citadel/mime_parser.c

   1 /*
   2  * $Id$
   3  *
   4  * This is the MIME parser for Citadel.  Sometimes it actually works.
   5  *
   6  * Copyright (c) 1998-2002 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
  25 #include "citadel.h"
  26 #include "server.h"
  27 #include "serv_extensions.h"
  28 #include "sysdep_decls.h"
  29 #include "tools.h"
  30
  31 #include "mime_parser.h"
  32
  33
  34 void extract_key(char *target, char *source, char *key)
  35 {
  36         int a, b;
  37
  38         strcpy(target, source);
  39         for (a = 0; a < strlen(target); ++a) {
  40                 if ((!strncasecmp(&target[a], key, strlen(key)))
  41                     && (target[a + strlen(key)] == '=')) {
  42                         strcpy(target, &target[a + strlen(key) + 1]);
  43                         if (target[0] == 34)
  44                                 strcpy(target, &target[1]);
  45                         for (b = 0; b < strlen(target); ++b)
  46                                 if (target[b] == 34)
  47                                         target[b] = 0;
  48                         return;
  49                 }
  50         }
  51         strcpy(target, "");
  52 }
  53
  54
  55 /*
  56  * For non-multipart messages, we need to generate a quickie partnum of "1"
  57  * to return to callback functions.  Some callbacks demand it.
  58  */
  59 char *fixed_partnum(char *supplied_partnum) {
  60         if (supplied_partnum == NULL) return "1";
  61         if (strlen(supplied_partnum)==0) return "1";
  62         return supplied_partnum;
  63 }
  64
  65
  66
  67 /*
  68  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
  69  */
  70 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
  71         char buf[SIZ];
  72         int buf_length = 0;
  73         int soft_line_break = 0;
  74         int ch;
  75         int decoded_length = 0;
  76         int i;
  77
  78         decoded[0] = 0;
  79         decoded_length = 0;
  80         buf[0] = 0;
  81         buf_length = 0;
  82
  83         for (i = 0; i < sourcelen; ++i) {
  84
  85                 buf[buf_length++] = encoded[i];
  86
  87                 if ( (encoded[i] == '\n')
  88                    || (encoded[i] == 0)
  89                    || (i == (sourcelen-1)) ) {
  90                         buf[buf_length++] = 0;
  91
  92                         /*** begin -- process one line ***/
  93
  94                         if (buf[strlen(buf)-1] == '\n') {
  95                                 buf[strlen(buf)-1] = 0;
  96                         }
  97                         if (buf[strlen(buf)-1] == '\r') {
  98                                 buf[strlen(buf)-1] = 0;
  99                         }
 100                         while (isspace(buf[strlen(buf)-1])) {
 101                                 buf[strlen(buf)-1] = 0;
 102                         }
 103                         soft_line_break = 0;
 104
 105                         while (strlen(buf) > 0) {
 106                                 if (!strcmp(buf, "=")) {
 107                                         soft_line_break = 1;
 108                                         strcpy(buf, "");
 109                                 } else if ((strlen(buf)>=3) && (buf[0]=='=')) {
 110                                         sscanf(&buf[1], "%02x", &ch);
 111                                         decoded[decoded_length++] = ch;
 112                                         strcpy(buf, &buf[3]);
 113                                 } else {
 114                                         decoded[decoded_length++] = buf[0];
 115                                         strcpy(buf, &buf[1]);
 116                                 }
 117                         }
 118                         if (soft_line_break == 0) {
 119                                 decoded[decoded_length++] = '\r';
 120                                 decoded[decoded_length++] = '\n';
 121                         }
 122                         buf_length = 0;
 123                         /*** end -- process one line ***/
 124                 }
 125         }
 126
 127         decoded[decoded_length++] = 0;
 128         return(decoded_length);
 129 }
 130
 131 /*
 132  * Given a message or message-part body and a length, handle any necessary
 133  * decoding and pass the request up the stack.
 134  */
 135 void mime_decode(char *partnum,
 136                  char *part_start, size_t length,
 137                  char *content_type, char *encoding,
 138                  char *disposition,
 139                  char *name, char *filename,
 140                  void (*CallBack)
 141                   (char *cbname,
 142                    char *cbfilename,
 143                    char *cbpartnum,
 144                    char *cbdisp,
 145                    void *cbcontent,
 146                    char *cbtype,
 147                    size_t cblength,
 148                    char *cbencoding,
 149                    void *cbuserdata),
 150                  void (*PreMultiPartCallBack)
 151                   (char *cbname,
 152                    char *cbfilename,
 153                    char *cbpartnum,
 154                    char *cbdisp,
 155                    void *cbcontent,
 156                    char *cbtype,
 157                    size_t cblength,
 158                    char *cbencoding,
 159                    void *cbuserdata),
 160                  void (*PostMultiPartCallBack)
 161                   (char *cbname,
 162                    char *cbfilename,
 163                    char *cbpartnum,
 164                    char *cbdisp,
 165                    void *cbcontent,
 166                    char *cbtype,
 167                    size_t cblength,
 168                    char *cbencoding,
 169                    void *cbuserdata),
 170                   void *userdata,
 171                   int dont_decode
 172 )
 173 {
 174
 175         char *decoded;
 176         size_t bytes_decoded = 0;
 177
 178         /* Some encodings aren't really encodings */
 179         if (!strcasecmp(encoding, "7bit"))
 180                 strcpy(encoding, "");
 181         if (!strcasecmp(encoding, "8bit"))
 182                 strcpy(encoding, "");
 183         if (!strcasecmp(encoding, "binary"))
 184                 strcpy(encoding, "");
 185
 186         /* If this part is not encoded, send as-is */
 187         if ( (strlen(encoding) == 0) || (dont_decode)) {
 188                 if (CallBack != NULL) {
 189                         CallBack(name, filename, fixed_partnum(partnum),
 190                                 disposition, part_start,
 191                                 content_type, length, encoding, userdata);
 192                         }
 193                 return;
 194         }
 195
 196         if ((strcasecmp(encoding, "base64"))
 197             && (strcasecmp(encoding, "quoted-printable"))) {
 198                 return;
 199         }
 200         /*
 201          * Allocate a buffer for the decoded data.  The output buffer is the
 202          * same size as the input buffer; this assumes that the decoded data
 203          * will never be larger than the encoded data.  This is a safe
 204          * assumption with base64, uuencode, and quoted-printable.
 205          */
 206         decoded = mallok(length+2048);
 207         if (decoded == NULL) {
 208                 return;
 209         }
 210
 211         if (!strcasecmp(encoding, "base64")) {
 212                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 213         }
 214         else if (!strcasecmp(encoding, "quoted-printable")) {
 215                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded,
 216                                                         part_start, length);
 217         }
 218
 219         if (bytes_decoded > 0) if (CallBack != NULL) {
 220                 CallBack(name, filename, fixed_partnum(partnum),
 221                         disposition, decoded,
 222                         content_type, bytes_decoded, "binary", userdata);
 223         }
 224
 225         phree(decoded);
 226 }
 227
 228 /*
 229  * Break out the components of a multipart message
 230  * (This function expects to be fed HEADERS + CONTENT)
 231  * Note: NULL can be supplied as content_end; in this case, the message is
 232  * considered to have ended when the parser encounters a 0x00 byte.
 233  */
 234 void the_mime_parser(char *partnum,
 235                      char *content_start, char *content_end,
 236                      void (*CallBack)
 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 (*PreMultiPartCallBack)
 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 (*PostMultiPartCallBack)
 257                       (char *cbname,
 258                        char *cbfilename,
 259                        char *cbpartnum,
 260                        char *cbdisp,
 261                        void *cbcontent,
 262                        char *cbtype,
 263                        size_t cblength,
 264                        char *cbencoding,
 265                        void *cbuserdata),
 266                       void *userdata,
 267                       int dont_decode
 268 )
 269 {
 270
 271         char *ptr;
 272         char *part_start, *part_end = NULL;
 273         char buf[SIZ];
 274         char *header;
 275         char *boundary;
 276         char *startary;
 277         char *endary;
 278         char *content_type;
 279         size_t content_length;
 280         char *encoding;
 281         char *disposition;
 282         char *name = NULL;
 283         char *content_type_name;
 284         char *content_disposition_name;
 285         char *filename;
 286         int is_multipart;
 287         int part_seq = 0;
 288         int i;
 289         size_t length;
 290         char nested_partnum[SIZ];
 291
 292         ptr = content_start;
 293         content_length = 0;
 294
 295         boundary = mallok(SIZ);
 296         memset(boundary, 0, SIZ);
 297
 298         startary = mallok(SIZ);
 299         memset(startary, 0, SIZ);
 300
 301         endary = mallok(SIZ);
 302         memset(endary, 0, SIZ);
 303
 304         header = mallok(SIZ);
 305         memset(header, 0, SIZ);
 306
 307         content_type = mallok(SIZ);
 308         memset(content_type, 0, SIZ);
 309
 310         encoding = mallok(SIZ);
 311         memset(encoding, 0, SIZ);
 312
 313         content_type_name = mallok(SIZ);
 314         memset(content_type_name, 0, SIZ);
 315
 316         content_disposition_name = mallok(SIZ);
 317         memset(content_disposition_name, 0, SIZ);
 318
 319         filename = mallok(SIZ);
 320         memset(filename, 0, SIZ);
 321
 322         disposition = mallok(SIZ);
 323         memset(disposition, 0, SIZ);
 324
 325         /* If the caller didn't supply an endpointer, generate one by measure */
 326         if (content_end == NULL) {
 327                 content_end = &content_start[strlen(content_start)];
 328         }
 329
 330         /* Learn interesting things from the headers */
 331         strcpy(header, "");
 332         do {
 333                 ptr = memreadline(ptr, buf, SIZ);
 334                 if (ptr >= content_end) {
 335                         goto end_parser;
 336                 }
 337
 338                 for (i = 0; i < strlen(buf); ++i)
 339                         if (isspace(buf[i]))
 340                                 buf[i] = ' ';
 341                 if (!isspace(buf[0])) {
 342                         if (!strncasecmp(header, "Content-type: ", 14)) {
 343                                 strcpy(content_type, &header[14]);
 344                                 extract_key(content_type_name, content_type, "name");
 345                                 /* Deal with weird headers */
 346                                 if (strchr(content_type, ' '))
 347                                         *(strchr(content_type, ' ')) = '\0';
 348                                 if (strchr(content_type, ';'))
 349                                         *(strchr(content_type, ';')) = '\0';
 350                         }
 351                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 352                                 strcpy(disposition, &header[21]);
 353                                 extract_key(content_disposition_name, disposition, "name");
 354                                 extract_key(filename, disposition, "filename");
 355                         }
 356                         if (!strncasecmp(header, "Content-length: ", 16)) {
 357                                 content_length = (size_t) atol(&header[16]);
 358                         }
 359                         if (!strncasecmp(header,
 360                                       "Content-transfer-encoding: ", 27))
 361                                 strcpy(encoding, &header[27]);
 362                         if (strlen(boundary) == 0)
 363                                 extract_key(boundary, header, "boundary");
 364                         strcpy(header, "");
 365                 }
 366                 if ((strlen(header) + strlen(buf) + 2) < SIZ)
 367                         strcat(header, buf);
 368         } while ((strlen(buf) > 0) && (*ptr != 0));
 369
 370         if (strchr(disposition, ';'))
 371                 *(strchr(disposition, ';')) = '\0';
 372         striplt(disposition);
 373         if (strchr(content_type, ';'))
 374                 *(strchr(content_type, ';')) = '\0';
 375         striplt(content_type);
 376
 377         if (strlen(boundary) > 0) {
 378                 is_multipart = 1;
 379         } else {
 380                 is_multipart = 0;
 381         }
 382
 383         /* If this is a multipart message, then recursively process it */
 384         part_start = NULL;
 385         if (is_multipart) {
 386
 387                 /* Tell the client about this message's multipartedness */
 388                 if (PreMultiPartCallBack != NULL) {
 389                         PreMultiPartCallBack("", "", partnum, "",
 390                                 NULL, content_type,
 391                                 0, encoding, userdata);
 392                 }
 393
 394                 /* Figure out where the boundaries are */
 395                 snprintf(startary, SIZ, "--%s", boundary);
 396                 snprintf(endary, SIZ, "--%s--", boundary);
 397                 do {
 398                         if ( (!strncasecmp(ptr, startary, strlen(startary)))
 399                            || (!strncasecmp(ptr, endary, strlen(endary))) ) {
 400                                 if (part_start != NULL) {
 401                                         if (strlen(partnum) > 0) {
 402                                                 snprintf(nested_partnum,
 403                                                          sizeof nested_partnum,
 404                                                          "%s.%d", partnum,
 405                                                          ++part_seq);
 406                                         }
 407                                         else {
 408                                                 snprintf(nested_partnum,
 409                                                          sizeof nested_partnum,
 410                                                          "%d", ++part_seq);
 411                                         }
 412                                         the_mime_parser(nested_partnum,
 413                                                     part_start, part_end,
 414                                                         CallBack,
 415                                                         PreMultiPartCallBack,
 416                                                         PostMultiPartCallBack,
 417                                                         userdata,
 418                                                         dont_decode);
 419                                 }
 420                                 ptr = memreadline(ptr, buf, SIZ);
 421                                 part_start = ptr;
 422                         }
 423                         else {
 424                                 part_end = ptr;
 425                                 ++ptr;
 426                         }
 427                         /* If we pass out of scope in the MIME multipart (by
 428                          * hitting the end boundary), force the pointer out
 429                          * of scope so this loop ends.
 430                          */
 431                         if (ptr < content_end) {
 432                                 if (!strcasecmp(ptr, endary)) {
 433                                         ptr = content_end++;
 434                                 }
 435                         }
 436                 } while (ptr <= content_end);
 437                 if (PostMultiPartCallBack != NULL) {
 438                         PostMultiPartCallBack("", "", partnum, "", NULL,
 439                                 content_type, 0, encoding, userdata);
 440                 }
 441                 goto end_parser;
 442         }
 443
 444         /* If it's not a multipart message, then do something with it */
 445         if (!is_multipart) {
 446                 part_start = ptr;
 447                 length = 0;
 448                 while (ptr < content_end) {
 449                         ++ptr;
 450                         ++length;
 451                 }
 452                 part_end = content_end;
 453                 /* fix an off-by-one error */
 454                 --part_end;
 455                 --length;
 456
 457                 /* Truncate if the header told us to */
 458                 if ( (content_length > 0) && (length > content_length) ) {
 459                         length = content_length;
 460                 }
 461
 462                 /* Sometimes the "name" field is tacked on to Content-type,
 463                  * and sometimes it's tacked on to Content-disposition.  Use
 464                  * whichever one we have.
 465                  */
 466                 if (strlen(content_disposition_name) > strlen(content_type_name)) {
 467                         name = content_disposition_name;
 468                 }
 469                 else {
 470                         name = content_type_name;
 471                 }
 472
 473                 mime_decode(partnum,
 474                             part_start, length,
 475                             content_type, encoding, disposition,
 476                             name, filename,
 477                             CallBack, NULL, NULL,
 478                             userdata, dont_decode);
 479         }
 480
 481 end_parser:     /* free the buffers!  end the oppression!! */
 482         phree(boundary);
 483         phree(startary);
 484         phree(endary);
 485         phree(header);
 486         phree(content_type);
 487         phree(encoding);
 488         phree(content_type_name);
 489         phree(content_disposition_name);
 490         phree(filename);
 491         phree(disposition);
 492 }
 493
 494
 495
 496 /*
 497  * Entry point for the MIME parser.
 498  * (This function expects to be fed HEADERS + CONTENT)
 499  * Note: NULL can be supplied as content_end; in this case, the message is
 500  * considered to have ended when the parser encounters a 0x00 byte.
 501  */
 502 void mime_parser(char *content_start,
 503                 char *content_end,
 504
 505                  void (*CallBack)
 506                   (char *cbname,
 507                    char *cbfilename,
 508                    char *cbpartnum,
 509                    char *cbdisp,
 510                    void *cbcontent,
 511                    char *cbtype,
 512                    size_t cblength,
 513                    char *cbencoding,
 514                    void *cbuserdata),
 515
 516                  void (*PreMultiPartCallBack)
 517                   (char *cbname,
 518                    char *cbfilename,
 519                    char *cbpartnum,
 520                    char *cbdisp,
 521                    void *cbcontent,
 522                    char *cbtype,
 523                    size_t cblength,
 524                    char *cbencoding,
 525                    void *cbuserdata),
 526
 527                  void (*PostMultiPartCallBack)
 528                   (char *cbname,
 529                    char *cbfilename,
 530                    char *cbpartnum,
 531                    char *cbdisp,
 532                    void *cbcontent,
 533                    char *cbtype,
 534                    size_t cblength,
 535                    char *cbencoding,
 536                    void *cbuserdata),
 537
 538                   void *userdata,
 539                   int dont_decode
 540 )
 541 {
 542
 543         the_mime_parser("", content_start, content_end,
 544                         CallBack,
 545                         PreMultiPartCallBack,
 546                         PostMultiPartCallBack,
 547                         userdata, dont_decode);
 548 }