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 *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                 CallBack(name, filename, partnum, disposition, part_start,
 125                          content_type, length, encoding, userdata);
 126                 return;
 127         }
 128         if ((strcasecmp(encoding, "base64"))
 129             && (strcasecmp(encoding, "quoted-printable"))) {
 130                 lprintf(5, "ERROR: unknown MIME encoding '%s'\n", encoding);
 131                 return;
 132         }
 133         /*
 134          * Allocate a buffer for the decoded data.  The output buffer is the
 135          * same size as the input buffer; this assumes that the decoded data
 136          * will never be larger than the encoded data.  This is a safe
 137          * assumption with base64, uuencode, and quoted-printable.  Just to
 138          * be safe, we still pad the buffer a bit.
 139          */
 140         decoded = mallok(length + 1024);
 141         if (decoded == NULL) {
 142                 lprintf(5, "ERROR: cannot allocate memory.\n");
 143                 return;
 144         }
 145         if (pipe(sendpipe) != 0)
 146                 return;
 147         if (pipe(recvpipe) != 0)
 148                 return;
 149
 150         childpid = fork();
 151         if (childpid < 0) {
 152                 phree(decoded);
 153                 return;
 154         }
 155         if (childpid == 0) {
 156                 close(2);
 157                 /* send stdio to the pipes */
 158                 if (dup2(sendpipe[0], 0) < 0)
 159                         lprintf(5, "ERROR dup2()\n");
 160                 if (dup2(recvpipe[1], 1) < 0)
 161                         lprintf(5, "ERROR dup2()\n");
 162                 close(sendpipe[1]);     /* Close the ends we're not using */
 163                 close(recvpipe[0]);
 164                 if (!strcasecmp(encoding, "base64"))
 165                         execlp("./base64", "base64", "-d", NULL);
 166                 else if (!strcasecmp(encoding, "quoted-printable"))
 167                         execlp("./qpdecode", "qpdecode", NULL);
 168                 lprintf(5, "ERROR: cannot exec decoder for %s\n", encoding);
 169                 exit(1);
 170         }
 171         close(sendpipe[0]);     /* Close the ends we're not using  */
 172         close(recvpipe[1]);
 173
 174         while ((bytes_sent < length) && (write_error == 0)) {
 175                 /* Empty the input pipe FIRST */
 176                 while (fstat(recvpipe[0], &statbuf), (statbuf.st_size > 0)) {
 177                         blocksize = read(recvpipe[0], &decoded[bytes_recv],
 178                                          statbuf.st_size);
 179                         if (blocksize < 0)
 180                                 lprintf(5, "ERROR: cannot read from pipe\n");
 181                         else
 182                                 bytes_recv = bytes_recv + blocksize;
 183                 }
 184                 /* Then put some data into the output pipe */
 185                 blocksize = length - bytes_sent;
 186                 if (blocksize > 2048)
 187                         blocksize = 2048;
 188                 if (write(sendpipe[1], &part_start[bytes_sent], blocksize) < 0) {
 189                         lprintf(5, "ERROR: cannot write to pipe: %s\n",
 190                                 strerror(errno));
 191                         write_error = 1;
 192                 }
 193                 bytes_sent = bytes_sent + blocksize;
 194         }
 195         close(sendpipe[1]);
 196         /* Empty the input pipe */
 197         while ((blocksize = read(recvpipe[0], &decoded[bytes_recv], 1)),
 198                (blocksize > 0)) {
 199                 bytes_recv = bytes_recv + blocksize;
 200         }
 201
 202         if (bytes_recv > 0)
 203                 CallBack(name, filename, partnum, disposition, decoded,
 204                          content_type, bytes_recv, "binary", userdata);
 205
 206         phree(decoded);
 207 }
 208
 209 /*
 210  * Break out the components of a multipart message
 211  * (This function expects to be fed HEADERS + CONTENT)
 212  * Note: NULL can be supplied as content_end; in this case, the message is
 213  * considered to have ended when the parser encounters a 0x00 byte.
 214  */
 215 void the_mime_parser(char *partnum,
 216                      char *content_start, char *content_end,
 217                      void (*CallBack)
 218                       (char *cbname,
 219                        char *cbfilename,
 220                        char *cbpartnum,
 221                        char *cbdisp,
 222                        void *cbcontent,
 223                        char *cbtype,
 224                        size_t cblength,
 225                        char *cbencoding,
 226                        void *cbuserdata),
 227                       void *userdata,
 228                       int dont_decode
 229 )
 230 {
 231
 232         char *ptr;
 233         char *part_start, *part_end;
 234         char buf[SIZ];
 235         char header[SIZ];
 236         char boundary[SIZ];
 237         char startary[SIZ];
 238         char endary[SIZ];
 239         char content_type[SIZ];
 240         char encoding[SIZ];
 241         char disposition[SIZ];
 242         char name[SIZ];
 243         char filename[SIZ];
 244         int is_multipart;
 245         int part_seq = 0;
 246         int i;
 247         size_t length;
 248         char nested_partnum[SIZ];
 249
 250         lprintf(9, "the_mime_parser() called\n");
 251         ptr = content_start;
 252         memset(boundary, 0, sizeof boundary);
 253         memset(content_type, 0, sizeof content_type);
 254         memset(encoding, 0, sizeof encoding);
 255         memset(name, 0, sizeof name);
 256         memset(filename, 0, sizeof filename);
 257         memset(disposition, 0, sizeof disposition);
 258
 259         /* Learn interesting things from the headers */
 260         strcpy(header, "");
 261         do {
 262                 ptr = memreadline(ptr, buf, sizeof buf);
 263                 if (*ptr == 0)
 264                         return; /* premature end of message */
 265                 if (content_end != NULL)
 266                         if (ptr >= content_end)
 267                                 return;
 268
 269                 for (i = 0; i < strlen(buf); ++i)
 270                         if (isspace(buf[i]))
 271                                 buf[i] = ' ';
 272                 if (!isspace(buf[0])) {
 273                         if (!strncasecmp(header, "Content-type: ", 14)) {
 274                                 strcpy(content_type, &header[14]);
 275                                 extract_key(name, content_type, "name");
 276                         }
 277                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 278                                 strcpy(disposition, &header[21]);
 279                                 extract_key(filename, disposition, "filename");
 280                         }
 281                         if (!strncasecmp(header,
 282                                       "Content-transfer-encoding: ", 27))
 283                                 strcpy(encoding, &header[27]);
 284                         if (strlen(boundary) == 0)
 285                                 extract_key(boundary, header, "boundary");
 286                         strcpy(header, "");
 287                 }
 288                 if ((strlen(header) + strlen(buf) + 2) < sizeof(header))
 289                         strcat(header, buf);
 290         } while ((strlen(buf) > 0) && (*ptr != 0));
 291
 292         for (i = 0; i < strlen(disposition); ++i)
 293                 if (disposition[i] == ';')
 294                         disposition[i] = 0;
 295         while (isspace(disposition[0]))
 296                 strcpy(disposition, &disposition[1]);
 297         for (i = 0; i < strlen(content_type); ++i)
 298                 if (content_type[i] == ';')
 299                         content_type[i] = 0;
 300         while (isspace(content_type[0]))
 301                 strcpy(content_type, &content_type[1]);
 302
 303         if (strlen(boundary) > 0) {
 304                 is_multipart = 1;
 305         } else {
 306                 is_multipart = 0;
 307         }
 308
 309         /* If this is a multipart message, then recursively process it */
 310         part_start = NULL;
 311         if (is_multipart) {
 312
 313                 /* Tell the client about this message's multipartedness */
 314                 CallBack("", "", partnum, "", NULL, content_type, 0,
 315                                 encoding, userdata);
 316
 317                 /* Figure out where the boundaries are */
 318                 sprintf(startary, "--%s", boundary);
 319                 sprintf(endary, "--%s--", boundary);
 320                 do {
 321                         part_end = ptr;
 322                         ptr = memreadline(ptr, buf, sizeof buf);
 323                         if (*ptr == 0)
 324                                 return;         /* premature end of message */
 325                         if (content_end != NULL)
 326                                 if (ptr >= content_end)
 327                                         return;
 328                         if ((!strcasecmp(buf, startary))
 329                             || (!strcasecmp(buf, endary))) {
 330                                 if (part_start != NULL) {
 331                                         if (strlen(partnum) > 0) {
 332                                                 sprintf(nested_partnum, "%s.%d",
 333                                                         partnum, ++part_seq);
 334                                         }
 335                                         else {
 336                                                 sprintf(nested_partnum, "%d",
 337                                                         ++part_seq);
 338                                         }
 339                                         the_mime_parser(nested_partnum,
 340                                                     part_start, part_end,
 341                                                         CallBack, userdata,
 342                                                         dont_decode);
 343                                 }
 344                                 part_start = ptr;
 345                         }
 346                 } while (strcasecmp(buf, endary));
 347         }
 348
 349         /* If it's not a multipart message, then do something with it */
 350         if (!is_multipart) {
 351                 part_start = ptr;
 352                 length = 0;
 353                 while ((*ptr != 0)
 354                       && ((content_end == NULL) || (ptr < content_end))) {
 355                         ++length;
 356                         part_end = ptr++;
 357                 }
 358                 mime_decode(partnum,
 359                             part_start, length,
 360                             content_type, encoding, disposition,
 361                             name, filename, CallBack, userdata, dont_decode);
 362         }
 363
 364
 365 }
 366
 367
 368
 369 /*
 370  * Entry point for the MIME parser.
 371  * (This function expects to be fed HEADERS + CONTENT)
 372  * Note: NULL can be supplied as content_end; in this case, the message is
 373  * considered to have ended when the parser encounters a 0x00 byte.
 374  */
 375 void mime_parser(char *content_start, char *content_end,
 376                  void (*CallBack)
 377                   (char *cbname,
 378                    char *cbfilename,
 379                    char *cbpartnum,
 380                    char *cbdisp,
 381                    void *cbcontent,
 382                    char *cbtype,
 383                    size_t cblength,
 384                    char *cbencoding,
 385                    void *cbuserdata),
 386                   void *userdata,
 387                   int dont_decode
 388 )
 389 {
 390
 391         lprintf(9, "mime_parser() called\n");
 392         the_mime_parser("", content_start, content_end, CallBack,
 393                         userdata, dont_decode);
 394 }