122697Sdist /*
234920Sbostic  * Copyright (c) 1983 Eric P. Allman
362522Sbostic  * Copyright (c) 1988, 1993
462522Sbostic  *	The Regents of the University of California.  All rights reserved.
533728Sbostic  *
642824Sbostic  * %sccs.include.redist.c%
733728Sbostic  */
822697Sdist 
922697Sdist #ifndef lint
10*66765Seric static char sccsid[] = "@(#)collect.c	8.10 (Berkeley) 04/12/94";
1133728Sbostic #endif /* not lint */
1222697Sdist 
131439Seric # include <errno.h>
143309Seric # include "sendmail.h"
151392Seric 
161392Seric /*
172969Seric **  COLLECT -- read & parse message header & make temp file.
181392Seric **
191392Seric **	Creates a temporary file name and copies the standard
209371Seric **	input to that file.  Leading UNIX-style "From" lines are
219371Seric **	stripped off (after important information is extracted).
221392Seric **
231392Seric **	Parameters:
2452106Seric **		smtpmode -- if set, we are running SMTP: give an RFC821
2552105Seric **			style message to say we are ready to collect
2652105Seric **			input, and never ignore a single dot to mean
2752105Seric **			end of message.
2858929Seric **		requeueflag -- this message will be requeued later, so
2958929Seric **			don't do final processing on it.
3058929Seric **		e -- the current envelope.
311392Seric **
321392Seric **	Returns:
334162Seric **		none.
341392Seric **
351392Seric **	Side Effects:
361392Seric **		Temp file is created and filled.
374162Seric **		The from person may be set.
381392Seric */
391392Seric 
4058929Seric collect(smtpmode, requeueflag, e)
4152105Seric 	bool smtpmode;
4258929Seric 	bool requeueflag;
4355012Seric 	register ENVELOPE *e;
441392Seric {
451392Seric 	register FILE *tf;
4652105Seric 	bool ignrdot = smtpmode ? FALSE : IgnrDot;
4757135Seric 	char buf[MAXLINE], buf2[MAXLINE];
4840965Sbostic 	register char *workbuf, *freebuf;
4964718Seric 	bool inputerr = FALSE;
502900Seric 	extern char *hvalue();
5140965Sbostic 	extern bool isheader(), flusheol();
521392Seric 
531392Seric 	/*
541392Seric 	**  Create the temp file name and create the file.
551392Seric 	*/
561392Seric 
5764086Seric 	e->e_df = queuename(e, 'd');
5864086Seric 	e->e_df = newstr(e->e_df);
5959745Seric 	if ((tf = dfopen(e->e_df, O_WRONLY|O_CREAT, FileMode)) == NULL)
601392Seric 	{
6155012Seric 		syserr("Cannot create %s", e->e_df);
625366Seric 		NoReturn = TRUE;
635366Seric 		finis();
641392Seric 	}
651392Seric 
664316Seric 	/*
674322Seric 	**  Tell ARPANET to go ahead.
684322Seric 	*/
694322Seric 
7052105Seric 	if (smtpmode)
7158151Seric 		message("354 Enter mail, end with \".\" on a line by itself");
724322Seric 
73*66765Seric 	/* set global timer to monitor progress */
74*66765Seric 	sfgetset(TimeOuts.to_datablock);
75*66765Seric 
764322Seric 	/*
774316Seric 	**  Try to read a UNIX-style From line
784316Seric 	*/
794316Seric 
8061093Seric 	if (sfgets(buf, MAXLINE, InChannel, TimeOuts.to_datablock,
8161093Seric 			"initial message read") == NULL)
8240965Sbostic 		goto readerr;
834557Seric 	fixcrlf(buf, FALSE);
844321Seric # ifndef NOTUNIX
854322Seric 	if (!SaveFrom && strncmp(buf, "From ", 5) == 0)
862900Seric 	{
8740965Sbostic 		if (!flusheol(buf, InChannel))
8840965Sbostic 			goto readerr;
8955012Seric 		eatfrom(buf, e);
9061093Seric 		if (sfgets(buf, MAXLINE, InChannel, TimeOuts.to_datablock,
9161093Seric 				"message header read") == NULL)
9240965Sbostic 			goto readerr;
934557Seric 		fixcrlf(buf, FALSE);
942900Seric 	}
9556795Seric # endif /* NOTUNIX */
962900Seric 
971392Seric 	/*
985975Seric 	**  Copy InChannel to temp file & do message editing.
991392Seric 	**	To keep certain mailers from getting confused,
1001392Seric 	**	and to keep the output clean, lines that look
10113932Seric 	**	like UNIX "From" lines are deleted in the header.
1021392Seric 	*/
1031392Seric 
10440965Sbostic 	workbuf = buf;		/* `workbuf' contains a header field */
10540965Sbostic 	freebuf = buf2;		/* `freebuf' can be used for read-ahead */
10640965Sbostic 	for (;;)
1071392Seric 	{
10857135Seric 		char *curbuf;
10957135Seric 		int curbuffree;
11057135Seric 		register int curbuflen;
11157135Seric 		char *p;
11257135Seric 
11340965Sbostic 		/* first, see if the header is over */
11440965Sbostic 		if (!isheader(workbuf))
11540965Sbostic 		{
11640965Sbostic 			fixcrlf(workbuf, TRUE);
11719036Seric 			break;
11840965Sbostic 		}
11919036Seric 
1207681Seric 		/* if the line is too long, throw the rest away */
12140965Sbostic 		if (!flusheol(workbuf, InChannel))
12240965Sbostic 			goto readerr;
1237681Seric 
12440965Sbostic 		/* it's okay to toss '\n' now (flusheol() needed it) */
12540965Sbostic 		fixcrlf(workbuf, TRUE);
1264557Seric 
12757135Seric 		curbuf = workbuf;
12857135Seric 		curbuflen = strlen(curbuf);
12957135Seric 		curbuffree = MAXLINE - curbuflen;
13057135Seric 		p = curbuf + curbuflen;
1312900Seric 
1322900Seric 		/* get the rest of this field */
13340965Sbostic 		for (;;)
1341392Seric 		{
13557135Seric 			int clen;
13657135Seric 
13761093Seric 			if (sfgets(freebuf, MAXLINE, InChannel,
13861093Seric 					TimeOuts.to_datablock,
13961093Seric 					"message header read") == NULL)
14064916Seric 			{
14164916Seric 				freebuf[0] = '\0';
14264916Seric 				break;
14364916Seric 			}
14440965Sbostic 
14540965Sbostic 			/* is this a continuation line? */
14640965Sbostic 			if (*freebuf != ' ' && *freebuf != '\t')
1472900Seric 				break;
14840965Sbostic 
14940965Sbostic 			if (!flusheol(freebuf, InChannel))
15040965Sbostic 				goto readerr;
15140965Sbostic 
15257135Seric 			fixcrlf(freebuf, TRUE);
15357135Seric 			clen = strlen(freebuf) + 1;
15457135Seric 
15557135Seric 			/* if insufficient room, dynamically allocate buffer */
15657135Seric 			if (clen >= curbuffree)
15740965Sbostic 			{
15857135Seric 				/* reallocate buffer */
15957135Seric 				int nbuflen = ((p - curbuf) + clen) * 2;
16057135Seric 				char *nbuf = xalloc(nbuflen);
16140965Sbostic 
16257135Seric 				p = nbuf + curbuflen;
16357135Seric 				curbuffree = nbuflen - curbuflen;
16457135Seric 				bcopy(curbuf, nbuf, curbuflen);
16557135Seric 				if (curbuf != buf && curbuf != buf2)
16657135Seric 					free(curbuf);
16757135Seric 				curbuf = nbuf;
16840965Sbostic 			}
16957135Seric 			*p++ = '\n';
17057135Seric 			bcopy(freebuf, p, clen - 1);
17157135Seric 			p += clen - 1;
17257135Seric 			curbuffree -= clen;
17357135Seric 			curbuflen += clen;
1741392Seric 		}
17557135Seric 		*p++ = '\0';
1761392Seric 
17757135Seric 		e->e_msgsize += curbuflen;
1781392Seric 
1792900Seric 		/*
18040965Sbostic 		**  The working buffer now becomes the free buffer, since
18140965Sbostic 		**  the free buffer contains a new header field.
18240965Sbostic 		**
18340965Sbostic 		**  This is premature, since we still havent called
18440965Sbostic 		**  chompheader() to process the field we just created
18540965Sbostic 		**  (so the call to chompheader() will use `freebuf').
18640965Sbostic 		**  This convolution is necessary so that if we break out
18740965Sbostic 		**  of the loop due to H_EOH, `workbuf' will always be
18840965Sbostic 		**  the next unprocessed buffer.
18940965Sbostic 		*/
19040965Sbostic 
19140965Sbostic 		{
19240965Sbostic 			register char *tmp = workbuf;
19340965Sbostic 			workbuf = freebuf;
19440965Sbostic 			freebuf = tmp;
19540965Sbostic 		}
19640965Sbostic 
19740965Sbostic 		/*
1982900Seric 		**  Snarf header away.
1992900Seric 		*/
2002900Seric 
20157135Seric 		if (bitset(H_EOH, chompheader(curbuf, FALSE, e)))
2023058Seric 			break;
20357135Seric 
20457135Seric 		/*
20557135Seric 		**  If the buffer was dynamically allocated, free it.
20657135Seric 		*/
20757135Seric 
20857135Seric 		if (curbuf != buf && curbuf != buf2)
20957135Seric 			free(curbuf);
21040965Sbostic 	}
2111392Seric 
2127673Seric 	if (tTd(30, 1))
2132900Seric 		printf("EOH\n");
2142900Seric 
21540965Sbostic 	if (*workbuf == '\0')
21640965Sbostic 	{
21740965Sbostic 		/* throw away a blank line */
21861093Seric 		if (sfgets(buf, MAXLINE, InChannel, TimeOuts.to_datablock,
21961093Seric 				"message separator read") == NULL)
22040965Sbostic 			goto readerr;
22140965Sbostic 	}
22240965Sbostic 	else if (workbuf == buf2)	/* guarantee `buf' contains data */
22340965Sbostic 		(void) strcpy(buf, buf2);
2242900Seric 
2252900Seric 	/*
2262900Seric 	**  Collect the body of the message.
2272900Seric 	*/
2282900Seric 
22964718Seric 	for (;;)
2302900Seric 	{
2314551Seric 		register char *bp = buf;
2324156Seric 
2337852Seric 		fixcrlf(buf, TRUE);
2344557Seric 
2352900Seric 		/* check for end-of-message */
23652105Seric 		if (!ignrdot && buf[0] == '.' && (buf[1] == '\n' || buf[1] == '\0'))
2372900Seric 			break;
2382900Seric 
2394551Seric 		/* check for transparent dot */
24065580Seric 		if ((OpMode == MD_SMTP || OpMode == MD_DAEMON) &&
24165580Seric 		    bp[0] == '.' && bp[1] == '.')
2424551Seric 			bp++;
2434551Seric 
2444156Seric 		/*
2454156Seric 		**  Figure message length, output the line to the temp
2464156Seric 		**  file, and insert a newline if missing.
2474156Seric 		*/
2484156Seric 
24955012Seric 		e->e_msgsize += strlen(bp) + 1;
2504551Seric 		fputs(bp, tf);
2517852Seric 		fputs("\n", tf);
2521392Seric 		if (ferror(tf))
25355012Seric 			tferror(tf, e);
25464718Seric 		if (sfgets(buf, MAXLINE, InChannel, TimeOuts.to_datablock,
25564718Seric 				"message body read") == NULL)
25664718Seric 			goto readerr;
25764718Seric 	}
25840965Sbostic 
25964718Seric 	if (feof(InChannel) || ferror(InChannel))
26064718Seric 	{
26140965Sbostic readerr:
26264916Seric 		if (tTd(30, 1))
26364916Seric 			printf("collect: read error\n");
26464718Seric 		inputerr = TRUE;
26564718Seric 	}
26664718Seric 
267*66765Seric 	/* reset global timer */
268*66765Seric 	sfgetset((time_t) 0);
269*66765Seric 
27011544Seric 	if (fflush(tf) != 0)
27155012Seric 		tferror(tf, e);
27264762Seric 	if (fsync(fileno(tf)) < 0 || fclose(tf) < 0)
27364762Seric 	{
27464762Seric 		syserr("cannot sync message data to disk (%s)", e->e_df);
27564762Seric 		finis();
27664762Seric 	}
2772900Seric 
27811145Seric 	/* An EOF when running SMTP is an error */
27965580Seric 	if (inputerr && (OpMode == MD_SMTP || OpMode == MD_DAEMON))
28016136Seric 	{
28158308Seric 		char *host;
28264718Seric 		char *problem;
28358082Seric 
28458308Seric 		host = RealHostName;
28558308Seric 		if (host == NULL)
28658308Seric 			host = "localhost";
28758308Seric 
28864718Seric 		if (feof(InChannel))
28964718Seric 			problem = "unexpected close";
29064718Seric 		else if (ferror(InChannel))
29164718Seric 			problem = "I/O error";
29264718Seric 		else
29364718Seric 			problem = "read timeout";
29436233Skarels # ifdef LOG
29558308Seric 		if (LogLevel > 0 && feof(InChannel))
29636230Skarels 			syslog(LOG_NOTICE,
29764718Seric 			    "collect: %s on connection from %s, sender=%s: %m\n",
29864718Seric 			    problem, host, e->e_from.q_paddr);
29936233Skarels # endif
30065951Seric 		if (feof(InChannel))
30165951Seric 			usrerr("451 collect: %s on connection from %s, from=%s",
30264718Seric 				problem, host, e->e_from.q_paddr);
30365951Seric 		else
30465951Seric 			syserr("451 collect: %s on connection from %s, from=%s",
30565951Seric 				problem, host, e->e_from.q_paddr);
30611145Seric 
30716136Seric 		/* don't return an error indication */
30855012Seric 		e->e_to = NULL;
30955012Seric 		e->e_flags &= ~EF_FATALERRS;
31064124Seric 		e->e_flags |= EF_CLRQUEUE;
31116136Seric 
31216136Seric 		/* and don't try to deliver the partial message either */
31364718Seric 		if (InChild)
31464718Seric 			ExitStat = EX_QUIT;
31516136Seric 		finis();
31616136Seric 	}
31716136Seric 
3182900Seric 	/*
3192900Seric 	**  Find out some information from the headers.
3203386Seric 	**	Examples are who is the from person & the date.
3212900Seric 	*/
3222900Seric 
32358929Seric 	eatheader(e, !requeueflag);
3247673Seric 
32564068Seric 	/* collect statistics */
32664068Seric 	if (OpMode != MD_VERIFY)
32764068Seric 		markstats(e, (ADDRESS *) NULL);
32864068Seric 
3297782Seric 	/*
3307782Seric 	**  Add an Apparently-To: line if we have no recipient lines.
3317782Seric 	*/
3324622Seric 
33355012Seric 	if (hvalue("to", e) == NULL && hvalue("cc", e) == NULL &&
33455012Seric 	    hvalue("bcc", e) == NULL && hvalue("apparently-to", e) == NULL)
3357367Seric 	{
3367367Seric 		register ADDRESS *q;
3377367Seric 
3387367Seric 		/* create an Apparently-To: field */
3397367Seric 		/*    that or reject the message.... */
34055012Seric 		for (q = e->e_sendqueue; q != NULL; q = q->q_next)
3417367Seric 		{
3427389Seric 			if (q->q_alias != NULL)
3437389Seric 				continue;
3447673Seric 			if (tTd(30, 3))
3457367Seric 				printf("Adding Apparently-To: %s\n", q->q_paddr);
34659579Seric 			addheader("Apparently-To", q->q_paddr, e);
3477367Seric 		}
3487367Seric 	}
3497367Seric 
35059320Seric 	/* check for message too large */
35159320Seric 	if (MaxMessageSize > 0 && e->e_msgsize > MaxMessageSize)
35259320Seric 	{
35359320Seric 		usrerr("552 Message exceeds maximum fixed size (%ld)",
35459320Seric 			MaxMessageSize);
35559320Seric 	}
35659320Seric 
35755012Seric 	if ((e->e_dfp = fopen(e->e_df, "r")) == NULL)
35858690Seric 	{
35958690Seric 		/* we haven't acked receipt yet, so just chuck this */
36055012Seric 		syserr("Cannot reopen %s", e->e_df);
36158690Seric 		finis();
36258690Seric 	}
3631392Seric }
3641392Seric /*
36540965Sbostic **  FLUSHEOL -- if not at EOL, throw away rest of input line.
36640965Sbostic **
36740965Sbostic **	Parameters:
36840965Sbostic **		buf -- last line read in (checked for '\n'),
36940965Sbostic **		fp -- file to be read from.
37040965Sbostic **
37140965Sbostic **	Returns:
37240965Sbostic **		FALSE on error from sfgets(), TRUE otherwise.
37340965Sbostic **
37440965Sbostic **	Side Effects:
37540965Sbostic **		none.
37640965Sbostic */
37740965Sbostic 
37840965Sbostic bool
37940965Sbostic flusheol(buf, fp)
38040965Sbostic 	char *buf;
38140965Sbostic 	FILE *fp;
38240965Sbostic {
38340965Sbostic 	register char *p = buf;
38457134Seric 	bool printmsg = TRUE;
38557134Seric 	char junkbuf[MAXLINE];
38640965Sbostic 
38757134Seric 	while (strchr(p, '\n') == NULL)
38857134Seric 	{
38957134Seric 		if (printmsg)
39058151Seric 			usrerr("553 header line too long");
39157134Seric 		printmsg = FALSE;
39261093Seric 		if (sfgets(junkbuf, MAXLINE, fp, TimeOuts.to_datablock,
39361093Seric 				"long line flush") == NULL)
39457134Seric 			return (FALSE);
39540965Sbostic 		p = junkbuf;
39640965Sbostic 	}
39740965Sbostic 
39857134Seric 	return (TRUE);
39940965Sbostic }
40040965Sbostic /*
40111544Seric **  TFERROR -- signal error on writing the temporary file.
40211544Seric **
40311544Seric **	Parameters:
40411544Seric **		tf -- the file pointer for the temporary file.
40511544Seric **
40611544Seric **	Returns:
40711544Seric **		none.
40811544Seric **
40911544Seric **	Side Effects:
41011544Seric **		Gives an error message.
41111544Seric **		Arranges for following output to go elsewhere.
41211544Seric */
41311544Seric 
41455012Seric tferror(tf, e)
41511544Seric 	FILE *tf;
41655012Seric 	register ENVELOPE *e;
41711544Seric {
41811544Seric 	if (errno == ENOSPC)
41911544Seric 	{
42055012Seric 		(void) freopen(e->e_df, "w", tf);
42111544Seric 		fputs("\nMAIL DELETED BECAUSE OF LACK OF DISK SPACE\n\n", tf);
42211544Seric 		usrerr("452 Out of disk space for temp file");
42311544Seric 	}
42411544Seric 	else
42555012Seric 		syserr("collect: Cannot write %s", e->e_df);
42611544Seric 	(void) freopen("/dev/null", "w", tf);
42711544Seric }
42811544Seric /*
4292900Seric **  EATFROM -- chew up a UNIX style from line and process
4302900Seric **
4312900Seric **	This does indeed make some assumptions about the format
4322900Seric **	of UNIX messages.
4332900Seric **
4342900Seric **	Parameters:
4352900Seric **		fm -- the from line.
4362900Seric **
4372900Seric **	Returns:
4382900Seric **		none.
4392900Seric **
4402900Seric **	Side Effects:
4412900Seric **		extracts what information it can from the header,
4423386Seric **		such as the date.
4432900Seric */
4442900Seric 
4454321Seric # ifndef NOTUNIX
4464321Seric 
4474203Seric char	*DowList[] =
4484203Seric {
4494203Seric 	"Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat", NULL
4504203Seric };
4514203Seric 
4522900Seric char	*MonthList[] =
4532900Seric {
4542900Seric 	"Jan", "Feb", "Mar", "Apr", "May", "Jun",
4552900Seric 	"Jul", "Aug", "Sep", "Oct", "Nov", "Dec",
4562900Seric 	NULL
4572900Seric };
4582900Seric 
45955012Seric eatfrom(fm, e)
4602900Seric 	char *fm;
46155012Seric 	register ENVELOPE *e;
4622900Seric {
4632900Seric 	register char *p;
4642900Seric 	register char **dt;
4652900Seric 
4667673Seric 	if (tTd(30, 2))
4674203Seric 		printf("eatfrom(%s)\n", fm);
4684203Seric 
4692900Seric 	/* find the date part */
4702900Seric 	p = fm;
4712900Seric 	while (*p != '\0')
4722900Seric 	{
4732900Seric 		/* skip a word */
4742900Seric 		while (*p != '\0' && *p != ' ')
47516896Seric 			p++;
4762900Seric 		while (*p == ' ')
47716896Seric 			p++;
47858050Seric 		if (!(isascii(*p) && isupper(*p)) ||
47958050Seric 		    p[3] != ' ' || p[13] != ':' || p[16] != ':')
4802900Seric 			continue;
4812900Seric 
4822900Seric 		/* we have a possible date */
4834203Seric 		for (dt = DowList; *dt != NULL; dt++)
4842900Seric 			if (strncmp(*dt, p, 3) == 0)
4852900Seric 				break;
4864203Seric 		if (*dt == NULL)
4874203Seric 			continue;
4882900Seric 
4894203Seric 		for (dt = MonthList; *dt != NULL; dt++)
4904203Seric 			if (strncmp(*dt, &p[4], 3) == 0)
4914203Seric 				break;
4922900Seric 		if (*dt != NULL)
4932900Seric 			break;
4942900Seric 	}
4952900Seric 
49660502Seric 	if (*p != '\0')
4972900Seric 	{
4983386Seric 		char *q;
4995366Seric 		extern char *arpadate();
5003386Seric 
5012900Seric 		/* we have found a date */
5023386Seric 		q = xalloc(25);
50323103Seric 		(void) strncpy(q, p, 25);
5043386Seric 		q[24] = '\0';
5055366Seric 		q = arpadate(q);
50655012Seric 		define('a', newstr(q), e);
5072900Seric 	}
5082900Seric }
5094321Seric 
51056795Seric # endif /* NOTUNIX */
511