3 * mf.c -- mail filter subroutines
7 * This code is Copyright (c) 2002, by the authors of nmh. See the
8 * COPYRIGHT file in the root directory of the nmh distribution for
9 * complete copyright information.
19 static char *getcpy (char *);
20 static char *add (char *, char *);
21 static void compress (char *, char *);
22 static int isat (char *);
23 static int parse_address (void);
24 static int phrase (char *);
25 static int route_addr (char *);
26 static int local_part (char *);
27 static int domain (char *);
28 static int route (char *);
29 static int my_lex (char *);
38 /* causes compiles to blow up because the symbol _cleanup is undefined
39 where did this ever come from? */
45 if ((p = malloc ((size_t) (strlen (s) + 2))))
52 add (char *s1, char *s2)
59 if ((p = malloc ((size_t) (strlen (s1) + strlen (s2) + 2))))
60 sprintf (p, "%s%s", s2, s1);
68 return (strncmp (string, "From ", 5) == 0
69 || strncmp (string, ">From ", 6) == 0);
74 lequal (char *a, char *b)
80 char c1 = islower (*a) ? toupper (*a) : *a;
81 char c2 = islower (*b) ? toupper (*b) : *b;
91 * seekadrx() is tricky. We want to cover both UUCP-style and ARPA-style
92 * addresses, so for each list of addresses we see if we can find some
93 * character to give us a hint.
97 #define CHKADR 0 /* undertermined address style */
98 #define UNIXDR 1 /* UNIX-style address */
99 #define ARPADR 2 /* ARPAnet-style address */
102 static char *punctuators = ";<>.()[]";
103 static char *vp = NULL;
104 static char *tp = NULL;
106 static struct adrx adrxs1;
110 seekadrx (char *addrs)
112 static int state = CHKADR;
114 register struct adrx *adrxp;
117 for (state = UNIXDR, cp = addrs; *cp; cp++)
118 if (strchr(punctuators, *cp)) {
125 adrxp = uucpadrx (addrs);
130 adrxp = getadrx (addrs);
142 * uucpadrx() implements a partial UUCP-style address parser. It's based
143 * on the UUCP notion that addresses are separated by spaces or commas.
148 uucpadrx (char *addrs)
150 register char *cp, *wp, *xp, *yp, *zp;
151 register struct adrx *adrxp = &adrxs1;
154 vp = tp = getcpy (addrs);
155 compress (addrs, vp);
164 for (cp = tp; isspace (*cp); cp++)
172 if ((wp = strchr(cp, ',')) == NULL) {
173 if ((wp = strchr(cp, ' ')) != NULL) {
175 while (isspace (*xp))
177 if (*xp != 0 && isat (--xp)) {
179 while (isspace (*yp))
182 if ((zp = strchr(yp, ' ')) != NULL)
201 adrxp->text = getcpy (cp);
203 adrxp->host = adrxp->path = NULL;
204 if ((wp = strrchr(cp, '@')) != NULL) {
206 adrxp->host = *wp ? wp : NULL;
209 for (wp = cp + strlen (cp) - 4; wp >= cp; wp--)
212 adrxp->host = wp + 3;
215 adrxp->pers = adrxp->grp = adrxp->note = adrxp->err = NULL;
223 compress (char *fp, char *tp)
225 register char c, *cp;
227 for (c = ' ', cp = tp; (*tp = *fp++) != 0;)
235 if (c == ' ' && cp < tp)
243 return (strncmp (p, " AT ", 4)
244 && strncmp (p, " At ", 4)
245 && strncmp (p, " aT ", 4)
246 && strncmp (p, " at ", 4) ? FALSE : TRUE);
252 * getadrx() implements a partial 822-style address parser. The parser
253 * is neither complete nor correct. It does however recognize nearly all
254 * of the 822 address syntax. In addition it handles the majority of the
255 * 733 syntax as well. Most problems arise from trying to accomodate both.
257 * In terms of 822, the route-specification in
259 * "<" [route] local-part "@" domain ">"
261 * is parsed and returned unchanged. Multiple at-signs are compressed
262 * via source-routing. Recursive groups are not allowed as per the
265 * In terms of 733, " at " is recognized as equivalent to "@".
267 * In terms of both the parser will not complain about missing hosts.
271 * We should not allow addresses like
273 * Marshall T. Rose <MRose@UCI>
275 * but should insist on
277 * "Marshall T. Rose" <MRose@UCI>
279 * Unfortunately, a lot of mailers stupidly let people get away with this.
283 * We should not allow addresses like
287 * but should insist on
291 * Unfortunately, a lot of mailers stupidly let people's UAs get away with
296 * We should not allow addresses like
298 * @UCI:MRose@UCI-750a
300 * but should insist on
302 * Marshall Rose <@UCI:MRose@UCI-750a>
304 * Unfortunately, a lot of mailers stupidly do this.
328 static struct specials special[] = {
345 static int glevel = 0;
346 static int ingrp = 0;
347 static int last_lex = LX_END;
349 static char *dp = NULL;
350 static char *cp = NULL;
351 static char *ap = NULL;
352 static char *pers = NULL;
353 static char *mbox = NULL;
354 static char *host = NULL;
355 static char *path = NULL;
356 static char *grp = NULL;
357 static char *note = NULL;
358 static char err[BUFSIZ];
359 static char adr[BUFSIZ];
361 static struct adrx adrxs2;
365 getadrx (char *addrs)
368 register struct adrx *adrxp = &adrxs2;
382 pers = mbox = host = path = grp = note = NULL;
386 dp = cp = getcpy (addrs ? addrs : "");
396 switch (parse_address ()) {
408 default: /* catch trailing comments */
433 while (isspace (*ap))
436 sprintf (adr, "%.*s", cp - ap, ap);
439 bp = adr + strlen (adr) - 1;
440 if (*bp == ',' || *bp == ';' || *bp == '\n')
449 adrxp->ingrp = ingrp;
451 adrxp->err = err[0] ? err : NULL;
464 switch (my_lex (buffer)) {
467 pers = getcpy (buffer);
472 strcpy (err, "extraneous semi-colon");
485 case LX_LBRK: /* sigh (2) */
488 case LX_AT: /* sigh (3) */
490 if (route_addr (buffer) == NOTOK)
492 return OK; /* why be choosy? */
495 sprintf (err, "illegal address construct (%s)", buffer);
499 switch (my_lex (buffer)) {
502 pers = add (buffer, add (" ", pers));
503 more_phrase: ; /* sigh (1) */
504 if (phrase (buffer) == NOTOK)
510 if (route_addr (buffer) == NOTOK)
512 if (last_lex == LX_RBRK)
514 sprintf (err, "missing right-bracket (%s)", buffer);
520 sprintf (err, "nested groups not allowed (%s)", pers);
523 grp = add (": ", pers);
529 switch (my_lex (buffer)) {
531 case LX_END: /* tsk, tsk */
540 return parse_address ();
544 case LX_DOT: /* sigh (1) */
545 pers = add (".", pers);
549 sprintf (err, "no mailbox in address, only a phrase (%s%s)",
561 mbox = add (buffer, pers);
563 if (route_addr (buffer) == NOTOK)
571 if (domain (buffer) == NOTOK)
577 strcpy (err, "extraneous semi-colon");
585 sprintf (err, "junk after local@domain (%s)", buffer);
589 case LX_SEMI: /* no host */
593 if (last_lex == LX_SEMI && glevel-- <= 0) {
594 strcpy (err, "extraneous semi-colon");
602 sprintf (err, "missing mailbox (%s)", buffer);
609 phrase (char *buffer)
612 switch (my_lex (buffer)) {
615 pers = add (buffer, add (" ", pers));
625 route_addr (char *buffer)
627 register char *pp = cp;
629 if (my_lex (buffer) == LX_AT) {
630 if (route (buffer) == NOTOK)
636 if (local_part (buffer) == NOTOK)
641 return domain (buffer);
643 case LX_SEMI: /* if in group */
644 case LX_RBRK: /* no host */
650 sprintf (err, "no at-sign after local-part (%s)", buffer);
657 local_part (char *buffer)
662 switch (my_lex (buffer)) {
665 mbox = add (buffer, mbox);
669 sprintf (err, "no mailbox in local-part (%s)", buffer);
673 switch (my_lex (buffer)) {
675 mbox = add (buffer, mbox);
686 domain (char *buffer)
689 switch (my_lex (buffer)) {
692 host = add (buffer, host);
696 sprintf (err, "no sub-domain in domain-part of address (%s)", buffer);
700 switch (my_lex (buffer)) {
702 host = add (buffer, host);
705 case LX_AT: /* sigh (0) */
706 mbox = add (host, add ("%", mbox));
724 switch (my_lex (buffer)) {
727 path = add (buffer, path);
731 sprintf (err, "no sub-domain in domain-part of address (%s)", buffer);
734 switch (my_lex (buffer)) {
736 path = add (buffer, path);
738 switch (my_lex (buffer)) {
743 path = add (buffer, path);
747 sprintf (err, "no at-sign found for next domain in route (%s)",
754 case LX_AT: /* XXX */
756 path = add (buffer, path);
760 path = add (buffer, path);
764 sprintf (err, "no colon found to terminate route (%s)", buffer);
772 my_lex (char *buffer)
775 register char c, *bp;
780 return (last_lex = LX_END);
788 return (last_lex = LX_END);
792 for (*bp++ = c, i = 0;;)
796 return (last_lex = LX_ERR);
799 if ((c = *cp++) == 0) {
801 return (last_lex = LX_ERR);
814 note = note ? add (buffer, add (" ", note))
816 return my_lex (buffer);
825 return (last_lex = LX_ERR);
828 if ((c = *cp++) == 0) {
830 return (last_lex = LX_ERR);
838 return (last_lex = LX_QSTR);
846 return (last_lex = LX_ERR);
849 if ((c = *cp++) == 0) {
851 return (last_lex = LX_ERR);
859 return (last_lex = LX_DLIT);
864 for (i = 0; special[i].lx_chr != 0; i++)
865 if (c == special[i].lx_chr)
866 return (last_lex = special[i].lx_val);
869 return (last_lex = LX_ERR);
872 if ((c = *cp++) == 0)
874 for (i = 0; special[i].lx_chr != 0; i++)
875 if (c == special[i].lx_chr)
877 if (iscntrl (c) || isspace (c))
887 last_lex = !gotat || cp == NULL || strchr(cp, '<') != NULL
894 legal_person (char *p)
898 static char buffer[BUFSIZ];
902 for (cp = p; *cp; cp++)
903 for (i = 0; special[i].lx_chr; i++)
904 if (*cp == special[i].lx_chr) {
905 sprintf (buffer, "\"%s\"", p);
914 mfgets (FILE *in, char **bp)
917 register char *cp, *dp, *ep;
919 static char *pp = NULL;
922 if (!(pp = malloc ((size_t) (len = BUFSIZ))))
925 for (ep = (cp = pp) + len - 2;;) {
926 switch (i = getc (in)) {
944 if (cp == pp) /* end of headers, gobble it */
946 switch (i = getc (in)) {
947 default: /* end of line */
948 case '\n': /* end of headers, save for next call */
952 case ' ': /* continue headers */
956 } /* fall into default case */
963 if (!(dp = realloc (pp, (size_t) (len += BUFSIZ)))) {
969 cp += dp - pp, ep = (pp = cp) + len - 2;