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  * For non-multipart messages, we need to generate a quickie partnum of "1"
  53  * to return to callback functions.  Some callbacks demand it.
  54  */
  55 char *fixed_partnum(char *supplied_partnum) {
  56         if (supplied_partnum == NULL) return "1";
  57         if (strlen(supplied_partnum)==0) return "1";
  58         return supplied_partnum;
  59 }
  60
  61
  62 /*
  63  * Given a message or message-part body and a length, handle any necessary
  64  * decoding and pass the request up the stack.
  65  */
  66 void mime_decode(char *partnum,
  67                  char *part_start, size_t length,
  68                  char *content_type, char *encoding,
  69                  char *disposition,
  70                  char *name, char *filename,
  71                  void (*CallBack)
  72                   (char *cbname,
  73                    char *cbfilename,
  74                    char *cbpartnum,
  75                    char *cbdisp,
  76                    void *cbcontent,
  77                    char *cbtype,
  78                    size_t cblength,
  79                    char *cbencoding,
  80                    void *cbuserdata),
  81                  void (*PreMultiPartCallBack)
  82                   (char *cbname,
  83                    char *cbfilename,
  84                    char *cbpartnum,
  85                    char *cbdisp,
  86                    void *cbcontent,
  87                    char *cbtype,
  88                    size_t cblength,
  89                    char *cbencoding,
  90                    void *cbuserdata),
  91                  void (*PostMultiPartCallBack)
  92                   (char *cbname,
  93                    char *cbfilename,
  94                    char *cbpartnum,
  95                    char *cbdisp,
  96                    void *cbcontent,
  97                    char *cbtype,
  98                    size_t cblength,
  99                    char *cbencoding,
 100                    void *cbuserdata),
 101                   void *userdata,
 102                   int dont_decode
 103 )
 104 {
 105
 106         char *decoded;
 107         struct stat statbuf;
 108         int sendpipe[2];
 109         int recvpipe[2];
 110         int childpid;
 111         size_t bytes_sent = 0;
 112         size_t bytes_recv = 0;
 113         size_t blocksize;
 114         int write_error = 0;
 115
 116         lprintf(9, "mime_decode() called\n");
 117
 118         /* Some encodings aren't really encodings */
 119         if (!strcasecmp(encoding, "7bit"))
 120                 strcpy(encoding, "");
 121         if (!strcasecmp(encoding, "8bit"))
 122                 strcpy(encoding, "");
 123         if (!strcasecmp(encoding, "binary"))
 124                 strcpy(encoding, "");
 125
 126         /* If this part is not encoded, send as-is */
 127         if ( (strlen(encoding) == 0) || (dont_decode)) {
 128                 if (CallBack != NULL) {
 129                         CallBack(name, filename, fixed_partnum(partnum),
 130                                 disposition, part_start,
 131                                 content_type, length, encoding, userdata);
 132                         }
 133                 return;
 134         }
 135         if ((strcasecmp(encoding, "base64"))
 136             && (strcasecmp(encoding, "quoted-printable"))) {
 137                 lprintf(5, "ERROR: unknown MIME encoding '%s'\n", encoding);
 138                 return;
 139         }
 140         /*
 141          * Allocate a buffer for the decoded data.  The output buffer is the
 142          * same size as the input buffer; this assumes that the decoded data
 143          * will never be larger than the encoded data.  This is a safe
 144          * assumption with base64, uuencode, and quoted-printable.  Just to
 145          * be safe, we still pad the buffer a bit.
 146          */
 147         decoded = mallok(length + 1024);
 148         if (decoded == NULL) {
 149                 lprintf(5, "ERROR: cannot allocate memory.\n");
 150                 return;
 151         }
 152         if (pipe(sendpipe) != 0)
 153                 return;
 154         if (pipe(recvpipe) != 0)
 155                 return;
 156
 157         childpid = fork();
 158         if (childpid < 0) {
 159                 phree(decoded);
 160                 return;
 161         }
 162         if (childpid == 0) {
 163                 close(2);
 164                 /* send stdio to the pipes */
 165                 if (dup2(sendpipe[0], 0) < 0)
 166                         lprintf(5, "ERROR dup2()\n");
 167                 if (dup2(recvpipe[1], 1) < 0)
 168                         lprintf(5, "ERROR dup2()\n");
 169                 close(sendpipe[1]);     /* Close the ends we're not using */
 170                 close(recvpipe[0]);
 171                 if (!strcasecmp(encoding, "base64"))
 172                         execlp("./base64", "base64", "-d", NULL);
 173                 else if (!strcasecmp(encoding, "quoted-printable"))
 174                         execlp("./qpdecode", "qpdecode", NULL);
 175                 lprintf(5, "ERROR: cannot exec decoder for %s\n", encoding);
 176                 exit(1);
 177         }
 178         close(sendpipe[0]);     /* Close the ends we're not using  */
 179         close(recvpipe[1]);
 180
 181         while ((bytes_sent < length) && (write_error == 0)) {
 182                 /* Empty the input pipe FIRST */
 183                 while (fstat(recvpipe[0], &statbuf), (statbuf.st_size > 0)) {
 184                         blocksize = read(recvpipe[0], &decoded[bytes_recv],
 185                                          statbuf.st_size);
 186                         if (blocksize < 0)
 187                                 lprintf(5, "ERROR: cannot read from pipe\n");
 188                         else
 189                                 bytes_recv = bytes_recv + blocksize;
 190                 }
 191                 /* Then put some data into the output pipe */
 192                 blocksize = length - bytes_sent;
 193                 if (blocksize > 2048)
 194                         blocksize = 2048;
 195                 if (write(sendpipe[1], &part_start[bytes_sent], blocksize) < 0) {
 196                         lprintf(5, "ERROR: cannot write to pipe: %s\n",
 197                                 strerror(errno));
 198                         write_error = 1;
 199                 }
 200                 bytes_sent = bytes_sent + blocksize;
 201         }
 202         close(sendpipe[1]);
 203         /* Empty the input pipe */
 204         while ((blocksize = read(recvpipe[0], &decoded[bytes_recv], 1)),
 205                (blocksize > 0)) {
 206                 bytes_recv = bytes_recv + blocksize;
 207         }
 208
 209         if (bytes_recv > 0) if (CallBack != NULL) {
 210                 CallBack(name, filename, fixed_partnum(partnum),
 211                         disposition, decoded,
 212                         content_type, bytes_recv, "binary", userdata);
 213         }
 214
 215         phree(decoded);
 216 }
 217
 218 /*
 219  * Break out the components of a multipart message
 220  * (This function expects to be fed HEADERS + CONTENT)
 221  * Note: NULL can be supplied as content_end; in this case, the message is
 222  * considered to have ended when the parser encounters a 0x00 byte.
 223  */
 224 void the_mime_parser(char *partnum,
 225                      char *content_start, char *content_end,
 226                      void (*CallBack)
 227                       (char *cbname,
 228                        char *cbfilename,
 229                        char *cbpartnum,
 230                        char *cbdisp,
 231                        void *cbcontent,
 232                        char *cbtype,
 233                        size_t cblength,
 234                        char *cbencoding,
 235                        void *cbuserdata),
 236                      void (*PreMultiPartCallBack)
 237                       (char *cbname,
 238                        char *cbfilename,
 239                        char *cbpartnum,
 240                        char *cbdisp,
 241                        void *cbcontent,
 242                        char *cbtype,
 243                        size_t cblength,
 244                        char *cbencoding,
 245                        void *cbuserdata),
 246                      void (*PostMultiPartCallBack)
 247                       (char *cbname,
 248                        char *cbfilename,
 249                        char *cbpartnum,
 250                        char *cbdisp,
 251                        void *cbcontent,
 252                        char *cbtype,
 253                        size_t cblength,
 254                        char *cbencoding,
 255                        void *cbuserdata),
 256                       void *userdata,
 257                       int dont_decode
 258 )
 259 {
 260
 261         char *ptr;
 262         char *part_start, *part_end;
 263         char buf[SIZ];
 264         char header[SIZ];
 265         char boundary[SIZ];
 266         char startary[SIZ];
 267         char endary[SIZ];
 268         char content_type[SIZ];
 269         char encoding[SIZ];
 270         char disposition[SIZ];
 271         char name[SIZ];
 272         char filename[SIZ];
 273         int is_multipart;
 274         int part_seq = 0;
 275         int i;
 276         size_t length;
 277         char nested_partnum[SIZ];
 278
 279         lprintf(9, "the_mime_parser() called\n");
 280         ptr = content_start;
 281         memset(boundary, 0, sizeof boundary);
 282         memset(content_type, 0, sizeof content_type);
 283         memset(encoding, 0, sizeof encoding);
 284         memset(name, 0, sizeof name);
 285         memset(filename, 0, sizeof filename);
 286         memset(disposition, 0, sizeof disposition);
 287
 288         /* Learn interesting things from the headers */
 289         strcpy(header, "");
 290         do {
 291                 ptr = memreadline(ptr, buf, sizeof buf);
 292                 if (*ptr == 0)
 293                         return; /* premature end of message */
 294                 if (content_end != NULL)
 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,
 311                                       "Content-transfer-encoding: ", 27))
 312                                 strcpy(encoding, &header[27]);
 313                         if (strlen(boundary) == 0)
 314                                 extract_key(boundary, header, "boundary");
 315                         strcpy(header, "");
 316                 }
 317                 if ((strlen(header) + strlen(buf) + 2) < sizeof(header))
 318                         strcat(header, buf);
 319         } while ((strlen(buf) > 0) && (*ptr != 0));
 320
 321         for (i = 0; i < strlen(disposition); ++i)
 322                 if (disposition[i] == ';')
 323                         disposition[i] = 0;
 324         while (isspace(disposition[0]))
 325                 strcpy(disposition, &disposition[1]);
 326         for (i = 0; i < strlen(content_type); ++i)
 327                 if (content_type[i] == ';')
 328                         content_type[i] = 0;
 329         while (isspace(content_type[0]))
 330                 strcpy(content_type, &content_type[1]);
 331
 332         if (strlen(boundary) > 0) {
 333                 is_multipart = 1;
 334         } else {
 335                 is_multipart = 0;
 336         }
 337
 338         /* If this is a multipart message, then recursively process it */
 339         part_start = NULL;
 340         if (is_multipart) {
 341
 342                 /* Tell the client about this message's multipartedness */
 343                 if (PreMultiPartCallBack != NULL) {
 344                         PreMultiPartCallBack("", "", partnum, "",
 345                                 NULL, content_type,
 346                                 0, encoding, userdata);
 347                 }
 348                 /*
 349                 if (CallBack != NULL) {
 350                         CallBack("", "", fixed_partnum(partnum),
 351                                 "", NULL, content_type,
 352                                 0, encoding, userdata);
 353                 }
 354                  */
 355
 356                 /* Figure out where the boundaries are */
 357                 sprintf(startary, "--%s", boundary);
 358                 sprintf(endary, "--%s--", boundary);
 359                 do {
 360                         part_end = ptr;
 361                         ptr = memreadline(ptr, buf, sizeof buf);
 362                         if (content_end != NULL)
 363                                 if (ptr >= content_end) goto END_MULTI;
 364
 365                         if ( (!strcasecmp(buf, startary))
 366                            || (!strcasecmp(buf, endary)) ) {
 367                                 if (part_start != NULL) {
 368                                         if (strlen(partnum) > 0) {
 369                                                 sprintf(nested_partnum, "%s.%d",
 370                                                         partnum, ++part_seq);
 371                                         }
 372                                         else {
 373                                                 sprintf(nested_partnum, "%d",
 374                                                         ++part_seq);
 375                                         }
 376                                         the_mime_parser(nested_partnum,
 377                                                     part_start, part_end,
 378                                                         CallBack,
 379                                                         PreMultiPartCallBack,
 380                                                         PostMultiPartCallBack,
 381                                                         userdata,
 382                                                         dont_decode);
 383                                 }
 384                                 part_start = ptr;
 385                         }
 386                 } while ( (strcasecmp(buf, endary)) && (ptr != 0) );
 387 END_MULTI:      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                 part_start = ptr;
 397                 length = 0;
 398                 while ((*ptr != 0)
 399                       && ((content_end == NULL) || (ptr < content_end))) {
 400                         ++length;
 401                         part_end = ptr++;
 402                 }
 403                 mime_decode(partnum,
 404                             part_start, length,
 405                             content_type, encoding, disposition,
 406                             name, filename,
 407                             CallBack, NULL, NULL,
 408                             userdata, dont_decode);
 409         }
 410
 411
 412 }
 413
 414
 415
 416 /*
 417  * Entry point for the MIME parser.
 418  * (This function expects to be fed HEADERS + CONTENT)
 419  * Note: NULL can be supplied as content_end; in this case, the message is
 420  * considered to have ended when the parser encounters a 0x00 byte.
 421  */
 422 void mime_parser(char *content_start,
 423                 char *content_end,
 424
 425                  void (*CallBack)
 426                   (char *cbname,
 427                    char *cbfilename,
 428                    char *cbpartnum,
 429                    char *cbdisp,
 430                    void *cbcontent,
 431                    char *cbtype,
 432                    size_t cblength,
 433                    char *cbencoding,
 434                    void *cbuserdata),
 435
 436                  void (*PreMultiPartCallBack)
 437                   (char *cbname,
 438                    char *cbfilename,
 439                    char *cbpartnum,
 440                    char *cbdisp,
 441                    void *cbcontent,
 442                    char *cbtype,
 443                    size_t cblength,
 444                    char *cbencoding,
 445                    void *cbuserdata),
 446
 447                  void (*PostMultiPartCallBack)
 448                   (char *cbname,
 449                    char *cbfilename,
 450                    char *cbpartnum,
 451                    char *cbdisp,
 452                    void *cbcontent,
 453                    char *cbtype,
 454                    size_t cblength,
 455                    char *cbencoding,
 456                    void *cbuserdata),
 457
 458                   void *userdata,
 459                   int dont_decode
 460 )
 461 {
 462
 463         lprintf(9, "mime_parser() called\n");
 464         the_mime_parser("", content_start, content_end,
 465                         CallBack,
 466                         PreMultiPartCallBack,
 467                         PostMultiPartCallBack,
 468                         userdata, dont_decode);
 469 }