122976Smiriam /*
234921Sbostic  * Copyright (c) 1983 Eric P. Allman
333730Sbostic  * Copyright (c) 1988 Regents of the University of California.
433730Sbostic  * All rights reserved.
533730Sbostic  *
6*42828Sbostic  * %sccs.include.redist.c%
733730Sbostic  */
822976Smiriam 
922976Smiriam #ifndef lint
10*42828Sbostic static char sccsid[] = "@(#)parseaddr.c	5.13 (Berkeley) 06/01/90";
1133730Sbostic #endif /* not lint */
1222976Smiriam 
133312Seric # include "sendmail.h"
14297Seric 
15297Seric /*
169888Seric **  PARSEADDR -- Parse an address
17297Seric **
18297Seric **	Parses an address and breaks it up into three parts: a
19297Seric **	net to transmit the message on, the host to transmit it
20297Seric **	to, and a user on that host.  These are loaded into an
212973Seric **	ADDRESS header with the values squirreled away if necessary.
22297Seric **	The "user" part may not be a real user; the process may
23297Seric **	just reoccur on that machine.  For example, on a machine
24297Seric **	with an arpanet connection, the address
25297Seric **		csvax.bill@berkeley
26297Seric **	will break up to a "user" of 'csvax.bill' and a host
27297Seric **	of 'berkeley' -- to be transmitted over the arpanet.
28297Seric **
29297Seric **	Parameters:
30297Seric **		addr -- the address to parse.
31297Seric **		a -- a pointer to the address descriptor buffer.
32297Seric **			If NULL, a header will be created.
33297Seric **		copyf -- determines what shall be copied:
34297Seric **			-1 -- don't copy anything.  The printname
35297Seric **				(q_paddr) is just addr, and the
36297Seric **				user & host are allocated internally
37297Seric **				to parse.
38297Seric **			0 -- copy out the parsed user & host, but
39297Seric **				don't copy the printname.
40297Seric **			+1 -- copy everything.
4111445Seric **		delim -- the character to terminate the address, passed
4211445Seric **			to prescan.
43297Seric **
44297Seric **	Returns:
45297Seric **		A pointer to the address descriptor header (`a' if
46297Seric **			`a' is non-NULL).
47297Seric **		NULL on error.
48297Seric **
49297Seric **	Side Effects:
50297Seric **		none
51297Seric */
52297Seric 
539374Seric /* following delimiters are inherent to the internal algorithms */
5416155Seric # define DELIMCHARS	"\001()<>,;\\\"\r\n"	/* word delimiters */
552091Seric 
562973Seric ADDRESS *
5711445Seric parseaddr(addr, a, copyf, delim)
58297Seric 	char *addr;
592973Seric 	register ADDRESS *a;
60297Seric 	int copyf;
6111445Seric 	char delim;
62297Seric {
633149Seric 	register char **pvp;
643149Seric 	register struct mailer *m;
6516914Seric 	char pvpbuf[PSBUFSIZE];
663149Seric 	extern char **prescan();
673149Seric 	extern ADDRESS *buildaddr();
68297Seric 
69297Seric 	/*
70297Seric 	**  Initialize and prescan address.
71297Seric 	*/
72297Seric 
736903Seric 	CurEnv->e_to = addr;
747675Seric 	if (tTd(20, 1))
759888Seric 		printf("\n--parseaddr(%s)\n", addr);
763188Seric 
7716914Seric 	pvp = prescan(addr, delim, pvpbuf);
783149Seric 	if (pvp == NULL)
79297Seric 		return (NULL);
80297Seric 
81297Seric 	/*
823149Seric 	**  Apply rewriting rules.
837889Seric 	**	Ruleset 0 does basic parsing.  It must resolve.
84297Seric 	*/
85297Seric 
868181Seric 	rewrite(pvp, 3);
874070Seric 	rewrite(pvp, 0);
88297Seric 
893149Seric 	/*
903149Seric 	**  See if we resolved to a real mailer.
913149Seric 	*/
92297Seric 
933149Seric 	if (pvp[0][0] != CANONNET)
943149Seric 	{
953149Seric 		setstat(EX_USAGE);
963149Seric 		usrerr("cannot resolve name");
973149Seric 		return (NULL);
98297Seric 	}
99297Seric 
100297Seric 	/*
1013149Seric 	**  Build canonical address from pvp.
102297Seric 	*/
103297Seric 
1043149Seric 	a = buildaddr(pvp, a);
1054279Seric 	if (a == NULL)
1064279Seric 		return (NULL);
1074598Seric 	m = a->q_mailer;
108297Seric 
109297Seric 	/*
1103149Seric 	**  Make local copies of the host & user and then
1113149Seric 	**  transport them out.
112297Seric 	*/
113297Seric 
114297Seric 	if (copyf > 0)
1158078Seric 	{
1168078Seric 		extern char *DelimChar;
1178078Seric 		char savec = *DelimChar;
1188078Seric 
1198078Seric 		*DelimChar = '\0';
1202973Seric 		a->q_paddr = newstr(addr);
1218078Seric 		*DelimChar = savec;
1228078Seric 	}
123297Seric 	else
124297Seric 		a->q_paddr = addr;
12524944Seric 
12624944Seric 	if (a->q_user == NULL)
12724944Seric 		a->q_user = "";
12824944Seric 	if (a->q_host == NULL)
12924944Seric 		a->q_host = "";
13024944Seric 
1313149Seric 	if (copyf >= 0)
132297Seric 	{
13324944Seric 		a->q_host = newstr(a->q_host);
1343149Seric 		if (a->q_user != a->q_paddr)
1353149Seric 			a->q_user = newstr(a->q_user);
136297Seric 	}
137297Seric 
138297Seric 	/*
13916202Seric 	**  Convert host name to lower case if requested.
14016202Seric 	**	User name will be done later.
14116202Seric 	*/
14216202Seric 
14316202Seric 	if (!bitnset(M_HST_UPPER, m->m_flags))
14416202Seric 		makelower(a->q_host);
14516202Seric 
14616202Seric 	/*
147297Seric 	**  Compute return value.
148297Seric 	*/
149297Seric 
1507675Seric 	if (tTd(20, 1))
1514443Seric 	{
1529888Seric 		printf("parseaddr-->");
1534443Seric 		printaddr(a, FALSE);
1544443Seric 	}
155297Seric 
156297Seric 	return (a);
157297Seric }
158297Seric /*
15916162Seric **  LOWERADDR -- map UPPER->lower case on addresses as requested.
16016162Seric **
16116162Seric **	Parameters:
16216162Seric **		a -- address to be mapped.
16316162Seric **
16416162Seric **	Returns:
16516162Seric **		none.
16616162Seric **
16716162Seric **	Side Effects:
16816162Seric **		none.
16916162Seric */
17016162Seric 
17116162Seric loweraddr(a)
17216162Seric 	register ADDRESS *a;
17316162Seric {
17416162Seric 	register MAILER *m = a->q_mailer;
17516162Seric 
17616162Seric 	if (!bitnset(M_USR_UPPER, m->m_flags))
17716162Seric 		makelower(a->q_user);
17816162Seric }
17916162Seric /*
180297Seric **  PRESCAN -- Prescan name and make it canonical
181297Seric **
1829374Seric **	Scans a name and turns it into a set of tokens.  This process
1839374Seric **	deletes blanks and comments (in parentheses).
184297Seric **
185297Seric **	This routine knows about quoted strings and angle brackets.
186297Seric **
187297Seric **	There are certain subtleties to this routine.  The one that
188297Seric **	comes to mind now is that backslashes on the ends of names
189297Seric **	are silently stripped off; this is intentional.  The problem
190297Seric **	is that some versions of sndmsg (like at LBL) set the kill
191297Seric **	character to something other than @ when reading addresses;
192297Seric **	so people type "csvax.eric\@berkeley" -- which screws up the
193297Seric **	berknet mailer.
194297Seric **
195297Seric **	Parameters:
196297Seric **		addr -- the name to chomp.
197297Seric **		delim -- the delimiter for the address, normally
198297Seric **			'\0' or ','; \0 is accepted in any case.
19915284Seric **			If '\t' then we are reading the .cf file.
20016914Seric **		pvpbuf -- place to put the saved text -- note that
20116914Seric **			the pointers are static.
202297Seric **
203297Seric **	Returns:
2043149Seric **		A pointer to a vector of tokens.
205297Seric **		NULL on error.
206297Seric **
207297Seric **	Side Effects:
20825279Seric **		sets DelimChar to point to the character matching 'delim'.
209297Seric */
210297Seric 
2118078Seric /* states and character types */
2128078Seric # define OPR		0	/* operator */
2138078Seric # define ATM		1	/* atom */
2148078Seric # define QST		2	/* in quoted string */
2158078Seric # define SPC		3	/* chewing up spaces */
2168078Seric # define ONE		4	/* pick up one character */
2173149Seric 
2188078Seric # define NSTATES	5	/* number of states */
2198078Seric # define TYPE		017	/* mask to select state type */
2208078Seric 
2218078Seric /* meta bits for table */
2228078Seric # define M		020	/* meta character; don't pass through */
2238078Seric # define B		040	/* cause a break */
2248078Seric # define MB		M|B	/* meta-break */
2258078Seric 
2268078Seric static short StateTab[NSTATES][NSTATES] =
2278078Seric {
2288087Seric    /*	oldst	chtype>	OPR	ATM	QST	SPC	ONE	*/
2299051Seric 	/*OPR*/		OPR|B,	ATM|B,	QST|B,	SPC|MB,	ONE|B,
2309051Seric 	/*ATM*/		OPR|B,	ATM,	QST|B,	SPC|MB,	ONE|B,
2319051Seric 	/*QST*/		QST,	QST,	OPR,	QST,	QST,
2328078Seric 	/*SPC*/		OPR,	ATM,	QST,	SPC|M,	ONE,
2338078Seric 	/*ONE*/		OPR,	OPR,	OPR,	OPR,	OPR,
2348078Seric };
2358078Seric 
2368078Seric # define NOCHAR		-1	/* signal nothing in lookahead token */
2378078Seric 
2388078Seric char	*DelimChar;		/* set to point to the delimiter */
2398078Seric 
2403149Seric char **
24116914Seric prescan(addr, delim, pvpbuf)
242297Seric 	char *addr;
243297Seric 	char delim;
24416914Seric 	char pvpbuf[];
245297Seric {
246297Seric 	register char *p;
2478078Seric 	register char *q;
2489346Seric 	register int c;
2493149Seric 	char **avp;
250297Seric 	bool bslashmode;
251297Seric 	int cmntcnt;
2528423Seric 	int anglecnt;
2533149Seric 	char *tok;
2548078Seric 	int state;
2558078Seric 	int newstate;
2568078Seric 	static char *av[MAXATOM+1];
25715253Seric 	extern int errno;
258297Seric 
25915253Seric 	/* make sure error messages don't have garbage on them */
26015253Seric 	errno = 0;
26115253Seric 
26216914Seric 	q = pvpbuf;
2633149Seric 	bslashmode = FALSE;
2647800Seric 	cmntcnt = 0;
2658423Seric 	anglecnt = 0;
2663149Seric 	avp = av;
2678078Seric 	state = OPR;
2688078Seric 	c = NOCHAR;
2698078Seric 	p = addr;
2708078Seric 	if (tTd(22, 45))
271297Seric 	{
2728078Seric 		printf("prescan: ");
2738078Seric 		xputs(p);
27423109Seric 		(void) putchar('\n');
2758078Seric 	}
2768078Seric 
2778078Seric 	do
2788078Seric 	{
2793149Seric 		/* read a token */
2803149Seric 		tok = q;
2818078Seric 		for (;;)
282297Seric 		{
2838078Seric 			/* store away any old lookahead character */
2848078Seric 			if (c != NOCHAR)
2858078Seric 			{
28615284Seric 				/* see if there is room */
28716914Seric 				if (q >= &pvpbuf[PSBUFSIZE - 5])
2888078Seric 				{
2898078Seric 					usrerr("Address too long");
2908078Seric 					DelimChar = p;
2918078Seric 					return (NULL);
2928078Seric 				}
29315284Seric 
29415284Seric 				/* squirrel it away */
2958078Seric 				*q++ = c;
2968078Seric 			}
2978078Seric 
2988078Seric 			/* read a new input character */
2998078Seric 			c = *p++;
3008078Seric 			if (c == '\0')
3018078Seric 				break;
30215284Seric 			c &= ~0200;
30315284Seric 
3048078Seric 			if (tTd(22, 101))
3058078Seric 				printf("c=%c, s=%d; ", c, state);
3068078Seric 
3073149Seric 			/* chew up special characters */
3083149Seric 			*q = '\0';
3093149Seric 			if (bslashmode)
3103149Seric 			{
31124944Seric 				/* kludge \! for naive users */
31224944Seric 				if (c != '!')
31324944Seric 					c |= 0200;
3143149Seric 				bslashmode = FALSE;
3153149Seric 			}
3163149Seric 			else if (c == '\\')
3173149Seric 			{
3183149Seric 				bslashmode = TRUE;
3198078Seric 				c = NOCHAR;
3203149Seric 			}
3218514Seric 			else if (state == QST)
3228514Seric 			{
3238514Seric 				/* do nothing, just avoid next clauses */
3248514Seric 			}
3258078Seric 			else if (c == '(')
3264100Seric 			{
3278078Seric 				cmntcnt++;
3288078Seric 				c = NOCHAR;
3294100Seric 			}
3308078Seric 			else if (c == ')')
3313149Seric 			{
3328078Seric 				if (cmntcnt <= 0)
3333149Seric 				{
3348078Seric 					usrerr("Unbalanced ')'");
3358078Seric 					DelimChar = p;
3368078Seric 					return (NULL);
3373149Seric 				}
3388078Seric 				else
3398078Seric 					cmntcnt--;
3408078Seric 			}
3418078Seric 			else if (cmntcnt > 0)
3428078Seric 				c = NOCHAR;
3438423Seric 			else if (c == '<')
3448423Seric 				anglecnt++;
3458423Seric 			else if (c == '>')
3468423Seric 			{
3478423Seric 				if (anglecnt <= 0)
3488423Seric 				{
3498423Seric 					usrerr("Unbalanced '>'");
3508423Seric 					DelimChar = p;
3518423Seric 					return (NULL);
3528423Seric 				}
3538423Seric 				anglecnt--;
3548423Seric 			}
35511423Seric 			else if (delim == ' ' && isspace(c))
35611423Seric 				c = ' ';
3573149Seric 
3588078Seric 			if (c == NOCHAR)
3598078Seric 				continue;
3603149Seric 
3618078Seric 			/* see if this is end of input */
36211405Seric 			if (c == delim && anglecnt <= 0 && state != QST)
3633149Seric 				break;
3643149Seric 
3658078Seric 			newstate = StateTab[state][toktype(c)];
3668078Seric 			if (tTd(22, 101))
3678078Seric 				printf("ns=%02o\n", newstate);
3688078Seric 			state = newstate & TYPE;
3698078Seric 			if (bitset(M, newstate))
3708078Seric 				c = NOCHAR;
3718078Seric 			if (bitset(B, newstate))
3724228Seric 				break;
373297Seric 		}
3743149Seric 
3753149Seric 		/* new token */
3768078Seric 		if (tok != q)
3771378Seric 		{
3788078Seric 			*q++ = '\0';
3798078Seric 			if (tTd(22, 36))
380297Seric 			{
3818078Seric 				printf("tok=");
3828078Seric 				xputs(tok);
38323109Seric 				(void) putchar('\n');
384297Seric 			}
3858078Seric 			if (avp >= &av[MAXATOM])
386297Seric 			{
3878078Seric 				syserr("prescan: too many tokens");
3888078Seric 				DelimChar = p;
3898078Seric 				return (NULL);
390297Seric 			}
3918078Seric 			*avp++ = tok;
392297Seric 		}
3938423Seric 	} while (c != '\0' && (c != delim || anglecnt > 0));
3943149Seric 	*avp = NULL;
3958078Seric 	DelimChar = --p;
3963149Seric 	if (cmntcnt > 0)
3973149Seric 		usrerr("Unbalanced '('");
3988423Seric 	else if (anglecnt > 0)
3998423Seric 		usrerr("Unbalanced '<'");
4008078Seric 	else if (state == QST)
4013149Seric 		usrerr("Unbalanced '\"'");
4023149Seric 	else if (av[0] != NULL)
4033149Seric 		return (av);
4043149Seric 	return (NULL);
4053149Seric }
4063149Seric /*
4073149Seric **  TOKTYPE -- return token type
4083149Seric **
4093149Seric **	Parameters:
4103149Seric **		c -- the character in question.
4113149Seric **
4123149Seric **	Returns:
4133149Seric **		Its type.
4143149Seric **
4153149Seric **	Side Effects:
4163149Seric **		none.
4173149Seric */
418297Seric 
4193149Seric toktype(c)
4203149Seric 	register char c;
4213149Seric {
4223380Seric 	static char buf[50];
4233382Seric 	static bool firstime = TRUE;
4243380Seric 
4253382Seric 	if (firstime)
4263380Seric 	{
4273382Seric 		firstime = FALSE;
42816155Seric 		expand("\001o", buf, &buf[sizeof buf - 1], CurEnv);
4297005Seric 		(void) strcat(buf, DELIMCHARS);
4303380Seric 	}
4319585Seric 	if (c == MATCHCLASS || c == MATCHREPL || c == MATCHNCLASS)
4328078Seric 		return (ONE);
4338078Seric 	if (c == '"')
4348078Seric 		return (QST);
4354100Seric 	if (!isascii(c))
4368078Seric 		return (ATM);
4378078Seric 	if (isspace(c) || c == ')')
4388078Seric 		return (SPC);
4393380Seric 	if (iscntrl(c) || index(buf, c) != NULL)
4408078Seric 		return (OPR);
4418078Seric 	return (ATM);
4423149Seric }
4433149Seric /*
4443149Seric **  REWRITE -- apply rewrite rules to token vector.
4453149Seric **
4464476Seric **	This routine is an ordered production system.  Each rewrite
4474476Seric **	rule has a LHS (called the pattern) and a RHS (called the
4484476Seric **	rewrite); 'rwr' points the the current rewrite rule.
4494476Seric **
4504476Seric **	For each rewrite rule, 'avp' points the address vector we
4514476Seric **	are trying to match against, and 'pvp' points to the pattern.
4528058Seric **	If pvp points to a special match value (MATCHZANY, MATCHANY,
4539585Seric **	MATCHONE, MATCHCLASS, MATCHNCLASS) then the address in avp
4549585Seric **	matched is saved away in the match vector (pointed to by 'mvp').
4554476Seric **
4564476Seric **	When a match between avp & pvp does not match, we try to
4579585Seric **	back out.  If we back up over MATCHONE, MATCHCLASS, or MATCHNCLASS
4584476Seric **	we must also back out the match in mvp.  If we reach a
4598058Seric **	MATCHANY or MATCHZANY we just extend the match and start
4608058Seric **	over again.
4614476Seric **
4624476Seric **	When we finally match, we rewrite the address vector
4634476Seric **	and try over again.
4644476Seric **
4653149Seric **	Parameters:
4663149Seric **		pvp -- pointer to token vector.
4673149Seric **
4683149Seric **	Returns:
4693149Seric **		none.
4703149Seric **
4713149Seric **	Side Effects:
4723149Seric **		pvp is modified.
4733149Seric */
4742091Seric 
4753149Seric struct match
4763149Seric {
4774468Seric 	char	**first;	/* first token matched */
4784468Seric 	char	**last;		/* last token matched */
4793149Seric };
4803149Seric 
4814468Seric # define MAXMATCH	9	/* max params per rewrite */
4823149Seric 
4833149Seric 
4844070Seric rewrite(pvp, ruleset)
4853149Seric 	char **pvp;
4864070Seric 	int ruleset;
4873149Seric {
4883149Seric 	register char *ap;		/* address pointer */
4893149Seric 	register char *rp;		/* rewrite pointer */
4903149Seric 	register char **avp;		/* address vector pointer */
4913149Seric 	register char **rvp;		/* rewrite vector pointer */
4928058Seric 	register struct match *mlp;	/* cur ptr into mlist */
4938058Seric 	register struct rewrite *rwr;	/* pointer to current rewrite rule */
4944468Seric 	struct match mlist[MAXMATCH];	/* stores match on LHS */
4953149Seric 	char *npvp[MAXATOM+1];		/* temporary space for rebuild */
4963149Seric 
4979279Seric 	if (OpMode == MD_TEST || tTd(21, 2))
4983149Seric 	{
4998959Seric 		printf("rewrite: ruleset %2d   input:", ruleset);
5003149Seric 		printav(pvp);
5013149Seric 	}
5028423Seric 	if (pvp == NULL)
5038423Seric 		return;
5043149Seric 
5053149Seric 	/*
5063149Seric 	**  Run through the list of rewrite rules, applying
5073149Seric 	**	any that match.
5083149Seric 	*/
5093149Seric 
5104070Seric 	for (rwr = RewriteRules[ruleset]; rwr != NULL; )
5113149Seric 	{
5127675Seric 		if (tTd(21, 12))
513297Seric 		{
5148069Seric 			printf("-----trying rule:");
5153149Seric 			printav(rwr->r_lhs);
5163149Seric 		}
5173149Seric 
5183149Seric 		/* try to match on this rule */
5194468Seric 		mlp = mlist;
5208058Seric 		rvp = rwr->r_lhs;
5218058Seric 		avp = pvp;
5228058Seric 		while ((ap = *avp) != NULL || *rvp != NULL)
5233149Seric 		{
5243149Seric 			rp = *rvp;
5258058Seric 			if (tTd(21, 35))
5268058Seric 			{
5278069Seric 				printf("ap=");
5288058Seric 				xputs(ap);
5298069Seric 				printf(", rp=");
5308058Seric 				xputs(rp);
5318069Seric 				printf("\n");
5328058Seric 			}
5333149Seric 			if (rp == NULL)
534297Seric 			{
5353149Seric 				/* end-of-pattern before end-of-address */
5368058Seric 				goto backup;
537297Seric 			}
5388058Seric 			if (ap == NULL && *rp != MATCHZANY)
5398058Seric 			{
5408058Seric 				/* end-of-input */
5418058Seric 				break;
5428058Seric 			}
5433149Seric 
5443149Seric 			switch (*rp)
5453149Seric 			{
5464060Seric 				register STAB *s;
5474060Seric 
5484060Seric 			  case MATCHCLASS:
5499585Seric 			  case MATCHNCLASS:
5509585Seric 				/* match any token in (not in) a class */
5514100Seric 				s = stab(ap, ST_CLASS, ST_FIND);
55210690Seric 				if (s == NULL || !bitnset(rp[1], s->s_class))
5539585Seric 				{
5549585Seric 					if (*rp == MATCHCLASS)
5559585Seric 						goto backup;
5569585Seric 				}
5579585Seric 				else if (*rp == MATCHNCLASS)
5588058Seric 					goto backup;
5594468Seric 
5604476Seric 				/* explicit fall-through */
5614476Seric 
5624476Seric 			  case MATCHONE:
5634476Seric 			  case MATCHANY:
5644476Seric 				/* match exactly one token */
5658058Seric 				mlp->first = avp;
5668058Seric 				mlp->last = avp++;
5674468Seric 				mlp++;
5684060Seric 				break;
5694060Seric 
5708058Seric 			  case MATCHZANY:
5718058Seric 				/* match zero or more tokens */
5728058Seric 				mlp->first = avp;
5738058Seric 				mlp->last = avp - 1;
5748058Seric 				mlp++;
5758058Seric 				break;
5768058Seric 
5773149Seric 			  default:
5783149Seric 				/* must have exact match */
57933725Sbostic 				if (strcasecmp(rp, ap))
5808058Seric 					goto backup;
5814468Seric 				avp++;
5823149Seric 				break;
5833149Seric 			}
5843149Seric 
5853149Seric 			/* successful match on this token */
5863149Seric 			rvp++;
5873149Seric 			continue;
5883149Seric 
5898058Seric 		  backup:
5903149Seric 			/* match failed -- back up */
5913149Seric 			while (--rvp >= rwr->r_lhs)
5923149Seric 			{
5933149Seric 				rp = *rvp;
5948058Seric 				if (*rp == MATCHANY || *rp == MATCHZANY)
5954468Seric 				{
5964476Seric 					/* extend binding and continue */
5978058Seric 					avp = ++mlp[-1].last;
5988058Seric 					avp++;
5994476Seric 					rvp++;
6003149Seric 					break;
6014468Seric 				}
6024476Seric 				avp--;
6039585Seric 				if (*rp == MATCHONE || *rp == MATCHCLASS ||
6049585Seric 				    *rp == MATCHNCLASS)
6053149Seric 				{
6064468Seric 					/* back out binding */
6074468Seric 					mlp--;
6083149Seric 				}
6093149Seric 			}
6103149Seric 
6113149Seric 			if (rvp < rwr->r_lhs)
6123149Seric 			{
6133149Seric 				/* total failure to match */
6143149Seric 				break;
6153149Seric 			}
616297Seric 		}
6173149Seric 
6183149Seric 		/*
6193149Seric 		**  See if we successfully matched
6203149Seric 		*/
6213149Seric 
6229374Seric 		if (rvp < rwr->r_lhs || *rvp != NULL)
6233149Seric 		{
6249374Seric 			if (tTd(21, 10))
6259374Seric 				printf("----- rule fails\n");
6269374Seric 			rwr = rwr->r_next;
6279374Seric 			continue;
6289374Seric 		}
6293149Seric 
6309374Seric 		rvp = rwr->r_rhs;
6319374Seric 		if (tTd(21, 12))
6329374Seric 		{
6339374Seric 			printf("-----rule matches:");
6349374Seric 			printav(rvp);
6359374Seric 		}
6369374Seric 
6379374Seric 		rp = *rvp;
6389374Seric 		if (*rp == CANONUSER)
6399374Seric 		{
6409374Seric 			rvp++;
6419374Seric 			rwr = rwr->r_next;
6429374Seric 		}
6439374Seric 		else if (*rp == CANONHOST)
6449374Seric 		{
6459374Seric 			rvp++;
6469374Seric 			rwr = NULL;
6479374Seric 		}
6489374Seric 		else if (*rp == CANONNET)
6499374Seric 			rwr = NULL;
6509374Seric 
6519374Seric 		/* substitute */
6529374Seric 		for (avp = npvp; *rvp != NULL; rvp++)
6539374Seric 		{
6549374Seric 			register struct match *m;
6559374Seric 			register char **pp;
6569374Seric 
6578058Seric 			rp = *rvp;
65816914Seric 			if (*rp == MATCHREPL)
6598058Seric 			{
66016914Seric 				/* substitute from LHS */
66116914Seric 				m = &mlist[rp[1] - '1'];
66216914Seric 				if (m >= mlp)
6639374Seric 				{
66416914Seric 					syserr("rewrite: ruleset %d: replacement out of bounds", ruleset);
6659374Seric 					return;
6669374Seric 				}
66716914Seric 				if (tTd(21, 15))
66816914Seric 				{
66916914Seric 					printf("$%c:", rp[1]);
67016914Seric 					pp = m->first;
67116914Seric 					while (pp <= m->last)
67216914Seric 					{
67316914Seric 						printf(" %x=\"", *pp);
67416914Seric 						(void) fflush(stdout);
67516914Seric 						printf("%s\"", *pp++);
67616914Seric 					}
67716914Seric 					printf("\n");
67816914Seric 				}
6799374Seric 				pp = m->first;
6809374Seric 				while (pp <= m->last)
6813149Seric 				{
68216914Seric 					if (avp >= &npvp[MAXATOM])
68316914Seric 					{
68416914Seric 						syserr("rewrite: expansion too long");
68516914Seric 						return;
68616914Seric 					}
68716914Seric 					*avp++ = *pp++;
6883149Seric 				}
6893149Seric 			}
69016914Seric 			else
6918226Seric 			{
69216914Seric 				/* vanilla replacement */
6939374Seric 				if (avp >= &npvp[MAXATOM])
69416889Seric 				{
69516914Seric 	toolong:
69616889Seric 					syserr("rewrite: expansion too long");
69716889Seric 					return;
69816889Seric 				}
69916914Seric 				*avp++ = rp;
7008226Seric 			}
7019374Seric 		}
7029374Seric 		*avp++ = NULL;
70316914Seric 
70416914Seric 		/*
70516914Seric 		**  Check for any hostname lookups.
70616914Seric 		*/
70716914Seric 
70816914Seric 		for (rvp = npvp; *rvp != NULL; rvp++)
70916914Seric 		{
71016914Seric 			char **hbrvp;
71116914Seric 			char **xpvp;
71216914Seric 			int trsize;
71317473Seric 			char *olddelimchar;
71416920Seric 			char buf[MAXNAME + 1];
71516914Seric 			char *pvpb1[MAXATOM + 1];
71617174Seric 			char pvpbuf[PSBUFSIZE];
71717473Seric 			extern char *DelimChar;
71816914Seric 
71916914Seric 			if (**rvp != HOSTBEGIN)
72016914Seric 				continue;
72116914Seric 
72216914Seric 			/*
72316914Seric 			**  Got a hostname lookup.
72416914Seric 			**
72516914Seric 			**	This could be optimized fairly easily.
72616914Seric 			*/
72716914Seric 
72816914Seric 			hbrvp = rvp;
72916914Seric 
73016914Seric 			/* extract the match part */
73116914Seric 			while (*++rvp != NULL && **rvp != HOSTEND)
73216914Seric 				continue;
73316914Seric 			if (*rvp != NULL)
73416914Seric 				*rvp++ = NULL;
73516914Seric 
73616914Seric 			/* save the remainder of the input string */
73716914Seric 			trsize = (int) (avp - rvp + 1) * sizeof *rvp;
73816914Seric 			bcopy((char *) rvp, (char *) pvpb1, trsize);
73916914Seric 
74016914Seric 			/* look it up */
74116914Seric 			cataddr(++hbrvp, buf, sizeof buf);
74216914Seric 			maphostname(buf, sizeof buf);
74316914Seric 
74416914Seric 			/* scan the new host name */
74517473Seric 			olddelimchar = DelimChar;
74616914Seric 			xpvp = prescan(buf, '\0', pvpbuf);
74717473Seric 			DelimChar = olddelimchar;
74816914Seric 			if (xpvp == NULL)
74916914Seric 			{
75016914Seric 				syserr("rewrite: cannot prescan canonical hostname: %s", buf);
75122976Smiriam 				return;
75216914Seric 			}
75316914Seric 
75416914Seric 			/* append it to the token list */
75517174Seric 			for (avp = --hbrvp; *xpvp != NULL; xpvp++)
75617174Seric 			{
75717174Seric 				*avp++ = newstr(*xpvp);
75816920Seric 				if (avp >= &npvp[MAXATOM])
75916914Seric 					goto toolong;
76017174Seric 			}
76116914Seric 
76216914Seric 			/* restore the old trailing information */
76317177Seric 			for (xpvp = pvpb1; (*avp++ = *xpvp++) != NULL; )
76416920Seric 				if (avp >= &npvp[MAXATOM])
76516914Seric 					goto toolong;
76617174Seric 
76717174Seric 			break;
76816914Seric 		}
76916914Seric 
77016914Seric 		/*
77116914Seric 		**  Check for subroutine calls.
77216914Seric 		*/
77316914Seric 
77424944Seric 		if (*npvp != NULL && **npvp == CALLSUBR)
7759374Seric 		{
77616889Seric 			bcopy((char *) &npvp[2], (char *) pvp,
77716900Seric 				(int) (avp - npvp - 2) * sizeof *avp);
77816889Seric 			if (tTd(21, 3))
77916889Seric 				printf("-----callsubr %s\n", npvp[1]);
78016889Seric 			rewrite(pvp, atoi(npvp[1]));
7813149Seric 		}
7823149Seric 		else
7833149Seric 		{
78417348Seric 			bcopy((char *) npvp, (char *) pvp,
78516900Seric 				(int) (avp - npvp) * sizeof *avp);
7869374Seric 		}
7879374Seric 		if (tTd(21, 4))
7889374Seric 		{
7899374Seric 			printf("rewritten as:");
7909374Seric 			printav(pvp);
7919374Seric 		}
792297Seric 	}
7938069Seric 
7949279Seric 	if (OpMode == MD_TEST || tTd(21, 2))
7958069Seric 	{
7968959Seric 		printf("rewrite: ruleset %2d returns:", ruleset);
7978069Seric 		printav(pvp);
7988069Seric 	}
7993149Seric }
8003149Seric /*
8013149Seric **  BUILDADDR -- build address from token vector.
8023149Seric **
8033149Seric **	Parameters:
8043149Seric **		tv -- token vector.
8053149Seric **		a -- pointer to address descriptor to fill.
8063149Seric **			If NULL, one will be allocated.
8073149Seric **
8083149Seric **	Returns:
8094279Seric **		NULL if there was an error.
8104279Seric **		'a' otherwise.
8113149Seric **
8123149Seric **	Side Effects:
8133149Seric **		fills in 'a'
8143149Seric */
8153149Seric 
8163149Seric ADDRESS *
8173149Seric buildaddr(tv, a)
8183149Seric 	register char **tv;
8193149Seric 	register ADDRESS *a;
8203149Seric {
8213149Seric 	static char buf[MAXNAME];
8223149Seric 	struct mailer **mp;
8233149Seric 	register struct mailer *m;
8243149Seric 
8253149Seric 	if (a == NULL)
8263149Seric 		a = (ADDRESS *) xalloc(sizeof *a);
82716889Seric 	bzero((char *) a, sizeof *a);
8283149Seric 
8293149Seric 	/* figure out what net/mailer to use */
8303149Seric 	if (**tv != CANONNET)
8314279Seric 	{
8323149Seric 		syserr("buildaddr: no net");
8334279Seric 		return (NULL);
8344279Seric 	}
8353149Seric 	tv++;
83633725Sbostic 	if (!strcasecmp(*tv, "error"))
8374279Seric 	{
83810183Seric 		if (**++tv == CANONHOST)
83910183Seric 		{
84010183Seric 			setstat(atoi(*++tv));
84110183Seric 			tv++;
84210183Seric 		}
84310183Seric 		if (**tv != CANONUSER)
8444279Seric 			syserr("buildaddr: error: no user");
8454279Seric 		buf[0] = '\0';
8464279Seric 		while (*++tv != NULL)
8474279Seric 		{
8484279Seric 			if (buf[0] != '\0')
8497005Seric 				(void) strcat(buf, " ");
8507005Seric 			(void) strcat(buf, *tv);
8514279Seric 		}
8524279Seric 		usrerr(buf);
8534279Seric 		return (NULL);
8544279Seric 	}
8554598Seric 	for (mp = Mailer; (m = *mp++) != NULL; )
8563149Seric 	{
85733725Sbostic 		if (!strcasecmp(m->m_name, *tv))
8583149Seric 			break;
8593149Seric 	}
8603149Seric 	if (m == NULL)
8614279Seric 	{
86224944Seric 		syserr("buildaddr: unknown mailer %s", *tv);
8634279Seric 		return (NULL);
8644279Seric 	}
8654598Seric 	a->q_mailer = m;
8663149Seric 
8673149Seric 	/* figure out what host (if any) */
8683149Seric 	tv++;
86910690Seric 	if (!bitnset(M_LOCAL, m->m_flags))
8703149Seric 	{
8715704Seric 		if (**tv++ != CANONHOST)
8724279Seric 		{
8733149Seric 			syserr("buildaddr: no host");
8744279Seric 			return (NULL);
8754279Seric 		}
8765704Seric 		buf[0] = '\0';
8775704Seric 		while (*tv != NULL && **tv != CANONUSER)
8787005Seric 			(void) strcat(buf, *tv++);
8795704Seric 		a->q_host = newstr(buf);
8803149Seric 	}
8813149Seric 	else
8823149Seric 		a->q_host = NULL;
8833149Seric 
8843149Seric 	/* figure out the user */
88536615Sbostic 	if (*tv == NULL || **tv != CANONUSER)
8864279Seric 	{
8873149Seric 		syserr("buildaddr: no user");
8884279Seric 		return (NULL);
8894279Seric 	}
89019040Seric 
89119040Seric 	/* rewrite according recipient mailer rewriting rules */
89219040Seric 	rewrite(++tv, 2);
89319040Seric 	if (m->m_r_rwset > 0)
89419040Seric 		rewrite(tv, m->m_r_rwset);
89519040Seric 	rewrite(tv, 4);
89619040Seric 
89719040Seric 	/* save the result for the command line/RCPT argument */
89811278Seric 	cataddr(tv, buf, sizeof buf);
8993149Seric 	a->q_user = buf;
9003149Seric 
9013149Seric 	return (a);
9023149Seric }
9033188Seric /*
9044228Seric **  CATADDR -- concatenate pieces of addresses (putting in <LWSP> subs)
9054228Seric **
9064228Seric **	Parameters:
9074228Seric **		pvp -- parameter vector to rebuild.
9084228Seric **		buf -- buffer to build the string into.
9094228Seric **		sz -- size of buf.
9104228Seric **
9114228Seric **	Returns:
9124228Seric **		none.
9134228Seric **
9144228Seric **	Side Effects:
9154228Seric **		Destroys buf.
9164228Seric */
9174228Seric 
9184228Seric cataddr(pvp, buf, sz)
9194228Seric 	char **pvp;
9204228Seric 	char *buf;
9214228Seric 	register int sz;
9224228Seric {
9234228Seric 	bool oatomtok = FALSE;
9244228Seric 	bool natomtok = FALSE;
9254228Seric 	register int i;
9264228Seric 	register char *p;
9274228Seric 
9288423Seric 	if (pvp == NULL)
9298423Seric 	{
93023109Seric 		(void) strcpy(buf, "");
9318423Seric 		return;
9328423Seric 	}
9334228Seric 	p = buf;
93411156Seric 	sz -= 2;
9354228Seric 	while (*pvp != NULL && (i = strlen(*pvp)) < sz)
9364228Seric 	{
9378078Seric 		natomtok = (toktype(**pvp) == ATM);
9384228Seric 		if (oatomtok && natomtok)
9399042Seric 			*p++ = SpaceSub;
9404228Seric 		(void) strcpy(p, *pvp);
9414228Seric 		oatomtok = natomtok;
9424228Seric 		p += i;
94311156Seric 		sz -= i + 1;
9444228Seric 		pvp++;
9454228Seric 	}
9464228Seric 	*p = '\0';
9474228Seric }
9484228Seric /*
9493188Seric **  SAMEADDR -- Determine if two addresses are the same
9503188Seric **
9513188Seric **	This is not just a straight comparison -- if the mailer doesn't
9523188Seric **	care about the host we just ignore it, etc.
9533188Seric **
9543188Seric **	Parameters:
9553188Seric **		a, b -- pointers to the internal forms to compare.
9563188Seric **
9573188Seric **	Returns:
9583188Seric **		TRUE -- they represent the same mailbox.
9593188Seric **		FALSE -- they don't.
9603188Seric **
9613188Seric **	Side Effects:
9623188Seric **		none.
9633188Seric */
9643188Seric 
9653188Seric bool
9669374Seric sameaddr(a, b)
9673188Seric 	register ADDRESS *a;
9683188Seric 	register ADDRESS *b;
9693188Seric {
9703188Seric 	/* if they don't have the same mailer, forget it */
9713188Seric 	if (a->q_mailer != b->q_mailer)
9723188Seric 		return (FALSE);
9733188Seric 
9743188Seric 	/* if the user isn't the same, we can drop out */
9759374Seric 	if (strcmp(a->q_user, b->q_user) != 0)
9763188Seric 		return (FALSE);
9773188Seric 
9783188Seric 	/* if the mailer ignores hosts, we have succeeded! */
97910690Seric 	if (bitnset(M_LOCAL, a->q_mailer->m_flags))
9803188Seric 		return (TRUE);
9813188Seric 
9823188Seric 	/* otherwise compare hosts (but be careful for NULL ptrs) */
9833188Seric 	if (a->q_host == NULL || b->q_host == NULL)
9843188Seric 		return (FALSE);
9853188Seric 	if (strcmp(a->q_host, b->q_host) != 0)
9863188Seric 		return (FALSE);
9873188Seric 
9883188Seric 	return (TRUE);
9893188Seric }
9903234Seric /*
9913234Seric **  PRINTADDR -- print address (for debugging)
9923234Seric **
9933234Seric **	Parameters:
9943234Seric **		a -- the address to print
9953234Seric **		follow -- follow the q_next chain.
9963234Seric **
9973234Seric **	Returns:
9983234Seric **		none.
9993234Seric **
10003234Seric **	Side Effects:
10013234Seric **		none.
10023234Seric */
10033234Seric 
10043234Seric printaddr(a, follow)
10053234Seric 	register ADDRESS *a;
10063234Seric 	bool follow;
10073234Seric {
10085001Seric 	bool first = TRUE;
10095001Seric 
10103234Seric 	while (a != NULL)
10113234Seric 	{
10125001Seric 		first = FALSE;
10134443Seric 		printf("%x=", a);
10144085Seric 		(void) fflush(stdout);
101540973Sbostic 		printf("%s: mailer %d (%s), host `%s', user `%s', ruser `%s'\n",
101640973Sbostic 		       a->q_paddr, a->q_mailer->m_mno, a->q_mailer->m_name,
101740973Sbostic 		       a->q_host, a->q_user, a->q_ruser? a->q_ruser: "<null>");
10188181Seric 		printf("\tnext=%x, flags=%o, alias %x\n", a->q_next, a->q_flags,
10198181Seric 		       a->q_alias);
10208181Seric 		printf("\thome=\"%s\", fullname=\"%s\"\n", a->q_home,
10218181Seric 		       a->q_fullname);
10224996Seric 
10233234Seric 		if (!follow)
10243234Seric 			return;
10254996Seric 		a = a->q_next;
10263234Seric 	}
10275001Seric 	if (first)
10284443Seric 		printf("[NULL]\n");
10293234Seric }
10304317Seric 
10317682Seric /*
10327682Seric **  REMOTENAME -- return the name relative to the current mailer
10337682Seric **
10347682Seric **	Parameters:
10357682Seric **		name -- the name to translate.
10368069Seric **		m -- the mailer that we want to do rewriting relative
10378069Seric **			to.
10388069Seric **		senderaddress -- if set, uses the sender rewriting rules
10398069Seric **			rather than the recipient rewriting rules.
104010310Seric **		canonical -- if set, strip out any comment information,
104110310Seric **			etc.
10427682Seric **
10437682Seric **	Returns:
10447682Seric **		the text string representing this address relative to
10457682Seric **			the receiving mailer.
10467682Seric **
10477682Seric **	Side Effects:
10487682Seric **		none.
10497682Seric **
10507682Seric **	Warnings:
10517682Seric **		The text string returned is tucked away locally;
10527682Seric **			copy it if you intend to save it.
10537682Seric */
10547682Seric 
10557682Seric char *
105610310Seric remotename(name, m, senderaddress, canonical)
10577682Seric 	char *name;
10587682Seric 	struct mailer *m;
10598069Seric 	bool senderaddress;
106010310Seric 	bool canonical;
10617682Seric {
10628069Seric 	register char **pvp;
10638069Seric 	char *fancy;
10648069Seric 	extern char *macvalue();
10658181Seric 	char *oldg = macvalue('g', CurEnv);
10667682Seric 	static char buf[MAXNAME];
10677682Seric 	char lbuf[MAXNAME];
106816914Seric 	char pvpbuf[PSBUFSIZE];
10697682Seric 	extern char **prescan();
10707889Seric 	extern char *crackaddr();
10717682Seric 
10727755Seric 	if (tTd(12, 1))
10737755Seric 		printf("remotename(%s)\n", name);
10747755Seric 
107510177Seric 	/* don't do anything if we are tagging it as special */
107610177Seric 	if ((senderaddress ? m->m_s_rwset : m->m_r_rwset) < 0)
107710177Seric 		return (name);
107810177Seric 
10797682Seric 	/*
10808181Seric 	**  Do a heuristic crack of this name to extract any comment info.
10818181Seric 	**	This will leave the name as a comment and a $g macro.
10827889Seric 	*/
10837889Seric 
108410310Seric 	if (canonical)
108516155Seric 		fancy = "\001g";
108610310Seric 	else
108710310Seric 		fancy = crackaddr(name);
10887889Seric 
10898181Seric 	/*
10908181Seric 	**  Turn the name into canonical form.
10918181Seric 	**	Normally this will be RFC 822 style, i.e., "user@domain".
10928181Seric 	**	If this only resolves to "user", and the "C" flag is
10938181Seric 	**	specified in the sending mailer, then the sender's
10948181Seric 	**	domain will be appended.
10958181Seric 	*/
10968181Seric 
109716914Seric 	pvp = prescan(name, '\0', pvpbuf);
10987889Seric 	if (pvp == NULL)
10997889Seric 		return (name);
11008181Seric 	rewrite(pvp, 3);
11018181Seric 	if (CurEnv->e_fromdomain != NULL)
11028181Seric 	{
11038181Seric 		/* append from domain to this address */
11048181Seric 		register char **pxp = pvp;
11058181Seric 
11069594Seric 		/* see if there is an "@domain" in the current name */
11078181Seric 		while (*pxp != NULL && strcmp(*pxp, "@") != 0)
11088181Seric 			pxp++;
11098181Seric 		if (*pxp == NULL)
11108181Seric 		{
11119594Seric 			/* no.... append the "@domain" from the sender */
11128181Seric 			register char **qxq = CurEnv->e_fromdomain;
11138181Seric 
11149594Seric 			while ((*pxp++ = *qxq++) != NULL)
11159594Seric 				continue;
111611726Seric 			rewrite(pvp, 3);
11178181Seric 		}
11188181Seric 	}
11198181Seric 
11208181Seric 	/*
11218959Seric 	**  Do more specific rewriting.
11228181Seric 	**	Rewrite using ruleset 1 or 2 depending on whether this is
11238181Seric 	**		a sender address or not.
11248181Seric 	**	Then run it through any receiving-mailer-specific rulesets.
11258181Seric 	*/
11268181Seric 
11278069Seric 	if (senderaddress)
11287755Seric 	{
11297889Seric 		rewrite(pvp, 1);
11308069Seric 		if (m->m_s_rwset > 0)
11318069Seric 			rewrite(pvp, m->m_s_rwset);
11328069Seric 	}
11338069Seric 	else
11348069Seric 	{
11357889Seric 		rewrite(pvp, 2);
11368069Seric 		if (m->m_r_rwset > 0)
11378069Seric 			rewrite(pvp, m->m_r_rwset);
11387682Seric 	}
11397682Seric 
11408181Seric 	/*
11418959Seric 	**  Do any final sanitation the address may require.
11428959Seric 	**	This will normally be used to turn internal forms
11438959Seric 	**	(e.g., user@host.LOCAL) into external form.  This
11448959Seric 	**	may be used as a default to the above rules.
11458959Seric 	*/
11468959Seric 
11478959Seric 	rewrite(pvp, 4);
11488959Seric 
11498959Seric 	/*
11508181Seric 	**  Now restore the comment information we had at the beginning.
11518181Seric 	*/
11528181Seric 
11537682Seric 	cataddr(pvp, lbuf, sizeof lbuf);
11549374Seric 	define('g', lbuf, CurEnv);
11557889Seric 	expand(fancy, buf, &buf[sizeof buf - 1], CurEnv);
11569374Seric 	define('g', oldg, CurEnv);
11577682Seric 
11587682Seric 	if (tTd(12, 1))
11597755Seric 		printf("remotename => `%s'\n", buf);
11607682Seric 	return (buf);
11617682Seric }
1162