webcit/mime_parser.c

   1 /*
   2  * $Id$
   3  *
   4  * This is the MIME parser for Citadel.  Sometimes it actually works.
   5  *
   6  * Copyright (c) 1998-2003 by Art Cancro
   7  * This code is distributed under the terms of the GNU General Public License.
   8  *
   9  */
  10
  11
  12 #include <stdlib.h>
  13 #include <unistd.h>
  14 #include <stdio.h>
  15 #include <signal.h>
  16 #include <sys/types.h>
  17 #include <ctype.h>
  18 #include <string.h>
  19 #include <sys/stat.h>
  20 #include <errno.h>
  21
  22 #include "webcit.h"
  23 #include "mime_parser.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 /*
  60  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
  61  */
  62 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
  63         char buf[SIZ];
  64         int buf_length = 0;
  65         int soft_line_break = 0;
  66         int ch;
  67         int decoded_length = 0;
  68         int i;
  69
  70         decoded[0] = 0;
  71         decoded_length = 0;
  72         buf[0] = 0;
  73         buf_length = 0;
  74
  75         for (i = 0; i < sourcelen; ++i) {
  76
  77                 buf[buf_length++] = encoded[i];
  78
  79                 if ( (encoded[i] == '\n')
  80                    || (encoded[i] == 0)
  81                    || (i == (sourcelen-1)) ) {
  82                         buf[buf_length++] = 0;
  83
  84                         /*** begin -- process one line ***/
  85
  86                         if (buf[strlen(buf)-1] == '\n') {
  87                                 buf[strlen(buf)-1] = 0;
  88                         }
  89                         if (buf[strlen(buf)-1] == '\r') {
  90                                 buf[strlen(buf)-1] = 0;
  91                         }
  92                         while (isspace(buf[strlen(buf)-1])) {
  93                                 buf[strlen(buf)-1] = 0;
  94                         }
  95                         soft_line_break = 0;
  96
  97                         while (strlen(buf) > 0) {
  98                                 if (!strcmp(buf, "=")) {
  99                                         soft_line_break = 1;
 100                                         strcpy(buf, "");
 101                                 } else if ((strlen(buf)>=3) && (buf[0]=='=')) {
 102                                         sscanf(&buf[1], "%02x", &ch);
 103                                         decoded[decoded_length++] = ch;
 104                                         strcpy(buf, &buf[3]);
 105                                 } else {
 106                                         decoded[decoded_length++] = buf[0];
 107                                         strcpy(buf, &buf[1]);
 108                                 }
 109                         }
 110                         if (soft_line_break == 0) {
 111                                 decoded[decoded_length++] = '\r';
 112                                 decoded[decoded_length++] = '\n';
 113                         }
 114                         buf_length = 0;
 115                         /*** end -- process one line ***/
 116                 }
 117         }
 118
 119         decoded[decoded_length++] = 0;
 120         return(decoded_length);
 121 }
 122
 123 /*
 124  * Given a message or message-part body and a length, handle any necessary
 125  * decoding and pass the request up the stack.
 126  */
 127 void mime_decode(char *partnum,
 128                  char *part_start, size_t length,
 129                  char *content_type, char *encoding,
 130                  char *disposition,
 131                  char *name, char *filename,
 132                  void (*CallBack)
 133                   (char *cbname,
 134                    char *cbfilename,
 135                    char *cbpartnum,
 136                    char *cbdisp,
 137                    void *cbcontent,
 138                    char *cbtype,
 139                    size_t cblength,
 140                    char *cbencoding,
 141                    void *cbuserdata),
 142                  void (*PreMultiPartCallBack)
 143                   (char *cbname,
 144                    char *cbfilename,
 145                    char *cbpartnum,
 146                    char *cbdisp,
 147                    void *cbcontent,
 148                    char *cbtype,
 149                    size_t cblength,
 150                    char *cbencoding,
 151                    void *cbuserdata),
 152                  void (*PostMultiPartCallBack)
 153                   (char *cbname,
 154                    char *cbfilename,
 155                    char *cbpartnum,
 156                    char *cbdisp,
 157                    void *cbcontent,
 158                    char *cbtype,
 159                    size_t cblength,
 160                    char *cbencoding,
 161                    void *cbuserdata),
 162                   void *userdata,
 163                   int dont_decode
 164 )
 165 {
 166
 167         char *decoded;
 168         size_t bytes_decoded = 0;
 169
 170         /* Some encodings aren't really encodings */
 171         if (!strcasecmp(encoding, "7bit"))
 172                 strcpy(encoding, "");
 173         if (!strcasecmp(encoding, "8bit"))
 174                 strcpy(encoding, "");
 175         if (!strcasecmp(encoding, "binary"))
 176                 strcpy(encoding, "");
 177
 178         /* If this part is not encoded, send as-is */
 179         if ( (strlen(encoding) == 0) || (dont_decode)) {
 180                 if (CallBack != NULL) {
 181                         CallBack(name, filename, fixed_partnum(partnum),
 182                                 disposition, part_start,
 183                                 content_type, length, encoding, userdata);
 184                         }
 185                 return;
 186         }
 187
 188         if ((strcasecmp(encoding, "base64"))
 189             && (strcasecmp(encoding, "quoted-printable"))) {
 190                 return;
 191         }
 192         /*
 193          * Allocate a buffer for the decoded data.  The output buffer is the
 194          * same size as the input buffer; this assumes that the decoded data
 195          * will never be larger than the encoded data.  This is a safe
 196          * assumption with base64, uuencode, and quoted-printable.
 197          */
 198         decoded = mallok(length+2048);
 199         if (decoded == NULL) {
 200                 return;
 201         }
 202
 203         if (!strcasecmp(encoding, "base64")) {
 204                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 205         }
 206         else if (!strcasecmp(encoding, "quoted-printable")) {
 207                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded,
 208                                                         part_start, length);
 209         }
 210
 211         if (bytes_decoded > 0) if (CallBack != NULL) {
 212                 CallBack(name, filename, fixed_partnum(partnum),
 213                         disposition, decoded,
 214                         content_type, bytes_decoded, "binary", userdata);
 215         }
 216
 217         phree(decoded);
 218 }
 219
 220 /*
 221  * Break out the components of a multipart message
 222  * (This function expects to be fed HEADERS + CONTENT)
 223  * Note: NULL can be supplied as content_end; in this case, the message is
 224  * considered to have ended when the parser encounters a 0x00 byte.
 225  */
 226 void the_mime_parser(char *partnum,
 227                      char *content_start, char *content_end,
 228                      void (*CallBack)
 229                       (char *cbname,
 230                        char *cbfilename,
 231                        char *cbpartnum,
 232                        char *cbdisp,
 233                        void *cbcontent,
 234                        char *cbtype,
 235                        size_t cblength,
 236                        char *cbencoding,
 237                        void *cbuserdata),
 238                      void (*PreMultiPartCallBack)
 239                       (char *cbname,
 240                        char *cbfilename,
 241                        char *cbpartnum,
 242                        char *cbdisp,
 243                        void *cbcontent,
 244                        char *cbtype,
 245                        size_t cblength,
 246                        char *cbencoding,
 247                        void *cbuserdata),
 248                      void (*PostMultiPartCallBack)
 249                       (char *cbname,
 250                        char *cbfilename,
 251                        char *cbpartnum,
 252                        char *cbdisp,
 253                        void *cbcontent,
 254                        char *cbtype,
 255                        size_t cblength,
 256                        char *cbencoding,
 257                        void *cbuserdata),
 258                       void *userdata,
 259                       int dont_decode
 260 )
 261 {
 262
 263         char *ptr;
 264         char *part_start, *part_end = NULL;
 265         char buf[SIZ];
 266         char *header;
 267         char *boundary;
 268         char *startary;
 269         char *endary;
 270         char *content_type;
 271         size_t content_length;
 272         char *encoding;
 273         char *disposition;
 274         char *name = NULL;
 275         char *content_type_name;
 276         char *content_disposition_name;
 277         char *filename;
 278         int is_multipart;
 279         int part_seq = 0;
 280         int i;
 281         size_t length;
 282         char nested_partnum[SIZ];
 283
 284         ptr = content_start;
 285         content_length = 0;
 286
 287         boundary = mallok(SIZ);
 288         memset(boundary, 0, SIZ);
 289
 290         startary = mallok(SIZ);
 291         memset(startary, 0, SIZ);
 292
 293         endary = mallok(SIZ);
 294         memset(endary, 0, SIZ);
 295
 296         header = mallok(SIZ);
 297         memset(header, 0, SIZ);
 298
 299         content_type = mallok(SIZ);
 300         memset(content_type, 0, SIZ);
 301
 302         encoding = mallok(SIZ);
 303         memset(encoding, 0, SIZ);
 304
 305         content_type_name = mallok(SIZ);
 306         memset(content_type_name, 0, SIZ);
 307
 308         content_disposition_name = mallok(SIZ);
 309         memset(content_disposition_name, 0, SIZ);
 310
 311         filename = mallok(SIZ);
 312         memset(filename, 0, SIZ);
 313
 314         disposition = mallok(SIZ);
 315         memset(disposition, 0, SIZ);
 316
 317         /* If the caller didn't supply an endpointer, generate one by measure */
 318         if (content_end == NULL) {
 319                 content_end = &content_start[strlen(content_start)];
 320         }
 321
 322         /* Learn interesting things from the headers */
 323         strcpy(header, "");
 324         do {
 325                 ptr = memreadline(ptr, buf, SIZ);
 326                 if (ptr >= content_end) {
 327                         goto end_parser;
 328                 }
 329
 330                 for (i = 0; i < strlen(buf); ++i)
 331                         if (isspace(buf[i]))
 332                                 buf[i] = ' ';
 333                 if (!isspace(buf[0])) {
 334                         if (!strncasecmp(header, "Content-type: ", 14)) {
 335                                 strcpy(content_type, &header[14]);
 336                                 extract_key(content_type_name, content_type, "name");
 337                                 /* Deal with weird headers */
 338                                 if (strchr(content_type, ' '))
 339                                         *(strchr(content_type, ' ')) = '\0';
 340                                 if (strchr(content_type, ';'))
 341                                         *(strchr(content_type, ';')) = '\0';
 342                         }
 343                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 344                                 strcpy(disposition, &header[21]);
 345                                 extract_key(content_disposition_name, disposition, "name");
 346                                 extract_key(filename, disposition, "filename");
 347                         }
 348                         if (!strncasecmp(header, "Content-length: ", 16)) {
 349                                 content_length = (size_t) atol(&header[16]);
 350                         }
 351                         if (!strncasecmp(header,
 352                                       "Content-transfer-encoding: ", 27))
 353                                 strcpy(encoding, &header[27]);
 354                         if (strlen(boundary) == 0)
 355                                 extract_key(boundary, header, "boundary");
 356                         strcpy(header, "");
 357                 }
 358                 if ((strlen(header) + strlen(buf) + 2) < SIZ)
 359                         strcat(header, buf);
 360         } while ((strlen(buf) > 0) && (*ptr != 0));
 361
 362         if (strchr(disposition, ';'))
 363                 *(strchr(disposition, ';')) = '\0';
 364         striplt(disposition);
 365         if (strchr(content_type, ';'))
 366                 *(strchr(content_type, ';')) = '\0';
 367         striplt(content_type);
 368
 369         if (strlen(boundary) > 0) {
 370                 is_multipart = 1;
 371         } else {
 372                 is_multipart = 0;
 373         }
 374
 375         /* If this is a multipart message, then recursively process it */
 376         part_start = NULL;
 377         if (is_multipart) {
 378
 379                 /* Tell the client about this message's multipartedness */
 380                 if (PreMultiPartCallBack != NULL) {
 381                         PreMultiPartCallBack("", "", partnum, "",
 382                                 NULL, content_type,
 383                                 0, encoding, userdata);
 384                 }
 385
 386                 /* Figure out where the boundaries are */
 387                 snprintf(startary, SIZ, "--%s", boundary);
 388                 snprintf(endary, SIZ, "--%s--", boundary);
 389                 do {
 390                         if ( (!strncasecmp(ptr, startary, strlen(startary)))
 391                            || (!strncasecmp(ptr, endary, strlen(endary))) ) {
 392                                 if (part_start != NULL) {
 393                                         if (strlen(partnum) > 0) {
 394                                                 snprintf(nested_partnum,
 395                                                          sizeof nested_partnum,
 396                                                          "%s.%d", partnum,
 397                                                          ++part_seq);
 398                                         }
 399                                         else {
 400                                                 snprintf(nested_partnum,
 401                                                          sizeof nested_partnum,
 402                                                          "%d", ++part_seq);
 403                                         }
 404                                         the_mime_parser(nested_partnum,
 405                                                     part_start, part_end,
 406                                                         CallBack,
 407                                                         PreMultiPartCallBack,
 408                                                         PostMultiPartCallBack,
 409                                                         userdata,
 410                                                         dont_decode);
 411                                 }
 412                                 ptr = memreadline(ptr, buf, SIZ);
 413                                 part_start = ptr;
 414                         }
 415                         else {
 416                                 part_end = ptr;
 417                                 ++ptr;
 418                         }
 419                         /* If we pass out of scope in the MIME multipart (by
 420                          * hitting the end boundary), force the pointer out
 421                          * of scope so this loop ends.
 422                          */
 423                         if (ptr < content_end) {
 424                                 if (!strcasecmp(ptr, endary)) {
 425                                         ptr = content_end++;
 426                                 }
 427                         }
 428                 } while (ptr <= content_end);
 429                 if (PostMultiPartCallBack != NULL) {
 430                         PostMultiPartCallBack("", "", partnum, "", NULL,
 431                                 content_type, 0, encoding, userdata);
 432                 }
 433                 goto end_parser;
 434         }
 435
 436         /* If it's not a multipart message, then do something with it */
 437         if (!is_multipart) {
 438                 part_start = ptr;
 439                 length = 0;
 440                 while (ptr < content_end) {
 441                         ++ptr;
 442                         ++length;
 443                 }
 444                 part_end = content_end;
 445                 /* fix an off-by-one error */
 446                 --part_end;
 447                 --length;
 448
 449                 /* Truncate if the header told us to */
 450                 if ( (content_length > 0) && (length > content_length) ) {
 451                         length = content_length;
 452                 }
 453
 454                 /* Sometimes the "name" field is tacked on to Content-type,
 455                  * and sometimes it's tacked on to Content-disposition.  Use
 456                  * whichever one we have.
 457                  */
 458                 if (strlen(content_disposition_name) > strlen(content_type_name)) {
 459                         name = content_disposition_name;
 460                 }
 461                 else {
 462                         name = content_type_name;
 463                 }
 464
 465                 mime_decode(partnum,
 466                             part_start, length,
 467                             content_type, encoding, disposition,
 468                             name, filename,
 469                             CallBack, NULL, NULL,
 470                             userdata, dont_decode);
 471         }
 472
 473 end_parser:     /* free the buffers!  end the oppression!! */
 474         phree(boundary);
 475         phree(startary);
 476         phree(endary);
 477         phree(header);
 478         phree(content_type);
 479         phree(encoding);
 480         phree(content_type_name);
 481         phree(content_disposition_name);
 482         phree(filename);
 483         phree(disposition);
 484 }
 485
 486
 487
 488 /*
 489  * Entry point for the MIME parser.
 490  * (This function expects to be fed HEADERS + CONTENT)
 491  * Note: NULL can be supplied as content_end; in this case, the message is
 492  * considered to have ended when the parser encounters a 0x00 byte.
 493  */
 494 void mime_parser(char *content_start,
 495                 char *content_end,
 496
 497                  void (*CallBack)
 498                   (char *cbname,
 499                    char *cbfilename,
 500                    char *cbpartnum,
 501                    char *cbdisp,
 502                    void *cbcontent,
 503                    char *cbtype,
 504                    size_t cblength,
 505                    char *cbencoding,
 506                    void *cbuserdata),
 507
 508                  void (*PreMultiPartCallBack)
 509                   (char *cbname,
 510                    char *cbfilename,
 511                    char *cbpartnum,
 512                    char *cbdisp,
 513                    void *cbcontent,
 514                    char *cbtype,
 515                    size_t cblength,
 516                    char *cbencoding,
 517                    void *cbuserdata),
 518
 519                  void (*PostMultiPartCallBack)
 520                   (char *cbname,
 521                    char *cbfilename,
 522                    char *cbpartnum,
 523                    char *cbdisp,
 524                    void *cbcontent,
 525                    char *cbtype,
 526                    size_t cblength,
 527                    char *cbencoding,
 528                    void *cbuserdata),
 529
 530                   void *userdata,
 531                   int dont_decode
 532 )
 533 {
 534
 535         the_mime_parser("", content_start, content_end,
 536                         CallBack,
 537                         PreMultiPartCallBack,
 538                         PostMultiPartCallBack,
 539                         userdata, dont_decode);
 540 }
 541