1 # include "sendmail.h" 2 3 static char SccsId[] = "@(#)parseaddr.c 3.29 10/06/81"; 4 5 /* 6 ** PARSE -- Parse an address 7 ** 8 ** Parses an address and breaks it up into three parts: a 9 ** net to transmit the message on, the host to transmit it 10 ** to, and a user on that host. These are loaded into an 11 ** ADDRESS header with the values squirreled away if necessary. 12 ** The "user" part may not be a real user; the process may 13 ** just reoccur on that machine. For example, on a machine 14 ** with an arpanet connection, the address 15 ** csvax.bill@berkeley 16 ** will break up to a "user" of 'csvax.bill' and a host 17 ** of 'berkeley' -- to be transmitted over the arpanet. 18 ** 19 ** Parameters: 20 ** addr -- the address to parse. 21 ** a -- a pointer to the address descriptor buffer. 22 ** If NULL, a header will be created. 23 ** copyf -- determines what shall be copied: 24 ** -1 -- don't copy anything. The printname 25 ** (q_paddr) is just addr, and the 26 ** user & host are allocated internally 27 ** to parse. 28 ** 0 -- copy out the parsed user & host, but 29 ** don't copy the printname. 30 ** +1 -- copy everything. 31 ** 32 ** Returns: 33 ** A pointer to the address descriptor header (`a' if 34 ** `a' is non-NULL). 35 ** NULL on error. 36 ** 37 ** Side Effects: 38 ** none 39 */ 40 41 # define DELIMCHARS "$()<>,;\\\"\r\n" /* word delimiters */ 42 43 ADDRESS * 44 parse(addr, a, copyf) 45 char *addr; 46 register ADDRESS *a; 47 int copyf; 48 { 49 register char **pvp; 50 register struct mailer *m; 51 extern char **prescan(); 52 extern ADDRESS *buildaddr(); 53 54 /* 55 ** Initialize and prescan address. 56 */ 57 58 To = addr; 59 # ifdef DEBUG 60 if (Debug) 61 printf("\n--parse(%s)\n", addr); 62 # endif DEBUG 63 64 pvp = prescan(addr, '\0'); 65 if (pvp == NULL) 66 return (NULL); 67 68 /* 69 ** Apply rewriting rules. 70 */ 71 72 rewrite(pvp, 0); 73 74 /* 75 ** See if we resolved to a real mailer. 76 */ 77 78 if (pvp[0][0] != CANONNET) 79 { 80 setstat(EX_USAGE); 81 usrerr("cannot resolve name"); 82 return (NULL); 83 } 84 85 /* 86 ** Build canonical address from pvp. 87 */ 88 89 a = buildaddr(pvp, a); 90 if (a == NULL) 91 return (NULL); 92 m = Mailer[a->q_mailer]; 93 94 /* 95 ** Make local copies of the host & user and then 96 ** transport them out. 97 */ 98 99 if (copyf > 0) 100 a->q_paddr = newstr(addr); 101 else 102 a->q_paddr = addr; 103 104 if (copyf >= 0) 105 { 106 if (a->q_host != NULL) 107 a->q_host = newstr(a->q_host); 108 else 109 a->q_host = ""; 110 if (a->q_user != a->q_paddr) 111 a->q_user = newstr(a->q_user); 112 } 113 114 /* 115 ** Do UPPER->lower case mapping unless inhibited. 116 */ 117 118 if (!bitset(M_HST_UPPER, m->m_flags)) 119 makelower(a->q_host); 120 if (!bitset(M_USR_UPPER, m->m_flags)) 121 makelower(a->q_user); 122 123 /* 124 ** Compute return value. 125 */ 126 127 # ifdef DEBUG 128 if (Debug) 129 { 130 printf("parse-->"); 131 printaddr(a, FALSE); 132 } 133 # endif DEBUG 134 135 return (a); 136 } 137 /* 138 ** PRESCAN -- Prescan name and make it canonical 139 ** 140 ** Scans a name and turns it into canonical form. This involves 141 ** deleting blanks, comments (in parentheses), and turning the 142 ** word "at" into an at-sign ("@"). The name is copied as this 143 ** is done; it is legal to copy a name onto itself, since this 144 ** process can only make things smaller. 145 ** 146 ** This routine knows about quoted strings and angle brackets. 147 ** 148 ** There are certain subtleties to this routine. The one that 149 ** comes to mind now is that backslashes on the ends of names 150 ** are silently stripped off; this is intentional. The problem 151 ** is that some versions of sndmsg (like at LBL) set the kill 152 ** character to something other than @ when reading addresses; 153 ** so people type "csvax.eric\@berkeley" -- which screws up the 154 ** berknet mailer. 155 ** 156 ** Parameters: 157 ** addr -- the name to chomp. 158 ** delim -- the delimiter for the address, normally 159 ** '\0' or ','; \0 is accepted in any case. 160 ** are moving in place; set buflim to high core. 161 ** 162 ** Returns: 163 ** A pointer to a vector of tokens. 164 ** NULL on error. 165 ** 166 ** Side Effects: 167 ** none. 168 */ 169 170 # define OPER 1 171 # define ATOM 2 172 # define EOTOK 3 173 # define QSTRING 4 174 # define SPACE 5 175 # define DOLLAR 6 176 # define GETONE 7 177 # define MACRO 8 178 179 char ** 180 prescan(addr, delim) 181 char *addr; 182 char delim; 183 { 184 register char *p; 185 static char buf[MAXNAME+MAXATOM]; 186 static char *av[MAXATOM+1]; 187 char **avp; 188 bool bslashmode; 189 int cmntcnt; 190 int brccnt; 191 register char c; 192 char *tok; 193 register char *q; 194 register int state; 195 int nstate; 196 extern char lower(); 197 198 q = buf; 199 bslashmode = FALSE; 200 cmntcnt = brccnt = 0; 201 avp = av; 202 state = OPER; 203 for (p = addr; *p != '\0' && *p != delim; ) 204 { 205 /* read a token */ 206 tok = q; 207 while ((c = *p++) != '\0' && c != delim) 208 { 209 /* chew up special characters */ 210 c &= ~0200; 211 *q = '\0'; 212 if (bslashmode) 213 { 214 c |= 0200; 215 bslashmode = FALSE; 216 } 217 else if (c == '\\') 218 { 219 bslashmode = TRUE; 220 continue; 221 } 222 else if (c == '"') 223 { 224 if (state == QSTRING) 225 state = OPER; 226 else 227 state = QSTRING; 228 break; 229 } 230 231 if (c == '$' && delim == '\t') 232 nstate = DOLLAR; 233 else 234 nstate = toktype(c); 235 switch (state) 236 { 237 case QSTRING: /* in quoted string */ 238 break; 239 240 case ATOM: /* regular atom */ 241 if (nstate != ATOM) 242 { 243 state = EOTOK; 244 p--; 245 } 246 break; 247 248 case GETONE: /* grab one character */ 249 state = OPER; 250 break; 251 252 case EOTOK: /* after atom or q-string */ 253 state = nstate; 254 if (state == SPACE) 255 continue; 256 break; 257 258 case SPACE: /* linear white space */ 259 state = nstate; 260 break; 261 262 case OPER: /* operator */ 263 if (nstate == SPACE) 264 continue; 265 state = nstate; 266 break; 267 268 case DOLLAR: /* $- etc. */ 269 state = OPER; 270 if (isascii(c) && isdigit(c)) 271 { 272 /* replacement */ 273 c = MATCHREPL; 274 state = GETONE; 275 p--; 276 break; 277 } 278 switch (c) 279 { 280 case '$': /* literal $ */ 281 break; 282 283 case '+': /* match anything */ 284 c = MATCHANY; 285 break; 286 287 case '-': /* match one token */ 288 c = MATCHONE; 289 break; 290 291 case '=': /* match one token of class */ 292 c = MATCHCLASS; 293 state = GETONE; 294 break; 295 296 case '#': /* canonical net name */ 297 c = CANONNET; 298 break; 299 300 case '@': /* canonical host name */ 301 c = CANONHOST; 302 break; 303 304 case ':': /* canonical user name */ 305 c = CANONUSER; 306 break; 307 308 default: 309 state = MACRO; 310 break; 311 } 312 break; 313 314 default: 315 syserr("prescan: unknown state %d", state); 316 } 317 318 if (state == EOTOK || state == SPACE) 319 break; 320 if (state == DOLLAR) 321 continue; 322 323 /* squirrel it away */ 324 if (q >= &buf[sizeof buf - 5]) 325 { 326 usrerr("Address too long"); 327 return (NULL); 328 } 329 if (state == MACRO) 330 { 331 char mbuf[3]; 332 333 mbuf[0] = '$'; 334 mbuf[1] = c; 335 mbuf[2] = '\0'; 336 (void) expand(mbuf, q, &buf[sizeof buf - 5]); 337 q += strlen(q); 338 state = EOTOK; 339 break; 340 } 341 *q++ = c; 342 343 /* decide whether this represents end of token */ 344 if (state == OPER) 345 break; 346 } 347 if (c == '\0' || c == delim) 348 p--; 349 350 /* new token */ 351 if (tok == q) 352 continue; 353 *q++ = '\0'; 354 355 c = tok[0]; 356 if (c == '(') 357 { 358 cmntcnt++; 359 continue; 360 } 361 else if (c == ')') 362 { 363 if (cmntcnt <= 0) 364 { 365 usrerr("Unbalanced ')'"); 366 return (NULL); 367 } 368 else 369 { 370 cmntcnt--; 371 continue; 372 } 373 } 374 else if (cmntcnt > 0) 375 continue; 376 377 /* we prefer <> specs */ 378 if (c == '<') 379 { 380 if (brccnt < 0) 381 { 382 usrerr("multiple < spec"); 383 return (NULL); 384 } 385 brccnt++; 386 if (brccnt == 1) 387 { 388 /* we prefer using machine readable name */ 389 q = buf; 390 *q = '\0'; 391 avp = av; 392 continue; 393 } 394 } 395 else if (c == '>') 396 { 397 if (brccnt <= 0) 398 { 399 usrerr("Unbalanced `>'"); 400 return (NULL); 401 } 402 else 403 brccnt--; 404 if (brccnt <= 0) 405 { 406 brccnt = -1; 407 continue; 408 } 409 } 410 411 if (avp >= &av[MAXATOM]) 412 { 413 syserr("prescan: too many tokens"); 414 return (NULL); 415 } 416 *avp++ = tok; 417 } 418 *avp = NULL; 419 if (cmntcnt > 0) 420 usrerr("Unbalanced '('"); 421 else if (brccnt > 0) 422 usrerr("Unbalanced '<'"); 423 else if (state == QSTRING) 424 usrerr("Unbalanced '\"'"); 425 else if (av[0] != NULL) 426 return (av); 427 return (NULL); 428 } 429 /* 430 ** TOKTYPE -- return token type 431 ** 432 ** Parameters: 433 ** c -- the character in question. 434 ** 435 ** Returns: 436 ** Its type. 437 ** 438 ** Side Effects: 439 ** none. 440 */ 441 442 toktype(c) 443 register char c; 444 { 445 static char buf[50]; 446 static bool firstime = TRUE; 447 448 if (firstime) 449 { 450 firstime = FALSE; 451 (void) expand("$o", buf, &buf[sizeof buf - 1]); 452 strcat(buf, DELIMCHARS); 453 } 454 if (!isascii(c)) 455 return (ATOM); 456 if (isspace(c)) 457 return (SPACE); 458 if (iscntrl(c) || index(buf, c) != NULL) 459 return (OPER); 460 return (ATOM); 461 } 462 /* 463 ** REWRITE -- apply rewrite rules to token vector. 464 ** 465 ** Parameters: 466 ** pvp -- pointer to token vector. 467 ** 468 ** Returns: 469 ** none. 470 ** 471 ** Side Effects: 472 ** pvp is modified. 473 */ 474 475 struct match 476 { 477 char **first; /* first token matched */ 478 char **last; /* last token matched */ 479 }; 480 481 # define MAXMATCH 9 /* max params per rewrite */ 482 483 484 rewrite(pvp, ruleset) 485 char **pvp; 486 int ruleset; 487 { 488 register char *ap; /* address pointer */ 489 register char *rp; /* rewrite pointer */ 490 register char **avp; /* address vector pointer */ 491 char **avfp; /* first word in current match */ 492 register char **rvp; /* rewrite vector pointer */ 493 struct rewrite *rwr; /* pointer to current rewrite rule */ 494 struct match mlist[MAXMATCH]; /* stores match on LHS */ 495 struct match *mlp; /* cur ptr into mlist */ 496 char *npvp[MAXATOM+1]; /* temporary space for rebuild */ 497 extern bool sameword(); 498 499 # ifdef DEBUG 500 if (Debug > 9) 501 { 502 printf("rewrite: original pvp:\n"); 503 printav(pvp); 504 } 505 # endif DEBUG 506 507 /* 508 ** Run through the list of rewrite rules, applying 509 ** any that match. 510 */ 511 512 for (rwr = RewriteRules[ruleset]; rwr != NULL; ) 513 { 514 # ifdef DEBUG 515 if (Debug > 10) 516 { 517 printf("-----trying rule:\n"); 518 printav(rwr->r_lhs); 519 } 520 # endif DEBUG 521 522 /* try to match on this rule */ 523 mlp = mlist; 524 for (rvp = rwr->r_lhs, avfp = avp = pvp; *avp != NULL; ) 525 { 526 ap = *avp; 527 rp = *rvp; 528 529 if (rp == NULL) 530 { 531 /* end-of-pattern before end-of-address */ 532 goto fail; 533 } 534 535 switch (*rp) 536 { 537 register STAB *s; 538 register int class; 539 540 case MATCHONE: 541 /* match exactly one token */ 542 mlp->first = mlp->last = avp++; 543 mlp++; 544 avfp = avp; 545 break; 546 547 case MATCHANY: 548 /* match any number of tokens */ 549 mlp->first = avfp; 550 mlp->last = avp++; 551 mlp++; 552 break; 553 554 case MATCHCLASS: 555 /* match any token in a class */ 556 class = rp[1]; 557 if (!isalpha(class)) 558 goto fail; 559 if (isupper(class)) 560 class -= 'A'; 561 else 562 class -= 'a'; 563 s = stab(ap, ST_CLASS, ST_FIND); 564 if (s == NULL || (s->s_class & (1 << class)) == 0) 565 goto fail; 566 567 /* mark match */ 568 mlp->first = mlp->last = avp++; 569 mlp++; 570 avfp = avp; 571 break; 572 573 default: 574 /* must have exact match */ 575 if (!sameword(rp, ap)) 576 goto fail; 577 avp++; 578 avfp = avp; 579 break; 580 } 581 582 /* successful match on this token */ 583 rvp++; 584 continue; 585 586 fail: 587 /* match failed -- back up */ 588 while (--rvp >= rwr->r_lhs) 589 { 590 rp = *rvp; 591 if (*rp == MATCHANY) 592 { 593 avfp = mlp->first; 594 break; 595 } 596 else if (*rp == MATCHONE || *rp == MATCHCLASS) 597 { 598 /* back out binding */ 599 avp--; 600 avfp = avp; 601 mlp--; 602 } 603 } 604 605 if (rvp < rwr->r_lhs) 606 { 607 /* total failure to match */ 608 break; 609 } 610 } 611 612 /* 613 ** See if we successfully matched 614 */ 615 616 if (rvp >= rwr->r_lhs && *rvp == NULL) 617 { 618 # ifdef DEBUG 619 if (Debug > 10) 620 { 621 printf("-----rule matches:\n"); 622 printav(rwr->r_rhs); 623 } 624 # endif DEBUG 625 626 /* substitute */ 627 for (rvp = rwr->r_rhs, avp = npvp; *rvp != NULL; rvp++) 628 { 629 rp = *rvp; 630 if (*rp == MATCHREPL) 631 { 632 register struct match *m; 633 register char **pp; 634 635 m = &mlist[rp[1] - '1']; 636 pp = m->first; 637 do 638 { 639 if (avp >= &npvp[MAXATOM]) 640 { 641 syserr("rewrite: expansion too long"); 642 return; 643 } 644 *avp++ = *pp; 645 } while (pp++ != m->last); 646 } 647 else 648 { 649 if (avp >= &npvp[MAXATOM]) 650 { 651 syserr("rewrite: expansion too long"); 652 return; 653 } 654 *avp++ = rp; 655 } 656 } 657 *avp++ = NULL; 658 bmove((char *) npvp, (char *) pvp, (avp - npvp) * sizeof *avp); 659 # ifdef DEBUG 660 if (Debug > 3) 661 { 662 char **vp; 663 664 printf("rewritten as `"); 665 for (vp = pvp; *vp != NULL; vp++) 666 { 667 if (vp != pvp) 668 printf("_"); 669 xputs(*vp); 670 } 671 printf("'\n"); 672 } 673 # endif DEBUG 674 if (pvp[0][0] == CANONNET) 675 break; 676 } 677 else 678 { 679 # ifdef DEBUG 680 if (Debug > 10) 681 printf("----- rule fails\n"); 682 # endif DEBUG 683 rwr = rwr->r_next; 684 } 685 } 686 } 687 /* 688 ** BUILDADDR -- build address from token vector. 689 ** 690 ** Parameters: 691 ** tv -- token vector. 692 ** a -- pointer to address descriptor to fill. 693 ** If NULL, one will be allocated. 694 ** 695 ** Returns: 696 ** NULL if there was an error. 697 ** 'a' otherwise. 698 ** 699 ** Side Effects: 700 ** fills in 'a' 701 */ 702 703 ADDRESS * 704 buildaddr(tv, a) 705 register char **tv; 706 register ADDRESS *a; 707 { 708 register int i; 709 static char buf[MAXNAME]; 710 struct mailer **mp; 711 register struct mailer *m; 712 713 if (a == NULL) 714 a = (ADDRESS *) xalloc(sizeof *a); 715 a->q_flags = 0; 716 a->q_home = NULL; 717 718 /* figure out what net/mailer to use */ 719 if (**tv != CANONNET) 720 { 721 syserr("buildaddr: no net"); 722 return (NULL); 723 } 724 tv++; 725 if (strcmp(*tv, "error") == 0) 726 { 727 if (**++tv != CANONUSER) 728 syserr("buildaddr: error: no user"); 729 buf[0] = '\0'; 730 while (*++tv != NULL) 731 { 732 if (buf[0] != '\0') 733 strcat(buf, " "); 734 strcat(buf, *tv); 735 } 736 usrerr(buf); 737 return (NULL); 738 } 739 for (mp = Mailer, i = 0; (m = *mp++) != NULL; i++) 740 { 741 if (strcmp(m->m_name, *tv) == 0) 742 break; 743 } 744 if (m == NULL) 745 { 746 syserr("buildaddr: unknown net %s", *tv); 747 return (NULL); 748 } 749 a->q_mailer = i; 750 751 /* figure out what host (if any) */ 752 tv++; 753 if (!bitset(M_LOCAL, m->m_flags)) 754 { 755 if (**tv != CANONHOST) 756 { 757 syserr("buildaddr: no host"); 758 return (NULL); 759 } 760 tv++; 761 a->q_host = *tv; 762 tv++; 763 } 764 else 765 a->q_host = NULL; 766 767 /* figure out the user */ 768 if (**tv != CANONUSER) 769 { 770 syserr("buildaddr: no user"); 771 return (NULL); 772 } 773 cataddr(++tv, buf, sizeof buf); 774 a->q_user = buf; 775 776 return (a); 777 } 778 /* 779 ** CATADDR -- concatenate pieces of addresses (putting in <LWSP> subs) 780 ** 781 ** Parameters: 782 ** pvp -- parameter vector to rebuild. 783 ** buf -- buffer to build the string into. 784 ** sz -- size of buf. 785 ** 786 ** Returns: 787 ** none. 788 ** 789 ** Side Effects: 790 ** Destroys buf. 791 */ 792 793 cataddr(pvp, buf, sz) 794 char **pvp; 795 char *buf; 796 register int sz; 797 { 798 bool oatomtok = FALSE; 799 bool natomtok = FALSE; 800 register int i; 801 register char *p; 802 803 p = buf; 804 sz--; 805 while (*pvp != NULL && (i = strlen(*pvp)) < sz) 806 { 807 natomtok = (toktype(**pvp) == ATOM); 808 if (oatomtok && natomtok) 809 *p++ = SPACESUB; 810 (void) strcpy(p, *pvp); 811 oatomtok = natomtok; 812 p += i; 813 sz -= i; 814 pvp++; 815 } 816 *p = '\0'; 817 } 818 /* 819 ** SAMEADDR -- Determine if two addresses are the same 820 ** 821 ** This is not just a straight comparison -- if the mailer doesn't 822 ** care about the host we just ignore it, etc. 823 ** 824 ** Parameters: 825 ** a, b -- pointers to the internal forms to compare. 826 ** wildflg -- if TRUE, 'a' may have no user specified, 827 ** in which case it is to match anything. 828 ** 829 ** Returns: 830 ** TRUE -- they represent the same mailbox. 831 ** FALSE -- they don't. 832 ** 833 ** Side Effects: 834 ** none. 835 */ 836 837 bool 838 sameaddr(a, b, wildflg) 839 register ADDRESS *a; 840 register ADDRESS *b; 841 bool wildflg; 842 { 843 /* if they don't have the same mailer, forget it */ 844 if (a->q_mailer != b->q_mailer) 845 return (FALSE); 846 847 /* if the user isn't the same, we can drop out */ 848 if ((!wildflg || a->q_user[0] != '\0') && strcmp(a->q_user, b->q_user) != 0) 849 return (FALSE); 850 851 /* if the mailer ignores hosts, we have succeeded! */ 852 if (bitset(M_LOCAL, Mailer[a->q_mailer]->m_flags)) 853 return (TRUE); 854 855 /* otherwise compare hosts (but be careful for NULL ptrs) */ 856 if (a->q_host == NULL || b->q_host == NULL) 857 return (FALSE); 858 if (strcmp(a->q_host, b->q_host) != 0) 859 return (FALSE); 860 861 return (TRUE); 862 } 863 /* 864 ** PRINTADDR -- print address (for debugging) 865 ** 866 ** Parameters: 867 ** a -- the address to print 868 ** follow -- follow the q_next chain. 869 ** 870 ** Returns: 871 ** none. 872 ** 873 ** Side Effects: 874 ** none. 875 */ 876 877 # ifdef DEBUG 878 879 printaddr(a, follow) 880 register ADDRESS *a; 881 bool follow; 882 { 883 while (a != NULL) 884 { 885 printf("%x=", a); 886 (void) fflush(stdout); 887 printf("%s: mailer %d (%s), host `%s', user `%s'\n", a->q_paddr, 888 a->q_mailer, Mailer[a->q_mailer]->m_name, a->q_host, a->q_user); 889 printf("\tnext=%x, flags=%o, rmailer %d\n", a->q_next, 890 a->q_flags, a->q_rmailer); 891 892 if (!follow) 893 return; 894 a = a->q_next; 895 } 896 if (!follow) 897 printf("[NULL]\n"); 898 } 899 900 # endif DEBUG 901