13312Seric # include "sendmail.h"
2297Seric 
3*15284Seric SCCSID(@(#)parseaddr.c	4.3		10/23/83);
4407Seric 
5297Seric /*
69888Seric **  PARSEADDR -- Parse an address
7297Seric **
8297Seric **	Parses an address and breaks it up into three parts: a
9297Seric **	net to transmit the message on, the host to transmit it
10297Seric **	to, and a user on that host.  These are loaded into an
112973Seric **	ADDRESS header with the values squirreled away if necessary.
12297Seric **	The "user" part may not be a real user; the process may
13297Seric **	just reoccur on that machine.  For example, on a machine
14297Seric **	with an arpanet connection, the address
15297Seric **		csvax.bill@berkeley
16297Seric **	will break up to a "user" of 'csvax.bill' and a host
17297Seric **	of 'berkeley' -- to be transmitted over the arpanet.
18297Seric **
19297Seric **	Parameters:
20297Seric **		addr -- the address to parse.
21297Seric **		a -- a pointer to the address descriptor buffer.
22297Seric **			If NULL, a header will be created.
23297Seric **		copyf -- determines what shall be copied:
24297Seric **			-1 -- don't copy anything.  The printname
25297Seric **				(q_paddr) is just addr, and the
26297Seric **				user & host are allocated internally
27297Seric **				to parse.
28297Seric **			0 -- copy out the parsed user & host, but
29297Seric **				don't copy the printname.
30297Seric **			+1 -- copy everything.
3111445Seric **		delim -- the character to terminate the address, passed
3211445Seric **			to prescan.
33297Seric **
34297Seric **	Returns:
35297Seric **		A pointer to the address descriptor header (`a' if
36297Seric **			`a' is non-NULL).
37297Seric **		NULL on error.
38297Seric **
39297Seric **	Side Effects:
40297Seric **		none
41297Seric */
42297Seric 
439374Seric /* following delimiters are inherent to the internal algorithms */
443380Seric # define DELIMCHARS	"$()<>,;\\\"\r\n"	/* word delimiters */
452091Seric 
462973Seric ADDRESS *
4711445Seric parseaddr(addr, a, copyf, delim)
48297Seric 	char *addr;
492973Seric 	register ADDRESS *a;
50297Seric 	int copyf;
5111445Seric 	char delim;
52297Seric {
533149Seric 	register char **pvp;
543149Seric 	register struct mailer *m;
553149Seric 	extern char **prescan();
563149Seric 	extern ADDRESS *buildaddr();
57297Seric 
58297Seric 	/*
59297Seric 	**  Initialize and prescan address.
60297Seric 	*/
61297Seric 
626903Seric 	CurEnv->e_to = addr;
633188Seric # ifdef DEBUG
647675Seric 	if (tTd(20, 1))
659888Seric 		printf("\n--parseaddr(%s)\n", addr);
663188Seric # endif DEBUG
673188Seric 
6811445Seric 	pvp = prescan(addr, delim);
693149Seric 	if (pvp == NULL)
70297Seric 		return (NULL);
71297Seric 
72297Seric 	/*
733149Seric 	**  Apply rewriting rules.
747889Seric 	**	Ruleset 0 does basic parsing.  It must resolve.
75297Seric 	*/
76297Seric 
778181Seric 	rewrite(pvp, 3);
784070Seric 	rewrite(pvp, 0);
79297Seric 
803149Seric 	/*
813149Seric 	**  See if we resolved to a real mailer.
823149Seric 	*/
83297Seric 
843149Seric 	if (pvp[0][0] != CANONNET)
853149Seric 	{
863149Seric 		setstat(EX_USAGE);
873149Seric 		usrerr("cannot resolve name");
883149Seric 		return (NULL);
89297Seric 	}
90297Seric 
91297Seric 	/*
923149Seric 	**  Build canonical address from pvp.
93297Seric 	*/
94297Seric 
953149Seric 	a = buildaddr(pvp, a);
964279Seric 	if (a == NULL)
974279Seric 		return (NULL);
984598Seric 	m = a->q_mailer;
99297Seric 
100297Seric 	/*
1013149Seric 	**  Make local copies of the host & user and then
1023149Seric 	**  transport them out.
103297Seric 	*/
104297Seric 
105297Seric 	if (copyf > 0)
1068078Seric 	{
1078078Seric 		extern char *DelimChar;
1088078Seric 		char savec = *DelimChar;
1098078Seric 
1108078Seric 		*DelimChar = '\0';
1112973Seric 		a->q_paddr = newstr(addr);
1128078Seric 		*DelimChar = savec;
1138078Seric 	}
114297Seric 	else
115297Seric 		a->q_paddr = addr;
1163149Seric 	if (copyf >= 0)
117297Seric 	{
1183149Seric 		if (a->q_host != NULL)
1193149Seric 			a->q_host = newstr(a->q_host);
120297Seric 		else
1213149Seric 			a->q_host = "";
1223149Seric 		if (a->q_user != a->q_paddr)
1233149Seric 			a->q_user = newstr(a->q_user);
124297Seric 	}
125297Seric 
126297Seric 	/*
127297Seric 	**  Do UPPER->lower case mapping unless inhibited.
128297Seric 	*/
129297Seric 
13010690Seric 	if (!bitnset(M_HST_UPPER, m->m_flags))
131297Seric 		makelower(a->q_host);
13210690Seric 	if (!bitnset(M_USR_UPPER, m->m_flags))
133297Seric 		makelower(a->q_user);
134297Seric 
135297Seric 	/*
136297Seric 	**  Compute return value.
137297Seric 	*/
138297Seric 
139297Seric # ifdef DEBUG
1407675Seric 	if (tTd(20, 1))
1414443Seric 	{
1429888Seric 		printf("parseaddr-->");
1434443Seric 		printaddr(a, FALSE);
1444443Seric 	}
145297Seric # endif DEBUG
146297Seric 
147297Seric 	return (a);
148297Seric }
149297Seric /*
150297Seric **  PRESCAN -- Prescan name and make it canonical
151297Seric **
1529374Seric **	Scans a name and turns it into a set of tokens.  This process
1539374Seric **	deletes blanks and comments (in parentheses).
154297Seric **
155297Seric **	This routine knows about quoted strings and angle brackets.
156297Seric **
157297Seric **	There are certain subtleties to this routine.  The one that
158297Seric **	comes to mind now is that backslashes on the ends of names
159297Seric **	are silently stripped off; this is intentional.  The problem
160297Seric **	is that some versions of sndmsg (like at LBL) set the kill
161297Seric **	character to something other than @ when reading addresses;
162297Seric **	so people type "csvax.eric\@berkeley" -- which screws up the
163297Seric **	berknet mailer.
164297Seric **
165297Seric **	Parameters:
166297Seric **		addr -- the name to chomp.
167297Seric **		delim -- the delimiter for the address, normally
168297Seric **			'\0' or ','; \0 is accepted in any case.
169*15284Seric **			If '\t' then we are reading the .cf file.
170297Seric **
171297Seric **	Returns:
1723149Seric **		A pointer to a vector of tokens.
173297Seric **		NULL on error.
174297Seric **
175297Seric **	Side Effects:
1763149Seric **		none.
177297Seric */
178297Seric 
1798078Seric /* states and character types */
1808078Seric # define OPR		0	/* operator */
1818078Seric # define ATM		1	/* atom */
1828078Seric # define QST		2	/* in quoted string */
1838078Seric # define SPC		3	/* chewing up spaces */
1848078Seric # define ONE		4	/* pick up one character */
1853149Seric 
1868078Seric # define NSTATES	5	/* number of states */
1878078Seric # define TYPE		017	/* mask to select state type */
1888078Seric 
1898078Seric /* meta bits for table */
1908078Seric # define M		020	/* meta character; don't pass through */
1918078Seric # define B		040	/* cause a break */
1928078Seric # define MB		M|B	/* meta-break */
1938078Seric 
1948078Seric static short StateTab[NSTATES][NSTATES] =
1958078Seric {
1968087Seric    /*	oldst	chtype>	OPR	ATM	QST	SPC	ONE	*/
1979051Seric 	/*OPR*/		OPR|B,	ATM|B,	QST|B,	SPC|MB,	ONE|B,
1989051Seric 	/*ATM*/		OPR|B,	ATM,	QST|B,	SPC|MB,	ONE|B,
1999051Seric 	/*QST*/		QST,	QST,	OPR,	QST,	QST,
2008078Seric 	/*SPC*/		OPR,	ATM,	QST,	SPC|M,	ONE,
2018078Seric 	/*ONE*/		OPR,	OPR,	OPR,	OPR,	OPR,
2028078Seric };
2038078Seric 
2048078Seric # define NOCHAR		-1	/* signal nothing in lookahead token */
2058078Seric 
2068078Seric char	*DelimChar;		/* set to point to the delimiter */
2078078Seric 
2083149Seric char **
2093149Seric prescan(addr, delim)
210297Seric 	char *addr;
211297Seric 	char delim;
212297Seric {
213297Seric 	register char *p;
2148078Seric 	register char *q;
2159346Seric 	register int c;
2163149Seric 	char **avp;
217297Seric 	bool bslashmode;
218297Seric 	int cmntcnt;
2198423Seric 	int anglecnt;
2203149Seric 	char *tok;
2218078Seric 	int state;
2228078Seric 	int newstate;
2238078Seric 	static char buf[MAXNAME+MAXATOM];
2248078Seric 	static char *av[MAXATOM+1];
22515253Seric 	extern int errno;
226297Seric 
22715253Seric 	/* make sure error messages don't have garbage on them */
22815253Seric 	errno = 0;
22915253Seric 
230297Seric 	q = buf;
2313149Seric 	bslashmode = FALSE;
2327800Seric 	cmntcnt = 0;
2338423Seric 	anglecnt = 0;
2343149Seric 	avp = av;
2358078Seric 	state = OPR;
2368078Seric 	c = NOCHAR;
2378078Seric 	p = addr;
2388078Seric # ifdef DEBUG
2398078Seric 	if (tTd(22, 45))
240297Seric 	{
2418078Seric 		printf("prescan: ");
2428078Seric 		xputs(p);
2438078Seric 		putchar('\n');
2448078Seric 	}
2458078Seric # endif DEBUG
2468078Seric 
2478078Seric 	do
2488078Seric 	{
2493149Seric 		/* read a token */
2503149Seric 		tok = q;
2518078Seric 		for (;;)
252297Seric 		{
2538078Seric 			/* store away any old lookahead character */
2548078Seric 			if (c != NOCHAR)
2558078Seric 			{
256*15284Seric 				/* see if there is room */
2578078Seric 				if (q >= &buf[sizeof buf - 5])
2588078Seric 				{
2598078Seric 					usrerr("Address too long");
2608078Seric 					DelimChar = p;
2618078Seric 					return (NULL);
2628078Seric 				}
263*15284Seric 
264*15284Seric 				/* squirrel it away */
2658078Seric 				*q++ = c;
2668078Seric 			}
2678078Seric 
2688078Seric 			/* read a new input character */
2698078Seric 			c = *p++;
2708078Seric 			if (c == '\0')
2718078Seric 				break;
272*15284Seric 			c &= ~0200;
273*15284Seric 
2748078Seric # ifdef DEBUG
2758078Seric 			if (tTd(22, 101))
2768078Seric 				printf("c=%c, s=%d; ", c, state);
2778078Seric # endif DEBUG
2788078Seric 
2793149Seric 			/* chew up special characters */
2803149Seric 			*q = '\0';
2813149Seric 			if (bslashmode)
2823149Seric 			{
2833149Seric 				c |= 0200;
2843149Seric 				bslashmode = FALSE;
2853149Seric 			}
2863149Seric 			else if (c == '\\')
2873149Seric 			{
2883149Seric 				bslashmode = TRUE;
2898078Seric 				c = NOCHAR;
2903149Seric 			}
2918514Seric 			else if (state == QST)
2928514Seric 			{
2938514Seric 				/* do nothing, just avoid next clauses */
2948514Seric 			}
2958078Seric 			else if (c == '(')
2964100Seric 			{
2978078Seric 				cmntcnt++;
2988078Seric 				c = NOCHAR;
2994100Seric 			}
3008078Seric 			else if (c == ')')
3013149Seric 			{
3028078Seric 				if (cmntcnt <= 0)
3033149Seric 				{
3048078Seric 					usrerr("Unbalanced ')'");
3058078Seric 					DelimChar = p;
3068078Seric 					return (NULL);
3073149Seric 				}
3088078Seric 				else
3098078Seric 					cmntcnt--;
3108078Seric 			}
3118078Seric 			else if (cmntcnt > 0)
3128078Seric 				c = NOCHAR;
3138423Seric 			else if (c == '<')
3148423Seric 				anglecnt++;
3158423Seric 			else if (c == '>')
3168423Seric 			{
3178423Seric 				if (anglecnt <= 0)
3188423Seric 				{
3198423Seric 					usrerr("Unbalanced '>'");
3208423Seric 					DelimChar = p;
3218423Seric 					return (NULL);
3228423Seric 				}
3238423Seric 				anglecnt--;
3248423Seric 			}
32511423Seric 			else if (delim == ' ' && isspace(c))
32611423Seric 				c = ' ';
3273149Seric 
3288078Seric 			if (c == NOCHAR)
3298078Seric 				continue;
3303149Seric 
3318078Seric 			/* see if this is end of input */
33211405Seric 			if (c == delim && anglecnt <= 0 && state != QST)
3333149Seric 				break;
3343149Seric 
3358078Seric 			newstate = StateTab[state][toktype(c)];
3368078Seric # ifdef DEBUG
3378078Seric 			if (tTd(22, 101))
3388078Seric 				printf("ns=%02o\n", newstate);
3398078Seric # endif DEBUG
3408078Seric 			state = newstate & TYPE;
3418078Seric 			if (bitset(M, newstate))
3428078Seric 				c = NOCHAR;
3438078Seric 			if (bitset(B, newstate))
3444228Seric 				break;
345297Seric 		}
3463149Seric 
3473149Seric 		/* new token */
3488078Seric 		if (tok != q)
3491378Seric 		{
3508078Seric 			*q++ = '\0';
3518078Seric # ifdef DEBUG
3528078Seric 			if (tTd(22, 36))
353297Seric 			{
3548078Seric 				printf("tok=");
3558078Seric 				xputs(tok);
3568078Seric 				putchar('\n');
357297Seric 			}
3588078Seric # endif DEBUG
3598078Seric 			if (avp >= &av[MAXATOM])
360297Seric 			{
3618078Seric 				syserr("prescan: too many tokens");
3628078Seric 				DelimChar = p;
3638078Seric 				return (NULL);
364297Seric 			}
3658078Seric 			*avp++ = tok;
366297Seric 		}
3678423Seric 	} while (c != '\0' && (c != delim || anglecnt > 0));
3683149Seric 	*avp = NULL;
3698078Seric 	DelimChar = --p;
3703149Seric 	if (cmntcnt > 0)
3713149Seric 		usrerr("Unbalanced '('");
3728423Seric 	else if (anglecnt > 0)
3738423Seric 		usrerr("Unbalanced '<'");
3748078Seric 	else if (state == QST)
3753149Seric 		usrerr("Unbalanced '\"'");
3763149Seric 	else if (av[0] != NULL)
3773149Seric 		return (av);
3783149Seric 	return (NULL);
3793149Seric }
3803149Seric /*
3813149Seric **  TOKTYPE -- return token type
3823149Seric **
3833149Seric **	Parameters:
3843149Seric **		c -- the character in question.
3853149Seric **
3863149Seric **	Returns:
3873149Seric **		Its type.
3883149Seric **
3893149Seric **	Side Effects:
3903149Seric **		none.
3913149Seric */
392297Seric 
3933149Seric toktype(c)
3943149Seric 	register char c;
3953149Seric {
3963380Seric 	static char buf[50];
3973382Seric 	static bool firstime = TRUE;
3983380Seric 
3993382Seric 	if (firstime)
4003380Seric 	{
4013382Seric 		firstime = FALSE;
4026977Seric 		expand("$o", buf, &buf[sizeof buf - 1], CurEnv);
4037005Seric 		(void) strcat(buf, DELIMCHARS);
4043380Seric 	}
4059585Seric 	if (c == MATCHCLASS || c == MATCHREPL || c == MATCHNCLASS)
4068078Seric 		return (ONE);
4078078Seric 	if (c == '"')
4088078Seric 		return (QST);
4094100Seric 	if (!isascii(c))
4108078Seric 		return (ATM);
4118078Seric 	if (isspace(c) || c == ')')
4128078Seric 		return (SPC);
4133380Seric 	if (iscntrl(c) || index(buf, c) != NULL)
4148078Seric 		return (OPR);
4158078Seric 	return (ATM);
4163149Seric }
4173149Seric /*
4183149Seric **  REWRITE -- apply rewrite rules to token vector.
4193149Seric **
4204476Seric **	This routine is an ordered production system.  Each rewrite
4214476Seric **	rule has a LHS (called the pattern) and a RHS (called the
4224476Seric **	rewrite); 'rwr' points the the current rewrite rule.
4234476Seric **
4244476Seric **	For each rewrite rule, 'avp' points the address vector we
4254476Seric **	are trying to match against, and 'pvp' points to the pattern.
4268058Seric **	If pvp points to a special match value (MATCHZANY, MATCHANY,
4279585Seric **	MATCHONE, MATCHCLASS, MATCHNCLASS) then the address in avp
4289585Seric **	matched is saved away in the match vector (pointed to by 'mvp').
4294476Seric **
4304476Seric **	When a match between avp & pvp does not match, we try to
4319585Seric **	back out.  If we back up over MATCHONE, MATCHCLASS, or MATCHNCLASS
4324476Seric **	we must also back out the match in mvp.  If we reach a
4338058Seric **	MATCHANY or MATCHZANY we just extend the match and start
4348058Seric **	over again.
4354476Seric **
4364476Seric **	When we finally match, we rewrite the address vector
4374476Seric **	and try over again.
4384476Seric **
4393149Seric **	Parameters:
4403149Seric **		pvp -- pointer to token vector.
4413149Seric **
4423149Seric **	Returns:
4433149Seric **		none.
4443149Seric **
4453149Seric **	Side Effects:
4463149Seric **		pvp is modified.
4473149Seric */
4482091Seric 
4493149Seric struct match
4503149Seric {
4514468Seric 	char	**first;	/* first token matched */
4524468Seric 	char	**last;		/* last token matched */
4533149Seric };
4543149Seric 
4554468Seric # define MAXMATCH	9	/* max params per rewrite */
4563149Seric 
4573149Seric 
4584070Seric rewrite(pvp, ruleset)
4593149Seric 	char **pvp;
4604070Seric 	int ruleset;
4613149Seric {
4623149Seric 	register char *ap;		/* address pointer */
4633149Seric 	register char *rp;		/* rewrite pointer */
4643149Seric 	register char **avp;		/* address vector pointer */
4653149Seric 	register char **rvp;		/* rewrite vector pointer */
4668058Seric 	register struct match *mlp;	/* cur ptr into mlist */
4678058Seric 	register struct rewrite *rwr;	/* pointer to current rewrite rule */
4684468Seric 	struct match mlist[MAXMATCH];	/* stores match on LHS */
4693149Seric 	char *npvp[MAXATOM+1];		/* temporary space for rebuild */
4704060Seric 	extern bool sameword();
4713149Seric 
4729279Seric 	if (OpMode == MD_TEST || tTd(21, 2))
4733149Seric 	{
4748959Seric 		printf("rewrite: ruleset %2d   input:", ruleset);
4753149Seric 		printav(pvp);
4763149Seric 	}
4778423Seric 	if (pvp == NULL)
4788423Seric 		return;
4793149Seric 
4803149Seric 	/*
4813149Seric 	**  Run through the list of rewrite rules, applying
4823149Seric 	**	any that match.
4833149Seric 	*/
4843149Seric 
4854070Seric 	for (rwr = RewriteRules[ruleset]; rwr != NULL; )
4863149Seric 	{
4874100Seric # ifdef DEBUG
4887675Seric 		if (tTd(21, 12))
489297Seric 		{
4908069Seric 			printf("-----trying rule:");
4913149Seric 			printav(rwr->r_lhs);
4923149Seric 		}
4934100Seric # endif DEBUG
4943149Seric 
4953149Seric 		/* try to match on this rule */
4964468Seric 		mlp = mlist;
4978058Seric 		rvp = rwr->r_lhs;
4988058Seric 		avp = pvp;
4998058Seric 		while ((ap = *avp) != NULL || *rvp != NULL)
5003149Seric 		{
5013149Seric 			rp = *rvp;
5028058Seric # ifdef DEBUG
5038058Seric 			if (tTd(21, 35))
5048058Seric 			{
5058069Seric 				printf("ap=");
5068058Seric 				xputs(ap);
5078069Seric 				printf(", rp=");
5088058Seric 				xputs(rp);
5098069Seric 				printf("\n");
5108058Seric 			}
5118058Seric # endif DEBUG
5123149Seric 			if (rp == NULL)
513297Seric 			{
5143149Seric 				/* end-of-pattern before end-of-address */
5158058Seric 				goto backup;
516297Seric 			}
5178058Seric 			if (ap == NULL && *rp != MATCHZANY)
5188058Seric 			{
5198058Seric 				/* end-of-input */
5208058Seric 				break;
5218058Seric 			}
5223149Seric 
5233149Seric 			switch (*rp)
5243149Seric 			{
5254060Seric 				register STAB *s;
5264060Seric 
5274060Seric 			  case MATCHCLASS:
5289585Seric 			  case MATCHNCLASS:
5299585Seric 				/* match any token in (not in) a class */
5304100Seric 				s = stab(ap, ST_CLASS, ST_FIND);
53110690Seric 				if (s == NULL || !bitnset(rp[1], s->s_class))
5329585Seric 				{
5339585Seric 					if (*rp == MATCHCLASS)
5349585Seric 						goto backup;
5359585Seric 				}
5369585Seric 				else if (*rp == MATCHNCLASS)
5378058Seric 					goto backup;
5384468Seric 
5394476Seric 				/* explicit fall-through */
5404476Seric 
5414476Seric 			  case MATCHONE:
5424476Seric 			  case MATCHANY:
5434476Seric 				/* match exactly one token */
5448058Seric 				mlp->first = avp;
5458058Seric 				mlp->last = avp++;
5464468Seric 				mlp++;
5474060Seric 				break;
5484060Seric 
5498058Seric 			  case MATCHZANY:
5508058Seric 				/* match zero or more tokens */
5518058Seric 				mlp->first = avp;
5528058Seric 				mlp->last = avp - 1;
5538058Seric 				mlp++;
5548058Seric 				break;
5558058Seric 
5563149Seric 			  default:
5573149Seric 				/* must have exact match */
5584060Seric 				if (!sameword(rp, ap))
5598058Seric 					goto backup;
5604468Seric 				avp++;
5613149Seric 				break;
5623149Seric 			}
5633149Seric 
5643149Seric 			/* successful match on this token */
5653149Seric 			rvp++;
5663149Seric 			continue;
5673149Seric 
5688058Seric 		  backup:
5693149Seric 			/* match failed -- back up */
5703149Seric 			while (--rvp >= rwr->r_lhs)
5713149Seric 			{
5723149Seric 				rp = *rvp;
5738058Seric 				if (*rp == MATCHANY || *rp == MATCHZANY)
5744468Seric 				{
5754476Seric 					/* extend binding and continue */
5768058Seric 					avp = ++mlp[-1].last;
5778058Seric 					avp++;
5784476Seric 					rvp++;
5793149Seric 					break;
5804468Seric 				}
5814476Seric 				avp--;
5829585Seric 				if (*rp == MATCHONE || *rp == MATCHCLASS ||
5839585Seric 				    *rp == MATCHNCLASS)
5843149Seric 				{
5854468Seric 					/* back out binding */
5864468Seric 					mlp--;
5873149Seric 				}
5883149Seric 			}
5893149Seric 
5903149Seric 			if (rvp < rwr->r_lhs)
5913149Seric 			{
5923149Seric 				/* total failure to match */
5933149Seric 				break;
5943149Seric 			}
595297Seric 		}
5963149Seric 
5973149Seric 		/*
5983149Seric 		**  See if we successfully matched
5993149Seric 		*/
6003149Seric 
6019374Seric 		if (rvp < rwr->r_lhs || *rvp != NULL)
6023149Seric 		{
6034100Seric # ifdef DEBUG
6049374Seric 			if (tTd(21, 10))
6059374Seric 				printf("----- rule fails\n");
6064100Seric # endif DEBUG
6079374Seric 			rwr = rwr->r_next;
6089374Seric 			continue;
6099374Seric 		}
6103149Seric 
6119374Seric 		rvp = rwr->r_rhs;
6129374Seric # ifdef DEBUG
6139374Seric 		if (tTd(21, 12))
6149374Seric 		{
6159374Seric 			printf("-----rule matches:");
6169374Seric 			printav(rvp);
6179374Seric 		}
6189374Seric # endif DEBUG
6199374Seric 
6209374Seric 		rp = *rvp;
6219374Seric 		if (*rp == CANONUSER)
6229374Seric 		{
6239374Seric 			rvp++;
6249374Seric 			rwr = rwr->r_next;
6259374Seric 		}
6269374Seric 		else if (*rp == CANONHOST)
6279374Seric 		{
6289374Seric 			rvp++;
6299374Seric 			rwr = NULL;
6309374Seric 		}
6319374Seric 		else if (*rp == CANONNET)
6329374Seric 			rwr = NULL;
6339374Seric 
6349374Seric 		/* substitute */
6359374Seric 		for (avp = npvp; *rvp != NULL; rvp++)
6369374Seric 		{
6379374Seric 			register struct match *m;
6389374Seric 			register char **pp;
6399374Seric 
6408058Seric 			rp = *rvp;
6419374Seric 			if (*rp != MATCHREPL)
6428058Seric 			{
6439374Seric 				if (avp >= &npvp[MAXATOM])
6449374Seric 				{
6459374Seric 					syserr("rewrite: expansion too long");
6469374Seric 					return;
6479374Seric 				}
6489374Seric 				*avp++ = rp;
6499374Seric 				continue;
6508069Seric 			}
6518058Seric 
6529374Seric 			/* substitute from LHS */
6539374Seric 			m = &mlist[rp[1] - '1'];
6549374Seric # ifdef DEBUG
6559374Seric 			if (tTd(21, 15))
6563149Seric 			{
6579374Seric 				printf("$%c:", rp[1]);
6589374Seric 				pp = m->first;
6599374Seric 				while (pp <= m->last)
6603149Seric 				{
6619374Seric 					printf(" %x=\"", *pp);
6629374Seric 					(void) fflush(stdout);
6639374Seric 					printf("%s\"", *pp++);
6643149Seric 				}
6659374Seric 				printf("\n");
6663149Seric 			}
6678226Seric # endif DEBUG
6689374Seric 			pp = m->first;
6699374Seric 			while (pp <= m->last)
6708226Seric 			{
6719374Seric 				if (avp >= &npvp[MAXATOM])
6729374Seric 				{
6739374Seric 					syserr("rewrite: expansion too long");
6749374Seric 					return;
6759374Seric 				}
6769374Seric 				*avp++ = *pp++;
6778226Seric 			}
6789374Seric 		}
6799374Seric 		*avp++ = NULL;
6809374Seric 		if (**npvp == CALLSUBR)
6819374Seric 		{
6829374Seric 			bmove((char *) &npvp[2], (char *) pvp,
6839374Seric 				(avp - npvp - 2) * sizeof *avp);
6848226Seric # ifdef DEBUG
6859374Seric 			if (tTd(21, 3))
6869374Seric 				printf("-----callsubr %s\n", npvp[1]);
6873149Seric # endif DEBUG
6889374Seric 			rewrite(pvp, atoi(npvp[1]));
6893149Seric 		}
6903149Seric 		else
6913149Seric 		{
6929374Seric 			bmove((char *) npvp, (char *) pvp,
6939374Seric 				(avp - npvp) * sizeof *avp);
6949374Seric 		}
6954100Seric # ifdef DEBUG
6969374Seric 		if (tTd(21, 4))
6979374Seric 		{
6989374Seric 			printf("rewritten as:");
6999374Seric 			printav(pvp);
7009374Seric 		}
7014100Seric # endif DEBUG
702297Seric 	}
7038069Seric 
7049279Seric 	if (OpMode == MD_TEST || tTd(21, 2))
7058069Seric 	{
7068959Seric 		printf("rewrite: ruleset %2d returns:", ruleset);
7078069Seric 		printav(pvp);
7088069Seric 	}
7093149Seric }
7103149Seric /*
7113149Seric **  BUILDADDR -- build address from token vector.
7123149Seric **
7133149Seric **	Parameters:
7143149Seric **		tv -- token vector.
7153149Seric **		a -- pointer to address descriptor to fill.
7163149Seric **			If NULL, one will be allocated.
7173149Seric **
7183149Seric **	Returns:
7194279Seric **		NULL if there was an error.
7204279Seric **		'a' otherwise.
7213149Seric **
7223149Seric **	Side Effects:
7233149Seric **		fills in 'a'
7243149Seric */
7253149Seric 
7263149Seric ADDRESS *
7273149Seric buildaddr(tv, a)
7283149Seric 	register char **tv;
7293149Seric 	register ADDRESS *a;
7303149Seric {
7313149Seric 	static char buf[MAXNAME];
7323149Seric 	struct mailer **mp;
7333149Seric 	register struct mailer *m;
7344635Seric 	extern bool sameword();
7353149Seric 
7363149Seric 	if (a == NULL)
7373149Seric 		a = (ADDRESS *) xalloc(sizeof *a);
7384988Seric 	clear((char *) a, sizeof *a);
7393149Seric 
7403149Seric 	/* figure out what net/mailer to use */
7413149Seric 	if (**tv != CANONNET)
7424279Seric 	{
7433149Seric 		syserr("buildaddr: no net");
7444279Seric 		return (NULL);
7454279Seric 	}
7463149Seric 	tv++;
7474635Seric 	if (sameword(*tv, "error"))
7484279Seric 	{
74910183Seric 		if (**++tv == CANONHOST)
75010183Seric 		{
75110183Seric 			setstat(atoi(*++tv));
75210183Seric 			tv++;
75310183Seric 		}
75410183Seric 		if (**tv != CANONUSER)
7554279Seric 			syserr("buildaddr: error: no user");
7564279Seric 		buf[0] = '\0';
7574279Seric 		while (*++tv != NULL)
7584279Seric 		{
7594279Seric 			if (buf[0] != '\0')
7607005Seric 				(void) strcat(buf, " ");
7617005Seric 			(void) strcat(buf, *tv);
7624279Seric 		}
7634279Seric 		usrerr(buf);
7644279Seric 		return (NULL);
7654279Seric 	}
7664598Seric 	for (mp = Mailer; (m = *mp++) != NULL; )
7673149Seric 	{
7684635Seric 		if (sameword(m->m_name, *tv))
7693149Seric 			break;
7703149Seric 	}
7713149Seric 	if (m == NULL)
7724279Seric 	{
7733149Seric 		syserr("buildaddr: unknown net %s", *tv);
7744279Seric 		return (NULL);
7754279Seric 	}
7764598Seric 	a->q_mailer = m;
7773149Seric 
7783149Seric 	/* figure out what host (if any) */
7793149Seric 	tv++;
78010690Seric 	if (!bitnset(M_LOCAL, m->m_flags))
7813149Seric 	{
7825704Seric 		if (**tv++ != CANONHOST)
7834279Seric 		{
7843149Seric 			syserr("buildaddr: no host");
7854279Seric 			return (NULL);
7864279Seric 		}
7875704Seric 		buf[0] = '\0';
7885704Seric 		while (*tv != NULL && **tv != CANONUSER)
7897005Seric 			(void) strcat(buf, *tv++);
7905704Seric 		a->q_host = newstr(buf);
7913149Seric 	}
7923149Seric 	else
7933149Seric 		a->q_host = NULL;
7943149Seric 
7953149Seric 	/* figure out the user */
7963149Seric 	if (**tv != CANONUSER)
7974279Seric 	{
7983149Seric 		syserr("buildaddr: no user");
7994279Seric 		return (NULL);
8004279Seric 	}
80111278Seric 	rewrite(++tv, 4);
80211278Seric 	cataddr(tv, buf, sizeof buf);
8033149Seric 	a->q_user = buf;
8043149Seric 
8053149Seric 	return (a);
8063149Seric }
8073188Seric /*
8084228Seric **  CATADDR -- concatenate pieces of addresses (putting in <LWSP> subs)
8094228Seric **
8104228Seric **	Parameters:
8114228Seric **		pvp -- parameter vector to rebuild.
8124228Seric **		buf -- buffer to build the string into.
8134228Seric **		sz -- size of buf.
8144228Seric **
8154228Seric **	Returns:
8164228Seric **		none.
8174228Seric **
8184228Seric **	Side Effects:
8194228Seric **		Destroys buf.
8204228Seric */
8214228Seric 
8224228Seric cataddr(pvp, buf, sz)
8234228Seric 	char **pvp;
8244228Seric 	char *buf;
8254228Seric 	register int sz;
8264228Seric {
8274228Seric 	bool oatomtok = FALSE;
8284228Seric 	bool natomtok = FALSE;
8294228Seric 	register int i;
8304228Seric 	register char *p;
8314228Seric 
8328423Seric 	if (pvp == NULL)
8338423Seric 	{
8348423Seric 		strcpy(buf, "");
8358423Seric 		return;
8368423Seric 	}
8374228Seric 	p = buf;
83811156Seric 	sz -= 2;
8394228Seric 	while (*pvp != NULL && (i = strlen(*pvp)) < sz)
8404228Seric 	{
8418078Seric 		natomtok = (toktype(**pvp) == ATM);
8424228Seric 		if (oatomtok && natomtok)
8439042Seric 			*p++ = SpaceSub;
8444228Seric 		(void) strcpy(p, *pvp);
8454228Seric 		oatomtok = natomtok;
8464228Seric 		p += i;
84711156Seric 		sz -= i + 1;
8484228Seric 		pvp++;
8494228Seric 	}
8504228Seric 	*p = '\0';
8514228Seric }
8524228Seric /*
8533188Seric **  SAMEADDR -- Determine if two addresses are the same
8543188Seric **
8553188Seric **	This is not just a straight comparison -- if the mailer doesn't
8563188Seric **	care about the host we just ignore it, etc.
8573188Seric **
8583188Seric **	Parameters:
8593188Seric **		a, b -- pointers to the internal forms to compare.
8603188Seric **
8613188Seric **	Returns:
8623188Seric **		TRUE -- they represent the same mailbox.
8633188Seric **		FALSE -- they don't.
8643188Seric **
8653188Seric **	Side Effects:
8663188Seric **		none.
8673188Seric */
8683188Seric 
8693188Seric bool
8709374Seric sameaddr(a, b)
8713188Seric 	register ADDRESS *a;
8723188Seric 	register ADDRESS *b;
8733188Seric {
8743188Seric 	/* if they don't have the same mailer, forget it */
8753188Seric 	if (a->q_mailer != b->q_mailer)
8763188Seric 		return (FALSE);
8773188Seric 
8783188Seric 	/* if the user isn't the same, we can drop out */
8799374Seric 	if (strcmp(a->q_user, b->q_user) != 0)
8803188Seric 		return (FALSE);
8813188Seric 
8823188Seric 	/* if the mailer ignores hosts, we have succeeded! */
88310690Seric 	if (bitnset(M_LOCAL, a->q_mailer->m_flags))
8843188Seric 		return (TRUE);
8853188Seric 
8863188Seric 	/* otherwise compare hosts (but be careful for NULL ptrs) */
8873188Seric 	if (a->q_host == NULL || b->q_host == NULL)
8883188Seric 		return (FALSE);
8893188Seric 	if (strcmp(a->q_host, b->q_host) != 0)
8903188Seric 		return (FALSE);
8913188Seric 
8923188Seric 	return (TRUE);
8933188Seric }
8943234Seric /*
8953234Seric **  PRINTADDR -- print address (for debugging)
8963234Seric **
8973234Seric **	Parameters:
8983234Seric **		a -- the address to print
8993234Seric **		follow -- follow the q_next chain.
9003234Seric **
9013234Seric **	Returns:
9023234Seric **		none.
9033234Seric **
9043234Seric **	Side Effects:
9053234Seric **		none.
9063234Seric */
9073234Seric 
9084317Seric # ifdef DEBUG
9094317Seric 
9103234Seric printaddr(a, follow)
9113234Seric 	register ADDRESS *a;
9123234Seric 	bool follow;
9133234Seric {
9145001Seric 	bool first = TRUE;
9155001Seric 
9163234Seric 	while (a != NULL)
9173234Seric 	{
9185001Seric 		first = FALSE;
9194443Seric 		printf("%x=", a);
9204085Seric 		(void) fflush(stdout);
9213234Seric 		printf("%s: mailer %d (%s), host `%s', user `%s'\n", a->q_paddr,
9228181Seric 		       a->q_mailer->m_mno, a->q_mailer->m_name, a->q_host,
9238181Seric 		       a->q_user);
9248181Seric 		printf("\tnext=%x, flags=%o, alias %x\n", a->q_next, a->q_flags,
9258181Seric 		       a->q_alias);
9268181Seric 		printf("\thome=\"%s\", fullname=\"%s\"\n", a->q_home,
9278181Seric 		       a->q_fullname);
9284996Seric 
9293234Seric 		if (!follow)
9303234Seric 			return;
9314996Seric 		a = a->q_next;
9323234Seric 	}
9335001Seric 	if (first)
9344443Seric 		printf("[NULL]\n");
9353234Seric }
9364317Seric 
9374317Seric # endif DEBUG
9387682Seric /*
9397682Seric **  REMOTENAME -- return the name relative to the current mailer
9407682Seric **
9417682Seric **	Parameters:
9427682Seric **		name -- the name to translate.
9438069Seric **		m -- the mailer that we want to do rewriting relative
9448069Seric **			to.
9458069Seric **		senderaddress -- if set, uses the sender rewriting rules
9468069Seric **			rather than the recipient rewriting rules.
94710310Seric **		canonical -- if set, strip out any comment information,
94810310Seric **			etc.
9497682Seric **
9507682Seric **	Returns:
9517682Seric **		the text string representing this address relative to
9527682Seric **			the receiving mailer.
9537682Seric **
9547682Seric **	Side Effects:
9557682Seric **		none.
9567682Seric **
9577682Seric **	Warnings:
9587682Seric **		The text string returned is tucked away locally;
9597682Seric **			copy it if you intend to save it.
9607682Seric */
9617682Seric 
9627682Seric char *
96310310Seric remotename(name, m, senderaddress, canonical)
9647682Seric 	char *name;
9657682Seric 	struct mailer *m;
9668069Seric 	bool senderaddress;
96710310Seric 	bool canonical;
9687682Seric {
9698069Seric 	register char **pvp;
9708069Seric 	char *fancy;
971*15284Seric 	register char *p;
9728069Seric 	extern char *macvalue();
9738181Seric 	char *oldg = macvalue('g', CurEnv);
9747682Seric 	static char buf[MAXNAME];
9757682Seric 	char lbuf[MAXNAME];
9767682Seric 	extern char **prescan();
9777889Seric 	extern char *crackaddr();
9787682Seric 
9797755Seric # ifdef DEBUG
9807755Seric 	if (tTd(12, 1))
9817755Seric 		printf("remotename(%s)\n", name);
9827755Seric # endif DEBUG
9837755Seric 
98410177Seric 	/* don't do anything if we are tagging it as special */
98510177Seric 	if ((senderaddress ? m->m_s_rwset : m->m_r_rwset) < 0)
98610177Seric 		return (name);
98710177Seric 
9887682Seric 	/*
9898181Seric 	**  Do a heuristic crack of this name to extract any comment info.
9908181Seric 	**	This will leave the name as a comment and a $g macro.
9917889Seric 	*/
9927889Seric 
99310310Seric 	if (canonical)
99410310Seric 		fancy = "$g";
99510310Seric 	else
99610310Seric 		fancy = crackaddr(name);
9977889Seric 
9988181Seric 	/*
9998181Seric 	**  Turn the name into canonical form.
10008181Seric 	**	Normally this will be RFC 822 style, i.e., "user@domain".
10018181Seric 	**	If this only resolves to "user", and the "C" flag is
10028181Seric 	**	specified in the sending mailer, then the sender's
10038181Seric 	**	domain will be appended.
10048181Seric 	*/
10058181Seric 
10067889Seric 	pvp = prescan(name, '\0');
10077889Seric 	if (pvp == NULL)
10087889Seric 		return (name);
10098181Seric 	rewrite(pvp, 3);
10108181Seric 	if (CurEnv->e_fromdomain != NULL)
10118181Seric 	{
10128181Seric 		/* append from domain to this address */
10138181Seric 		register char **pxp = pvp;
10148181Seric 
10159594Seric 		/* see if there is an "@domain" in the current name */
10168181Seric 		while (*pxp != NULL && strcmp(*pxp, "@") != 0)
10178181Seric 			pxp++;
10188181Seric 		if (*pxp == NULL)
10198181Seric 		{
10209594Seric 			/* no.... append the "@domain" from the sender */
10218181Seric 			register char **qxq = CurEnv->e_fromdomain;
10228181Seric 
10239594Seric 			while ((*pxp++ = *qxq++) != NULL)
10249594Seric 				continue;
102511726Seric 			rewrite(pvp, 3);
10268181Seric 		}
10278181Seric 	}
10288181Seric 
10298181Seric 	/*
10308959Seric 	**  Do more specific rewriting.
10318181Seric 	**	Rewrite using ruleset 1 or 2 depending on whether this is
10328181Seric 	**		a sender address or not.
10338181Seric 	**	Then run it through any receiving-mailer-specific rulesets.
10348181Seric 	*/
10358181Seric 
10368069Seric 	if (senderaddress)
10377755Seric 	{
10387889Seric 		rewrite(pvp, 1);
10398069Seric 		if (m->m_s_rwset > 0)
10408069Seric 			rewrite(pvp, m->m_s_rwset);
10418069Seric 	}
10428069Seric 	else
10438069Seric 	{
10447889Seric 		rewrite(pvp, 2);
10458069Seric 		if (m->m_r_rwset > 0)
10468069Seric 			rewrite(pvp, m->m_r_rwset);
10477682Seric 	}
10487682Seric 
10498181Seric 	/*
10508959Seric 	**  Do any final sanitation the address may require.
10518959Seric 	**	This will normally be used to turn internal forms
10528959Seric 	**	(e.g., user@host.LOCAL) into external form.  This
10538959Seric 	**	may be used as a default to the above rules.
10548959Seric 	*/
10558959Seric 
10568959Seric 	rewrite(pvp, 4);
10578959Seric 
10588959Seric 	/*
10598181Seric 	**  Now restore the comment information we had at the beginning.
1060*15284Seric 	**	Make sure that any real '$' characters in the input are
1061*15284Seric 	**	not accidently interpreted as macro expansions by quoting
1062*15284Seric 	**	them before expansion.
10638181Seric 	*/
10648181Seric 
10657682Seric 	cataddr(pvp, lbuf, sizeof lbuf);
1066*15284Seric 	for (p = lbuf; *p != '\0'; p++)
1067*15284Seric 		if (*p == '$')
1068*15284Seric 			*p |= 0200;
10699374Seric 	define('g', lbuf, CurEnv);
10707889Seric 	expand(fancy, buf, &buf[sizeof buf - 1], CurEnv);
10719374Seric 	define('g', oldg, CurEnv);
1072*15284Seric 	stripquotes(buf, FALSE);
10737682Seric 
10747682Seric # ifdef DEBUG
10757682Seric 	if (tTd(12, 1))
10767755Seric 		printf("remotename => `%s'\n", buf);
10777682Seric # endif DEBUG
10787682Seric 	return (buf);
10797682Seric }
1080