tools/rfc822parse.c

   1 /* rfc822parse.c - Simple mail and MIME parser
   2  *      Copyright (C) 1999, 2000 Werner Koch, Duesseldorf
   3  *      Copyright (C) 2003, 2004 g10 Code GmbH
   4  *
   5  * This program is free software; you can redistribute it and/or
   6  * modify it under the terms of the GNU Lesser General Public License
   7  * as published by the Free Software Foundation; either version 3 of
   8  * the License, or (at your option) any later version.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  13  * GNU Lesser General Public License for more details.
  14  *
  15  * You should have received a copy of the GNU Lesser General Public
  16  * License along with this program; if not, see <http://www.gnu.org/licenses/>.
  17  */
  18
  19
  20 /* According to RFC822 binary zeroes are allowed at many places. We do
  21  * not handle this correct especially in the field parsing code.  It
  22  * should be easy to fix and the API provides a interfaces which
  23  * returns the length but in addition makes sure that returned strings
  24  * are always ended by a \0.
  25  *
  26  * Furthermore, the case of field names is changed and thus it is not
  27  * always a good idea to use these modified header
  28  * lines (e.g. signatures may break).
  29  */
  30
  31 #ifdef HAVE_CONFIG_H
  32 #include <config.h>
  33 #endif
  34
  35 #include <stdlib.h>
  36 #include <stdio.h>
  37 #include <string.h>
  38 #include <errno.h>
  39 #include <stdarg.h>
  40 #include <assert.h>
  41
  42 #include "rfc822parse.h"
  43
  44 enum token_type
  45   {
  46     tSPACE,
  47     tATOM,
  48     tQUOTED,
  49     tDOMAINLIT,
  50     tSPECIAL
  51   };
  52
  53 /* For now we directly use our TOKEN as the parse context */
  54 typedef struct rfc822parse_field_context *TOKEN;
  55 struct rfc822parse_field_context
  56 {
  57   TOKEN next;
  58   enum token_type type;
  59   struct {
  60     unsigned int cont:1;
  61     unsigned int lowered:1;
  62   } flags;
  63   /*TOKEN owner_pantry; */
  64   char data[1];
  65 };
  66
  67 struct hdr_line
  68 {
  69   struct hdr_line *next;
  70   int cont;     /* This is a continuation of the previous line. */
  71   unsigned char line[1];
  72 };
  73
  74 typedef struct hdr_line *HDR_LINE;
  75
  76
  77 struct part
  78 {
  79   struct part *right;     /* The next part. */
  80   struct part *down;      /* A contained part. */
  81   HDR_LINE hdr_lines;       /* Header lines os that part. */
  82   HDR_LINE *hdr_lines_tail; /* Helper for adding lines. */
  83   char *boundary;           /* Only used in the first part. */
  84 };
  85 typedef struct part *part_t;
  86
  87 struct rfc822parse_context
  88 {
  89   rfc822parse_cb_t callback;
  90   void *callback_value;
  91   int callback_error;
  92   int in_body;
  93   int in_preamble;      /* Wether we are before the first boundary. */
  94   part_t parts;         /* The tree of parts. */
  95   part_t current_part;  /* Whom we are processing (points into parts). */
  96   const char *boundary; /* Current boundary. */
  97 };
  98
  99 static HDR_LINE find_header (rfc822parse_t msg, const char *name,
 100                              int which, HDR_LINE * rprev);
 101
 102
 103 static size_t
 104 length_sans_trailing_ws (const unsigned char *line, size_t len)
 105 {
 106   const unsigned char *p, *mark;
 107   size_t n;
 108
 109   for (mark=NULL, p=line, n=0; n < len; n++, p++)
 110     {
 111       if (strchr (" \t\r\n", *p ))
 112         {
 113           if( !mark )
 114             mark = p;
 115         }
 116       else
 117         mark = NULL;
 118     }
 119
 120   if (mark)
 121     return mark - line;
 122   return len;
 123 }
 124
 125
 126 static void
 127 lowercase_string (unsigned char *string)
 128 {
 129   for (; *string; string++)
 130     if (*string >= 'A' && *string <= 'Z')
 131       *string = *string - 'A' + 'a';
 132 }
 133
 134 /* Transform a header name into a standard capitalized format; i.e
 135    "Content-Type".  Conversion stops at the colon.  As usual we don't
 136    use the localized versions of ctype.h.
 137  */
 138 static void
 139 capitalize_header_name (unsigned char *name)
 140 {
 141   int first = 1;
 142
 143   for (; *name && *name != ':'; name++)
 144     if (*name == '-')
 145       first = 1;
 146     else if (first)
 147       {
 148         if (*name >= 'a' && *name <= 'z')
 149           *name = *name - 'a' + 'A';
 150         first = 0;
 151       }
 152     else if (*name >= 'A' && *name <= 'Z')
 153       *name = *name - 'A' + 'a';
 154 }
 155
 156 #ifndef HAVE_STPCPY
 157 static char *
 158 stpcpy (char *a,const char *b)
 159 {
 160   while (*b)
 161     *a++ = *b++;
 162   *a = 0;
 163
 164   return (char*)a;
 165 }
 166 #endif
 167
 168
 169 /* If a callback has been registerd, call it for the event of type
 170    EVENT. */
 171 static int
 172 do_callback (rfc822parse_t msg, rfc822parse_event_t event)
 173 {
 174   int rc;
 175
 176   if (!msg->callback || msg->callback_error)
 177     return 0;
 178   rc = msg->callback (msg->callback_value, event, msg);
 179   if (rc)
 180     msg->callback_error = rc;
 181   return rc;
 182 }
 183
 184 static part_t
 185 new_part (void)
 186 {
 187   part_t part;
 188
 189   part = calloc (1, sizeof *part);
 190   if (part)
 191     {
 192       part->hdr_lines_tail = &part->hdr_lines;
 193     }
 194   return part;
 195 }
 196
 197
 198 static void
 199 release_part (part_t part)
 200 {
 201   part_t tmp;
 202   HDR_LINE hdr, hdr2;
 203
 204   for (; part; part = tmp)
 205     {
 206       tmp = part->right;
 207       if (part->down)
 208         release_part (part->down);
 209       for (hdr = part->hdr_lines; hdr; hdr = hdr2)
 210         {
 211           hdr2 = hdr->next;
 212           free (hdr);
 213         }
 214       free (part->boundary);
 215       free (part);
 216     }
 217 }
 218
 219
 220 static void
 221 release_handle_data (rfc822parse_t msg)
 222 {
 223   release_part (msg->parts);
 224   msg->parts = NULL;
 225   msg->current_part = NULL;
 226   msg->boundary = NULL;
 227 }
 228
 229
 230 /* Create a new parsing context for an entire rfc822 message and
 231    return it.  CB and CB_VALUE may be given to callback for certain
 232    events.  NULL is returned on error with errno set appropriately. */
 233 rfc822parse_t
 234 rfc822parse_open (rfc822parse_cb_t cb, void *cb_value)
 235 {
 236   rfc822parse_t msg = calloc (1, sizeof *msg);
 237   if (msg)
 238     {
 239       msg->parts = msg->current_part = new_part ();
 240       if (!msg->parts)
 241         {
 242           free (msg);
 243           msg = NULL;
 244         }
 245       else
 246         {
 247           msg->callback = cb;
 248           msg->callback_value = cb_value;
 249           if (do_callback (msg, RFC822PARSE_OPEN))
 250             {
 251               release_handle_data (msg);
 252               free (msg);
 253               msg = NULL;
 254             }
 255         }
 256     }
 257   return msg;
 258 }
 259
 260
 261 void
 262 rfc822parse_cancel (rfc822parse_t msg)
 263 {
 264   if (msg)
 265     {
 266       do_callback (msg, RFC822PARSE_CANCEL);
 267       release_handle_data (msg);
 268       free (msg);
 269     }
 270 }
 271
 272
 273 void
 274 rfc822parse_close (rfc822parse_t msg)
 275 {
 276   if (msg)
 277     {
 278       do_callback (msg, RFC822PARSE_CLOSE);
 279       release_handle_data (msg);
 280       free (msg);
 281     }
 282 }
 283
 284 static part_t
 285 find_parent (part_t tree, part_t target)
 286 {
 287   part_t part;
 288
 289   for (part = tree->down; part; part = part->right)
 290     {
 291       if (part == target)
 292         return tree; /* Found. */
 293       if (part->down)
 294         {
 295           part_t tmp = find_parent (part, target);
 296           if (tmp)
 297             return tmp;
 298         }
 299     }
 300   return NULL;
 301 }
 302
 303 static void
 304 set_current_part_to_parent (rfc822parse_t msg)
 305 {
 306   part_t parent;
 307
 308   assert (msg->current_part);
 309   parent = find_parent (msg->parts, msg->current_part);
 310   if (!parent)
 311     return; /* Already at the top. */
 312
 313 #ifndef NDEBUG
 314   {
 315     part_t part;
 316     for (part = parent->down; part; part = part->right)
 317       if (part == msg->current_part)
 318         break;
 319     assert (part);
 320   }
 321 #endif
 322   msg->current_part = parent;
 323
 324   parent = find_parent (msg->parts, parent);
 325   msg->boundary = parent? parent->boundary: NULL;
 326 }
 327
 328
 329
 330 /****************
 331  * We have read in all header lines and are about to receive the body
 332  * part.  The delimiter line has already been processed.
 333  *
 334  * FIXME: we's better return an error in case of memory failures.
 335  */
 336 static int
 337 transition_to_body (rfc822parse_t msg)
 338 {
 339   rfc822parse_field_t ctx;
 340   int rc;
 341
 342   rc = do_callback (msg, RFC822PARSE_T2BODY);
 343   if (!rc)
 344     {
 345       /* Store the boundary if we have multipart type. */
 346       ctx = rfc822parse_parse_field (msg, "Content-Type", -1);
 347       if (ctx)
 348         {
 349           const char *s;
 350
 351           s = rfc822parse_query_media_type (ctx, NULL);
 352           if (s && !strcmp (s,"multipart"))
 353             {
 354               s = rfc822parse_query_parameter (ctx, "boundary", 0);
 355               if (s)
 356                 {
 357                   assert (!msg->current_part->boundary);
 358                   msg->current_part->boundary = malloc (strlen (s) + 1);
 359                   if (msg->current_part->boundary)
 360                     {
 361                       part_t part;
 362
 363                       strcpy (msg->current_part->boundary, s);
 364                       msg->boundary = msg->current_part->boundary;
 365                       part = new_part ();
 366                       if (!part)
 367                         {
 368                           int save_errno = errno;
 369                           rfc822parse_release_field (ctx);
 370                           errno = save_errno;
 371                           return -1;
 372                         }
 373                       rc = do_callback (msg, RFC822PARSE_LEVEL_DOWN);
 374                       assert (!msg->current_part->down);
 375                       msg->current_part->down = part;
 376                       msg->current_part = part;
 377                       msg->in_preamble = 1;
 378                     }
 379                 }
 380             }
 381           rfc822parse_release_field (ctx);
 382         }
 383     }
 384
 385   return rc;
 386 }
 387
 388 /* We have just passed a MIME boundary and need to prepare for new part.
 389    headers. */
 390 static int
 391 transition_to_header (rfc822parse_t msg)
 392 {
 393   part_t part;
 394
 395   assert (msg->current_part);
 396   assert (!msg->current_part->right);
 397
 398   part = new_part ();
 399   if (!part)
 400     return -1;
 401
 402   msg->current_part->right = part;
 403   msg->current_part = part;
 404   return 0;
 405 }
 406
 407
 408 static int
 409 insert_header (rfc822parse_t msg, const unsigned char *line, size_t length)
 410 {
 411   HDR_LINE hdr;
 412
 413   assert (msg->current_part);
 414   if (!length)
 415     {
 416       msg->in_body = 1;
 417       return transition_to_body (msg);
 418     }
 419
 420   if (!msg->current_part->hdr_lines)
 421     do_callback (msg, RFC822PARSE_BEGIN_HEADER);
 422
 423   length = length_sans_trailing_ws (line, length);
 424   hdr = malloc (sizeof (*hdr) + length);
 425   if (!hdr)
 426     return -1;
 427   hdr->next = NULL;
 428   hdr->cont = (*line == ' ' || *line == '\t');
 429   memcpy (hdr->line, line, length);
 430   hdr->line[length] = 0; /* Make it a string. */
 431
 432   /* Transform a field name into canonical format. */
 433   if (!hdr->cont && strchr (line, ':'))
 434      capitalize_header_name (hdr->line);
 435
 436   *msg->current_part->hdr_lines_tail = hdr;
 437   msg->current_part->hdr_lines_tail = &hdr->next;
 438
 439   /* Lets help the caller to prevent mail loops and issue an event for
 440    * every Received header. */
 441   if (length >= 9 && !memcmp (line, "Received:", 9))
 442      do_callback (msg, RFC822PARSE_RCVD_SEEN);
 443   return 0;
 444 }
 445
 446
 447 /****************
 448  * Note: We handle the body transparent to allow binary zeroes in it.
 449  */
 450 static int
 451 insert_body (rfc822parse_t msg, const unsigned char *line, size_t length)
 452 {
 453   int rc = 0;
 454
 455   if (length > 2 && *line == '-' && line[1] == '-' && msg->boundary)
 456     {
 457       size_t blen = strlen (msg->boundary);
 458
 459       if (length == blen + 2
 460           && !memcmp (line+2, msg->boundary, blen))
 461         {
 462           rc = do_callback (msg, RFC822PARSE_BOUNDARY);
 463           msg->in_body = 0;
 464           if (!rc && !msg->in_preamble)
 465             rc = transition_to_header (msg);
 466           msg->in_preamble = 0;
 467         }
 468       else if (length == blen + 4
 469           && line[length-2] =='-' && line[length-1] == '-'
 470           && !memcmp (line+2, msg->boundary, blen))
 471         {
 472           rc = do_callback (msg, RFC822PARSE_LAST_BOUNDARY);
 473           msg->boundary = NULL; /* No current boundary anymore. */
 474           set_current_part_to_parent (msg);
 475
 476           /* Fixme: The next should actually be send right before the
 477              next boundary, so that we can mark the epilogue. */
 478           if (!rc)
 479             rc = do_callback (msg, RFC822PARSE_LEVEL_UP);
 480         }
 481     }
 482   if (msg->in_preamble && !rc)
 483     rc = do_callback (msg, RFC822PARSE_PREAMBLE);
 484
 485   return rc;
 486 }
 487
 488 /* Insert the next line into the parser. Return 0 on success or true
 489    on error with errno set appropriately. */
 490 int
 491 rfc822parse_insert (rfc822parse_t msg, const unsigned char *line, size_t length)
 492 {
 493   return (msg->in_body
 494           ? insert_body (msg, line, length)
 495           : insert_header (msg, line, length));
 496 }
 497
 498
 499 /* Tell the parser that we have finished the message. */
 500 int
 501 rfc822parse_finish (rfc822parse_t msg)
 502 {
 503   return do_callback (msg, RFC822PARSE_FINISH);
 504 }
 505
 506
 507
 508 /****************
 509  * Get a copy of a header line. The line is returned as one long
 510  * string with LF to separate the continuation line. Caller must free
 511  * the return buffer.  WHICH may be used to enumerate over all lines.
 512  * Wildcards are allowed.  This function works on the current headers;
 513  * i.e. the regular mail headers or the MIME headers of the current
 514  * part.
 515  *
 516  * WHICH gives the mode:
 517  *  -1 := Take the last occurence
 518  *   n := Take the n-th  one.
 519  *
 520  * Returns a newly allocated buffer or NULL on error.  errno is set in
 521  * case of a memory failure or set to 0 if the requested field is not
 522  * available.
 523  *
 524  * If VALUEOFF is not NULL it will receive the offset of the first non
 525  * space character in the value part of the line (i.e. after the first
 526  * colon).
 527  */
 528 char *
 529 rfc822parse_get_field (rfc822parse_t msg, const char *name, int which,
 530                        size_t *valueoff)
 531 {
 532   HDR_LINE h, h2;
 533   char *buf, *p;
 534   size_t n;
 535
 536   h = find_header (msg, name, which, NULL);
 537   if (!h)
 538     {
 539       errno = 0;
 540       return NULL; /* no such field */
 541     }
 542
 543   n = strlen (h->line) + 1;
 544   for (h2 = h->next; h2 && h2->cont; h2 = h2->next)
 545     n += strlen (h2->line) + 1;
 546
 547   buf = p = malloc (n);
 548   if (buf)
 549     {
 550       p = stpcpy (p, h->line);
 551       *p++ = '\n';
 552       for (h2 = h->next; h2 && h2->cont; h2 = h2->next)
 553         {
 554           p = stpcpy (p, h2->line);
 555           *p++ = '\n';
 556         }
 557       p[-1] = 0;
 558     }
 559
 560   if (valueoff)
 561     {
 562       p = strchr (buf, ':');
 563       if (!p)
 564         *valueoff = 0; /* Oops: should never happen. */
 565       else
 566         {
 567           p++;
 568           while (*p == ' ' || *p == '\t' || *p == '\r' || *p == '\n')
 569             p++;
 570           *valueoff = p - buf;
 571         }
 572     }
 573
 574   return buf;
 575 }
 576
 577
 578 /****************
 579  * Enumerate all header.  Caller has to provide the address of a pointer
 580  * which has to be initialzed to NULL, the caller should then never change this
 581  * pointer until he has closed the enumeration by passing again the address
 582  * of the pointer but with msg set to NULL.
 583  * The function returns pointers to all the header lines or NULL when
 584  * all lines have been enumerated or no headers are available.
 585  */
 586 const char *
 587 rfc822parse_enum_header_lines (rfc822parse_t msg, void **context)
 588 {
 589   HDR_LINE l;
 590
 591   if (!msg) /* Close. */
 592     return NULL;
 593
 594   if (*context == msg || !msg->current_part)
 595     return NULL;
 596
 597   l = *context ? (HDR_LINE) *context : msg->current_part->hdr_lines;
 598
 599   if (l)
 600     {
 601       *context = l->next ? (void *) (l->next) : (void *) msg;
 602       return l->line;
 603     }
 604   *context = msg; /* Mark end of list. */
 605   return NULL;
 606 }
 607
 608
 609
 610 /****************
 611  * Find a header field.  If the Name does end in an asterisk this is meant
 612  * to be a wildcard.
 613  *
 614  *  which  -1 : Retrieve the last field
 615  *         >0 : Retrieve the n-th field
 616
 617  * RPREV may be used to return the predecessor of the returned field;
 618  * which may be NULL for the very first one. It has to be initialzed
 619  * to either NULL in which case the search start at the first header line,
 620  * or it may point to a headerline, where the search should start
 621  */
 622 static HDR_LINE
 623 find_header (rfc822parse_t msg, const char *name, int which, HDR_LINE *rprev)
 624 {
 625   HDR_LINE hdr, prev = NULL, mark = NULL;
 626   unsigned char *p;
 627   size_t namelen, n;
 628   int found = 0;
 629   int glob = 0;
 630
 631   if (!msg->current_part)
 632     return NULL;
 633
 634   namelen = strlen (name);
 635   if (namelen && name[namelen - 1] == '*')
 636     {
 637       namelen--;
 638       glob = 1;
 639     }
 640
 641   hdr = msg->current_part->hdr_lines;
 642   if (rprev && *rprev)
 643     {
 644       /* spool forward to the requested starting place.
 645        * we cannot simply set this as we have to return
 646        * the previous list element too */
 647       for (; hdr && hdr != *rprev; prev = hdr, hdr = hdr->next)
 648         ;
 649     }
 650
 651   for (; hdr; prev = hdr, hdr = hdr->next)
 652     {
 653       if (hdr->cont)
 654         continue;
 655       if (!(p = strchr (hdr->line, ':')))
 656         continue;               /* invalid header, just skip it. */
 657       n = p - hdr->line;
 658       if (!n)
 659         continue;               /* invalid name */
 660       if ((glob ? (namelen <= n) : (namelen == n))
 661           && !memcmp (hdr->line, name, namelen))
 662         {
 663           found++;
 664           if (which == -1)
 665             mark = hdr;
 666           else if (found == which)
 667             {
 668               if (rprev)
 669                 *rprev = prev;
 670               return hdr;
 671             }
 672         }
 673     }
 674   if (mark && rprev)
 675     *rprev = prev;
 676   return mark;
 677 }
 678
 679
 680
 681 static const char *
 682 skip_ws (const char *s)
 683 {
 684   while (*s == ' ' || *s == '\t' || *s == '\r' || *s == '\n')
 685     s++;
 686   return s;
 687 }
 688
 689
 690 static void
 691 release_token_list (TOKEN t)
 692 {
 693   while (t)
 694     {
 695       TOKEN t2 = t->next;
 696       /* fixme: If we have owner_pantry, put the token back to
 697        * this pantry so that it can be reused later */
 698       free (t);
 699       t = t2;
 700     }
 701 }
 702
 703
 704 static TOKEN
 705 new_token (enum token_type type, const char *buf, size_t length)
 706 {
 707   TOKEN t;
 708
 709   /* fixme: look through our pantries to find a suitable
 710    * token for reuse */
 711   t = malloc (sizeof *t + length);
 712   if (t)
 713     {
 714       t->next = NULL;
 715       t->type = type;
 716       memset (&t->flags, 0, sizeof (t->flags));
 717       t->data[0] = 0;
 718       if (buf)
 719         {
 720           memcpy (t->data, buf, length);
 721           t->data[length] = 0;  /* Make sure it is a C string. */
 722         }
 723       else
 724         t->data[0] = 0;
 725     }
 726   return t;
 727 }
 728
 729 static TOKEN
 730 append_to_token (TOKEN old, const char *buf, size_t length)
 731 {
 732   size_t n = strlen (old->data);
 733   TOKEN t;
 734
 735   t = malloc (sizeof *t + n + length);
 736   if (t)
 737     {
 738       t->next = old->next;
 739       t->type = old->type;
 740       t->flags = old->flags;
 741       memcpy (t->data, old->data, n);
 742       memcpy (t->data + n, buf, length);
 743       t->data[n + length] = 0;
 744       old->next = NULL;
 745       release_token_list (old);
 746     }
 747   return t;
 748 }
 749
 750
 751
 752 /*
 753    Parse a field into tokens as defined by rfc822.
 754  */
 755 static TOKEN
 756 parse_field (HDR_LINE hdr)
 757 {
 758   static const char specials[] = "<>@.,;:\\[]\"()";
 759   static const char specials2[] = "<>@.,;:";
 760   static const char tspecials[] = "/?=<>@,;:\\[]\"()";
 761   static const char tspecials2[] = "/?=<>@.,;:";  /* FIXME: really
 762                                                      include '.'?*/
 763   static struct
 764   {
 765     const unsigned char *name;
 766     size_t namelen;
 767   } tspecial_header[] = {
 768     { "Content-Type", 12},
 769     { "Content-Transfer-Encoding", 25},
 770     { "Content-Disposition", 19},
 771     { NULL, 0}
 772   };
 773   const char *delimiters;
 774   const char *delimiters2;
 775   const unsigned char *line, *s, *s2;
 776   size_t n;
 777   int i, invalid = 0;
 778   TOKEN t, tok, *tok_tail;
 779
 780   errno = 0;
 781   if (!hdr)
 782     return NULL;
 783
 784   tok = NULL;
 785   tok_tail = &tok;
 786
 787   line = hdr->line;
 788   if (!(s = strchr (line, ':')))
 789     return NULL; /* oops */
 790
 791   n = s - line;
 792   if (!n)
 793     return NULL; /* oops: invalid name */
 794
 795   delimiters = specials;
 796   delimiters2 = specials2;
 797   for (i = 0; tspecial_header[i].name; i++)
 798     {
 799       if (n == tspecial_header[i].namelen
 800           && !memcmp (line, tspecial_header[i].name, n))
 801         {
 802           delimiters = tspecials;
 803           delimiters2 = tspecials2;
 804           break;
 805         }
 806     }
 807
 808   s++; /* Move over the colon. */
 809   for (;;)
 810     {
 811       if (!*s)
 812         {
 813           if (!hdr->next || !hdr->next->cont)
 814             break;
 815           hdr = hdr->next;
 816           s = hdr->line;
 817         }
 818
 819       if (*s == '(')
 820         {
 821           int level = 1;
 822           int in_quote = 0;
 823
 824           invalid = 0;
 825           for (s++;; s++)
 826             {
 827               if (!*s)
 828                 {
 829                   if (!hdr->next || !hdr->next->cont)
 830                     break;
 831                   hdr = hdr->next;
 832                   s = hdr->line;
 833                 }
 834
 835               if (in_quote)
 836                 {
 837                   if (*s == '\"')
 838                     in_quote = 0;
 839                   else if (*s == '\\' && s[1])  /* what about continuation? */
 840                     s++;
 841                 }
 842               else if (*s == ')')
 843                 {
 844                   if (!--level)
 845                     break;
 846                 }
 847               else if (*s == '(')
 848                 level++;
 849               else if (*s == '\"')
 850                 in_quote = 1;
 851             }
 852           if (!*s)
 853             ; /* Actually this is an error, but we don't care about it. */
 854           else
 855             s++;
 856         }
 857       else if (*s == '\"' || *s == '[')
 858         {
 859           /* We do not check for non-allowed nesting of domainliterals */
 860           int term = *s == '\"' ? '\"' : ']';
 861           invalid = 0;
 862           s++;
 863           t = NULL;
 864
 865           for (;;)
 866             {
 867               for (s2 = s; *s2; s2++)
 868                 {
 869                   if (*s2 == term)
 870                     break;
 871                   else if (*s2 == '\\' && s2[1]) /* what about continuation? */
 872                     s2++;
 873                 }
 874
 875               t = (t
 876                    ? append_to_token (t, s, s2 - s)
 877                    : new_token (term == '\"'? tQUOTED : tDOMAINLIT, s, s2 - s));
 878               if (!t)
 879                 goto failure;
 880
 881               if (*s2 || !hdr->next || !hdr->next->cont)
 882                 break;
 883               hdr = hdr->next;
 884               s = hdr->line;
 885             }
 886           *tok_tail = t;
 887           tok_tail = &t->next;
 888           s = s2;
 889           if (*s)
 890             s++; /* skip the delimiter */
 891         }
 892       else if ((s2 = strchr (delimiters2, *s)))
 893         { /* Special characters which are not handled above. */
 894           invalid = 0;
 895           t = new_token (tSPECIAL, s, 1);
 896           if (!t)
 897             goto failure;
 898           *tok_tail = t;
 899           tok_tail = &t->next;
 900           s++;
 901         }
 902       else if (*s == ' ' || *s == '\t' || *s == '\r' || *s == '\n')
 903         {
 904           invalid = 0;
 905           s = skip_ws (s + 1);
 906         }
 907       else if (*s > 0x20 && !(*s & 128))
 908         { /* Atom. */
 909           invalid = 0;
 910           for (s2 = s + 1; *s2 > 0x20
 911                && !(*s2 & 128) && !strchr (delimiters, *s2); s2++)
 912             ;
 913           t = new_token (tATOM, s, s2 - s);
 914           if (!t)
 915             goto failure;
 916           *tok_tail = t;
 917           tok_tail = &t->next;
 918           s = s2;
 919         }
 920       else
 921         { /* Invalid character. */
 922           if (!invalid)
 923             { /* For parsing we assume only one space. */
 924               t = new_token (tSPACE, NULL, 0);
 925               if (!t)
 926                 goto failure;
 927               *tok_tail = t;
 928               tok_tail = &t->next;
 929               invalid = 1;
 930             }
 931           s++;
 932         }
 933     }
 934
 935   return tok;
 936
 937  failure:
 938   {
 939     int save = errno;
 940     release_token_list (tok);
 941     errno = save;
 942   }
 943   return NULL;
 944 }
 945
 946
 947
 948
 949 /****************
 950  * Find and parse a header field.
 951  * WHICH indicates what to do if there are multiple instance of the same
 952  * field (like "Received"); the following value are defined:
 953  *  -1 := Take the last occurence
 954  *   0 := Reserved
 955  *   n := Take the n-th one.
 956  * Returns a handle for further operations on the parse context of the field
 957  * or NULL if the field was not found.
 958  */
 959 rfc822parse_field_t
 960 rfc822parse_parse_field (rfc822parse_t msg, const char *name, int which)
 961 {
 962   HDR_LINE hdr;
 963
 964   if (!which)
 965     return NULL;
 966
 967   hdr = find_header (msg, name, which, NULL);
 968   if (!hdr)
 969     return NULL;
 970   return parse_field (hdr);
 971 }
 972
 973 void
 974 rfc822parse_release_field (rfc822parse_field_t ctx)
 975 {
 976   if (ctx)
 977     release_token_list (ctx);
 978 }
 979
 980
 981
 982 /****************
 983  * Check whether T points to a parameter.
 984  * A parameter starts with a semicolon and it is assumed that t
 985  * points to exactly this one.
 986  */
 987 static int
 988 is_parameter (TOKEN t)
 989 {
 990   t = t->next;
 991   if (!t || t->type != tATOM)
 992     return 0;
 993   t = t->next;
 994   if (!t || !(t->type == tSPECIAL && t->data[0] == '='))
 995     return 0;
 996   t = t->next;
 997   if (!t)
 998     return 1; /* We assume that an non existing value is an empty one. */
 999   return t->type == tQUOTED || t->type == tATOM;
1000 }
1001
1002 /*
1003    Some header (Content-type) have a special syntax where attribute=value
1004    pairs are used after a leading semicolon.  The parse_field code
1005    knows about these fields and changes the parsing to the one defined
1006    in RFC2045.
1007    Returns a pointer to the value which is valid as long as the
1008    parse context is valid; NULL is returned in case that attr is not
1009    defined in the header, a missing value is reppresented by an empty string.
1010
1011    With LOWER_VALUE set to true, a matching field valuebe be
1012    lowercased.
1013
1014    Note, that ATTR should be lowercase.
1015  */
1016 const char *
1017 rfc822parse_query_parameter (rfc822parse_field_t ctx, const char *attr,
1018                              int lower_value)
1019 {
1020   TOKEN t, a;
1021
1022   for (t = ctx; t; t = t->next)
1023     {
1024       /* skip to the next semicolon */
1025       for (; t && !(t->type == tSPECIAL && t->data[0] == ';'); t = t->next)
1026         ;
1027       if (!t)
1028         return NULL;
1029       if (is_parameter (t))
1030         { /* Look closer. */
1031           a = t->next; /* We know that this is an atom */
1032           if ( !a->flags.lowered )
1033             {
1034               lowercase_string (a->data);
1035               a->flags.lowered = 1;
1036             }
1037           if (!strcmp (a->data, attr))
1038             { /* found */
1039               t = a->next->next;
1040               /* Either T is now an atom, a quoted string or NULL in
1041                * which case we return an empty string. */
1042
1043               if ( lower_value && t && !t->flags.lowered )
1044                 {
1045                   lowercase_string (t->data);
1046                   t->flags.lowered = 1;
1047                 }
1048               return t ? t->data : "";
1049             }
1050         }
1051     }
1052   return NULL;
1053 }
1054
1055 /****************
1056  * This function may be used for the Content-Type header to figure out
1057  * the media type and subtype.  Note, that the returned strings are
1058  * guaranteed to be lowercase as required by MIME.
1059  *
1060  * Returns: a pointer to the media type and if subtype is not NULL,
1061  *          a pointer to the subtype.
1062  */
1063 const char *
1064 rfc822parse_query_media_type (rfc822parse_field_t ctx, const char **subtype)
1065 {
1066   TOKEN t = ctx;
1067   const char *type;
1068
1069   if (t->type != tATOM)
1070     return NULL;
1071   if (!t->flags.lowered)
1072     {
1073       lowercase_string (t->data);
1074       t->flags.lowered = 1;
1075     }
1076   type = t->data;
1077   t = t->next;
1078   if (!t || t->type != tSPECIAL || t->data[0] != '/')
1079     return NULL;
1080   t = t->next;
1081   if (!t || t->type != tATOM)
1082     return NULL;
1083
1084   if (subtype)
1085     {
1086       if (!t->flags.lowered)
1087         {
1088           lowercase_string (t->data);
1089           t->flags.lowered = 1;
1090         }
1091       *subtype = t->data;
1092     }
1093   return type;
1094 }
1095
1096
1097
1098
1099
1100 #ifdef TESTING
1101
1102 /* Internal debug function to print the structure of the message. */
1103 static void
1104 dump_structure (rfc822parse_t msg, part_t part, int indent)
1105 {
1106   if (!part)
1107     {
1108       printf ("*** Structure of this message:\n");
1109       part = msg->parts;
1110     }
1111
1112   for (; part; part = part->right)
1113     {
1114       rfc822parse_field_t ctx;
1115       part_t save_part; /* ugly hack - we should have a function to
1116                            get part inforation. */
1117       const char *s;
1118
1119       save_part = msg->current_part;
1120       msg->current_part = part;
1121       ctx = rfc822parse_parse_field (msg, "Content-Type", -1);
1122       msg->current_part = save_part;
1123       if (ctx)
1124         {
1125           const char *s1, *s2;
1126           s1 = rfc822parse_query_media_type (ctx, &s2);
1127           if (s1)
1128             printf ("***   %*s %s/%s", indent*2, "", s1, s2);
1129           else
1130             printf ("***   %*s [not found]", indent*2, "");
1131
1132           s = rfc822parse_query_parameter (ctx, "boundary", 0);
1133           if (s)
1134             printf (" (boundary=\"%s\")", s);
1135           rfc822parse_release_field (ctx);
1136         }
1137       else
1138         printf ("***   %*s text/plain [assumed]", indent*2, "");
1139       putchar('\n');
1140
1141       if (part->down)
1142         dump_structure (msg, part->down, indent + 1);
1143     }
1144
1145 }
1146
1147
1148
1149 static void
1150 show_param (rfc822parse_field_t ctx, const char *name)
1151 {
1152   const char *s;
1153
1154   if (!ctx)
1155     return;
1156   s = rfc822parse_query_parameter (ctx, name, 0);
1157   if (s)
1158     printf ("***   %s: `%s'\n", name, s);
1159 }
1160
1161
1162
1163 static void
1164 show_event (rfc822parse_event_t event)
1165 {
1166   const char *s;
1167
1168   switch (event)
1169     {
1170     case RFC822PARSE_OPEN: s= "Open"; break;
1171     case RFC822PARSE_CLOSE: s= "Close"; break;
1172     case RFC822PARSE_CANCEL: s= "Cancel"; break;
1173     case RFC822PARSE_T2BODY: s= "T2Body"; break;
1174     case RFC822PARSE_FINISH: s= "Finish"; break;
1175     case RFC822PARSE_RCVD_SEEN: s= "Rcvd_Seen"; break;
1176     case RFC822PARSE_BOUNDARY: s= "Boundary"; break;
1177     case RFC822PARSE_LAST_BOUNDARY: s= "Last_Boundary"; break;
1178     default: s= "***invalid event***"; break;
1179     }
1180   printf ("*** got RFC822 event %s\n", s);
1181 }
1182
1183 static int
1184 msg_cb (void *dummy_arg, rfc822parse_event_t event, rfc822parse_t msg)
1185 {
1186   show_event (event);
1187   if (event == RFC822PARSE_T2BODY)
1188     {
1189       rfc822parse_field_t ctx;
1190       void *ectx;
1191       const char *line;
1192
1193       for (ectx=NULL; (line = rfc822parse_enum_header_lines (msg, &ectx)); )
1194         {
1195           printf ("*** HDR: %s\n", line);
1196         }
1197       rfc822parse_enum_header_lines (NULL, &ectx); /* Close enumerator. */
1198
1199       ctx = rfc822parse_parse_field (msg, "Content-Type", -1);
1200       if (ctx)
1201         {
1202           const char *s1, *s2;
1203           s1 = rfc822parse_query_media_type (ctx, &s2);
1204           if (s1)
1205             printf ("***   media: `%s/%s'\n", s1, s2);
1206           else
1207             printf ("***   media: [not found]\n");
1208           show_param (ctx, "boundary");
1209           show_param (ctx, "protocol");
1210           rfc822parse_release_field (ctx);
1211         }
1212       else
1213         printf ("***   media: text/plain [assumed]\n");
1214
1215     }
1216
1217
1218   return 0;
1219 }
1220
1221
1222
1223 int
1224 main (int argc, char **argv)
1225 {
1226   char line[5000];
1227   size_t length;
1228   rfc822parse_t msg;
1229
1230   msg = rfc822parse_open (msg_cb, NULL);
1231   if (!msg)
1232     abort ();
1233
1234   while (fgets (line, sizeof (line), stdin))
1235     {
1236       length = strlen (line);
1237       if (length && line[length - 1] == '\n')
1238         line[--length] = 0;
1239       if (length && line[length - 1] == '\r')
1240         line[--length] = 0;
1241       if (rfc822parse_insert (msg, line, length))
1242         abort ();
1243     }
1244
1245   dump_structure (msg, NULL, 0);
1246
1247   rfc822parse_close (msg);
1248   return 0;
1249 }
1250 #endif
1251
1252 /*
1253 Local Variables:
1254 compile-command: "gcc -Wall -Wno-pointer-sign -g -DTESTING -o rfc822parse rfc822parse.c"
1255 End:
1256 */