diplodocus.org Git - nmh/blob - sbr/mf.c

   1 /* mf.c -- mail filter subroutines
   2  *
   3  * This code is Copyright (c) 2002, by the authors of nmh.  See the
   4  * COPYRIGHT file in the root directory of the nmh distribution for
   5  * complete copyright information.
   6  */
   7
   8 #include <h/mh.h>
   9 #include <h/mf.h>
  10 #include <h/utils.h>
  11
  12 /*
  13  * static prototypes
  14  */
  15 static int isat (const char *);
  16 static int parse_address (void);
  17 static int phrase (char *);
  18 static int route_addr (char *);
  19 static int local_part (char *);
  20 static int domain (char *);
  21 static int route (char *);
  22 static int my_lex (char *);
  23
  24
  25 static int
  26 isat (const char *p)
  27 {
  28     return *p == ' ' &&
  29         (p[1] == 'a' || p[1] == 'A') &&
  30         (p[2] == 't' || p[2] == 'T') &&
  31         p[3] == ' ';
  32 }
  33
  34
  35 /*
  36  *
  37  * getadrx() implements a partial 822-style address parser.  The parser
  38  * is neither complete nor correct.  It does however recognize nearly all
  39  * of the 822 address syntax.  In addition it handles the majority of the
  40  * 733 syntax as well.  Most problems arise from trying to accommodate both.
  41  *
  42  * In terms of 822, the route-specification in
  43  *
  44  *               "<" [route] local-part "@" domain ">"
  45  *
  46  * is parsed and returned unchanged.  Multiple at-signs are compressed
  47  * via source-routing.  Recursive groups are not allowed as per the
  48  * standard.
  49  *
  50  * In terms of 733, " at " is recognized as equivalent to "@".
  51  *
  52  * In terms of both the parser will not complain about missing hosts.
  53  *
  54  * -----
  55  *
  56  * We should not allow addresses like
  57  *
  58  *              Marshall T. Rose <MRose@UCI>
  59  *
  60  * but should insist on
  61  *
  62  *              "Marshall T. Rose" <MRose@UCI>
  63  *
  64  * Unfortunately, a lot of mailers stupidly let people get away with this.
  65  *
  66  * -----
  67  *
  68  * We should not allow addresses like
  69  *
  70  *              <MRose@UCI>
  71  *
  72  * but should insist on
  73  *
  74  *              MRose@UCI
  75  *
  76  * Unfortunately, a lot of mailers stupidly let people's UAs get away with
  77  * this.
  78  *
  79  * -----
  80  *
  81  * We should not allow addresses like
  82  *
  83  *              @UCI:MRose@UCI-750a
  84  *
  85  * but should insist on
  86  *
  87  *              Marshall Rose <@UCI:MRose@UCI-750a>
  88  *
  89  * Unfortunately, a lot of mailers stupidly do this.
  90  *
  91  */
  92
  93 #define LX_END   0
  94 #define LX_ERR   1
  95 #define LX_ATOM  2
  96 #define LX_QSTR  3
  97 #define LX_DLIT  4
  98 #define LX_SEMI  5
  99 #define LX_COMA  6
 100 #define LX_LBRK  7
 101 #define LX_RBRK  8
 102 #define LX_COLN  9
 103 #define LX_DOT  10
 104 #define LX_AT   11
 105
 106 struct specials {
 107     char lx_chr;
 108     int  lx_val;
 109 };
 110
 111 static struct specials special[] = {
 112     { ';',   LX_SEMI },
 113     { ',',   LX_COMA },
 114     { '<',   LX_LBRK },
 115     { '>',   LX_RBRK },
 116     { ':',   LX_COLN },
 117     { '.',   LX_DOT },
 118     { '@',   LX_AT },
 119     { '(',   LX_ERR },
 120     { ')',   LX_ERR },
 121     { '\\',  LX_ERR },
 122     { '"',   LX_ERR },
 123     { '[',   LX_ERR },
 124     { ']',   LX_ERR },
 125     { 0,     0 }
 126 };
 127
 128 static int glevel = 0;
 129 static int ingrp = 0;
 130 static int last_lex = LX_END;
 131
 132 static char *dp = NULL;
 133 static char *cp = NULL;
 134 static char *ap = NULL;
 135 static char *pers = NULL;
 136 static char *mbox = NULL;
 137 static char *host = NULL;
 138 static char *routepath = NULL;
 139 static char *grp = NULL;
 140 static char *note = NULL;
 141 static char err[BUFSIZ];
 142 static char adr[BUFSIZ];
 143
 144 static struct adrx  adrxs2;
 145
 146
 147 /* eai = Email Address Internationalization */
 148 struct adrx *
 149 getadrx (const char *addrs, int eai)
 150 {
 151     int parse;
 152     char *bp;
 153     struct adrx *adrxp = &adrxs2;
 154
 155     free(pers);
 156     free(mbox);
 157     free(host);
 158     free(routepath);
 159     free(grp);
 160     free(note);
 161     pers = mbox = host = routepath = grp = note = NULL;
 162     err[0] = 0;
 163
 164     if (dp == NULL) {
 165         dp = cp = strdup (FENDNULL(addrs));
 166         glevel = 0;
 167     } else if (cp == NULL) {
 168         free (dp);
 169         dp = NULL;
 170         return NULL;
 171     }
 172
 173     parse = parse_address();
 174     if (parse == DONE) {
 175         free(dp);
 176         dp = cp = NULL;
 177         return NULL;
 178     }
 179     if (parse == OK && last_lex != LX_COMA && last_lex != LX_END) {
 180         /* catch trailing comments */
 181         bp = cp;
 182         my_lex(adr);
 183         cp = bp;
 184     }
 185
 186     /* Reject the address if key fields contain 8bit characters. */
 187     if (!eai &&
 188         (contains8bit(mbox, NULL) || contains8bit(host, NULL) ||
 189         contains8bit(routepath, NULL) || contains8bit(grp, NULL)))
 190         strcpy(err, "Address contains 8-bit characters");
 191
 192     if (err[0])
 193         while (last_lex != LX_COMA && last_lex != LX_END)
 194             my_lex(adr);
 195
 196     while (isspace ((unsigned char) *ap))
 197         ap++;
 198     if (cp)
 199         snprintf(adr, sizeof adr, "%.*s", (int)(cp - ap), ap);
 200     else
 201         strcpy (adr, ap);
 202     bp = adr + strlen (adr) - 1;
 203     if (*bp == ',' || *bp == ';' || *bp == '\n')
 204         *bp = 0;
 205
 206     adrxp->text = adr;
 207     adrxp->pers = pers;
 208     adrxp->mbox = mbox;
 209     adrxp->host = host;
 210     adrxp->path = routepath;
 211     adrxp->grp = grp;
 212     adrxp->ingrp = ingrp;
 213     adrxp->note = note;
 214     adrxp->err = err[0] ? err : NULL;
 215
 216     return adrxp;
 217 }
 218
 219
 220 static int
 221 parse_address (void)
 222 {
 223     char buffer[BUFSIZ];
 224
 225 again: ;
 226     ap = cp;
 227     switch (my_lex (buffer)) {
 228         case LX_ATOM:
 229         case LX_QSTR:
 230             pers = strdup (buffer);
 231             break;
 232
 233         case LX_SEMI:
 234             if (glevel-- <= 0) {
 235                 strcpy (err, "extraneous semi-colon");
 236                 return NOTOK;
 237             }
 238             /* FALLTHRU */
 239         case LX_COMA:
 240             free(note);
 241             note = NULL;
 242             goto again;
 243
 244         case LX_END:
 245             return DONE;
 246
 247         case LX_LBRK:           /* sigh (2) */
 248             goto get_addr;
 249
 250         case LX_AT:             /* sigh (3) */
 251             cp = ap;
 252             if (route_addr (buffer) == NOTOK)
 253                 return NOTOK;
 254             return OK;          /* why be choosy? */
 255
 256         default:
 257             snprintf(err, sizeof err, "illegal address construct (%s)", buffer);
 258             return NOTOK;
 259     }
 260
 261     switch (my_lex (buffer)) {
 262         case LX_ATOM:
 263         case LX_QSTR:
 264             pers = add (buffer, add (" ", pers));
 265     more_phrase: ;              /* sigh (1) */
 266             if (phrase (buffer) == NOTOK)
 267                 return NOTOK;
 268
 269             switch (last_lex) {
 270                 case LX_LBRK:
 271             get_addr: ;
 272                     if (route_addr (buffer) == NOTOK)
 273                         return NOTOK;
 274                     if (last_lex == LX_RBRK)
 275                         return OK;
 276                     snprintf(err, sizeof err, "missing right-bracket (%s)", buffer);
 277                     return NOTOK;
 278
 279                 case LX_COLN:
 280             get_group: ;
 281                     if (glevel++ > 0) {
 282                         snprintf(err, sizeof err, "nested groups not allowed (%s)", pers);
 283                         return NOTOK;
 284                     }
 285                     grp = add (": ", pers);
 286                     pers = NULL;
 287                     {
 288                         char   *pp = cp;
 289
 290                         for (;;)
 291                             switch (my_lex (buffer)) {
 292                                 case LX_SEMI:
 293                                 case LX_END: /* tsk, tsk */
 294                                     glevel--;
 295                                     return OK;
 296
 297                                 case LX_COMA:
 298                                     continue;
 299
 300                                 default:
 301                                     cp = pp;
 302                                     return parse_address ();
 303                             }
 304                     }
 305
 306                 case LX_DOT:    /* sigh (1) */
 307                     pers = add (".", pers);
 308                     goto more_phrase;
 309
 310                 default:
 311                     snprintf(err, sizeof err, "no mailbox in address, only a phrase (%s%s)",
 312                             pers, buffer);
 313                     return NOTOK;
 314             }
 315
 316         case LX_LBRK:
 317             goto get_addr;
 318
 319         case LX_COLN:
 320             goto get_group;
 321
 322         case LX_DOT:
 323             mbox = add (buffer, pers);
 324             pers = NULL;
 325             if (route_addr (buffer) == NOTOK)
 326                 return NOTOK;
 327             goto check_end;
 328
 329         case LX_AT:
 330             ingrp = glevel;
 331             mbox = pers;
 332             pers = NULL;
 333             if (domain (buffer) == NOTOK)
 334                 return NOTOK;
 335     check_end: ;
 336             switch (last_lex) {
 337                 case LX_SEMI:
 338                     if (glevel-- <= 0) {
 339                         strcpy (err, "extraneous semi-colon");
 340                         return NOTOK;
 341                     }
 342                     return OK;
 343                 case LX_COMA:
 344                 case LX_END:
 345                     return OK;
 346
 347                 default:
 348                     snprintf(err, sizeof err, "junk after local@domain (%s)", buffer);
 349                     return NOTOK;
 350             }
 351
 352         case LX_SEMI:           /* no host */
 353         case LX_COMA:
 354         case LX_END:
 355             ingrp = glevel;
 356             if (last_lex == LX_SEMI && glevel-- <= 0) {
 357                 strcpy (err, "extraneous semi-colon");
 358                 return NOTOK;
 359             }
 360             mbox = pers;
 361             pers = NULL;
 362             return OK;
 363
 364         default:
 365             snprintf(err, sizeof err, "missing mailbox (%s)", buffer);
 366             return NOTOK;
 367     }
 368 }
 369
 370
 371 static int
 372 phrase (char *buffer)
 373 {
 374     int lex;
 375
 376     while ((lex = my_lex(buffer)) == LX_ATOM || lex == LX_QSTR)
 377         pers = add(buffer, add(" ", pers));
 378
 379     return OK;
 380 }
 381
 382
 383 static int
 384 route_addr (char *buffer)
 385 {
 386     char *pp = cp;
 387
 388     if (my_lex (buffer) == LX_AT) {
 389         if (route (buffer) == NOTOK)
 390             return NOTOK;
 391     }
 392     else
 393         cp = pp;
 394
 395     if (local_part (buffer) == NOTOK)
 396         return NOTOK;
 397
 398     switch (last_lex) {
 399         case LX_AT:
 400             return domain (buffer);
 401
 402         case LX_SEMI:   /* if in group */
 403         case LX_RBRK:           /* no host */
 404         case LX_COMA:
 405         case LX_END:
 406             return OK;
 407
 408         default:
 409             snprintf(err, sizeof err, "no at-sign after local-part (%s)", buffer);
 410             return NOTOK;
 411     }
 412 }
 413
 414
 415 static int
 416 local_part (char *buffer)
 417 {
 418     ingrp = glevel;
 419
 420     for (;;) {
 421         switch (my_lex (buffer)) {
 422             case LX_ATOM:
 423             case LX_QSTR:
 424                 mbox = add (buffer, mbox);
 425                 break;
 426
 427             default:
 428                 snprintf(err, sizeof err, "no mailbox in local-part (%s)", buffer);
 429                 return NOTOK;
 430         }
 431
 432         switch (my_lex (buffer)) {
 433             case LX_DOT:
 434                 mbox = add (buffer, mbox);
 435                 continue;
 436
 437             default:
 438                 return OK;
 439         }
 440     }
 441 }
 442
 443
 444 static int
 445 domain (char *buffer)
 446 {
 447     for (;;) {
 448         switch (my_lex (buffer)) {
 449             case LX_ATOM:
 450             case LX_DLIT:
 451                 host = add (buffer, host);
 452                 break;
 453
 454             default:
 455                 snprintf(err, sizeof err, "no sub-domain in domain-part of address (%s)", buffer);
 456                 return NOTOK;
 457         }
 458
 459         switch (my_lex (buffer)) {
 460             case LX_DOT:
 461                 host = add (buffer, host);
 462                 continue;
 463
 464             case LX_AT:         /* sigh (0) */
 465                 mbox = add (host, add ("%", mbox));
 466                 free (host);
 467                 host = NULL;
 468                 continue;
 469
 470             default:
 471                 return OK;
 472         }
 473     }
 474 }
 475
 476
 477 static int
 478 route (char *buffer)
 479 {
 480     routepath = mh_xstrdup ("@");
 481
 482     for (;;) {
 483         switch (my_lex (buffer)) {
 484             case LX_ATOM:
 485             case LX_DLIT:
 486                 routepath = add (buffer, routepath);
 487                 break;
 488
 489             default:
 490                 snprintf(err, sizeof err, "no sub-domain in domain-part of address (%s)", buffer);
 491                 return NOTOK;
 492         }
 493         switch (my_lex (buffer)) {
 494             case LX_COMA:
 495                 routepath = add (buffer, routepath);
 496                 for (;;) {
 497                     switch (my_lex (buffer)) {
 498                         case LX_COMA:
 499                             continue;
 500
 501                         case LX_AT:
 502                             routepath = add (buffer, routepath);
 503                             break;
 504
 505                         default:
 506                             snprintf(err, sizeof err, "no at-sign found for next domain in route (%s)",
 507                                      buffer);
 508                     }
 509                     break;
 510                 }
 511                 continue;
 512
 513             case LX_AT:         /* XXX */
 514             case LX_DOT:
 515                 routepath = add (buffer, routepath);
 516                 continue;
 517
 518             case LX_COLN:
 519                 routepath = add (buffer, routepath);
 520                 return OK;
 521
 522             default:
 523                 snprintf(err, sizeof err, "no colon found to terminate route (%s)", buffer);
 524                 return NOTOK;
 525         }
 526     }
 527 }
 528
 529
 530 static int
 531 my_lex (char *buffer)
 532 {
 533     /* buffer should be at least BUFSIZ bytes long */
 534     int i, gotat = 0;
 535     char c, *bp;
 536
 537 /* Add C to the buffer bp. After use of this macro *bp is guaranteed to be within the buffer. */
 538 #define ADDCHR(C) do { *bp++ = (C); if ((bp - buffer) == (BUFSIZ-1)) goto my_lex_buffull; } while (0)
 539
 540     bp = buffer;
 541     *bp = 0;
 542     if (!cp)
 543         return (last_lex = LX_END);
 544
 545     gotat = isat (cp);
 546     c = *cp++;
 547     while (isspace ((unsigned char) c))
 548         c = *cp++;
 549     if (c == 0) {
 550         cp = NULL;
 551         return (last_lex = LX_END);
 552     }
 553
 554     if (c == '(') {
 555         ADDCHR(c);
 556         for (i = 0;;)
 557             switch (c = *cp++) {
 558                 case 0:
 559                     cp = NULL;
 560                     return (last_lex = LX_ERR);
 561                 case '\\':
 562                     ADDCHR(c);
 563                     if ((c = *cp++) == 0) {
 564                         cp = NULL;
 565                         return (last_lex = LX_ERR);
 566                     }
 567                     ADDCHR(c);
 568                     continue;
 569                 case '(':
 570                     i++;
 571                     /* FALLTHRU */
 572                 default:
 573                     ADDCHR(c);
 574                     continue;
 575                 case ')':
 576                     ADDCHR(c);
 577                     if (--i < 0) {
 578                         *bp = 0;
 579                         note = note ? add (buffer, add (" ", note))
 580                             : strdup (buffer);
 581                         return my_lex (buffer);
 582                     }
 583             }
 584     }
 585
 586     if (c == '"') {
 587         ADDCHR(c);
 588         for (;;)
 589             switch (c = *cp++) {
 590                 case 0:
 591                     cp = NULL;
 592                     return (last_lex = LX_ERR);
 593                 case '\\':
 594                     ADDCHR(c);
 595                     if ((c = *cp++) == 0) {
 596                         cp = NULL;
 597                         return (last_lex = LX_ERR);
 598                     }
 599                     /* FALLTHRU */
 600                 default:
 601                     ADDCHR(c);
 602                     continue;
 603                 case '"':
 604                     ADDCHR(c);
 605                     *bp = 0;
 606                     return (last_lex = LX_QSTR);
 607             }
 608     }
 609
 610     if (c == '[') {
 611         ADDCHR(c);
 612         for (;;)
 613             switch (c = *cp++) {
 614                 case 0:
 615                     cp = NULL;
 616                     return (last_lex = LX_ERR);
 617                 case '\\':
 618                     ADDCHR(c);
 619                     if ((c = *cp++) == 0) {
 620                         cp = NULL;
 621                         return (last_lex = LX_ERR);
 622                     }
 623                     /* FALLTHRU */
 624                 default:
 625                     ADDCHR(c);
 626                     continue;
 627                 case ']':
 628                     ADDCHR(c);
 629                     *bp = 0;
 630                     return (last_lex = LX_DLIT);
 631             }
 632     }
 633
 634     ADDCHR(c);
 635     *bp = 0;
 636     for (i = 0; special[i].lx_chr != 0; i++)
 637         if (c == special[i].lx_chr)
 638             return (last_lex = special[i].lx_val);
 639
 640     if (iscntrl ((unsigned char) c))
 641         return (last_lex = LX_ERR);
 642
 643     for (;;) {
 644         if ((c = *cp++) == 0)
 645             break;
 646         for (i = 0; special[i].lx_chr != 0; i++)
 647             if (c == special[i].lx_chr)
 648                 goto got_atom;
 649         if (iscntrl ((unsigned char) c) || isspace ((unsigned char) c))
 650             break;
 651         ADDCHR(c);
 652     }
 653 got_atom: ;
 654     if (c == 0)
 655         cp = NULL;
 656     else
 657         cp--;
 658     *bp = 0;
 659     last_lex = !gotat || cp == NULL || strchr(cp, '<') != NULL
 660         ? LX_ATOM : LX_AT;
 661     return last_lex;
 662
 663  my_lex_buffull:
 664     /* Out of buffer space. *bp is the last byte in the buffer */
 665     *bp = 0;
 666     return (last_lex = LX_ERR);
 667 }
 668
 669
 670 char *
 671 legal_person (const char *p)
 672 {
 673     int i;
 674     const char *cp;
 675     static char buffer[BUFSIZ];
 676
 677     if (*p == '"')
 678         return (char *) p;
 679     for (cp = p; *cp; cp++)
 680         for (i = 0; special[i].lx_chr; i++)
 681             if (*cp == special[i].lx_chr) {
 682                 snprintf(buffer, sizeof buffer, "\"%s\"", p);
 683                 return buffer;
 684             }
 685
 686     return (char *) p;
 687 }