1*22976Smiriam 
2*22976Smiriam /*
3*22976Smiriam **  Sendmail
4*22976Smiriam **  Copyright (c) 1983  Eric P. Allman
5*22976Smiriam **  Berkeley, California
6*22976Smiriam **
7*22976Smiriam **  Copyright (c) 1983 Regents of the University of California.
8*22976Smiriam **  All rights reserved.  The Berkeley software License Agreement
9*22976Smiriam **  specifies the terms and conditions for redistribution.
10*22976Smiriam */
11*22976Smiriam 
12*22976Smiriam #ifndef lint
13*22976Smiriam static char	SccsId[] = "@(#)parseaddr.c	4.16 (Berkeley) 06/07/85";
14*22976Smiriam #endif not lint
15*22976Smiriam 
163312Seric # include "sendmail.h"
17297Seric 
18297Seric /*
199888Seric **  PARSEADDR -- Parse an address
20297Seric **
21297Seric **	Parses an address and breaks it up into three parts: a
22297Seric **	net to transmit the message on, the host to transmit it
23297Seric **	to, and a user on that host.  These are loaded into an
242973Seric **	ADDRESS header with the values squirreled away if necessary.
25297Seric **	The "user" part may not be a real user; the process may
26297Seric **	just reoccur on that machine.  For example, on a machine
27297Seric **	with an arpanet connection, the address
28297Seric **		csvax.bill@berkeley
29297Seric **	will break up to a "user" of 'csvax.bill' and a host
30297Seric **	of 'berkeley' -- to be transmitted over the arpanet.
31297Seric **
32297Seric **	Parameters:
33297Seric **		addr -- the address to parse.
34297Seric **		a -- a pointer to the address descriptor buffer.
35297Seric **			If NULL, a header will be created.
36297Seric **		copyf -- determines what shall be copied:
37297Seric **			-1 -- don't copy anything.  The printname
38297Seric **				(q_paddr) is just addr, and the
39297Seric **				user & host are allocated internally
40297Seric **				to parse.
41297Seric **			0 -- copy out the parsed user & host, but
42297Seric **				don't copy the printname.
43297Seric **			+1 -- copy everything.
4411445Seric **		delim -- the character to terminate the address, passed
4511445Seric **			to prescan.
46297Seric **
47297Seric **	Returns:
48297Seric **		A pointer to the address descriptor header (`a' if
49297Seric **			`a' is non-NULL).
50297Seric **		NULL on error.
51297Seric **
52297Seric **	Side Effects:
53297Seric **		none
54297Seric */
55297Seric 
569374Seric /* following delimiters are inherent to the internal algorithms */
5716155Seric # define DELIMCHARS	"\001()<>,;\\\"\r\n"	/* word delimiters */
582091Seric 
592973Seric ADDRESS *
6011445Seric parseaddr(addr, a, copyf, delim)
61297Seric 	char *addr;
622973Seric 	register ADDRESS *a;
63297Seric 	int copyf;
6411445Seric 	char delim;
65297Seric {
663149Seric 	register char **pvp;
673149Seric 	register struct mailer *m;
6816914Seric 	char pvpbuf[PSBUFSIZE];
693149Seric 	extern char **prescan();
703149Seric 	extern ADDRESS *buildaddr();
71297Seric 
72297Seric 	/*
73297Seric 	**  Initialize and prescan address.
74297Seric 	*/
75297Seric 
766903Seric 	CurEnv->e_to = addr;
773188Seric # ifdef DEBUG
787675Seric 	if (tTd(20, 1))
799888Seric 		printf("\n--parseaddr(%s)\n", addr);
803188Seric # endif DEBUG
813188Seric 
8216914Seric 	pvp = prescan(addr, delim, pvpbuf);
833149Seric 	if (pvp == NULL)
84297Seric 		return (NULL);
85297Seric 
86297Seric 	/*
873149Seric 	**  Apply rewriting rules.
887889Seric 	**	Ruleset 0 does basic parsing.  It must resolve.
89297Seric 	*/
90297Seric 
918181Seric 	rewrite(pvp, 3);
924070Seric 	rewrite(pvp, 0);
93297Seric 
943149Seric 	/*
953149Seric 	**  See if we resolved to a real mailer.
963149Seric 	*/
97297Seric 
983149Seric 	if (pvp[0][0] != CANONNET)
993149Seric 	{
1003149Seric 		setstat(EX_USAGE);
1013149Seric 		usrerr("cannot resolve name");
1023149Seric 		return (NULL);
103297Seric 	}
104297Seric 
105297Seric 	/*
1063149Seric 	**  Build canonical address from pvp.
107297Seric 	*/
108297Seric 
1093149Seric 	a = buildaddr(pvp, a);
1104279Seric 	if (a == NULL)
1114279Seric 		return (NULL);
1124598Seric 	m = a->q_mailer;
113297Seric 
114297Seric 	/*
1153149Seric 	**  Make local copies of the host & user and then
1163149Seric 	**  transport them out.
117297Seric 	*/
118297Seric 
119297Seric 	if (copyf > 0)
1208078Seric 	{
1218078Seric 		extern char *DelimChar;
1228078Seric 		char savec = *DelimChar;
1238078Seric 
1248078Seric 		*DelimChar = '\0';
1252973Seric 		a->q_paddr = newstr(addr);
1268078Seric 		*DelimChar = savec;
1278078Seric 	}
128297Seric 	else
129297Seric 		a->q_paddr = addr;
1303149Seric 	if (copyf >= 0)
131297Seric 	{
1323149Seric 		if (a->q_host != NULL)
1333149Seric 			a->q_host = newstr(a->q_host);
134297Seric 		else
1353149Seric 			a->q_host = "";
1363149Seric 		if (a->q_user != a->q_paddr)
1373149Seric 			a->q_user = newstr(a->q_user);
138297Seric 	}
139297Seric 
140297Seric 	/*
14116202Seric 	**  Convert host name to lower case if requested.
14216202Seric 	**	User name will be done later.
14316202Seric 	*/
14416202Seric 
14516202Seric 	if (!bitnset(M_HST_UPPER, m->m_flags))
14616202Seric 		makelower(a->q_host);
14716202Seric 
14816202Seric 	/*
149297Seric 	**  Compute return value.
150297Seric 	*/
151297Seric 
152297Seric # ifdef DEBUG
1537675Seric 	if (tTd(20, 1))
1544443Seric 	{
1559888Seric 		printf("parseaddr-->");
1564443Seric 		printaddr(a, FALSE);
1574443Seric 	}
158297Seric # endif DEBUG
159297Seric 
160297Seric 	return (a);
161297Seric }
162297Seric /*
16316162Seric **  LOWERADDR -- map UPPER->lower case on addresses as requested.
16416162Seric **
16516162Seric **	Parameters:
16616162Seric **		a -- address to be mapped.
16716162Seric **
16816162Seric **	Returns:
16916162Seric **		none.
17016162Seric **
17116162Seric **	Side Effects:
17216162Seric **		none.
17316162Seric */
17416162Seric 
17516162Seric loweraddr(a)
17616162Seric 	register ADDRESS *a;
17716162Seric {
17816162Seric 	register MAILER *m = a->q_mailer;
17916162Seric 
18016162Seric 	if (!bitnset(M_USR_UPPER, m->m_flags))
18116162Seric 		makelower(a->q_user);
18216162Seric }
18316162Seric /*
184297Seric **  PRESCAN -- Prescan name and make it canonical
185297Seric **
1869374Seric **	Scans a name and turns it into a set of tokens.  This process
1879374Seric **	deletes blanks and comments (in parentheses).
188297Seric **
189297Seric **	This routine knows about quoted strings and angle brackets.
190297Seric **
191297Seric **	There are certain subtleties to this routine.  The one that
192297Seric **	comes to mind now is that backslashes on the ends of names
193297Seric **	are silently stripped off; this is intentional.  The problem
194297Seric **	is that some versions of sndmsg (like at LBL) set the kill
195297Seric **	character to something other than @ when reading addresses;
196297Seric **	so people type "csvax.eric\@berkeley" -- which screws up the
197297Seric **	berknet mailer.
198297Seric **
199297Seric **	Parameters:
200297Seric **		addr -- the name to chomp.
201297Seric **		delim -- the delimiter for the address, normally
202297Seric **			'\0' or ','; \0 is accepted in any case.
20315284Seric **			If '\t' then we are reading the .cf file.
20416914Seric **		pvpbuf -- place to put the saved text -- note that
20516914Seric **			the pointers are static.
206297Seric **
207297Seric **	Returns:
2083149Seric **		A pointer to a vector of tokens.
209297Seric **		NULL on error.
210297Seric **
211297Seric **	Side Effects:
2123149Seric **		none.
213297Seric */
214297Seric 
2158078Seric /* states and character types */
2168078Seric # define OPR		0	/* operator */
2178078Seric # define ATM		1	/* atom */
2188078Seric # define QST		2	/* in quoted string */
2198078Seric # define SPC		3	/* chewing up spaces */
2208078Seric # define ONE		4	/* pick up one character */
2213149Seric 
2228078Seric # define NSTATES	5	/* number of states */
2238078Seric # define TYPE		017	/* mask to select state type */
2248078Seric 
2258078Seric /* meta bits for table */
2268078Seric # define M		020	/* meta character; don't pass through */
2278078Seric # define B		040	/* cause a break */
2288078Seric # define MB		M|B	/* meta-break */
2298078Seric 
2308078Seric static short StateTab[NSTATES][NSTATES] =
2318078Seric {
2328087Seric    /*	oldst	chtype>	OPR	ATM	QST	SPC	ONE	*/
2339051Seric 	/*OPR*/		OPR|B,	ATM|B,	QST|B,	SPC|MB,	ONE|B,
2349051Seric 	/*ATM*/		OPR|B,	ATM,	QST|B,	SPC|MB,	ONE|B,
2359051Seric 	/*QST*/		QST,	QST,	OPR,	QST,	QST,
2368078Seric 	/*SPC*/		OPR,	ATM,	QST,	SPC|M,	ONE,
2378078Seric 	/*ONE*/		OPR,	OPR,	OPR,	OPR,	OPR,
2388078Seric };
2398078Seric 
2408078Seric # define NOCHAR		-1	/* signal nothing in lookahead token */
2418078Seric 
2428078Seric char	*DelimChar;		/* set to point to the delimiter */
2438078Seric 
2443149Seric char **
24516914Seric prescan(addr, delim, pvpbuf)
246297Seric 	char *addr;
247297Seric 	char delim;
24816914Seric 	char pvpbuf[];
249297Seric {
250297Seric 	register char *p;
2518078Seric 	register char *q;
2529346Seric 	register int c;
2533149Seric 	char **avp;
254297Seric 	bool bslashmode;
255297Seric 	int cmntcnt;
2568423Seric 	int anglecnt;
2573149Seric 	char *tok;
2588078Seric 	int state;
2598078Seric 	int newstate;
2608078Seric 	static char *av[MAXATOM+1];
26115253Seric 	extern int errno;
262297Seric 
26315253Seric 	/* make sure error messages don't have garbage on them */
26415253Seric 	errno = 0;
26515253Seric 
26616914Seric 	q = pvpbuf;
2673149Seric 	bslashmode = FALSE;
2687800Seric 	cmntcnt = 0;
2698423Seric 	anglecnt = 0;
2703149Seric 	avp = av;
2718078Seric 	state = OPR;
2728078Seric 	c = NOCHAR;
2738078Seric 	p = addr;
2748078Seric # ifdef DEBUG
2758078Seric 	if (tTd(22, 45))
276297Seric 	{
2778078Seric 		printf("prescan: ");
2788078Seric 		xputs(p);
2798078Seric 		putchar('\n');
2808078Seric 	}
2818078Seric # endif DEBUG
2828078Seric 
2838078Seric 	do
2848078Seric 	{
2853149Seric 		/* read a token */
2863149Seric 		tok = q;
2878078Seric 		for (;;)
288297Seric 		{
2898078Seric 			/* store away any old lookahead character */
2908078Seric 			if (c != NOCHAR)
2918078Seric 			{
29215284Seric 				/* see if there is room */
29316914Seric 				if (q >= &pvpbuf[PSBUFSIZE - 5])
2948078Seric 				{
2958078Seric 					usrerr("Address too long");
2968078Seric 					DelimChar = p;
2978078Seric 					return (NULL);
2988078Seric 				}
29915284Seric 
30015284Seric 				/* squirrel it away */
3018078Seric 				*q++ = c;
3028078Seric 			}
3038078Seric 
3048078Seric 			/* read a new input character */
3058078Seric 			c = *p++;
3068078Seric 			if (c == '\0')
3078078Seric 				break;
30815284Seric 			c &= ~0200;
30915284Seric 
3108078Seric # ifdef DEBUG
3118078Seric 			if (tTd(22, 101))
3128078Seric 				printf("c=%c, s=%d; ", c, state);
3138078Seric # endif DEBUG
3148078Seric 
3153149Seric 			/* chew up special characters */
3163149Seric 			*q = '\0';
3173149Seric 			if (bslashmode)
3183149Seric 			{
3193149Seric 				c |= 0200;
3203149Seric 				bslashmode = FALSE;
3213149Seric 			}
3223149Seric 			else if (c == '\\')
3233149Seric 			{
3243149Seric 				bslashmode = TRUE;
3258078Seric 				c = NOCHAR;
3263149Seric 			}
3278514Seric 			else if (state == QST)
3288514Seric 			{
3298514Seric 				/* do nothing, just avoid next clauses */
3308514Seric 			}
3318078Seric 			else if (c == '(')
3324100Seric 			{
3338078Seric 				cmntcnt++;
3348078Seric 				c = NOCHAR;
3354100Seric 			}
3368078Seric 			else if (c == ')')
3373149Seric 			{
3388078Seric 				if (cmntcnt <= 0)
3393149Seric 				{
3408078Seric 					usrerr("Unbalanced ')'");
3418078Seric 					DelimChar = p;
3428078Seric 					return (NULL);
3433149Seric 				}
3448078Seric 				else
3458078Seric 					cmntcnt--;
3468078Seric 			}
3478078Seric 			else if (cmntcnt > 0)
3488078Seric 				c = NOCHAR;
3498423Seric 			else if (c == '<')
3508423Seric 				anglecnt++;
3518423Seric 			else if (c == '>')
3528423Seric 			{
3538423Seric 				if (anglecnt <= 0)
3548423Seric 				{
3558423Seric 					usrerr("Unbalanced '>'");
3568423Seric 					DelimChar = p;
3578423Seric 					return (NULL);
3588423Seric 				}
3598423Seric 				anglecnt--;
3608423Seric 			}
36111423Seric 			else if (delim == ' ' && isspace(c))
36211423Seric 				c = ' ';
3633149Seric 
3648078Seric 			if (c == NOCHAR)
3658078Seric 				continue;
3663149Seric 
3678078Seric 			/* see if this is end of input */
36811405Seric 			if (c == delim && anglecnt <= 0 && state != QST)
3693149Seric 				break;
3703149Seric 
3718078Seric 			newstate = StateTab[state][toktype(c)];
3728078Seric # ifdef DEBUG
3738078Seric 			if (tTd(22, 101))
3748078Seric 				printf("ns=%02o\n", newstate);
3758078Seric # endif DEBUG
3768078Seric 			state = newstate & TYPE;
3778078Seric 			if (bitset(M, newstate))
3788078Seric 				c = NOCHAR;
3798078Seric 			if (bitset(B, newstate))
3804228Seric 				break;
381297Seric 		}
3823149Seric 
3833149Seric 		/* new token */
3848078Seric 		if (tok != q)
3851378Seric 		{
3868078Seric 			*q++ = '\0';
3878078Seric # ifdef DEBUG
3888078Seric 			if (tTd(22, 36))
389297Seric 			{
3908078Seric 				printf("tok=");
3918078Seric 				xputs(tok);
3928078Seric 				putchar('\n');
393297Seric 			}
3948078Seric # endif DEBUG
3958078Seric 			if (avp >= &av[MAXATOM])
396297Seric 			{
3978078Seric 				syserr("prescan: too many tokens");
3988078Seric 				DelimChar = p;
3998078Seric 				return (NULL);
400297Seric 			}
4018078Seric 			*avp++ = tok;
402297Seric 		}
4038423Seric 	} while (c != '\0' && (c != delim || anglecnt > 0));
4043149Seric 	*avp = NULL;
4058078Seric 	DelimChar = --p;
4063149Seric 	if (cmntcnt > 0)
4073149Seric 		usrerr("Unbalanced '('");
4088423Seric 	else if (anglecnt > 0)
4098423Seric 		usrerr("Unbalanced '<'");
4108078Seric 	else if (state == QST)
4113149Seric 		usrerr("Unbalanced '\"'");
4123149Seric 	else if (av[0] != NULL)
4133149Seric 		return (av);
4143149Seric 	return (NULL);
4153149Seric }
4163149Seric /*
4173149Seric **  TOKTYPE -- return token type
4183149Seric **
4193149Seric **	Parameters:
4203149Seric **		c -- the character in question.
4213149Seric **
4223149Seric **	Returns:
4233149Seric **		Its type.
4243149Seric **
4253149Seric **	Side Effects:
4263149Seric **		none.
4273149Seric */
428297Seric 
4293149Seric toktype(c)
4303149Seric 	register char c;
4313149Seric {
4323380Seric 	static char buf[50];
4333382Seric 	static bool firstime = TRUE;
4343380Seric 
4353382Seric 	if (firstime)
4363380Seric 	{
4373382Seric 		firstime = FALSE;
43816155Seric 		expand("\001o", buf, &buf[sizeof buf - 1], CurEnv);
4397005Seric 		(void) strcat(buf, DELIMCHARS);
4403380Seric 	}
4419585Seric 	if (c == MATCHCLASS || c == MATCHREPL || c == MATCHNCLASS)
4428078Seric 		return (ONE);
4438078Seric 	if (c == '"')
4448078Seric 		return (QST);
4454100Seric 	if (!isascii(c))
4468078Seric 		return (ATM);
4478078Seric 	if (isspace(c) || c == ')')
4488078Seric 		return (SPC);
4493380Seric 	if (iscntrl(c) || index(buf, c) != NULL)
4508078Seric 		return (OPR);
4518078Seric 	return (ATM);
4523149Seric }
4533149Seric /*
4543149Seric **  REWRITE -- apply rewrite rules to token vector.
4553149Seric **
4564476Seric **	This routine is an ordered production system.  Each rewrite
4574476Seric **	rule has a LHS (called the pattern) and a RHS (called the
4584476Seric **	rewrite); 'rwr' points the the current rewrite rule.
4594476Seric **
4604476Seric **	For each rewrite rule, 'avp' points the address vector we
4614476Seric **	are trying to match against, and 'pvp' points to the pattern.
4628058Seric **	If pvp points to a special match value (MATCHZANY, MATCHANY,
4639585Seric **	MATCHONE, MATCHCLASS, MATCHNCLASS) then the address in avp
4649585Seric **	matched is saved away in the match vector (pointed to by 'mvp').
4654476Seric **
4664476Seric **	When a match between avp & pvp does not match, we try to
4679585Seric **	back out.  If we back up over MATCHONE, MATCHCLASS, or MATCHNCLASS
4684476Seric **	we must also back out the match in mvp.  If we reach a
4698058Seric **	MATCHANY or MATCHZANY we just extend the match and start
4708058Seric **	over again.
4714476Seric **
4724476Seric **	When we finally match, we rewrite the address vector
4734476Seric **	and try over again.
4744476Seric **
4753149Seric **	Parameters:
4763149Seric **		pvp -- pointer to token vector.
4773149Seric **
4783149Seric **	Returns:
4793149Seric **		none.
4803149Seric **
4813149Seric **	Side Effects:
4823149Seric **		pvp is modified.
4833149Seric */
4842091Seric 
4853149Seric struct match
4863149Seric {
4874468Seric 	char	**first;	/* first token matched */
4884468Seric 	char	**last;		/* last token matched */
4893149Seric };
4903149Seric 
4914468Seric # define MAXMATCH	9	/* max params per rewrite */
4923149Seric 
4933149Seric 
4944070Seric rewrite(pvp, ruleset)
4953149Seric 	char **pvp;
4964070Seric 	int ruleset;
4973149Seric {
4983149Seric 	register char *ap;		/* address pointer */
4993149Seric 	register char *rp;		/* rewrite pointer */
5003149Seric 	register char **avp;		/* address vector pointer */
5013149Seric 	register char **rvp;		/* rewrite vector pointer */
5028058Seric 	register struct match *mlp;	/* cur ptr into mlist */
5038058Seric 	register struct rewrite *rwr;	/* pointer to current rewrite rule */
5044468Seric 	struct match mlist[MAXMATCH];	/* stores match on LHS */
5053149Seric 	char *npvp[MAXATOM+1];		/* temporary space for rebuild */
5064060Seric 	extern bool sameword();
5073149Seric 
5089279Seric 	if (OpMode == MD_TEST || tTd(21, 2))
5093149Seric 	{
5108959Seric 		printf("rewrite: ruleset %2d   input:", ruleset);
5113149Seric 		printav(pvp);
5123149Seric 	}
5138423Seric 	if (pvp == NULL)
5148423Seric 		return;
5153149Seric 
5163149Seric 	/*
5173149Seric 	**  Run through the list of rewrite rules, applying
5183149Seric 	**	any that match.
5193149Seric 	*/
5203149Seric 
5214070Seric 	for (rwr = RewriteRules[ruleset]; rwr != NULL; )
5223149Seric 	{
5234100Seric # ifdef DEBUG
5247675Seric 		if (tTd(21, 12))
525297Seric 		{
5268069Seric 			printf("-----trying rule:");
5273149Seric 			printav(rwr->r_lhs);
5283149Seric 		}
5294100Seric # endif DEBUG
5303149Seric 
5313149Seric 		/* try to match on this rule */
5324468Seric 		mlp = mlist;
5338058Seric 		rvp = rwr->r_lhs;
5348058Seric 		avp = pvp;
5358058Seric 		while ((ap = *avp) != NULL || *rvp != NULL)
5363149Seric 		{
5373149Seric 			rp = *rvp;
5388058Seric # ifdef DEBUG
5398058Seric 			if (tTd(21, 35))
5408058Seric 			{
5418069Seric 				printf("ap=");
5428058Seric 				xputs(ap);
5438069Seric 				printf(", rp=");
5448058Seric 				xputs(rp);
5458069Seric 				printf("\n");
5468058Seric 			}
5478058Seric # endif DEBUG
5483149Seric 			if (rp == NULL)
549297Seric 			{
5503149Seric 				/* end-of-pattern before end-of-address */
5518058Seric 				goto backup;
552297Seric 			}
5538058Seric 			if (ap == NULL && *rp != MATCHZANY)
5548058Seric 			{
5558058Seric 				/* end-of-input */
5568058Seric 				break;
5578058Seric 			}
5583149Seric 
5593149Seric 			switch (*rp)
5603149Seric 			{
5614060Seric 				register STAB *s;
5624060Seric 
5634060Seric 			  case MATCHCLASS:
5649585Seric 			  case MATCHNCLASS:
5659585Seric 				/* match any token in (not in) a class */
5664100Seric 				s = stab(ap, ST_CLASS, ST_FIND);
56710690Seric 				if (s == NULL || !bitnset(rp[1], s->s_class))
5689585Seric 				{
5699585Seric 					if (*rp == MATCHCLASS)
5709585Seric 						goto backup;
5719585Seric 				}
5729585Seric 				else if (*rp == MATCHNCLASS)
5738058Seric 					goto backup;
5744468Seric 
5754476Seric 				/* explicit fall-through */
5764476Seric 
5774476Seric 			  case MATCHONE:
5784476Seric 			  case MATCHANY:
5794476Seric 				/* match exactly one token */
5808058Seric 				mlp->first = avp;
5818058Seric 				mlp->last = avp++;
5824468Seric 				mlp++;
5834060Seric 				break;
5844060Seric 
5858058Seric 			  case MATCHZANY:
5868058Seric 				/* match zero or more tokens */
5878058Seric 				mlp->first = avp;
5888058Seric 				mlp->last = avp - 1;
5898058Seric 				mlp++;
5908058Seric 				break;
5918058Seric 
5923149Seric 			  default:
5933149Seric 				/* must have exact match */
5944060Seric 				if (!sameword(rp, ap))
5958058Seric 					goto backup;
5964468Seric 				avp++;
5973149Seric 				break;
5983149Seric 			}
5993149Seric 
6003149Seric 			/* successful match on this token */
6013149Seric 			rvp++;
6023149Seric 			continue;
6033149Seric 
6048058Seric 		  backup:
6053149Seric 			/* match failed -- back up */
6063149Seric 			while (--rvp >= rwr->r_lhs)
6073149Seric 			{
6083149Seric 				rp = *rvp;
6098058Seric 				if (*rp == MATCHANY || *rp == MATCHZANY)
6104468Seric 				{
6114476Seric 					/* extend binding and continue */
6128058Seric 					avp = ++mlp[-1].last;
6138058Seric 					avp++;
6144476Seric 					rvp++;
6153149Seric 					break;
6164468Seric 				}
6174476Seric 				avp--;
6189585Seric 				if (*rp == MATCHONE || *rp == MATCHCLASS ||
6199585Seric 				    *rp == MATCHNCLASS)
6203149Seric 				{
6214468Seric 					/* back out binding */
6224468Seric 					mlp--;
6233149Seric 				}
6243149Seric 			}
6253149Seric 
6263149Seric 			if (rvp < rwr->r_lhs)
6273149Seric 			{
6283149Seric 				/* total failure to match */
6293149Seric 				break;
6303149Seric 			}
631297Seric 		}
6323149Seric 
6333149Seric 		/*
6343149Seric 		**  See if we successfully matched
6353149Seric 		*/
6363149Seric 
6379374Seric 		if (rvp < rwr->r_lhs || *rvp != NULL)
6383149Seric 		{
6394100Seric # ifdef DEBUG
6409374Seric 			if (tTd(21, 10))
6419374Seric 				printf("----- rule fails\n");
6424100Seric # endif DEBUG
6439374Seric 			rwr = rwr->r_next;
6449374Seric 			continue;
6459374Seric 		}
6463149Seric 
6479374Seric 		rvp = rwr->r_rhs;
6489374Seric # ifdef DEBUG
6499374Seric 		if (tTd(21, 12))
6509374Seric 		{
6519374Seric 			printf("-----rule matches:");
6529374Seric 			printav(rvp);
6539374Seric 		}
6549374Seric # endif DEBUG
6559374Seric 
6569374Seric 		rp = *rvp;
6579374Seric 		if (*rp == CANONUSER)
6589374Seric 		{
6599374Seric 			rvp++;
6609374Seric 			rwr = rwr->r_next;
6619374Seric 		}
6629374Seric 		else if (*rp == CANONHOST)
6639374Seric 		{
6649374Seric 			rvp++;
6659374Seric 			rwr = NULL;
6669374Seric 		}
6679374Seric 		else if (*rp == CANONNET)
6689374Seric 			rwr = NULL;
6699374Seric 
6709374Seric 		/* substitute */
6719374Seric 		for (avp = npvp; *rvp != NULL; rvp++)
6729374Seric 		{
6739374Seric 			register struct match *m;
6749374Seric 			register char **pp;
6759374Seric 
6768058Seric 			rp = *rvp;
67716914Seric 			if (*rp == MATCHREPL)
6788058Seric 			{
67916914Seric 				/* substitute from LHS */
68016914Seric 				m = &mlist[rp[1] - '1'];
68116914Seric 				if (m >= mlp)
6829374Seric 				{
68316914Seric 					syserr("rewrite: ruleset %d: replacement out of bounds", ruleset);
6849374Seric 					return;
6859374Seric 				}
6869374Seric # ifdef DEBUG
68716914Seric 				if (tTd(21, 15))
68816914Seric 				{
68916914Seric 					printf("$%c:", rp[1]);
69016914Seric 					pp = m->first;
69116914Seric 					while (pp <= m->last)
69216914Seric 					{
69316914Seric 						printf(" %x=\"", *pp);
69416914Seric 						(void) fflush(stdout);
69516914Seric 						printf("%s\"", *pp++);
69616914Seric 					}
69716914Seric 					printf("\n");
69816914Seric 				}
69916914Seric # endif DEBUG
7009374Seric 				pp = m->first;
7019374Seric 				while (pp <= m->last)
7023149Seric 				{
70316914Seric 					if (avp >= &npvp[MAXATOM])
70416914Seric 					{
70516914Seric 						syserr("rewrite: expansion too long");
70616914Seric 						return;
70716914Seric 					}
70816914Seric 					*avp++ = *pp++;
7093149Seric 				}
7103149Seric 			}
71116914Seric 			else
7128226Seric 			{
71316914Seric 				/* vanilla replacement */
7149374Seric 				if (avp >= &npvp[MAXATOM])
71516889Seric 				{
71616914Seric 	toolong:
71716889Seric 					syserr("rewrite: expansion too long");
71816889Seric 					return;
71916889Seric 				}
72016914Seric 				*avp++ = rp;
7218226Seric 			}
7229374Seric 		}
7239374Seric 		*avp++ = NULL;
72416914Seric 
72516914Seric 		/*
72616914Seric 		**  Check for any hostname lookups.
72716914Seric 		*/
72816914Seric 
72916914Seric 		for (rvp = npvp; *rvp != NULL; rvp++)
73016914Seric 		{
73116914Seric 			char **hbrvp;
73216914Seric 			char **xpvp;
73316914Seric 			int trsize;
73417473Seric 			char *olddelimchar;
73516920Seric 			char buf[MAXNAME + 1];
73616914Seric 			char *pvpb1[MAXATOM + 1];
73717174Seric 			char pvpbuf[PSBUFSIZE];
73817473Seric 			extern char *DelimChar;
73916914Seric 
74016914Seric 			if (**rvp != HOSTBEGIN)
74116914Seric 				continue;
74216914Seric 
74316914Seric 			/*
74416914Seric 			**  Got a hostname lookup.
74516914Seric 			**
74616914Seric 			**	This could be optimized fairly easily.
74716914Seric 			*/
74816914Seric 
74916914Seric 			hbrvp = rvp;
75016914Seric 
75116914Seric 			/* extract the match part */
75216914Seric 			while (*++rvp != NULL && **rvp != HOSTEND)
75316914Seric 				continue;
75416914Seric 			if (*rvp != NULL)
75516914Seric 				*rvp++ = NULL;
75616914Seric 
75716914Seric 			/* save the remainder of the input string */
75816914Seric 			trsize = (int) (avp - rvp + 1) * sizeof *rvp;
75916914Seric 			bcopy((char *) rvp, (char *) pvpb1, trsize);
76016914Seric 
76116914Seric 			/* look it up */
76216914Seric 			cataddr(++hbrvp, buf, sizeof buf);
76316914Seric 			maphostname(buf, sizeof buf);
76416914Seric 
76516914Seric 			/* scan the new host name */
76617473Seric 			olddelimchar = DelimChar;
76716914Seric 			xpvp = prescan(buf, '\0', pvpbuf);
76817473Seric 			DelimChar = olddelimchar;
76916914Seric 			if (xpvp == NULL)
77016914Seric 			{
77116914Seric 				syserr("rewrite: cannot prescan canonical hostname: %s", buf);
772*22976Smiriam 				return;
77316914Seric 			}
77416914Seric 
77516914Seric 			/* append it to the token list */
77617174Seric 			for (avp = --hbrvp; *xpvp != NULL; xpvp++)
77717174Seric 			{
77817174Seric 				*avp++ = newstr(*xpvp);
77916920Seric 				if (avp >= &npvp[MAXATOM])
78016914Seric 					goto toolong;
78117174Seric 			}
78216914Seric 
78316914Seric 			/* restore the old trailing information */
78417177Seric 			for (xpvp = pvpb1; (*avp++ = *xpvp++) != NULL; )
78516920Seric 				if (avp >= &npvp[MAXATOM])
78616914Seric 					goto toolong;
78717174Seric 
78817174Seric 			break;
78916914Seric 		}
79016914Seric 
79116914Seric 		/*
79216914Seric 		**  Check for subroutine calls.
79316914Seric 		*/
79416914Seric 
79516889Seric 		if (**npvp == CALLSUBR)
7969374Seric 		{
79716889Seric 			bcopy((char *) &npvp[2], (char *) pvp,
79816900Seric 				(int) (avp - npvp - 2) * sizeof *avp);
79916889Seric # ifdef DEBUG
80016889Seric 			if (tTd(21, 3))
80116889Seric 				printf("-----callsubr %s\n", npvp[1]);
80216889Seric # endif DEBUG
80316889Seric 			rewrite(pvp, atoi(npvp[1]));
8043149Seric 		}
8053149Seric 		else
8063149Seric 		{
80717348Seric 			bcopy((char *) npvp, (char *) pvp,
80816900Seric 				(int) (avp - npvp) * sizeof *avp);
8099374Seric 		}
8104100Seric # ifdef DEBUG
8119374Seric 		if (tTd(21, 4))
8129374Seric 		{
8139374Seric 			printf("rewritten as:");
8149374Seric 			printav(pvp);
8159374Seric 		}
8164100Seric # endif DEBUG
817297Seric 	}
8188069Seric 
8199279Seric 	if (OpMode == MD_TEST || tTd(21, 2))
8208069Seric 	{
8218959Seric 		printf("rewrite: ruleset %2d returns:", ruleset);
8228069Seric 		printav(pvp);
8238069Seric 	}
8243149Seric }
8253149Seric /*
8263149Seric **  BUILDADDR -- build address from token vector.
8273149Seric **
8283149Seric **	Parameters:
8293149Seric **		tv -- token vector.
8303149Seric **		a -- pointer to address descriptor to fill.
8313149Seric **			If NULL, one will be allocated.
8323149Seric **
8333149Seric **	Returns:
8344279Seric **		NULL if there was an error.
8354279Seric **		'a' otherwise.
8363149Seric **
8373149Seric **	Side Effects:
8383149Seric **		fills in 'a'
8393149Seric */
8403149Seric 
8413149Seric ADDRESS *
8423149Seric buildaddr(tv, a)
8433149Seric 	register char **tv;
8443149Seric 	register ADDRESS *a;
8453149Seric {
8463149Seric 	static char buf[MAXNAME];
8473149Seric 	struct mailer **mp;
8483149Seric 	register struct mailer *m;
8494635Seric 	extern bool sameword();
8503149Seric 
8513149Seric 	if (a == NULL)
8523149Seric 		a = (ADDRESS *) xalloc(sizeof *a);
85316889Seric 	bzero((char *) a, sizeof *a);
8543149Seric 
8553149Seric 	/* figure out what net/mailer to use */
8563149Seric 	if (**tv != CANONNET)
8574279Seric 	{
8583149Seric 		syserr("buildaddr: no net");
8594279Seric 		return (NULL);
8604279Seric 	}
8613149Seric 	tv++;
8624635Seric 	if (sameword(*tv, "error"))
8634279Seric 	{
86410183Seric 		if (**++tv == CANONHOST)
86510183Seric 		{
86610183Seric 			setstat(atoi(*++tv));
86710183Seric 			tv++;
86810183Seric 		}
86910183Seric 		if (**tv != CANONUSER)
8704279Seric 			syserr("buildaddr: error: no user");
8714279Seric 		buf[0] = '\0';
8724279Seric 		while (*++tv != NULL)
8734279Seric 		{
8744279Seric 			if (buf[0] != '\0')
8757005Seric 				(void) strcat(buf, " ");
8767005Seric 			(void) strcat(buf, *tv);
8774279Seric 		}
8784279Seric 		usrerr(buf);
8794279Seric 		return (NULL);
8804279Seric 	}
8814598Seric 	for (mp = Mailer; (m = *mp++) != NULL; )
8823149Seric 	{
8834635Seric 		if (sameword(m->m_name, *tv))
8843149Seric 			break;
8853149Seric 	}
8863149Seric 	if (m == NULL)
8874279Seric 	{
8883149Seric 		syserr("buildaddr: unknown net %s", *tv);
8894279Seric 		return (NULL);
8904279Seric 	}
8914598Seric 	a->q_mailer = m;
8923149Seric 
8933149Seric 	/* figure out what host (if any) */
8943149Seric 	tv++;
89510690Seric 	if (!bitnset(M_LOCAL, m->m_flags))
8963149Seric 	{
8975704Seric 		if (**tv++ != CANONHOST)
8984279Seric 		{
8993149Seric 			syserr("buildaddr: no host");
9004279Seric 			return (NULL);
9014279Seric 		}
9025704Seric 		buf[0] = '\0';
9035704Seric 		while (*tv != NULL && **tv != CANONUSER)
9047005Seric 			(void) strcat(buf, *tv++);
9055704Seric 		a->q_host = newstr(buf);
9063149Seric 	}
9073149Seric 	else
9083149Seric 		a->q_host = NULL;
9093149Seric 
9103149Seric 	/* figure out the user */
9113149Seric 	if (**tv != CANONUSER)
9124279Seric 	{
9133149Seric 		syserr("buildaddr: no user");
9144279Seric 		return (NULL);
9154279Seric 	}
91619040Seric 
91719040Seric 	/* rewrite according recipient mailer rewriting rules */
91819040Seric 	rewrite(++tv, 2);
91919040Seric 	if (m->m_r_rwset > 0)
92019040Seric 		rewrite(tv, m->m_r_rwset);
92119040Seric 	rewrite(tv, 4);
92219040Seric 
92319040Seric 	/* save the result for the command line/RCPT argument */
92411278Seric 	cataddr(tv, buf, sizeof buf);
9253149Seric 	a->q_user = buf;
9263149Seric 
9273149Seric 	return (a);
9283149Seric }
9293188Seric /*
9304228Seric **  CATADDR -- concatenate pieces of addresses (putting in <LWSP> subs)
9314228Seric **
9324228Seric **	Parameters:
9334228Seric **		pvp -- parameter vector to rebuild.
9344228Seric **		buf -- buffer to build the string into.
9354228Seric **		sz -- size of buf.
9364228Seric **
9374228Seric **	Returns:
9384228Seric **		none.
9394228Seric **
9404228Seric **	Side Effects:
9414228Seric **		Destroys buf.
9424228Seric */
9434228Seric 
9444228Seric cataddr(pvp, buf, sz)
9454228Seric 	char **pvp;
9464228Seric 	char *buf;
9474228Seric 	register int sz;
9484228Seric {
9494228Seric 	bool oatomtok = FALSE;
9504228Seric 	bool natomtok = FALSE;
9514228Seric 	register int i;
9524228Seric 	register char *p;
9534228Seric 
9548423Seric 	if (pvp == NULL)
9558423Seric 	{
9568423Seric 		strcpy(buf, "");
9578423Seric 		return;
9588423Seric 	}
9594228Seric 	p = buf;
96011156Seric 	sz -= 2;
9614228Seric 	while (*pvp != NULL && (i = strlen(*pvp)) < sz)
9624228Seric 	{
9638078Seric 		natomtok = (toktype(**pvp) == ATM);
9644228Seric 		if (oatomtok && natomtok)
9659042Seric 			*p++ = SpaceSub;
9664228Seric 		(void) strcpy(p, *pvp);
9674228Seric 		oatomtok = natomtok;
9684228Seric 		p += i;
96911156Seric 		sz -= i + 1;
9704228Seric 		pvp++;
9714228Seric 	}
9724228Seric 	*p = '\0';
9734228Seric }
9744228Seric /*
9753188Seric **  SAMEADDR -- Determine if two addresses are the same
9763188Seric **
9773188Seric **	This is not just a straight comparison -- if the mailer doesn't
9783188Seric **	care about the host we just ignore it, etc.
9793188Seric **
9803188Seric **	Parameters:
9813188Seric **		a, b -- pointers to the internal forms to compare.
9823188Seric **
9833188Seric **	Returns:
9843188Seric **		TRUE -- they represent the same mailbox.
9853188Seric **		FALSE -- they don't.
9863188Seric **
9873188Seric **	Side Effects:
9883188Seric **		none.
9893188Seric */
9903188Seric 
9913188Seric bool
9929374Seric sameaddr(a, b)
9933188Seric 	register ADDRESS *a;
9943188Seric 	register ADDRESS *b;
9953188Seric {
9963188Seric 	/* if they don't have the same mailer, forget it */
9973188Seric 	if (a->q_mailer != b->q_mailer)
9983188Seric 		return (FALSE);
9993188Seric 
10003188Seric 	/* if the user isn't the same, we can drop out */
10019374Seric 	if (strcmp(a->q_user, b->q_user) != 0)
10023188Seric 		return (FALSE);
10033188Seric 
10043188Seric 	/* if the mailer ignores hosts, we have succeeded! */
100510690Seric 	if (bitnset(M_LOCAL, a->q_mailer->m_flags))
10063188Seric 		return (TRUE);
10073188Seric 
10083188Seric 	/* otherwise compare hosts (but be careful for NULL ptrs) */
10093188Seric 	if (a->q_host == NULL || b->q_host == NULL)
10103188Seric 		return (FALSE);
10113188Seric 	if (strcmp(a->q_host, b->q_host) != 0)
10123188Seric 		return (FALSE);
10133188Seric 
10143188Seric 	return (TRUE);
10153188Seric }
10163234Seric /*
10173234Seric **  PRINTADDR -- print address (for debugging)
10183234Seric **
10193234Seric **	Parameters:
10203234Seric **		a -- the address to print
10213234Seric **		follow -- follow the q_next chain.
10223234Seric **
10233234Seric **	Returns:
10243234Seric **		none.
10253234Seric **
10263234Seric **	Side Effects:
10273234Seric **		none.
10283234Seric */
10293234Seric 
10304317Seric # ifdef DEBUG
10314317Seric 
10323234Seric printaddr(a, follow)
10333234Seric 	register ADDRESS *a;
10343234Seric 	bool follow;
10353234Seric {
10365001Seric 	bool first = TRUE;
10375001Seric 
10383234Seric 	while (a != NULL)
10393234Seric 	{
10405001Seric 		first = FALSE;
10414443Seric 		printf("%x=", a);
10424085Seric 		(void) fflush(stdout);
10433234Seric 		printf("%s: mailer %d (%s), host `%s', user `%s'\n", a->q_paddr,
10448181Seric 		       a->q_mailer->m_mno, a->q_mailer->m_name, a->q_host,
10458181Seric 		       a->q_user);
10468181Seric 		printf("\tnext=%x, flags=%o, alias %x\n", a->q_next, a->q_flags,
10478181Seric 		       a->q_alias);
10488181Seric 		printf("\thome=\"%s\", fullname=\"%s\"\n", a->q_home,
10498181Seric 		       a->q_fullname);
10504996Seric 
10513234Seric 		if (!follow)
10523234Seric 			return;
10534996Seric 		a = a->q_next;
10543234Seric 	}
10555001Seric 	if (first)
10564443Seric 		printf("[NULL]\n");
10573234Seric }
10584317Seric 
10594317Seric # endif DEBUG
10607682Seric /*
10617682Seric **  REMOTENAME -- return the name relative to the current mailer
10627682Seric **
10637682Seric **	Parameters:
10647682Seric **		name -- the name to translate.
10658069Seric **		m -- the mailer that we want to do rewriting relative
10668069Seric **			to.
10678069Seric **		senderaddress -- if set, uses the sender rewriting rules
10688069Seric **			rather than the recipient rewriting rules.
106910310Seric **		canonical -- if set, strip out any comment information,
107010310Seric **			etc.
10717682Seric **
10727682Seric **	Returns:
10737682Seric **		the text string representing this address relative to
10747682Seric **			the receiving mailer.
10757682Seric **
10767682Seric **	Side Effects:
10777682Seric **		none.
10787682Seric **
10797682Seric **	Warnings:
10807682Seric **		The text string returned is tucked away locally;
10817682Seric **			copy it if you intend to save it.
10827682Seric */
10837682Seric 
10847682Seric char *
108510310Seric remotename(name, m, senderaddress, canonical)
10867682Seric 	char *name;
10877682Seric 	struct mailer *m;
10888069Seric 	bool senderaddress;
108910310Seric 	bool canonical;
10907682Seric {
10918069Seric 	register char **pvp;
10928069Seric 	char *fancy;
10938069Seric 	extern char *macvalue();
10948181Seric 	char *oldg = macvalue('g', CurEnv);
10957682Seric 	static char buf[MAXNAME];
10967682Seric 	char lbuf[MAXNAME];
109716914Seric 	char pvpbuf[PSBUFSIZE];
10987682Seric 	extern char **prescan();
10997889Seric 	extern char *crackaddr();
11007682Seric 
11017755Seric # ifdef DEBUG
11027755Seric 	if (tTd(12, 1))
11037755Seric 		printf("remotename(%s)\n", name);
11047755Seric # endif DEBUG
11057755Seric 
110610177Seric 	/* don't do anything if we are tagging it as special */
110710177Seric 	if ((senderaddress ? m->m_s_rwset : m->m_r_rwset) < 0)
110810177Seric 		return (name);
110910177Seric 
11107682Seric 	/*
11118181Seric 	**  Do a heuristic crack of this name to extract any comment info.
11128181Seric 	**	This will leave the name as a comment and a $g macro.
11137889Seric 	*/
11147889Seric 
111510310Seric 	if (canonical)
111616155Seric 		fancy = "\001g";
111710310Seric 	else
111810310Seric 		fancy = crackaddr(name);
11197889Seric 
11208181Seric 	/*
11218181Seric 	**  Turn the name into canonical form.
11228181Seric 	**	Normally this will be RFC 822 style, i.e., "user@domain".
11238181Seric 	**	If this only resolves to "user", and the "C" flag is
11248181Seric 	**	specified in the sending mailer, then the sender's
11258181Seric 	**	domain will be appended.
11268181Seric 	*/
11278181Seric 
112816914Seric 	pvp = prescan(name, '\0', pvpbuf);
11297889Seric 	if (pvp == NULL)
11307889Seric 		return (name);
11318181Seric 	rewrite(pvp, 3);
11328181Seric 	if (CurEnv->e_fromdomain != NULL)
11338181Seric 	{
11348181Seric 		/* append from domain to this address */
11358181Seric 		register char **pxp = pvp;
11368181Seric 
11379594Seric 		/* see if there is an "@domain" in the current name */
11388181Seric 		while (*pxp != NULL && strcmp(*pxp, "@") != 0)
11398181Seric 			pxp++;
11408181Seric 		if (*pxp == NULL)
11418181Seric 		{
11429594Seric 			/* no.... append the "@domain" from the sender */
11438181Seric 			register char **qxq = CurEnv->e_fromdomain;
11448181Seric 
11459594Seric 			while ((*pxp++ = *qxq++) != NULL)
11469594Seric 				continue;
114711726Seric 			rewrite(pvp, 3);
11488181Seric 		}
11498181Seric 	}
11508181Seric 
11518181Seric 	/*
11528959Seric 	**  Do more specific rewriting.
11538181Seric 	**	Rewrite using ruleset 1 or 2 depending on whether this is
11548181Seric 	**		a sender address or not.
11558181Seric 	**	Then run it through any receiving-mailer-specific rulesets.
11568181Seric 	*/
11578181Seric 
11588069Seric 	if (senderaddress)
11597755Seric 	{
11607889Seric 		rewrite(pvp, 1);
11618069Seric 		if (m->m_s_rwset > 0)
11628069Seric 			rewrite(pvp, m->m_s_rwset);
11638069Seric 	}
11648069Seric 	else
11658069Seric 	{
11667889Seric 		rewrite(pvp, 2);
11678069Seric 		if (m->m_r_rwset > 0)
11688069Seric 			rewrite(pvp, m->m_r_rwset);
11697682Seric 	}
11707682Seric 
11718181Seric 	/*
11728959Seric 	**  Do any final sanitation the address may require.
11738959Seric 	**	This will normally be used to turn internal forms
11748959Seric 	**	(e.g., user@host.LOCAL) into external form.  This
11758959Seric 	**	may be used as a default to the above rules.
11768959Seric 	*/
11778959Seric 
11788959Seric 	rewrite(pvp, 4);
11798959Seric 
11808959Seric 	/*
11818181Seric 	**  Now restore the comment information we had at the beginning.
11828181Seric 	*/
11838181Seric 
11847682Seric 	cataddr(pvp, lbuf, sizeof lbuf);
11859374Seric 	define('g', lbuf, CurEnv);
11867889Seric 	expand(fancy, buf, &buf[sizeof buf - 1], CurEnv);
11879374Seric 	define('g', oldg, CurEnv);
11887682Seric 
11897682Seric # ifdef DEBUG
11907682Seric 	if (tTd(12, 1))
11917755Seric 		printf("remotename => `%s'\n", buf);
11927682Seric # endif DEBUG
11937682Seric 	return (buf);
11947682Seric }
1195