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                    void *cbuserdata),
  96                   void *userdata
  97 )
  98 {
  99
 100         char *decoded;
 101         struct stat statbuf;
 102         int sendpipe[2];
 103         int recvpipe[2];
 104         int childpid;
 105         size_t bytes_sent = 0;
 106         size_t bytes_recv = 0;
 107         size_t blocksize;
 108         int write_error = 0;
 109
 110         lprintf(9, "mime_decode() called\n");
 111
 112         /* Some encodings aren't really encodings */
 113         if (!strcasecmp(encoding, "7bit"))
 114                 strcpy(encoding, "");
 115         if (!strcasecmp(encoding, "8bit"))
 116                 strcpy(encoding, "");
 117         if (!strcasecmp(encoding, "binary"))
 118                 strcpy(encoding, "");
 119
 120         /* If this part is not encoded, send as-is */
 121         if (strlen(encoding) == 0) {
 122                 CallBack(name, filename, partnum, disposition, part_start,
 123                          content_type, length, userdata);
 124                 return;
 125         }
 126         if ((strcasecmp(encoding, "base64"))
 127             && (strcasecmp(encoding, "quoted-printable"))) {
 128                 lprintf(5, "ERROR: unknown MIME encoding '%s'\n", encoding);
 129                 return;
 130         }
 131         /*
 132          * Allocate a buffer for the decoded data.  The output buffer is the
 133          * same size as the input buffer; this assumes that the decoded data
 134          * will never be larger than the encoded data.  This is a safe
 135          * assumption with base64, uuencode, and quoted-printable.  Just to
 136          * be safe, we still pad the buffer a bit.
 137          */
 138         decoded = mallok(length + 1024);
 139         if (decoded == NULL) {
 140                 lprintf(5, "ERROR: cannot allocate memory.\n");
 141                 return;
 142         }
 143         if (pipe(sendpipe) != 0)
 144                 return;
 145         if (pipe(recvpipe) != 0)
 146                 return;
 147
 148         childpid = fork();
 149         if (childpid < 0) {
 150                 phree(decoded);
 151                 return;
 152         }
 153         if (childpid == 0) {
 154                 close(2);
 155                 /* send stdio to the pipes */
 156                 if (dup2(sendpipe[0], 0) < 0)
 157                         lprintf(5, "ERROR dup2()\n");
 158                 if (dup2(recvpipe[1], 1) < 0)
 159                         lprintf(5, "ERROR dup2()\n");
 160                 close(sendpipe[1]);     /* Close the ends we're not using */
 161                 close(recvpipe[0]);
 162                 if (!strcasecmp(encoding, "base64"))
 163                         execlp("./base64", "base64", "-d", NULL);
 164                 else if (!strcasecmp(encoding, "quoted-printable"))
 165                         execlp("./qpdecode", "qpdecode", NULL);
 166                 lprintf(5, "ERROR: cannot exec decoder for %s\n", encoding);
 167                 exit(1);
 168         }
 169         close(sendpipe[0]);     /* Close the ends we're not using  */
 170         close(recvpipe[1]);
 171
 172         while ((bytes_sent < length) && (write_error == 0)) {
 173                 /* Empty the input pipe FIRST */
 174                 while (fstat(recvpipe[0], &statbuf), (statbuf.st_size > 0)) {
 175                         blocksize = read(recvpipe[0], &decoded[bytes_recv],
 176                                          statbuf.st_size);
 177                         if (blocksize < 0)
 178                                 lprintf(5, "ERROR: cannot read from pipe\n");
 179                         else
 180                                 bytes_recv = bytes_recv + blocksize;
 181                 }
 182                 /* Then put some data into the output pipe */
 183                 blocksize = length - bytes_sent;
 184                 if (blocksize > 2048)
 185                         blocksize = 2048;
 186                 if (write(sendpipe[1], &part_start[bytes_sent], blocksize) < 0) {
 187                         lprintf(5, "ERROR: cannot write to pipe: %s\n",
 188                                 strerror(errno));
 189                         write_error = 1;
 190                 }
 191                 bytes_sent = bytes_sent + blocksize;
 192         }
 193         close(sendpipe[1]);
 194         /* Empty the input pipe */
 195         while ((blocksize = read(recvpipe[0], &decoded[bytes_recv], 1)),
 196                (blocksize > 0)) {
 197                 bytes_recv = bytes_recv + blocksize;
 198         }
 199
 200         if (bytes_recv > 0)
 201                 CallBack(name, filename, partnum, disposition, decoded,
 202                          content_type, bytes_recv, userdata);
 203
 204         phree(decoded);
 205 }
 206
 207 /*
 208  * Break out the components of a multipart message
 209  * (This function expects to be fed HEADERS + CONTENT)
 210  * Note: NULL can be supplied as content_end; in this case, the message is
 211  * considered to have ended when the parser encounters a 0x00 byte.
 212  */
 213 void the_mime_parser(char *partnum,
 214                      char *content_start, char *content_end,
 215                      void (*CallBack)
 216                       (char *cbname,
 217                        char *cbfilename,
 218                        char *cbpartnum,
 219                        char *cbdisp,
 220                        void *cbcontent,
 221                        char *cbtype,
 222                        size_t cblength,
 223                        void *cbuserdata),
 224                       void *userdata
 225 )
 226 {
 227
 228         char *ptr;
 229         char *part_start, *part_end;
 230         char buf[256];
 231         char header[256];
 232         char boundary[256];
 233         char startary[256];
 234         char endary[256];
 235         char content_type[256];
 236         char encoding[256];
 237         char disposition[256];
 238         char name[256];
 239         char filename[256];
 240         int is_multipart;
 241         int part_seq = 0;
 242         int i;
 243         size_t length;
 244         char nested_partnum[256];
 245
 246         lprintf(9, "the_mime_parser() called\n");
 247         ptr = content_start;
 248         memset(boundary, 0, sizeof boundary);
 249         memset(content_type, 0, sizeof content_type);
 250         memset(encoding, 0, sizeof encoding);
 251         memset(name, 0, sizeof name);
 252         memset(filename, 0, sizeof filename);
 253         memset(disposition, 0, sizeof disposition);
 254
 255         /* Learn interesting things from the headers */
 256         strcpy(header, "");
 257         do {
 258                 ptr = memreadline(ptr, buf, sizeof buf);
 259                 if (*ptr == 0)
 260                         return; /* premature end of message */
 261                 if (content_end != NULL)
 262                         if (ptr >= content_end)
 263                                 return;
 264
 265                 for (i = 0; i < strlen(buf); ++i)
 266                         if (isspace(buf[i]))
 267                                 buf[i] = ' ';
 268                 if (!isspace(buf[0])) {
 269                         if (!strncasecmp(header, "Content-type: ", 14)) {
 270                                 strcpy(content_type, &header[14]);
 271                                 extract_key(name, content_type, "name");
 272                         }
 273                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 274                                 strcpy(disposition, &header[21]);
 275                                 extract_key(filename, disposition, "filename");
 276                         }
 277                         if (!strncasecmp(header,
 278                                       "Content-transfer-encoding: ", 27))
 279                                 strcpy(encoding, &header[27]);
 280                         if (strlen(boundary) == 0)
 281                                 extract_key(boundary, header, "boundary");
 282                         strcpy(header, "");
 283                 }
 284                 if ((strlen(header) + strlen(buf) + 2) < sizeof(header))
 285                         strcat(header, buf);
 286         } while ((strlen(buf) > 0) && (*ptr != 0));
 287
 288         for (i = 0; i < strlen(disposition); ++i)
 289                 if (disposition[i] == ';')
 290                         disposition[i] = 0;
 291         while (isspace(disposition[0]))
 292                 strcpy(disposition, &disposition[1]);
 293         for (i = 0; i < strlen(content_type); ++i)
 294                 if (content_type[i] == ';')
 295                         content_type[i] = 0;
 296         while (isspace(content_type[0]))
 297                 strcpy(content_type, &content_type[1]);
 298
 299         if (strlen(boundary) > 0) {
 300                 is_multipart = 1;
 301         } else {
 302                 is_multipart = 0;
 303         }
 304
 305         /* If this is a multipart message, then recursively process it */
 306         part_start = NULL;
 307         if (is_multipart) {
 308
 309                 /* Tell the client about this message's multipartedness */
 310                 CallBack("", "", partnum, "", NULL, content_type, 0, userdata);
 311
 312                 /* Figure out where the boundaries are */
 313                 sprintf(startary, "--%s", boundary);
 314                 sprintf(endary, "--%s--", boundary);
 315                 do {
 316                         part_end = ptr;
 317                         ptr = memreadline(ptr, buf, sizeof buf);
 318                         if (*ptr == 0)
 319                                 return;         /* premature end of message */
 320                         if (content_end != NULL)
 321                                 if (ptr >= content_end)
 322                                         return;
 323                         if ((!strcasecmp(buf, startary))
 324                             || (!strcasecmp(buf, endary))) {
 325                                 if (part_start != NULL) {
 326                                         if (strlen(partnum) > 0) {
 327                                                 sprintf(nested_partnum, "%s.%d",
 328                                                         partnum, ++part_seq);
 329                                         }
 330                                         else {
 331                                                 sprintf(nested_partnum, "%d",
 332                                                         ++part_seq);
 333                                         }
 334                                         the_mime_parser(nested_partnum,
 335                                                     part_start, part_end,
 336                                                         CallBack, userdata);
 337                                 }
 338                                 part_start = ptr;
 339                         }
 340                 } while (strcasecmp(buf, endary));
 341         }
 342
 343         /* If it's not a multipart message, then do something with it */
 344         if (!is_multipart) {
 345                 part_start = ptr;
 346                 length = 0;
 347                 while ((*ptr != 0)
 348                       && ((content_end == NULL) || (ptr < content_end))) {
 349                         ++length;
 350                         part_end = ptr++;
 351                 }
 352                 mime_decode(partnum,
 353                             part_start, length,
 354                             content_type, encoding, disposition,
 355                             name, filename, CallBack, userdata);
 356         }
 357
 358
 359 }
 360
 361
 362
 363 /*
 364  * Entry point for the MIME parser.
 365  * (This function expects to be fed HEADERS + CONTENT)
 366  * Note: NULL can be supplied as content_end; in this case, the message is
 367  * considered to have ended when the parser encounters a 0x00 byte.
 368  */
 369 void mime_parser(char *content_start, char *content_end,
 370                  void (*CallBack)
 371                   (char *cbname,
 372                    char *cbfilename,
 373                    char *cbpartnum,
 374                    char *cbdisp,
 375                    void *cbcontent,
 376                    char *cbtype,
 377                    size_t cblength,
 378                    void *cbuserdata),
 379                   void *userdata
 380 )
 381 {
 382
 383         lprintf(9, "mime_parser() called\n");
 384         the_mime_parser("", content_start, content_end, CallBack, userdata);
 385 }