webcit/mime_parser.c

   1 /*
   2  * $Id$
   3  *
   4  * This is the MIME parser for Citadel.  Sometimes it actually works.
   5  *
   6  * Copyright (c) 1998-2005 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
  24 #include "mime_parser.h"
  25
  26
  27 void extract_key(char *target, char *source, char *key)
  28 {
  29         int a, b;
  30
  31         strcpy(target, source);
  32         for (a = 0; a < strlen(target); ++a) {
  33                 if ((!strncasecmp(&target[a], key, strlen(key)))
  34                     && (target[a + strlen(key)] == '=')) {
  35                         strcpy(target, &target[a + strlen(key) + 1]);
  36                         if (target[0] == 34)
  37                                 strcpy(target, &target[1]);
  38                         for (b = 0; b < strlen(target); ++b)
  39                                 if (target[b] == 34)
  40                                         target[b] = 0;
  41                         return;
  42                 }
  43         }
  44         strcpy(target, "");
  45 }
  46
  47
  48 /*
  49  * For non-multipart messages, we need to generate a quickie partnum of "1"
  50  * to return to callback functions.  Some callbacks demand it.
  51  */
  52 char *fixed_partnum(char *supplied_partnum) {
  53         if (supplied_partnum == NULL) return "1";
  54         if (strlen(supplied_partnum)==0) return "1";
  55         return supplied_partnum;
  56 }
  57
  58
  59
  60 /*
  61  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
  62  */
  63 int CtdlDecodeQuotedPrintable(char *decoded, char *encoded, int sourcelen) {
  64         char buf[SIZ];
  65         int buf_length = 0;
  66         int soft_line_break = 0;
  67         unsigned int ch;
  68         int decoded_length = 0;
  69         int i;
  70
  71         decoded[0] = 0;
  72         decoded_length = 0;
  73         buf[0] = 0;
  74         buf_length = 0;
  75
  76         for (i = 0; i < sourcelen; ++i) {
  77
  78                 buf[buf_length++] = encoded[i];
  79
  80                 if ( (encoded[i] == '\n')
  81                    || (encoded[i] == 0)
  82                    || (i == (sourcelen-1)) ) {
  83                         buf[buf_length++] = 0;
  84
  85                         /*** begin -- process one line ***/
  86
  87                         if (buf[strlen(buf)-1] == '\n') {
  88                                 buf[strlen(buf)-1] = 0;
  89                         }
  90                         if (buf[strlen(buf)-1] == '\r') {
  91                                 buf[strlen(buf)-1] = 0;
  92                         }
  93                         while (isspace(buf[strlen(buf)-1])) {
  94                                 buf[strlen(buf)-1] = 0;
  95                         }
  96                         soft_line_break = 0;
  97
  98                         while (strlen(buf) > 0) {
  99                                 if (!strcmp(buf, "=")) {
 100                                         soft_line_break = 1;
 101                                         strcpy(buf, "");
 102                                 } else if ((strlen(buf)>=3) && (buf[0]=='=')) {
 103                                         sscanf(&buf[1], "%02x", &ch);
 104                                         decoded[decoded_length++] = ch;
 105                                         strcpy(buf, &buf[3]);
 106                                 } else {
 107                                         decoded[decoded_length++] = buf[0];
 108                                         strcpy(buf, &buf[1]);
 109                                 }
 110                         }
 111                         if (soft_line_break == 0) {
 112                                 decoded[decoded_length++] = '\r';
 113                                 decoded[decoded_length++] = '\n';
 114                         }
 115                         buf_length = 0;
 116                         /*** end -- process one line ***/
 117                 }
 118         }
 119
 120         decoded[decoded_length++] = 0;
 121         return(decoded_length);
 122 }
 123
 124 /*
 125  * Given a message or message-part body and a length, handle any necessary
 126  * decoding and pass the request up the stack.
 127  */
 128 void mime_decode(char *partnum,
 129                  char *part_start, size_t length,
 130                  char *content_type, char *charset, char *encoding,
 131                  char *disposition,
 132                  char *name, char *filename,
 133                  void (*CallBack)
 134                   (char *cbname,
 135                    char *cbfilename,
 136                    char *cbpartnum,
 137                    char *cbdisp,
 138                    void *cbcontent,
 139                    char *cbtype,
 140                    char *cbcharset,
 141                    size_t cblength,
 142                    char *cbencoding,
 143                    void *cbuserdata),
 144                  void (*PreMultiPartCallBack)
 145                   (char *cbname,
 146                    char *cbfilename,
 147                    char *cbpartnum,
 148                    char *cbdisp,
 149                    void *cbcontent,
 150                    char *cbtype,
 151                    char *cbcharset,
 152                    size_t cblength,
 153                    char *cbencoding,
 154                    void *cbuserdata),
 155                  void (*PostMultiPartCallBack)
 156                   (char *cbname,
 157                    char *cbfilename,
 158                    char *cbpartnum,
 159                    char *cbdisp,
 160                    void *cbcontent,
 161                    char *cbtype,
 162                    char *cbcharset,
 163                    size_t cblength,
 164                    char *cbencoding,
 165                    void *cbuserdata),
 166                   void *userdata,
 167                   int dont_decode
 168 )
 169 {
 170
 171         char *decoded;
 172         size_t bytes_decoded = 0;
 173
 174         /* Some encodings aren't really encodings */
 175         if (!strcasecmp(encoding, "7bit"))
 176                 strcpy(encoding, "");
 177         if (!strcasecmp(encoding, "8bit"))
 178                 strcpy(encoding, "");
 179         if (!strcasecmp(encoding, "binary"))
 180                 strcpy(encoding, "");
 181
 182         /* If this part is not encoded, send as-is */
 183         if ( (strlen(encoding) == 0) || (dont_decode)) {
 184                 if (CallBack != NULL) {
 185                         CallBack(name, filename, fixed_partnum(partnum),
 186                                 disposition, part_start,
 187                                 content_type, charset, length, encoding, userdata);
 188                         }
 189                 return;
 190         }
 191
 192         if ((strcasecmp(encoding, "base64"))
 193             && (strcasecmp(encoding, "quoted-printable"))) {
 194                 return;
 195         }
 196         /*
 197          * Allocate a buffer for the decoded data.  The output buffer is the
 198          * same size as the input buffer; this assumes that the decoded data
 199          * will never be larger than the encoded data.  This is a safe
 200          * assumption with base64, uuencode, and quoted-printable.
 201          */
 202         decoded = malloc(length+2048);
 203         if (decoded == NULL) {
 204                 return;
 205         }
 206
 207         if (!strcasecmp(encoding, "base64")) {
 208                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 209         }
 210         else if (!strcasecmp(encoding, "quoted-printable")) {
 211                 bytes_decoded = CtdlDecodeQuotedPrintable(decoded,
 212                                                         part_start, length);
 213         }
 214
 215         if (bytes_decoded > 0) if (CallBack != NULL) {
 216                 CallBack(name, filename, fixed_partnum(partnum),
 217                         disposition, decoded,
 218                         content_type, charset, bytes_decoded, "binary", userdata);
 219         }
 220
 221         free(decoded);
 222 }
 223
 224 /*
 225  * Break out the components of a multipart message
 226  * (This function expects to be fed HEADERS + CONTENT)
 227  * Note: NULL can be supplied as content_end; in this case, the message is
 228  * considered to have ended when the parser encounters a 0x00 byte.
 229  */
 230 void the_mime_parser(char *partnum,
 231                      char *content_start, char *content_end,
 232                      void (*CallBack)
 233                       (char *cbname,
 234                        char *cbfilename,
 235                        char *cbpartnum,
 236                        char *cbdisp,
 237                        void *cbcontent,
 238                        char *cbtype,
 239                        char *cbcharset,
 240                        size_t cblength,
 241                        char *cbencoding,
 242                        void *cbuserdata),
 243                      void (*PreMultiPartCallBack)
 244                       (char *cbname,
 245                        char *cbfilename,
 246                        char *cbpartnum,
 247                        char *cbdisp,
 248                        void *cbcontent,
 249                        char *cbtype,
 250                        char *cbcharset,
 251                        size_t cblength,
 252                        char *cbencoding,
 253                        void *cbuserdata),
 254                      void (*PostMultiPartCallBack)
 255                       (char *cbname,
 256                        char *cbfilename,
 257                        char *cbpartnum,
 258                        char *cbdisp,
 259                        void *cbcontent,
 260                        char *cbtype,
 261                        char *cbcharset,
 262                        size_t cblength,
 263                        char *cbencoding,
 264                        void *cbuserdata),
 265                       void *userdata,
 266                       int dont_decode
 267 )
 268 {
 269
 270         char *ptr;
 271         char *part_start, *part_end = NULL;
 272         char buf[SIZ];
 273         char *header;
 274         char *boundary;
 275         char *startary;
 276         char *endary;
 277         char *next_boundary;
 278         char *content_type;
 279         char *charset;
 280         size_t content_length;
 281         char *encoding;
 282         char *disposition;
 283         char *name = NULL;
 284         char *content_type_name;
 285         char *content_disposition_name;
 286         char *filename;
 287         int is_multipart;
 288         int part_seq = 0;
 289         int i;
 290         size_t length;
 291         char nested_partnum[SIZ];
 292
 293         ptr = content_start;
 294         content_length = 0;
 295
 296         boundary = malloc(SIZ);
 297         memset(boundary, 0, SIZ);
 298
 299         startary = malloc(SIZ);
 300         memset(startary, 0, SIZ);
 301
 302         endary = malloc(SIZ);
 303         memset(endary, 0, SIZ);
 304
 305         header = malloc(SIZ);
 306         memset(header, 0, SIZ);
 307
 308         content_type = malloc(SIZ);
 309         memset(content_type, 0, SIZ);
 310
 311         charset = malloc(SIZ);
 312         memset(charset, 0, SIZ);
 313
 314         encoding = malloc(SIZ);
 315         memset(encoding, 0, SIZ);
 316
 317         content_type_name = malloc(SIZ);
 318         memset(content_type_name, 0, SIZ);
 319
 320         content_disposition_name = malloc(SIZ);
 321         memset(content_disposition_name, 0, SIZ);
 322
 323         filename = malloc(SIZ);
 324         memset(filename, 0, SIZ);
 325
 326         disposition = malloc(SIZ);
 327         memset(disposition, 0, SIZ);
 328
 329         /* If the caller didn't supply an endpointer, generate one by measure */
 330         if (content_end == NULL) {
 331                 content_end = &content_start[strlen(content_start)];
 332         }
 333
 334         /* Learn interesting things from the headers */
 335         strcpy(header, "");
 336         do {
 337                 ptr = memreadline(ptr, buf, SIZ);
 338                 if (ptr >= content_end) {
 339                         goto end_parser;
 340                 }
 341
 342                 for (i = 0; i < strlen(buf); ++i) {
 343                         if (isspace(buf[i])) {
 344                                 buf[i] = ' ';
 345                         }
 346                 }
 347
 348                 if (!isspace(buf[0])) {
 349                         if (!strncasecmp(header, "Content-type: ", 14)) {
 350                                 strcpy(content_type, &header[14]);
 351                                 extract_key(content_type_name, content_type, "name");
 352                                 extract_key(charset, content_type, "charset");
 353                                 /* Deal with weird headers */
 354                                 if (strchr(content_type, ' '))
 355                                         *(strchr(content_type, ' ')) = '\0';
 356                                 if (strchr(content_type, ';'))
 357                                         *(strchr(content_type, ';')) = '\0';
 358                         }
 359                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 360                                 strcpy(disposition, &header[21]);
 361                                 extract_key(content_disposition_name, disposition, "name");
 362                                 extract_key(filename, disposition, "filename");
 363                         }
 364                         if (!strncasecmp(header, "Content-length: ", 16)) {
 365                                 content_length = (size_t) atol(&header[16]);
 366                         }
 367                         if (!strncasecmp(header,
 368                                       "Content-transfer-encoding: ", 27))
 369                                 strcpy(encoding, &header[27]);
 370                         if (strlen(boundary) == 0)
 371                                 extract_key(boundary, header, "boundary");
 372                         strcpy(header, "");
 373                 }
 374                 if ((strlen(header) + strlen(buf) + 2) < SIZ)
 375                         strcat(header, buf);
 376         } while ((strlen(buf) > 0) && (*ptr != 0));
 377
 378         if (strchr(disposition, ';'))
 379                 *(strchr(disposition, ';')) = '\0';
 380         striplt(disposition);
 381         if (strchr(content_type, ';'))
 382                 *(strchr(content_type, ';')) = '\0';
 383         striplt(content_type);
 384
 385         if (strlen(boundary) > 0) {
 386                 is_multipart = 1;
 387         } else {
 388                 is_multipart = 0;
 389         }
 390
 391         /* If this is a multipart message, then recursively process it */
 392         part_start = NULL;
 393         if (is_multipart) {
 394
 395                 /* Tell the client about this message's multipartedness */
 396                 if (PreMultiPartCallBack != NULL) {
 397                         PreMultiPartCallBack("", "", partnum, "",
 398                                 NULL, content_type, charset,
 399                                 0, encoding, userdata);
 400                 }
 401
 402                 /* Figure out where the boundaries are */
 403                 snprintf(startary, SIZ, "--%s", boundary);
 404                 snprintf(endary, SIZ, "--%s--", boundary);
 405
 406                 part_start = NULL;
 407                 do {
 408                         next_boundary = bmstrstr(ptr, startary, strncmp);
 409                         if ( (part_start != NULL) && (next_boundary != NULL) ) {
 410                                 part_end = next_boundary;
 411                                 --part_end;
 412
 413                                 if (strlen(partnum) > 0) {
 414                                         snprintf(nested_partnum,
 415                                                  sizeof nested_partnum,
 416                                                  "%s.%d", partnum,
 417                                                  ++part_seq);
 418                                 }
 419                                 else {
 420                                         snprintf(nested_partnum,
 421                                                  sizeof nested_partnum,
 422                                                  "%d", ++part_seq);
 423                                 }
 424                                 the_mime_parser(nested_partnum,
 425                                             part_start, part_end,
 426                                                 CallBack,
 427                                                 PreMultiPartCallBack,
 428                                                 PostMultiPartCallBack,
 429                                                 userdata,
 430                                                 dont_decode);
 431                         }
 432
 433                         if (next_boundary != NULL) {
 434                                 /* If we pass out of scope, don't attempt to read
 435                                  * past the end boundary. */
 436                                 if (!strcmp(next_boundary, endary)) {
 437                                         ptr = content_end;
 438                                 }
 439                                 else {
 440                                         /* Set up for the next part. */
 441                                         part_start = strstr(next_boundary, "\n");
 442                                         ++part_start;
 443                                         ptr = part_start;
 444                                 }
 445                         }
 446                         else {
 447                                 /* Invalid end of multipart.  Bail out! */
 448                                 ptr = content_end;
 449                         }
 450                 } while ( (ptr < content_end) && (next_boundary != NULL) );
 451
 452                 if (PostMultiPartCallBack != NULL) {
 453                         PostMultiPartCallBack("", "", partnum, "", NULL,
 454                                 content_type, charset, 0, encoding, userdata);
 455                 }
 456                 goto end_parser;
 457         }
 458
 459         /* If it's not a multipart message, then do something with it */
 460         if (!is_multipart) {
 461                 part_start = ptr;
 462                 length = 0;
 463                 while (ptr < content_end) {
 464                         ++ptr;
 465                         ++length;
 466                 }
 467                 part_end = content_end;
 468                 /* fix an off-by-one error */
 469                 --part_end;
 470                 --length;
 471
 472                 /* Truncate if the header told us to */
 473                 if ( (content_length > 0) && (length > content_length) ) {
 474                         length = content_length;
 475                 }
 476
 477                 /* Sometimes the "name" field is tacked on to Content-type,
 478                  * and sometimes it's tacked on to Content-disposition.  Use
 479                  * whichever one we have.
 480                  */
 481                 if (strlen(content_disposition_name) > strlen(content_type_name)) {
 482                         name = content_disposition_name;
 483                 }
 484                 else {
 485                         name = content_type_name;
 486                 }
 487
 488                 mime_decode(partnum,
 489                             part_start, length,
 490                             content_type, charset, encoding, disposition,
 491                             name, filename,
 492                             CallBack, NULL, NULL,
 493                             userdata, dont_decode);
 494         }
 495
 496 end_parser:     /* free the buffers!  end the oppression!! */
 497         free(boundary);
 498         free(startary);
 499         free(endary);
 500         free(header);
 501         free(content_type);
 502         free(charset);
 503         free(encoding);
 504         free(content_type_name);
 505         free(content_disposition_name);
 506         free(filename);
 507         free(disposition);
 508 }
 509
 510
 511
 512 /*
 513  * Entry point for the MIME parser.
 514  * (This function expects to be fed HEADERS + CONTENT)
 515  * Note: NULL can be supplied as content_end; in this case, the message is
 516  * considered to have ended when the parser encounters a 0x00 byte.
 517  */
 518 void mime_parser(char *content_start,
 519                 char *content_end,
 520
 521                  void (*CallBack)
 522                   (char *cbname,
 523                    char *cbfilename,
 524                    char *cbpartnum,
 525                    char *cbdisp,
 526                    void *cbcontent,
 527                    char *cbtype,
 528                    char *cbcharset,
 529                    size_t cblength,
 530                    char *cbencoding,
 531                    void *cbuserdata),
 532
 533                  void (*PreMultiPartCallBack)
 534                   (char *cbname,
 535                    char *cbfilename,
 536                    char *cbpartnum,
 537                    char *cbdisp,
 538                    void *cbcontent,
 539                    char *cbtype,
 540                    char *cbcharset,
 541                    size_t cblength,
 542                    char *cbencoding,
 543                    void *cbuserdata),
 544
 545                  void (*PostMultiPartCallBack)
 546                   (char *cbname,
 547                    char *cbfilename,
 548                    char *cbpartnum,
 549                    char *cbdisp,
 550                    void *cbcontent,
 551                    char *cbtype,
 552                    char *cbcharset,
 553                    size_t cblength,
 554                    char *cbencoding,
 555                    void *cbuserdata),
 556
 557                   void *userdata,
 558                   int dont_decode
 559 )
 560 {
 561
 562         the_mime_parser("", content_start, content_end,
 563                         CallBack,
 564                         PreMultiPartCallBack,
 565                         PostMultiPartCallBack,
 566                         userdata, dont_decode);
 567 }