webcit/mime_parser.c

   1 /*
   2  * $Id$
   3  *
   4  * This is the MIME parser for Citadel.  Sometimes it actually works.
   5  *
   6  * Copyright (c) 1998-2001 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 "webcit.h"
  26 #include "mime_parser.h"
  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  * For non-multipart messages, we need to generate a quickie partnum of "1"
  52  * to return to callback functions.  Some callbacks demand it.
  53  */
  54 char *fixed_partnum(char *supplied_partnum) {
  55         if (supplied_partnum == NULL) return "1";
  56         if (strlen(supplied_partnum)==0) return "1";
  57         return supplied_partnum;
  58 }
  59
  60
  61
  62 /*
  63  * Convert "quoted-printable" to binary.  Returns number of bytes decoded.
  64  */
  65 int decode_quoted_printable(char *decoded, char *encoded, int sourcelen) {
  66         char buf[SIZ];
  67         int buf_length = 0;
  68         int soft_line_break = 0;
  69         int ch;
  70         int decoded_length = 0;
  71         int i;
  72
  73         decoded[0] = 0;
  74         decoded_length = 0;
  75         buf[0] = 0;
  76         buf_length = 0;
  77
  78         for (i = 0; i < sourcelen; ++i) {
  79
  80                 buf[buf_length++] = encoded[i];
  81
  82                 if ( (encoded[i] == '\n')
  83                    || (encoded[i] == 0)
  84                    || (i == (sourcelen-1)) ) {
  85                         buf[buf_length++] = 0;
  86
  87                         /*** begin -- process one line ***/
  88
  89                         if (buf[strlen(buf)-1] == '\n') {
  90                                 buf[strlen(buf)-1] = 0;
  91                         }
  92                         if (buf[strlen(buf)-1] == '\r') {
  93                                 buf[strlen(buf)-1] = 0;
  94                         }
  95                         while (isspace(buf[strlen(buf)-1])) {
  96                                 buf[strlen(buf)-1] = 0;
  97                         }
  98                         soft_line_break = 0;
  99
 100                         while (strlen(buf) > 0) {
 101                                 if (!strcmp(buf, "=")) {
 102                                         soft_line_break = 1;
 103                                         strcpy(buf, "");
 104                                 } else if ((strlen(buf)>=3) && (buf[0]=='=')) {
 105                                         sscanf(&buf[1], "%02x", &ch);
 106                                         decoded[decoded_length++] = ch;
 107                                         strcpy(buf, &buf[3]);
 108                                 } else {
 109                                         decoded[decoded_length++] = buf[0];
 110                                         strcpy(buf, &buf[1]);
 111                                 }
 112                         }
 113                         if (soft_line_break == 0) {
 114                                 decoded[decoded_length++] = '\r';
 115                                 decoded[decoded_length++] = '\n';
 116                         }
 117                         buf_length = 0;
 118                         /*** end -- process one line ***/
 119                 }
 120         }
 121
 122         decoded[decoded_length++] = 0;
 123         return(decoded_length);
 124 }
 125
 126 /*
 127  * Given a message or message-part body and a length, handle any necessary
 128  * decoding and pass the request up the stack.
 129  */
 130 void mime_decode(char *partnum,
 131                  char *part_start, size_t length,
 132                  char *content_type, char *encoding,
 133                  char *disposition,
 134                  char *name, char *filename,
 135                  void (*CallBack)
 136                   (char *cbname,
 137                    char *cbfilename,
 138                    char *cbpartnum,
 139                    char *cbdisp,
 140                    void *cbcontent,
 141                    char *cbtype,
 142                    size_t cblength,
 143                    char *cbencoding,
 144                    void *cbuserdata),
 145                  void (*PreMultiPartCallBack)
 146                   (char *cbname,
 147                    char *cbfilename,
 148                    char *cbpartnum,
 149                    char *cbdisp,
 150                    void *cbcontent,
 151                    char *cbtype,
 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                    size_t cblength,
 163                    char *cbencoding,
 164                    void *cbuserdata),
 165                   void *userdata,
 166                   int dont_decode
 167 )
 168 {
 169
 170         char *decoded;
 171         size_t bytes_decoded = 0;
 172
 173         /* Some encodings aren't really encodings */
 174         if (!strcasecmp(encoding, "7bit"))
 175                 strcpy(encoding, "");
 176         if (!strcasecmp(encoding, "8bit"))
 177                 strcpy(encoding, "");
 178         if (!strcasecmp(encoding, "binary"))
 179                 strcpy(encoding, "");
 180
 181         /* If this part is not encoded, send as-is */
 182         if ( (strlen(encoding) == 0) || (dont_decode)) {
 183                 if (CallBack != NULL) {
 184                         CallBack(name, filename, fixed_partnum(partnum),
 185                                 disposition, part_start,
 186                                 content_type, length, encoding, userdata);
 187                         }
 188                 return;
 189         }
 190
 191         if ((strcasecmp(encoding, "base64"))
 192             && (strcasecmp(encoding, "quoted-printable"))) {
 193                 return;
 194         }
 195         /*
 196          * Allocate a buffer for the decoded data.  The output buffer is the
 197          * same size as the input buffer; this assumes that the decoded data
 198          * will never be larger than the encoded data.  This is a safe
 199          * assumption with base64, uuencode, and quoted-printable.
 200          */
 201         decoded = mallok(length+2048);
 202         if (decoded == NULL) {
 203                 return;
 204         }
 205
 206         if (!strcasecmp(encoding, "base64")) {
 207                 bytes_decoded = CtdlDecodeBase64(decoded, part_start, length);
 208         }
 209         else if (!strcasecmp(encoding, "quoted-printable")) {
 210                 bytes_decoded = decode_quoted_printable(decoded,
 211                                                         part_start, length);
 212         }
 213
 214         if (bytes_decoded > 0) if (CallBack != NULL) {
 215                 CallBack(name, filename, fixed_partnum(partnum),
 216                         disposition, decoded,
 217                         content_type, bytes_decoded, "binary", userdata);
 218         }
 219
 220         phree(decoded);
 221 }
 222
 223 /*
 224  * Break out the components of a multipart message
 225  * (This function expects to be fed HEADERS + CONTENT)
 226  * Note: NULL can be supplied as content_end; in this case, the message is
 227  * considered to have ended when the parser encounters a 0x00 byte.
 228  */
 229 void the_mime_parser(char *partnum,
 230                      char *content_start, char *content_end,
 231                      void (*CallBack)
 232                       (char *cbname,
 233                        char *cbfilename,
 234                        char *cbpartnum,
 235                        char *cbdisp,
 236                        void *cbcontent,
 237                        char *cbtype,
 238                        size_t cblength,
 239                        char *cbencoding,
 240                        void *cbuserdata),
 241                      void (*PreMultiPartCallBack)
 242                       (char *cbname,
 243                        char *cbfilename,
 244                        char *cbpartnum,
 245                        char *cbdisp,
 246                        void *cbcontent,
 247                        char *cbtype,
 248                        size_t cblength,
 249                        char *cbencoding,
 250                        void *cbuserdata),
 251                      void (*PostMultiPartCallBack)
 252                       (char *cbname,
 253                        char *cbfilename,
 254                        char *cbpartnum,
 255                        char *cbdisp,
 256                        void *cbcontent,
 257                        char *cbtype,
 258                        size_t cblength,
 259                        char *cbencoding,
 260                        void *cbuserdata),
 261                       void *userdata,
 262                       int dont_decode
 263 )
 264 {
 265
 266         char *ptr;
 267         char *part_start, *part_end = NULL;
 268         char buf[SIZ];
 269         char *header;
 270         char *boundary;
 271         char *startary;
 272         char *endary;
 273         char *content_type;
 274         size_t content_length;
 275         char *encoding;
 276         char *disposition;
 277         char *name;
 278         char *filename;
 279         int is_multipart;
 280         int part_seq = 0;
 281         int i;
 282         size_t length;
 283         char nested_partnum[SIZ];
 284
 285         ptr = content_start;
 286         content_length = 0;
 287
 288         boundary = mallok(SIZ);
 289         memset(boundary, 0, SIZ);
 290
 291         startary = mallok(SIZ);
 292         memset(startary, 0, SIZ);
 293
 294         endary = mallok(SIZ);
 295         memset(endary, 0, SIZ);
 296
 297         header = mallok(SIZ);
 298         memset(header, 0, SIZ);
 299
 300         content_type = mallok(SIZ);
 301         memset(content_type, 0, SIZ);
 302
 303         encoding = mallok(SIZ);
 304         memset(encoding, 0, SIZ);
 305
 306         name = mallok(SIZ);
 307         memset(name, 0, SIZ);
 308
 309         filename = mallok(SIZ);
 310         memset(filename, 0, SIZ);
 311
 312         disposition = mallok(SIZ);
 313         memset(disposition, 0, SIZ);
 314
 315         /* If the caller didn't supply an endpointer, generate one by measure */
 316         if (content_end == NULL) {
 317                 content_end = &content_start[strlen(content_start)];
 318         }
 319
 320         /* Learn interesting things from the headers */
 321         strcpy(header, "");
 322         do {
 323                 ptr = memreadline(ptr, buf, SIZ);
 324                 if (ptr >= content_end) {
 325                         goto end_parser;
 326                 }
 327
 328                 for (i = 0; i < strlen(buf); ++i)
 329                         if (isspace(buf[i]))
 330                                 buf[i] = ' ';
 331                 if (!isspace(buf[0])) {
 332                         if (!strncasecmp(header, "Content-type: ", 14)) {
 333                                 strcpy(content_type, &header[14]);
 334                                 extract_key(name, content_type, "name");
 335                                 /* Deal with weird headers */
 336                                 if (strchr(content_type, ' '))
 337                                         *(strchr(content_type, ' ')) = '\0';
 338                                 if (strchr(content_type, ';'))
 339                                         *(strchr(content_type, ';')) = '\0';
 340                         }
 341                         if (!strncasecmp(header, "Content-Disposition: ", 21)) {
 342                                 strcpy(disposition, &header[21]);
 343                                 extract_key(filename, disposition, "filename");
 344                         }
 345                         if (!strncasecmp(header, "Content-length: ", 16)) {
 346                                 content_length = (size_t) atol(&header[16]);
 347                         }
 348                         if (!strncasecmp(header,
 349                                       "Content-transfer-encoding: ", 27))
 350                                 strcpy(encoding, &header[27]);
 351                         if (strlen(boundary) == 0)
 352                                 extract_key(boundary, header, "boundary");
 353                         strcpy(header, "");
 354                 }
 355                 if ((strlen(header) + strlen(buf) + 2) < SIZ)
 356                         strcat(header, buf);
 357         } while ((strlen(buf) > 0) && (*ptr != 0));
 358
 359         if (strchr(disposition, ';'))
 360                 *(strchr(disposition, ';')) = '\0';
 361         striplt(disposition);
 362         if (strchr(content_type, ';'))
 363                 *(strchr(content_type, ';')) = '\0';
 364         striplt(content_type);
 365
 366         if (strlen(boundary) > 0) {
 367                 is_multipart = 1;
 368         } else {
 369                 is_multipart = 0;
 370         }
 371
 372         /* If this is a multipart message, then recursively process it */
 373         part_start = NULL;
 374         if (is_multipart) {
 375
 376                 /* Tell the client about this message's multipartedness */
 377                 if (PreMultiPartCallBack != NULL) {
 378                         PreMultiPartCallBack("", "", partnum, "",
 379                                 NULL, content_type,
 380                                 0, encoding, userdata);
 381                 }
 382
 383                 /* Figure out where the boundaries are */
 384                 snprintf(startary, SIZ, "--%s", boundary);
 385                 snprintf(endary, SIZ, "--%s--", boundary);
 386                 do {
 387                         if ( (!strncasecmp(ptr, startary, strlen(startary)))
 388                            || (!strncasecmp(ptr, endary, strlen(endary))) ) {
 389                                 if (part_start != NULL) {
 390                                         if (strlen(partnum) > 0) {
 391                                                 snprintf(nested_partnum,
 392                                                          sizeof nested_partnum,
 393                                                          "%s.%d", partnum,
 394                                                          ++part_seq);
 395                                         }
 396                                         else {
 397                                                 snprintf(nested_partnum,
 398                                                          sizeof nested_partnum,
 399                                                          "%d", ++part_seq);
 400                                         }
 401                                         the_mime_parser(nested_partnum,
 402                                                     part_start, part_end,
 403                                                         CallBack,
 404                                                         PreMultiPartCallBack,
 405                                                         PostMultiPartCallBack,
 406                                                         userdata,
 407                                                         dont_decode);
 408                                 }
 409                                 ptr = memreadline(ptr, buf, SIZ);
 410                                 part_start = ptr;
 411                         }
 412                         else {
 413                                 part_end = ptr;
 414                                 ++ptr;
 415                         }
 416                 } while ( (strcasecmp(ptr, endary)) && (ptr <= content_end) );
 417                 if (PostMultiPartCallBack != NULL) {
 418                         PostMultiPartCallBack("", "", partnum, "", NULL,
 419                                 content_type, 0, encoding, userdata);
 420                 }
 421                 goto end_parser;
 422         }
 423
 424         /* If it's not a multipart message, then do something with it */
 425         if (!is_multipart) {
 426                 part_start = ptr;
 427                 length = 0;
 428                 while (ptr < content_end) {
 429                         ++ptr;
 430                         ++length;
 431                 }
 432                 part_end = content_end;
 433
 434                 /* Truncate if the header told us to */
 435                 if ( (content_length > 0) && (length > content_length) ) {
 436                         length = content_length;
 437                 }
 438
 439                 mime_decode(partnum,
 440                             part_start, length,
 441                             content_type, encoding, disposition,
 442                             name, filename,
 443                             CallBack, NULL, NULL,
 444                             userdata, dont_decode);
 445         }
 446
 447 end_parser:     /* free the buffers!  end the oppression!! */
 448         phree(boundary);
 449         phree(startary);
 450         phree(endary);
 451         phree(header);
 452         phree(content_type);
 453         phree(encoding);
 454         phree(name);
 455         phree(filename);
 456         phree(disposition);
 457 }
 458
 459
 460
 461 /*
 462  * Entry point for the MIME parser.
 463  * (This function expects to be fed HEADERS + CONTENT)
 464  * Note: NULL can be supplied as content_end; in this case, the message is
 465  * considered to have ended when the parser encounters a 0x00 byte.
 466  */
 467 void mime_parser(char *content_start,
 468                 char *content_end,
 469
 470                  void (*CallBack)
 471                   (char *cbname,
 472                    char *cbfilename,
 473                    char *cbpartnum,
 474                    char *cbdisp,
 475                    void *cbcontent,
 476                    char *cbtype,
 477                    size_t cblength,
 478                    char *cbencoding,
 479                    void *cbuserdata),
 480
 481                  void (*PreMultiPartCallBack)
 482                   (char *cbname,
 483                    char *cbfilename,
 484                    char *cbpartnum,
 485                    char *cbdisp,
 486                    void *cbcontent,
 487                    char *cbtype,
 488                    size_t cblength,
 489                    char *cbencoding,
 490                    void *cbuserdata),
 491
 492                  void (*PostMultiPartCallBack)
 493                   (char *cbname,
 494                    char *cbfilename,
 495                    char *cbpartnum,
 496                    char *cbdisp,
 497                    void *cbcontent,
 498                    char *cbtype,
 499                    size_t cblength,
 500                    char *cbencoding,
 501                    void *cbuserdata),
 502
 503                   void *userdata,
 504                   int dont_decode
 505 )
 506 {
 507
 508         the_mime_parser("", content_start, content_end,
 509                         CallBack,
 510                         PreMultiPartCallBack,
 511                         PostMultiPartCallBack,
 512                         userdata, dont_decode);
 513 }