122976Smiriam /*
2*33730Sbostic  * Copyright (c) 1988 Regents of the University of California.
3*33730Sbostic  * All rights reserved.
4*33730Sbostic  *
5*33730Sbostic  * Redistribution and use in source and binary forms are permitted
6*33730Sbostic  * provided that this notice is preserved and that due credit is given
7*33730Sbostic  * to the University of California at Berkeley. The name of the University
8*33730Sbostic  * may not be used to endorse or promote products derived from this
9*33730Sbostic  * software without specific prior written permission. This software
10*33730Sbostic  * is provided ``as is'' without express or implied warranty.
11*33730Sbostic  *
12*33730Sbostic  *  Sendmail
13*33730Sbostic  *  Copyright (c) 1983  Eric P. Allman
14*33730Sbostic  *  Berkeley, California
15*33730Sbostic  */
1622976Smiriam 
1722976Smiriam #ifndef lint
18*33730Sbostic static char sccsid[] = "@(#)parseaddr.c	5.8 (Berkeley) 03/13/88";
19*33730Sbostic #endif /* not lint */
2022976Smiriam 
213312Seric # include "sendmail.h"
22297Seric 
23297Seric /*
249888Seric **  PARSEADDR -- Parse an address
25297Seric **
26297Seric **	Parses an address and breaks it up into three parts: a
27297Seric **	net to transmit the message on, the host to transmit it
28297Seric **	to, and a user on that host.  These are loaded into an
292973Seric **	ADDRESS header with the values squirreled away if necessary.
30297Seric **	The "user" part may not be a real user; the process may
31297Seric **	just reoccur on that machine.  For example, on a machine
32297Seric **	with an arpanet connection, the address
33297Seric **		csvax.bill@berkeley
34297Seric **	will break up to a "user" of 'csvax.bill' and a host
35297Seric **	of 'berkeley' -- to be transmitted over the arpanet.
36297Seric **
37297Seric **	Parameters:
38297Seric **		addr -- the address to parse.
39297Seric **		a -- a pointer to the address descriptor buffer.
40297Seric **			If NULL, a header will be created.
41297Seric **		copyf -- determines what shall be copied:
42297Seric **			-1 -- don't copy anything.  The printname
43297Seric **				(q_paddr) is just addr, and the
44297Seric **				user & host are allocated internally
45297Seric **				to parse.
46297Seric **			0 -- copy out the parsed user & host, but
47297Seric **				don't copy the printname.
48297Seric **			+1 -- copy everything.
4911445Seric **		delim -- the character to terminate the address, passed
5011445Seric **			to prescan.
51297Seric **
52297Seric **	Returns:
53297Seric **		A pointer to the address descriptor header (`a' if
54297Seric **			`a' is non-NULL).
55297Seric **		NULL on error.
56297Seric **
57297Seric **	Side Effects:
58297Seric **		none
59297Seric */
60297Seric 
619374Seric /* following delimiters are inherent to the internal algorithms */
6216155Seric # define DELIMCHARS	"\001()<>,;\\\"\r\n"	/* word delimiters */
632091Seric 
642973Seric ADDRESS *
6511445Seric parseaddr(addr, a, copyf, delim)
66297Seric 	char *addr;
672973Seric 	register ADDRESS *a;
68297Seric 	int copyf;
6911445Seric 	char delim;
70297Seric {
713149Seric 	register char **pvp;
723149Seric 	register struct mailer *m;
7316914Seric 	char pvpbuf[PSBUFSIZE];
743149Seric 	extern char **prescan();
753149Seric 	extern ADDRESS *buildaddr();
76297Seric 
77297Seric 	/*
78297Seric 	**  Initialize and prescan address.
79297Seric 	*/
80297Seric 
816903Seric 	CurEnv->e_to = addr;
823188Seric # ifdef DEBUG
837675Seric 	if (tTd(20, 1))
849888Seric 		printf("\n--parseaddr(%s)\n", addr);
853188Seric # endif DEBUG
863188Seric 
8716914Seric 	pvp = prescan(addr, delim, pvpbuf);
883149Seric 	if (pvp == NULL)
89297Seric 		return (NULL);
90297Seric 
91297Seric 	/*
923149Seric 	**  Apply rewriting rules.
937889Seric 	**	Ruleset 0 does basic parsing.  It must resolve.
94297Seric 	*/
95297Seric 
968181Seric 	rewrite(pvp, 3);
974070Seric 	rewrite(pvp, 0);
98297Seric 
993149Seric 	/*
1003149Seric 	**  See if we resolved to a real mailer.
1013149Seric 	*/
102297Seric 
1033149Seric 	if (pvp[0][0] != CANONNET)
1043149Seric 	{
1053149Seric 		setstat(EX_USAGE);
1063149Seric 		usrerr("cannot resolve name");
1073149Seric 		return (NULL);
108297Seric 	}
109297Seric 
110297Seric 	/*
1113149Seric 	**  Build canonical address from pvp.
112297Seric 	*/
113297Seric 
1143149Seric 	a = buildaddr(pvp, a);
1154279Seric 	if (a == NULL)
1164279Seric 		return (NULL);
1174598Seric 	m = a->q_mailer;
118297Seric 
119297Seric 	/*
1203149Seric 	**  Make local copies of the host & user and then
1213149Seric 	**  transport them out.
122297Seric 	*/
123297Seric 
124297Seric 	if (copyf > 0)
1258078Seric 	{
1268078Seric 		extern char *DelimChar;
1278078Seric 		char savec = *DelimChar;
1288078Seric 
1298078Seric 		*DelimChar = '\0';
1302973Seric 		a->q_paddr = newstr(addr);
1318078Seric 		*DelimChar = savec;
1328078Seric 	}
133297Seric 	else
134297Seric 		a->q_paddr = addr;
13524944Seric 
13624944Seric 	if (a->q_user == NULL)
13724944Seric 		a->q_user = "";
13824944Seric 	if (a->q_host == NULL)
13924944Seric 		a->q_host = "";
14024944Seric 
1413149Seric 	if (copyf >= 0)
142297Seric 	{
14324944Seric 		a->q_host = newstr(a->q_host);
1443149Seric 		if (a->q_user != a->q_paddr)
1453149Seric 			a->q_user = newstr(a->q_user);
146297Seric 	}
147297Seric 
148297Seric 	/*
14916202Seric 	**  Convert host name to lower case if requested.
15016202Seric 	**	User name will be done later.
15116202Seric 	*/
15216202Seric 
15316202Seric 	if (!bitnset(M_HST_UPPER, m->m_flags))
15416202Seric 		makelower(a->q_host);
15516202Seric 
15616202Seric 	/*
157297Seric 	**  Compute return value.
158297Seric 	*/
159297Seric 
160297Seric # ifdef DEBUG
1617675Seric 	if (tTd(20, 1))
1624443Seric 	{
1639888Seric 		printf("parseaddr-->");
1644443Seric 		printaddr(a, FALSE);
1654443Seric 	}
166297Seric # endif DEBUG
167297Seric 
168297Seric 	return (a);
169297Seric }
170297Seric /*
17116162Seric **  LOWERADDR -- map UPPER->lower case on addresses as requested.
17216162Seric **
17316162Seric **	Parameters:
17416162Seric **		a -- address to be mapped.
17516162Seric **
17616162Seric **	Returns:
17716162Seric **		none.
17816162Seric **
17916162Seric **	Side Effects:
18016162Seric **		none.
18116162Seric */
18216162Seric 
18316162Seric loweraddr(a)
18416162Seric 	register ADDRESS *a;
18516162Seric {
18616162Seric 	register MAILER *m = a->q_mailer;
18716162Seric 
18816162Seric 	if (!bitnset(M_USR_UPPER, m->m_flags))
18916162Seric 		makelower(a->q_user);
19016162Seric }
19116162Seric /*
192297Seric **  PRESCAN -- Prescan name and make it canonical
193297Seric **
1949374Seric **	Scans a name and turns it into a set of tokens.  This process
1959374Seric **	deletes blanks and comments (in parentheses).
196297Seric **
197297Seric **	This routine knows about quoted strings and angle brackets.
198297Seric **
199297Seric **	There are certain subtleties to this routine.  The one that
200297Seric **	comes to mind now is that backslashes on the ends of names
201297Seric **	are silently stripped off; this is intentional.  The problem
202297Seric **	is that some versions of sndmsg (like at LBL) set the kill
203297Seric **	character to something other than @ when reading addresses;
204297Seric **	so people type "csvax.eric\@berkeley" -- which screws up the
205297Seric **	berknet mailer.
206297Seric **
207297Seric **	Parameters:
208297Seric **		addr -- the name to chomp.
209297Seric **		delim -- the delimiter for the address, normally
210297Seric **			'\0' or ','; \0 is accepted in any case.
21115284Seric **			If '\t' then we are reading the .cf file.
21216914Seric **		pvpbuf -- place to put the saved text -- note that
21316914Seric **			the pointers are static.
214297Seric **
215297Seric **	Returns:
2163149Seric **		A pointer to a vector of tokens.
217297Seric **		NULL on error.
218297Seric **
219297Seric **	Side Effects:
22025279Seric **		sets DelimChar to point to the character matching 'delim'.
221297Seric */
222297Seric 
2238078Seric /* states and character types */
2248078Seric # define OPR		0	/* operator */
2258078Seric # define ATM		1	/* atom */
2268078Seric # define QST		2	/* in quoted string */
2278078Seric # define SPC		3	/* chewing up spaces */
2288078Seric # define ONE		4	/* pick up one character */
2293149Seric 
2308078Seric # define NSTATES	5	/* number of states */
2318078Seric # define TYPE		017	/* mask to select state type */
2328078Seric 
2338078Seric /* meta bits for table */
2348078Seric # define M		020	/* meta character; don't pass through */
2358078Seric # define B		040	/* cause a break */
2368078Seric # define MB		M|B	/* meta-break */
2378078Seric 
2388078Seric static short StateTab[NSTATES][NSTATES] =
2398078Seric {
2408087Seric    /*	oldst	chtype>	OPR	ATM	QST	SPC	ONE	*/
2419051Seric 	/*OPR*/		OPR|B,	ATM|B,	QST|B,	SPC|MB,	ONE|B,
2429051Seric 	/*ATM*/		OPR|B,	ATM,	QST|B,	SPC|MB,	ONE|B,
2439051Seric 	/*QST*/		QST,	QST,	OPR,	QST,	QST,
2448078Seric 	/*SPC*/		OPR,	ATM,	QST,	SPC|M,	ONE,
2458078Seric 	/*ONE*/		OPR,	OPR,	OPR,	OPR,	OPR,
2468078Seric };
2478078Seric 
2488078Seric # define NOCHAR		-1	/* signal nothing in lookahead token */
2498078Seric 
2508078Seric char	*DelimChar;		/* set to point to the delimiter */
2518078Seric 
2523149Seric char **
25316914Seric prescan(addr, delim, pvpbuf)
254297Seric 	char *addr;
255297Seric 	char delim;
25616914Seric 	char pvpbuf[];
257297Seric {
258297Seric 	register char *p;
2598078Seric 	register char *q;
2609346Seric 	register int c;
2613149Seric 	char **avp;
262297Seric 	bool bslashmode;
263297Seric 	int cmntcnt;
2648423Seric 	int anglecnt;
2653149Seric 	char *tok;
2668078Seric 	int state;
2678078Seric 	int newstate;
2688078Seric 	static char *av[MAXATOM+1];
26915253Seric 	extern int errno;
270297Seric 
27115253Seric 	/* make sure error messages don't have garbage on them */
27215253Seric 	errno = 0;
27315253Seric 
27416914Seric 	q = pvpbuf;
2753149Seric 	bslashmode = FALSE;
2767800Seric 	cmntcnt = 0;
2778423Seric 	anglecnt = 0;
2783149Seric 	avp = av;
2798078Seric 	state = OPR;
2808078Seric 	c = NOCHAR;
2818078Seric 	p = addr;
2828078Seric # ifdef DEBUG
2838078Seric 	if (tTd(22, 45))
284297Seric 	{
2858078Seric 		printf("prescan: ");
2868078Seric 		xputs(p);
28723109Seric 		(void) putchar('\n');
2888078Seric 	}
2898078Seric # endif DEBUG
2908078Seric 
2918078Seric 	do
2928078Seric 	{
2933149Seric 		/* read a token */
2943149Seric 		tok = q;
2958078Seric 		for (;;)
296297Seric 		{
2978078Seric 			/* store away any old lookahead character */
2988078Seric 			if (c != NOCHAR)
2998078Seric 			{
30015284Seric 				/* see if there is room */
30116914Seric 				if (q >= &pvpbuf[PSBUFSIZE - 5])
3028078Seric 				{
3038078Seric 					usrerr("Address too long");
3048078Seric 					DelimChar = p;
3058078Seric 					return (NULL);
3068078Seric 				}
30715284Seric 
30815284Seric 				/* squirrel it away */
3098078Seric 				*q++ = c;
3108078Seric 			}
3118078Seric 
3128078Seric 			/* read a new input character */
3138078Seric 			c = *p++;
3148078Seric 			if (c == '\0')
3158078Seric 				break;
31615284Seric 			c &= ~0200;
31715284Seric 
3188078Seric # ifdef DEBUG
3198078Seric 			if (tTd(22, 101))
3208078Seric 				printf("c=%c, s=%d; ", c, state);
3218078Seric # endif DEBUG
3228078Seric 
3233149Seric 			/* chew up special characters */
3243149Seric 			*q = '\0';
3253149Seric 			if (bslashmode)
3263149Seric 			{
32724944Seric 				/* kludge \! for naive users */
32824944Seric 				if (c != '!')
32924944Seric 					c |= 0200;
3303149Seric 				bslashmode = FALSE;
3313149Seric 			}
3323149Seric 			else if (c == '\\')
3333149Seric 			{
3343149Seric 				bslashmode = TRUE;
3358078Seric 				c = NOCHAR;
3363149Seric 			}
3378514Seric 			else if (state == QST)
3388514Seric 			{
3398514Seric 				/* do nothing, just avoid next clauses */
3408514Seric 			}
3418078Seric 			else if (c == '(')
3424100Seric 			{
3438078Seric 				cmntcnt++;
3448078Seric 				c = NOCHAR;
3454100Seric 			}
3468078Seric 			else if (c == ')')
3473149Seric 			{
3488078Seric 				if (cmntcnt <= 0)
3493149Seric 				{
3508078Seric 					usrerr("Unbalanced ')'");
3518078Seric 					DelimChar = p;
3528078Seric 					return (NULL);
3533149Seric 				}
3548078Seric 				else
3558078Seric 					cmntcnt--;
3568078Seric 			}
3578078Seric 			else if (cmntcnt > 0)
3588078Seric 				c = NOCHAR;
3598423Seric 			else if (c == '<')
3608423Seric 				anglecnt++;
3618423Seric 			else if (c == '>')
3628423Seric 			{
3638423Seric 				if (anglecnt <= 0)
3648423Seric 				{
3658423Seric 					usrerr("Unbalanced '>'");
3668423Seric 					DelimChar = p;
3678423Seric 					return (NULL);
3688423Seric 				}
3698423Seric 				anglecnt--;
3708423Seric 			}
37111423Seric 			else if (delim == ' ' && isspace(c))
37211423Seric 				c = ' ';
3733149Seric 
3748078Seric 			if (c == NOCHAR)
3758078Seric 				continue;
3763149Seric 
3778078Seric 			/* see if this is end of input */
37811405Seric 			if (c == delim && anglecnt <= 0 && state != QST)
3793149Seric 				break;
3803149Seric 
3818078Seric 			newstate = StateTab[state][toktype(c)];
3828078Seric # ifdef DEBUG
3838078Seric 			if (tTd(22, 101))
3848078Seric 				printf("ns=%02o\n", newstate);
3858078Seric # endif DEBUG
3868078Seric 			state = newstate & TYPE;
3878078Seric 			if (bitset(M, newstate))
3888078Seric 				c = NOCHAR;
3898078Seric 			if (bitset(B, newstate))
3904228Seric 				break;
391297Seric 		}
3923149Seric 
3933149Seric 		/* new token */
3948078Seric 		if (tok != q)
3951378Seric 		{
3968078Seric 			*q++ = '\0';
3978078Seric # ifdef DEBUG
3988078Seric 			if (tTd(22, 36))
399297Seric 			{
4008078Seric 				printf("tok=");
4018078Seric 				xputs(tok);
40223109Seric 				(void) putchar('\n');
403297Seric 			}
4048078Seric # endif DEBUG
4058078Seric 			if (avp >= &av[MAXATOM])
406297Seric 			{
4078078Seric 				syserr("prescan: too many tokens");
4088078Seric 				DelimChar = p;
4098078Seric 				return (NULL);
410297Seric 			}
4118078Seric 			*avp++ = tok;
412297Seric 		}
4138423Seric 	} while (c != '\0' && (c != delim || anglecnt > 0));
4143149Seric 	*avp = NULL;
4158078Seric 	DelimChar = --p;
4163149Seric 	if (cmntcnt > 0)
4173149Seric 		usrerr("Unbalanced '('");
4188423Seric 	else if (anglecnt > 0)
4198423Seric 		usrerr("Unbalanced '<'");
4208078Seric 	else if (state == QST)
4213149Seric 		usrerr("Unbalanced '\"'");
4223149Seric 	else if (av[0] != NULL)
4233149Seric 		return (av);
4243149Seric 	return (NULL);
4253149Seric }
4263149Seric /*
4273149Seric **  TOKTYPE -- return token type
4283149Seric **
4293149Seric **	Parameters:
4303149Seric **		c -- the character in question.
4313149Seric **
4323149Seric **	Returns:
4333149Seric **		Its type.
4343149Seric **
4353149Seric **	Side Effects:
4363149Seric **		none.
4373149Seric */
438297Seric 
4393149Seric toktype(c)
4403149Seric 	register char c;
4413149Seric {
4423380Seric 	static char buf[50];
4433382Seric 	static bool firstime = TRUE;
4443380Seric 
4453382Seric 	if (firstime)
4463380Seric 	{
4473382Seric 		firstime = FALSE;
44816155Seric 		expand("\001o", buf, &buf[sizeof buf - 1], CurEnv);
4497005Seric 		(void) strcat(buf, DELIMCHARS);
4503380Seric 	}
4519585Seric 	if (c == MATCHCLASS || c == MATCHREPL || c == MATCHNCLASS)
4528078Seric 		return (ONE);
4538078Seric 	if (c == '"')
4548078Seric 		return (QST);
4554100Seric 	if (!isascii(c))
4568078Seric 		return (ATM);
4578078Seric 	if (isspace(c) || c == ')')
4588078Seric 		return (SPC);
4593380Seric 	if (iscntrl(c) || index(buf, c) != NULL)
4608078Seric 		return (OPR);
4618078Seric 	return (ATM);
4623149Seric }
4633149Seric /*
4643149Seric **  REWRITE -- apply rewrite rules to token vector.
4653149Seric **
4664476Seric **	This routine is an ordered production system.  Each rewrite
4674476Seric **	rule has a LHS (called the pattern) and a RHS (called the
4684476Seric **	rewrite); 'rwr' points the the current rewrite rule.
4694476Seric **
4704476Seric **	For each rewrite rule, 'avp' points the address vector we
4714476Seric **	are trying to match against, and 'pvp' points to the pattern.
4728058Seric **	If pvp points to a special match value (MATCHZANY, MATCHANY,
4739585Seric **	MATCHONE, MATCHCLASS, MATCHNCLASS) then the address in avp
4749585Seric **	matched is saved away in the match vector (pointed to by 'mvp').
4754476Seric **
4764476Seric **	When a match between avp & pvp does not match, we try to
4779585Seric **	back out.  If we back up over MATCHONE, MATCHCLASS, or MATCHNCLASS
4784476Seric **	we must also back out the match in mvp.  If we reach a
4798058Seric **	MATCHANY or MATCHZANY we just extend the match and start
4808058Seric **	over again.
4814476Seric **
4824476Seric **	When we finally match, we rewrite the address vector
4834476Seric **	and try over again.
4844476Seric **
4853149Seric **	Parameters:
4863149Seric **		pvp -- pointer to token vector.
4873149Seric **
4883149Seric **	Returns:
4893149Seric **		none.
4903149Seric **
4913149Seric **	Side Effects:
4923149Seric **		pvp is modified.
4933149Seric */
4942091Seric 
4953149Seric struct match
4963149Seric {
4974468Seric 	char	**first;	/* first token matched */
4984468Seric 	char	**last;		/* last token matched */
4993149Seric };
5003149Seric 
5014468Seric # define MAXMATCH	9	/* max params per rewrite */
5023149Seric 
5033149Seric 
5044070Seric rewrite(pvp, ruleset)
5053149Seric 	char **pvp;
5064070Seric 	int ruleset;
5073149Seric {
5083149Seric 	register char *ap;		/* address pointer */
5093149Seric 	register char *rp;		/* rewrite pointer */
5103149Seric 	register char **avp;		/* address vector pointer */
5113149Seric 	register char **rvp;		/* rewrite vector pointer */
5128058Seric 	register struct match *mlp;	/* cur ptr into mlist */
5138058Seric 	register struct rewrite *rwr;	/* pointer to current rewrite rule */
5144468Seric 	struct match mlist[MAXMATCH];	/* stores match on LHS */
5153149Seric 	char *npvp[MAXATOM+1];		/* temporary space for rebuild */
5163149Seric 
5179279Seric 	if (OpMode == MD_TEST || tTd(21, 2))
5183149Seric 	{
5198959Seric 		printf("rewrite: ruleset %2d   input:", ruleset);
5203149Seric 		printav(pvp);
5213149Seric 	}
5228423Seric 	if (pvp == NULL)
5238423Seric 		return;
5243149Seric 
5253149Seric 	/*
5263149Seric 	**  Run through the list of rewrite rules, applying
5273149Seric 	**	any that match.
5283149Seric 	*/
5293149Seric 
5304070Seric 	for (rwr = RewriteRules[ruleset]; rwr != NULL; )
5313149Seric 	{
5324100Seric # ifdef DEBUG
5337675Seric 		if (tTd(21, 12))
534297Seric 		{
5358069Seric 			printf("-----trying rule:");
5363149Seric 			printav(rwr->r_lhs);
5373149Seric 		}
5384100Seric # endif DEBUG
5393149Seric 
5403149Seric 		/* try to match on this rule */
5414468Seric 		mlp = mlist;
5428058Seric 		rvp = rwr->r_lhs;
5438058Seric 		avp = pvp;
5448058Seric 		while ((ap = *avp) != NULL || *rvp != NULL)
5453149Seric 		{
5463149Seric 			rp = *rvp;
5478058Seric # ifdef DEBUG
5488058Seric 			if (tTd(21, 35))
5498058Seric 			{
5508069Seric 				printf("ap=");
5518058Seric 				xputs(ap);
5528069Seric 				printf(", rp=");
5538058Seric 				xputs(rp);
5548069Seric 				printf("\n");
5558058Seric 			}
5568058Seric # endif DEBUG
5573149Seric 			if (rp == NULL)
558297Seric 			{
5593149Seric 				/* end-of-pattern before end-of-address */
5608058Seric 				goto backup;
561297Seric 			}
5628058Seric 			if (ap == NULL && *rp != MATCHZANY)
5638058Seric 			{
5648058Seric 				/* end-of-input */
5658058Seric 				break;
5668058Seric 			}
5673149Seric 
5683149Seric 			switch (*rp)
5693149Seric 			{
5704060Seric 				register STAB *s;
5714060Seric 
5724060Seric 			  case MATCHCLASS:
5739585Seric 			  case MATCHNCLASS:
5749585Seric 				/* match any token in (not in) a class */
5754100Seric 				s = stab(ap, ST_CLASS, ST_FIND);
57610690Seric 				if (s == NULL || !bitnset(rp[1], s->s_class))
5779585Seric 				{
5789585Seric 					if (*rp == MATCHCLASS)
5799585Seric 						goto backup;
5809585Seric 				}
5819585Seric 				else if (*rp == MATCHNCLASS)
5828058Seric 					goto backup;
5834468Seric 
5844476Seric 				/* explicit fall-through */
5854476Seric 
5864476Seric 			  case MATCHONE:
5874476Seric 			  case MATCHANY:
5884476Seric 				/* match exactly one token */
5898058Seric 				mlp->first = avp;
5908058Seric 				mlp->last = avp++;
5914468Seric 				mlp++;
5924060Seric 				break;
5934060Seric 
5948058Seric 			  case MATCHZANY:
5958058Seric 				/* match zero or more tokens */
5968058Seric 				mlp->first = avp;
5978058Seric 				mlp->last = avp - 1;
5988058Seric 				mlp++;
5998058Seric 				break;
6008058Seric 
6013149Seric 			  default:
6023149Seric 				/* must have exact match */
60333725Sbostic 				if (strcasecmp(rp, ap))
6048058Seric 					goto backup;
6054468Seric 				avp++;
6063149Seric 				break;
6073149Seric 			}
6083149Seric 
6093149Seric 			/* successful match on this token */
6103149Seric 			rvp++;
6113149Seric 			continue;
6123149Seric 
6138058Seric 		  backup:
6143149Seric 			/* match failed -- back up */
6153149Seric 			while (--rvp >= rwr->r_lhs)
6163149Seric 			{
6173149Seric 				rp = *rvp;
6188058Seric 				if (*rp == MATCHANY || *rp == MATCHZANY)
6194468Seric 				{
6204476Seric 					/* extend binding and continue */
6218058Seric 					avp = ++mlp[-1].last;
6228058Seric 					avp++;
6234476Seric 					rvp++;
6243149Seric 					break;
6254468Seric 				}
6264476Seric 				avp--;
6279585Seric 				if (*rp == MATCHONE || *rp == MATCHCLASS ||
6289585Seric 				    *rp == MATCHNCLASS)
6293149Seric 				{
6304468Seric 					/* back out binding */
6314468Seric 					mlp--;
6323149Seric 				}
6333149Seric 			}
6343149Seric 
6353149Seric 			if (rvp < rwr->r_lhs)
6363149Seric 			{
6373149Seric 				/* total failure to match */
6383149Seric 				break;
6393149Seric 			}
640297Seric 		}
6413149Seric 
6423149Seric 		/*
6433149Seric 		**  See if we successfully matched
6443149Seric 		*/
6453149Seric 
6469374Seric 		if (rvp < rwr->r_lhs || *rvp != NULL)
6473149Seric 		{
6484100Seric # ifdef DEBUG
6499374Seric 			if (tTd(21, 10))
6509374Seric 				printf("----- rule fails\n");
6514100Seric # endif DEBUG
6529374Seric 			rwr = rwr->r_next;
6539374Seric 			continue;
6549374Seric 		}
6553149Seric 
6569374Seric 		rvp = rwr->r_rhs;
6579374Seric # ifdef DEBUG
6589374Seric 		if (tTd(21, 12))
6599374Seric 		{
6609374Seric 			printf("-----rule matches:");
6619374Seric 			printav(rvp);
6629374Seric 		}
6639374Seric # endif DEBUG
6649374Seric 
6659374Seric 		rp = *rvp;
6669374Seric 		if (*rp == CANONUSER)
6679374Seric 		{
6689374Seric 			rvp++;
6699374Seric 			rwr = rwr->r_next;
6709374Seric 		}
6719374Seric 		else if (*rp == CANONHOST)
6729374Seric 		{
6739374Seric 			rvp++;
6749374Seric 			rwr = NULL;
6759374Seric 		}
6769374Seric 		else if (*rp == CANONNET)
6779374Seric 			rwr = NULL;
6789374Seric 
6799374Seric 		/* substitute */
6809374Seric 		for (avp = npvp; *rvp != NULL; rvp++)
6819374Seric 		{
6829374Seric 			register struct match *m;
6839374Seric 			register char **pp;
6849374Seric 
6858058Seric 			rp = *rvp;
68616914Seric 			if (*rp == MATCHREPL)
6878058Seric 			{
68816914Seric 				/* substitute from LHS */
68916914Seric 				m = &mlist[rp[1] - '1'];
69016914Seric 				if (m >= mlp)
6919374Seric 				{
69216914Seric 					syserr("rewrite: ruleset %d: replacement out of bounds", ruleset);
6939374Seric 					return;
6949374Seric 				}
6959374Seric # ifdef DEBUG
69616914Seric 				if (tTd(21, 15))
69716914Seric 				{
69816914Seric 					printf("$%c:", rp[1]);
69916914Seric 					pp = m->first;
70016914Seric 					while (pp <= m->last)
70116914Seric 					{
70216914Seric 						printf(" %x=\"", *pp);
70316914Seric 						(void) fflush(stdout);
70416914Seric 						printf("%s\"", *pp++);
70516914Seric 					}
70616914Seric 					printf("\n");
70716914Seric 				}
70816914Seric # endif DEBUG
7099374Seric 				pp = m->first;
7109374Seric 				while (pp <= m->last)
7113149Seric 				{
71216914Seric 					if (avp >= &npvp[MAXATOM])
71316914Seric 					{
71416914Seric 						syserr("rewrite: expansion too long");
71516914Seric 						return;
71616914Seric 					}
71716914Seric 					*avp++ = *pp++;
7183149Seric 				}
7193149Seric 			}
72016914Seric 			else
7218226Seric 			{
72216914Seric 				/* vanilla replacement */
7239374Seric 				if (avp >= &npvp[MAXATOM])
72416889Seric 				{
72516914Seric 	toolong:
72616889Seric 					syserr("rewrite: expansion too long");
72716889Seric 					return;
72816889Seric 				}
72916914Seric 				*avp++ = rp;
7308226Seric 			}
7319374Seric 		}
7329374Seric 		*avp++ = NULL;
73316914Seric 
73416914Seric 		/*
73516914Seric 		**  Check for any hostname lookups.
73616914Seric 		*/
73716914Seric 
73816914Seric 		for (rvp = npvp; *rvp != NULL; rvp++)
73916914Seric 		{
74016914Seric 			char **hbrvp;
74116914Seric 			char **xpvp;
74216914Seric 			int trsize;
74317473Seric 			char *olddelimchar;
74416920Seric 			char buf[MAXNAME + 1];
74516914Seric 			char *pvpb1[MAXATOM + 1];
74617174Seric 			char pvpbuf[PSBUFSIZE];
74717473Seric 			extern char *DelimChar;
74816914Seric 
74916914Seric 			if (**rvp != HOSTBEGIN)
75016914Seric 				continue;
75116914Seric 
75216914Seric 			/*
75316914Seric 			**  Got a hostname lookup.
75416914Seric 			**
75516914Seric 			**	This could be optimized fairly easily.
75616914Seric 			*/
75716914Seric 
75816914Seric 			hbrvp = rvp;
75916914Seric 
76016914Seric 			/* extract the match part */
76116914Seric 			while (*++rvp != NULL && **rvp != HOSTEND)
76216914Seric 				continue;
76316914Seric 			if (*rvp != NULL)
76416914Seric 				*rvp++ = NULL;
76516914Seric 
76616914Seric 			/* save the remainder of the input string */
76716914Seric 			trsize = (int) (avp - rvp + 1) * sizeof *rvp;
76816914Seric 			bcopy((char *) rvp, (char *) pvpb1, trsize);
76916914Seric 
77016914Seric 			/* look it up */
77116914Seric 			cataddr(++hbrvp, buf, sizeof buf);
77216914Seric 			maphostname(buf, sizeof buf);
77316914Seric 
77416914Seric 			/* scan the new host name */
77517473Seric 			olddelimchar = DelimChar;
77616914Seric 			xpvp = prescan(buf, '\0', pvpbuf);
77717473Seric 			DelimChar = olddelimchar;
77816914Seric 			if (xpvp == NULL)
77916914Seric 			{
78016914Seric 				syserr("rewrite: cannot prescan canonical hostname: %s", buf);
78122976Smiriam 				return;
78216914Seric 			}
78316914Seric 
78416914Seric 			/* append it to the token list */
78517174Seric 			for (avp = --hbrvp; *xpvp != NULL; xpvp++)
78617174Seric 			{
78717174Seric 				*avp++ = newstr(*xpvp);
78816920Seric 				if (avp >= &npvp[MAXATOM])
78916914Seric 					goto toolong;
79017174Seric 			}
79116914Seric 
79216914Seric 			/* restore the old trailing information */
79317177Seric 			for (xpvp = pvpb1; (*avp++ = *xpvp++) != NULL; )
79416920Seric 				if (avp >= &npvp[MAXATOM])
79516914Seric 					goto toolong;
79617174Seric 
79717174Seric 			break;
79816914Seric 		}
79916914Seric 
80016914Seric 		/*
80116914Seric 		**  Check for subroutine calls.
80216914Seric 		*/
80316914Seric 
80424944Seric 		if (*npvp != NULL && **npvp == CALLSUBR)
8059374Seric 		{
80616889Seric 			bcopy((char *) &npvp[2], (char *) pvp,
80716900Seric 				(int) (avp - npvp - 2) * sizeof *avp);
80816889Seric # ifdef DEBUG
80916889Seric 			if (tTd(21, 3))
81016889Seric 				printf("-----callsubr %s\n", npvp[1]);
81116889Seric # endif DEBUG
81216889Seric 			rewrite(pvp, atoi(npvp[1]));
8133149Seric 		}
8143149Seric 		else
8153149Seric 		{
81617348Seric 			bcopy((char *) npvp, (char *) pvp,
81716900Seric 				(int) (avp - npvp) * sizeof *avp);
8189374Seric 		}
8194100Seric # ifdef DEBUG
8209374Seric 		if (tTd(21, 4))
8219374Seric 		{
8229374Seric 			printf("rewritten as:");
8239374Seric 			printav(pvp);
8249374Seric 		}
8254100Seric # endif DEBUG
826297Seric 	}
8278069Seric 
8289279Seric 	if (OpMode == MD_TEST || tTd(21, 2))
8298069Seric 	{
8308959Seric 		printf("rewrite: ruleset %2d returns:", ruleset);
8318069Seric 		printav(pvp);
8328069Seric 	}
8333149Seric }
8343149Seric /*
8353149Seric **  BUILDADDR -- build address from token vector.
8363149Seric **
8373149Seric **	Parameters:
8383149Seric **		tv -- token vector.
8393149Seric **		a -- pointer to address descriptor to fill.
8403149Seric **			If NULL, one will be allocated.
8413149Seric **
8423149Seric **	Returns:
8434279Seric **		NULL if there was an error.
8444279Seric **		'a' otherwise.
8453149Seric **
8463149Seric **	Side Effects:
8473149Seric **		fills in 'a'
8483149Seric */
8493149Seric 
8503149Seric ADDRESS *
8513149Seric buildaddr(tv, a)
8523149Seric 	register char **tv;
8533149Seric 	register ADDRESS *a;
8543149Seric {
8553149Seric 	static char buf[MAXNAME];
8563149Seric 	struct mailer **mp;
8573149Seric 	register struct mailer *m;
8583149Seric 
8593149Seric 	if (a == NULL)
8603149Seric 		a = (ADDRESS *) xalloc(sizeof *a);
86116889Seric 	bzero((char *) a, sizeof *a);
8623149Seric 
8633149Seric 	/* figure out what net/mailer to use */
8643149Seric 	if (**tv != CANONNET)
8654279Seric 	{
8663149Seric 		syserr("buildaddr: no net");
8674279Seric 		return (NULL);
8684279Seric 	}
8693149Seric 	tv++;
87033725Sbostic 	if (!strcasecmp(*tv, "error"))
8714279Seric 	{
87210183Seric 		if (**++tv == CANONHOST)
87310183Seric 		{
87410183Seric 			setstat(atoi(*++tv));
87510183Seric 			tv++;
87610183Seric 		}
87710183Seric 		if (**tv != CANONUSER)
8784279Seric 			syserr("buildaddr: error: no user");
8794279Seric 		buf[0] = '\0';
8804279Seric 		while (*++tv != NULL)
8814279Seric 		{
8824279Seric 			if (buf[0] != '\0')
8837005Seric 				(void) strcat(buf, " ");
8847005Seric 			(void) strcat(buf, *tv);
8854279Seric 		}
8864279Seric 		usrerr(buf);
8874279Seric 		return (NULL);
8884279Seric 	}
8894598Seric 	for (mp = Mailer; (m = *mp++) != NULL; )
8903149Seric 	{
89133725Sbostic 		if (!strcasecmp(m->m_name, *tv))
8923149Seric 			break;
8933149Seric 	}
8943149Seric 	if (m == NULL)
8954279Seric 	{
89624944Seric 		syserr("buildaddr: unknown mailer %s", *tv);
8974279Seric 		return (NULL);
8984279Seric 	}
8994598Seric 	a->q_mailer = m;
9003149Seric 
9013149Seric 	/* figure out what host (if any) */
9023149Seric 	tv++;
90310690Seric 	if (!bitnset(M_LOCAL, m->m_flags))
9043149Seric 	{
9055704Seric 		if (**tv++ != CANONHOST)
9064279Seric 		{
9073149Seric 			syserr("buildaddr: no host");
9084279Seric 			return (NULL);
9094279Seric 		}
9105704Seric 		buf[0] = '\0';
9115704Seric 		while (*tv != NULL && **tv != CANONUSER)
9127005Seric 			(void) strcat(buf, *tv++);
9135704Seric 		a->q_host = newstr(buf);
9143149Seric 	}
9153149Seric 	else
9163149Seric 		a->q_host = NULL;
9173149Seric 
9183149Seric 	/* figure out the user */
9193149Seric 	if (**tv != CANONUSER)
9204279Seric 	{
9213149Seric 		syserr("buildaddr: no user");
9224279Seric 		return (NULL);
9234279Seric 	}
92419040Seric 
92519040Seric 	/* rewrite according recipient mailer rewriting rules */
92619040Seric 	rewrite(++tv, 2);
92719040Seric 	if (m->m_r_rwset > 0)
92819040Seric 		rewrite(tv, m->m_r_rwset);
92919040Seric 	rewrite(tv, 4);
93019040Seric 
93119040Seric 	/* save the result for the command line/RCPT argument */
93211278Seric 	cataddr(tv, buf, sizeof buf);
9333149Seric 	a->q_user = buf;
9343149Seric 
9353149Seric 	return (a);
9363149Seric }
9373188Seric /*
9384228Seric **  CATADDR -- concatenate pieces of addresses (putting in <LWSP> subs)
9394228Seric **
9404228Seric **	Parameters:
9414228Seric **		pvp -- parameter vector to rebuild.
9424228Seric **		buf -- buffer to build the string into.
9434228Seric **		sz -- size of buf.
9444228Seric **
9454228Seric **	Returns:
9464228Seric **		none.
9474228Seric **
9484228Seric **	Side Effects:
9494228Seric **		Destroys buf.
9504228Seric */
9514228Seric 
9524228Seric cataddr(pvp, buf, sz)
9534228Seric 	char **pvp;
9544228Seric 	char *buf;
9554228Seric 	register int sz;
9564228Seric {
9574228Seric 	bool oatomtok = FALSE;
9584228Seric 	bool natomtok = FALSE;
9594228Seric 	register int i;
9604228Seric 	register char *p;
9614228Seric 
9628423Seric 	if (pvp == NULL)
9638423Seric 	{
96423109Seric 		(void) strcpy(buf, "");
9658423Seric 		return;
9668423Seric 	}
9674228Seric 	p = buf;
96811156Seric 	sz -= 2;
9694228Seric 	while (*pvp != NULL && (i = strlen(*pvp)) < sz)
9704228Seric 	{
9718078Seric 		natomtok = (toktype(**pvp) == ATM);
9724228Seric 		if (oatomtok && natomtok)
9739042Seric 			*p++ = SpaceSub;
9744228Seric 		(void) strcpy(p, *pvp);
9754228Seric 		oatomtok = natomtok;
9764228Seric 		p += i;
97711156Seric 		sz -= i + 1;
9784228Seric 		pvp++;
9794228Seric 	}
9804228Seric 	*p = '\0';
9814228Seric }
9824228Seric /*
9833188Seric **  SAMEADDR -- Determine if two addresses are the same
9843188Seric **
9853188Seric **	This is not just a straight comparison -- if the mailer doesn't
9863188Seric **	care about the host we just ignore it, etc.
9873188Seric **
9883188Seric **	Parameters:
9893188Seric **		a, b -- pointers to the internal forms to compare.
9903188Seric **
9913188Seric **	Returns:
9923188Seric **		TRUE -- they represent the same mailbox.
9933188Seric **		FALSE -- they don't.
9943188Seric **
9953188Seric **	Side Effects:
9963188Seric **		none.
9973188Seric */
9983188Seric 
9993188Seric bool
10009374Seric sameaddr(a, b)
10013188Seric 	register ADDRESS *a;
10023188Seric 	register ADDRESS *b;
10033188Seric {
10043188Seric 	/* if they don't have the same mailer, forget it */
10053188Seric 	if (a->q_mailer != b->q_mailer)
10063188Seric 		return (FALSE);
10073188Seric 
10083188Seric 	/* if the user isn't the same, we can drop out */
10099374Seric 	if (strcmp(a->q_user, b->q_user) != 0)
10103188Seric 		return (FALSE);
10113188Seric 
10123188Seric 	/* if the mailer ignores hosts, we have succeeded! */
101310690Seric 	if (bitnset(M_LOCAL, a->q_mailer->m_flags))
10143188Seric 		return (TRUE);
10153188Seric 
10163188Seric 	/* otherwise compare hosts (but be careful for NULL ptrs) */
10173188Seric 	if (a->q_host == NULL || b->q_host == NULL)
10183188Seric 		return (FALSE);
10193188Seric 	if (strcmp(a->q_host, b->q_host) != 0)
10203188Seric 		return (FALSE);
10213188Seric 
10223188Seric 	return (TRUE);
10233188Seric }
10243234Seric /*
10253234Seric **  PRINTADDR -- print address (for debugging)
10263234Seric **
10273234Seric **	Parameters:
10283234Seric **		a -- the address to print
10293234Seric **		follow -- follow the q_next chain.
10303234Seric **
10313234Seric **	Returns:
10323234Seric **		none.
10333234Seric **
10343234Seric **	Side Effects:
10353234Seric **		none.
10363234Seric */
10373234Seric 
10384317Seric # ifdef DEBUG
10394317Seric 
10403234Seric printaddr(a, follow)
10413234Seric 	register ADDRESS *a;
10423234Seric 	bool follow;
10433234Seric {
10445001Seric 	bool first = TRUE;
10455001Seric 
10463234Seric 	while (a != NULL)
10473234Seric 	{
10485001Seric 		first = FALSE;
10494443Seric 		printf("%x=", a);
10504085Seric 		(void) fflush(stdout);
10513234Seric 		printf("%s: mailer %d (%s), host `%s', user `%s'\n", a->q_paddr,
10528181Seric 		       a->q_mailer->m_mno, a->q_mailer->m_name, a->q_host,
10538181Seric 		       a->q_user);
10548181Seric 		printf("\tnext=%x, flags=%o, alias %x\n", a->q_next, a->q_flags,
10558181Seric 		       a->q_alias);
10568181Seric 		printf("\thome=\"%s\", fullname=\"%s\"\n", a->q_home,
10578181Seric 		       a->q_fullname);
10584996Seric 
10593234Seric 		if (!follow)
10603234Seric 			return;
10614996Seric 		a = a->q_next;
10623234Seric 	}
10635001Seric 	if (first)
10644443Seric 		printf("[NULL]\n");
10653234Seric }
10664317Seric 
10674317Seric # endif DEBUG
10687682Seric /*
10697682Seric **  REMOTENAME -- return the name relative to the current mailer
10707682Seric **
10717682Seric **	Parameters:
10727682Seric **		name -- the name to translate.
10738069Seric **		m -- the mailer that we want to do rewriting relative
10748069Seric **			to.
10758069Seric **		senderaddress -- if set, uses the sender rewriting rules
10768069Seric **			rather than the recipient rewriting rules.
107710310Seric **		canonical -- if set, strip out any comment information,
107810310Seric **			etc.
10797682Seric **
10807682Seric **	Returns:
10817682Seric **		the text string representing this address relative to
10827682Seric **			the receiving mailer.
10837682Seric **
10847682Seric **	Side Effects:
10857682Seric **		none.
10867682Seric **
10877682Seric **	Warnings:
10887682Seric **		The text string returned is tucked away locally;
10897682Seric **			copy it if you intend to save it.
10907682Seric */
10917682Seric 
10927682Seric char *
109310310Seric remotename(name, m, senderaddress, canonical)
10947682Seric 	char *name;
10957682Seric 	struct mailer *m;
10968069Seric 	bool senderaddress;
109710310Seric 	bool canonical;
10987682Seric {
10998069Seric 	register char **pvp;
11008069Seric 	char *fancy;
11018069Seric 	extern char *macvalue();
11028181Seric 	char *oldg = macvalue('g', CurEnv);
11037682Seric 	static char buf[MAXNAME];
11047682Seric 	char lbuf[MAXNAME];
110516914Seric 	char pvpbuf[PSBUFSIZE];
11067682Seric 	extern char **prescan();
11077889Seric 	extern char *crackaddr();
11087682Seric 
11097755Seric # ifdef DEBUG
11107755Seric 	if (tTd(12, 1))
11117755Seric 		printf("remotename(%s)\n", name);
11127755Seric # endif DEBUG
11137755Seric 
111410177Seric 	/* don't do anything if we are tagging it as special */
111510177Seric 	if ((senderaddress ? m->m_s_rwset : m->m_r_rwset) < 0)
111610177Seric 		return (name);
111710177Seric 
11187682Seric 	/*
11198181Seric 	**  Do a heuristic crack of this name to extract any comment info.
11208181Seric 	**	This will leave the name as a comment and a $g macro.
11217889Seric 	*/
11227889Seric 
112310310Seric 	if (canonical)
112416155Seric 		fancy = "\001g";
112510310Seric 	else
112610310Seric 		fancy = crackaddr(name);
11277889Seric 
11288181Seric 	/*
11298181Seric 	**  Turn the name into canonical form.
11308181Seric 	**	Normally this will be RFC 822 style, i.e., "user@domain".
11318181Seric 	**	If this only resolves to "user", and the "C" flag is
11328181Seric 	**	specified in the sending mailer, then the sender's
11338181Seric 	**	domain will be appended.
11348181Seric 	*/
11358181Seric 
113616914Seric 	pvp = prescan(name, '\0', pvpbuf);
11377889Seric 	if (pvp == NULL)
11387889Seric 		return (name);
11398181Seric 	rewrite(pvp, 3);
11408181Seric 	if (CurEnv->e_fromdomain != NULL)
11418181Seric 	{
11428181Seric 		/* append from domain to this address */
11438181Seric 		register char **pxp = pvp;
11448181Seric 
11459594Seric 		/* see if there is an "@domain" in the current name */
11468181Seric 		while (*pxp != NULL && strcmp(*pxp, "@") != 0)
11478181Seric 			pxp++;
11488181Seric 		if (*pxp == NULL)
11498181Seric 		{
11509594Seric 			/* no.... append the "@domain" from the sender */
11518181Seric 			register char **qxq = CurEnv->e_fromdomain;
11528181Seric 
11539594Seric 			while ((*pxp++ = *qxq++) != NULL)
11549594Seric 				continue;
115511726Seric 			rewrite(pvp, 3);
11568181Seric 		}
11578181Seric 	}
11588181Seric 
11598181Seric 	/*
11608959Seric 	**  Do more specific rewriting.
11618181Seric 	**	Rewrite using ruleset 1 or 2 depending on whether this is
11628181Seric 	**		a sender address or not.
11638181Seric 	**	Then run it through any receiving-mailer-specific rulesets.
11648181Seric 	*/
11658181Seric 
11668069Seric 	if (senderaddress)
11677755Seric 	{
11687889Seric 		rewrite(pvp, 1);
11698069Seric 		if (m->m_s_rwset > 0)
11708069Seric 			rewrite(pvp, m->m_s_rwset);
11718069Seric 	}
11728069Seric 	else
11738069Seric 	{
11747889Seric 		rewrite(pvp, 2);
11758069Seric 		if (m->m_r_rwset > 0)
11768069Seric 			rewrite(pvp, m->m_r_rwset);
11777682Seric 	}
11787682Seric 
11798181Seric 	/*
11808959Seric 	**  Do any final sanitation the address may require.
11818959Seric 	**	This will normally be used to turn internal forms
11828959Seric 	**	(e.g., user@host.LOCAL) into external form.  This
11838959Seric 	**	may be used as a default to the above rules.
11848959Seric 	*/
11858959Seric 
11868959Seric 	rewrite(pvp, 4);
11878959Seric 
11888959Seric 	/*
11898181Seric 	**  Now restore the comment information we had at the beginning.
11908181Seric 	*/
11918181Seric 
11927682Seric 	cataddr(pvp, lbuf, sizeof lbuf);
11939374Seric 	define('g', lbuf, CurEnv);
11947889Seric 	expand(fancy, buf, &buf[sizeof buf - 1], CurEnv);
11959374Seric 	define('g', oldg, CurEnv);
11967682Seric 
11977682Seric # ifdef DEBUG
11987682Seric 	if (tTd(12, 1))
11997755Seric 		printf("remotename => `%s'\n", buf);
12007682Seric # endif DEBUG
12017682Seric 	return (buf);
12027682Seric }
1203