecmp(header, "Content-type: ", 14)) {
 303                                 strcpy(content_type, &header[14]);
 304                                 extract_key(name, content_type, "name");
 305                         }
 306                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 307                                 strcpy(disposition, &header[21]);
 308                                 extract_key(filename, disposition, "filename");
 309                         }
 310                         if (!strncasecmp(header, "Content-length: ", 16)) {
 311                                 content_length = (size_t) atol(&header[16]);
 312                         }
 313                         if (!strncasecmp(header,
 314                                       "Content-transfer-encoding: ", 27))
 315                                 strcpy(encoding, &header[27]);
 316                         if (strlen(boundary) == 0)
 317                                 extract_key(boundary, header, "boundary");
 318                         strcpy(header, "");
 319                 }
 320                 if ((strlen(header) + strlen(buf) + 2) < sizeof(header))
 321                         strcat(header, buf);
 322         } while ((strlen(buf) > 0) && (*ptr != 0));
 323
 324         for (i = 0; i < strlen(disposition); ++i)
 325                 if (disposition[i] == ';')
 326                         disposition[i] = 0;
 327         while (isspace(disposition[0]))
 328                 strcpy(disposition, &disposition[1]);
 329         for (i = 0; i < strlen(content_type); ++i)
 330                 if (content_type[i] == ';')
 331                         content_type[i] = 0;
 332         while (isspace(content_type[0]))
 333                 strcpy(content_type, &content_type[1]);
 334
 335         if (strlen(boundary) > 0) {
 336                 is_multipart = 1;
 337         } else {
 338                 is_multipart = 0;
 339         }
 340
 341         lprintf(9, "is_multipart=%d, boundary=<%s>\n",
 342                 is_multipart, boundary);
 343
 344         /* If this is a multipart message, then recursively process it */
 345         part_start = NULL;
 346         if (is_multipart) {
 347
 348                 /* Tell the client about this message's multipartedness */
 349                 if (PreMultiPartCallBack != NULL) {
 350                         PreMultiPartCallBack("", "", partnum, "",
 351                                 NULL, content_type,
 352                                 0, encoding, userdata);
 353                 }
 354
 355                 /* Figure out where the boundaries are */
 356                 sprintf(startary, "--%s", boundary);
 357                 sprintf(endary, "--%s--", boundary);
 358                 do {
 359                         if ( (!strncasecmp(ptr, startary, strlen(startary)))
 360                            || (!strncasecmp(ptr, endary, strlen(endary))) ) {
 361                                 lprintf(9, "hit boundary!\n");
 362                                 if (part_start != NULL) {
 363                                         if (strlen(partnum) > 0) {
 364                                                 sprintf(nested_partnum, "%s.%d",
 365                                                         partnum, ++part_seq);
 366                                         }
 367                                         else {
 368                                                 sprintf(nested_partnum, "%d",
 369                                                         ++part_seq);
 370                                         }
 371                                         the_mime_parser(nested_partnum,
 372                                                     part_start, part_end,
 373                                                         CallBack,
 374                                                         PreMultiPartCallBack,
 375                                                         PostMultiPartCallBack,
 376                                                         userdata,
 377                                                         dont_decode);
 378                                 }
 379                                 ptr = memreadline(ptr, buf, sizeof(buf));
 380                                 part_start = ptr;
 381                         }
 382                         else {
 383                                 part_end = ptr;
 384                                 ++ptr;
 385                         }
 386                 } while ( (strcasecmp(ptr, endary)) && (ptr <= content_end) );
 387                 if (PostMultiPartCallBack != NULL) {
 388                         PostMultiPartCallBack("", "", partnum, "", NULL,
 389                                 content_type, 0, encoding, userdata);
 390                 }
 391                 return;
 392         }
 393
 394         /* If it's not a multipart message, then do something with it */
 395         if (!is_multipart) {
 396                 lprintf(9, "doing non-multipart thing\n");
 397                 part_start = ptr;
 398                 length = 0;
 399                 while (ptr < content_end) {
 400                         ++ptr;
 401                         ++length;
 402                 }
 403                 part_end = content_end;
 404
 405                 /* Truncate if the header told us to */
 406                 if ( (content_length > 0) && (length > content_length) ) {
 407                         length = content_length;
 408                         lprintf(9, "truncated to %d\n", content_length);
 409                 }
 410
 411                 mime_decode(partnum,
 412                             part_start, length,
 413                             content_type, encoding, disposition,
 414                             name, filename,
 415                             CallBack, NULL, NULL,
 416                             userdata, dont_decode);
 417         }
 418 }
 419
 420
 421
 422 /*
 423  * Entry point for the MIME parser.
 424  * (This function expects to be fed HEADERS + CONTENT)
 425  * Note: NULL can be supplied as content_end; in this case, the message is
 426  * considered to have ended when the parser encounters a 0x00 byte.
 427  */
 428 void mime_parser(char *content_start,
 429                 char *content_end,
 430
 431                  void (*CallBack)
 432                   (char *cbname,
 433                    char *cbfilename,
 434                    char *cbpartnum,
 435                    char *cbdisp,
 436                    void *cbcontent,
 437                    char *cbtype,
 438                    size_t cblength,
 439                    char *cbencoding,
 440                    void *cbuserdata),
 441
 442                  void (*PreMultiPartCallBack)
 443                   (char *cbname,
 444                    char *cbfilename,
 445                    char *cbpartnum,
 446                    char *cbdisp,
 447                    void *cbcontent,
 448                    char *cbtype,
 449                    size_t cblength,
 450                    char *cbencoding,
 451                    void *cbuserdata),
 452
 453                  void (*PostMultiPartCallBack)
 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 *userdata,
 465                   int dont_decode
 466 )
 467 {
 468
 469         lprintf(9, "mime_parser() called\n");
 470         the_mime_parser("", content_start, content_end,
 471                         CallBack,
 472                         PreMultiPartCallBack,
 473                         PostMultiPartCallBack,
 474                         userdata, dont_decode);
 475 }