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*65580Seric static char sccsid[] = "@(#)collect.c	8.8 (Berkeley) 01/08/94";
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 
734322Seric 	/*
744316Seric 	**  Try to read a UNIX-style From line
754316Seric 	*/
764316Seric 
7761093Seric 	if (sfgets(buf, MAXLINE, InChannel, TimeOuts.to_datablock,
7861093Seric 			"initial message read") == NULL)
7940965Sbostic 		goto readerr;
804557Seric 	fixcrlf(buf, FALSE);
814321Seric # ifndef NOTUNIX
824322Seric 	if (!SaveFrom && strncmp(buf, "From ", 5) == 0)
832900Seric 	{
8440965Sbostic 		if (!flusheol(buf, InChannel))
8540965Sbostic 			goto readerr;
8655012Seric 		eatfrom(buf, e);
8761093Seric 		if (sfgets(buf, MAXLINE, InChannel, TimeOuts.to_datablock,
8861093Seric 				"message header read") == NULL)
8940965Sbostic 			goto readerr;
904557Seric 		fixcrlf(buf, FALSE);
912900Seric 	}
9256795Seric # endif /* NOTUNIX */
932900Seric 
941392Seric 	/*
955975Seric 	**  Copy InChannel to temp file & do message editing.
961392Seric 	**	To keep certain mailers from getting confused,
971392Seric 	**	and to keep the output clean, lines that look
9813932Seric 	**	like UNIX "From" lines are deleted in the header.
991392Seric 	*/
1001392Seric 
10140965Sbostic 	workbuf = buf;		/* `workbuf' contains a header field */
10240965Sbostic 	freebuf = buf2;		/* `freebuf' can be used for read-ahead */
10340965Sbostic 	for (;;)
1041392Seric 	{
10557135Seric 		char *curbuf;
10657135Seric 		int curbuffree;
10757135Seric 		register int curbuflen;
10857135Seric 		char *p;
10957135Seric 
11040965Sbostic 		/* first, see if the header is over */
11140965Sbostic 		if (!isheader(workbuf))
11240965Sbostic 		{
11340965Sbostic 			fixcrlf(workbuf, TRUE);
11419036Seric 			break;
11540965Sbostic 		}
11619036Seric 
1177681Seric 		/* if the line is too long, throw the rest away */
11840965Sbostic 		if (!flusheol(workbuf, InChannel))
11940965Sbostic 			goto readerr;
1207681Seric 
12140965Sbostic 		/* it's okay to toss '\n' now (flusheol() needed it) */
12240965Sbostic 		fixcrlf(workbuf, TRUE);
1234557Seric 
12457135Seric 		curbuf = workbuf;
12557135Seric 		curbuflen = strlen(curbuf);
12657135Seric 		curbuffree = MAXLINE - curbuflen;
12757135Seric 		p = curbuf + curbuflen;
1282900Seric 
1292900Seric 		/* get the rest of this field */
13040965Sbostic 		for (;;)
1311392Seric 		{
13257135Seric 			int clen;
13357135Seric 
13461093Seric 			if (sfgets(freebuf, MAXLINE, InChannel,
13561093Seric 					TimeOuts.to_datablock,
13661093Seric 					"message header read") == NULL)
13764916Seric 			{
13864916Seric 				freebuf[0] = '\0';
13964916Seric 				break;
14064916Seric 			}
14140965Sbostic 
14240965Sbostic 			/* is this a continuation line? */
14340965Sbostic 			if (*freebuf != ' ' && *freebuf != '\t')
1442900Seric 				break;
14540965Sbostic 
14640965Sbostic 			if (!flusheol(freebuf, InChannel))
14740965Sbostic 				goto readerr;
14840965Sbostic 
14957135Seric 			fixcrlf(freebuf, TRUE);
15057135Seric 			clen = strlen(freebuf) + 1;
15157135Seric 
15257135Seric 			/* if insufficient room, dynamically allocate buffer */
15357135Seric 			if (clen >= curbuffree)
15440965Sbostic 			{
15557135Seric 				/* reallocate buffer */
15657135Seric 				int nbuflen = ((p - curbuf) + clen) * 2;
15757135Seric 				char *nbuf = xalloc(nbuflen);
15840965Sbostic 
15957135Seric 				p = nbuf + curbuflen;
16057135Seric 				curbuffree = nbuflen - curbuflen;
16157135Seric 				bcopy(curbuf, nbuf, curbuflen);
16257135Seric 				if (curbuf != buf && curbuf != buf2)
16357135Seric 					free(curbuf);
16457135Seric 				curbuf = nbuf;
16540965Sbostic 			}
16657135Seric 			*p++ = '\n';
16757135Seric 			bcopy(freebuf, p, clen - 1);
16857135Seric 			p += clen - 1;
16957135Seric 			curbuffree -= clen;
17057135Seric 			curbuflen += clen;
1711392Seric 		}
17257135Seric 		*p++ = '\0';
1731392Seric 
17457135Seric 		e->e_msgsize += curbuflen;
1751392Seric 
1762900Seric 		/*
17740965Sbostic 		**  The working buffer now becomes the free buffer, since
17840965Sbostic 		**  the free buffer contains a new header field.
17940965Sbostic 		**
18040965Sbostic 		**  This is premature, since we still havent called
18140965Sbostic 		**  chompheader() to process the field we just created
18240965Sbostic 		**  (so the call to chompheader() will use `freebuf').
18340965Sbostic 		**  This convolution is necessary so that if we break out
18440965Sbostic 		**  of the loop due to H_EOH, `workbuf' will always be
18540965Sbostic 		**  the next unprocessed buffer.
18640965Sbostic 		*/
18740965Sbostic 
18840965Sbostic 		{
18940965Sbostic 			register char *tmp = workbuf;
19040965Sbostic 			workbuf = freebuf;
19140965Sbostic 			freebuf = tmp;
19240965Sbostic 		}
19340965Sbostic 
19440965Sbostic 		/*
1952900Seric 		**  Snarf header away.
1962900Seric 		*/
1972900Seric 
19857135Seric 		if (bitset(H_EOH, chompheader(curbuf, FALSE, e)))
1993058Seric 			break;
20057135Seric 
20157135Seric 		/*
20257135Seric 		**  If the buffer was dynamically allocated, free it.
20357135Seric 		*/
20457135Seric 
20557135Seric 		if (curbuf != buf && curbuf != buf2)
20657135Seric 			free(curbuf);
20740965Sbostic 	}
2081392Seric 
2097673Seric 	if (tTd(30, 1))
2102900Seric 		printf("EOH\n");
2112900Seric 
21240965Sbostic 	if (*workbuf == '\0')
21340965Sbostic 	{
21440965Sbostic 		/* throw away a blank line */
21561093Seric 		if (sfgets(buf, MAXLINE, InChannel, TimeOuts.to_datablock,
21661093Seric 				"message separator read") == NULL)
21740965Sbostic 			goto readerr;
21840965Sbostic 	}
21940965Sbostic 	else if (workbuf == buf2)	/* guarantee `buf' contains data */
22040965Sbostic 		(void) strcpy(buf, buf2);
2212900Seric 
2222900Seric 	/*
2232900Seric 	**  Collect the body of the message.
2242900Seric 	*/
2252900Seric 
22664718Seric 	for (;;)
2272900Seric 	{
2284551Seric 		register char *bp = buf;
2294156Seric 
2307852Seric 		fixcrlf(buf, TRUE);
2314557Seric 
2322900Seric 		/* check for end-of-message */
23352105Seric 		if (!ignrdot && buf[0] == '.' && (buf[1] == '\n' || buf[1] == '\0'))
2342900Seric 			break;
2352900Seric 
2364551Seric 		/* check for transparent dot */
237*65580Seric 		if ((OpMode == MD_SMTP || OpMode == MD_DAEMON) &&
238*65580Seric 		    bp[0] == '.' && bp[1] == '.')
2394551Seric 			bp++;
2404551Seric 
2414156Seric 		/*
2424156Seric 		**  Figure message length, output the line to the temp
2434156Seric 		**  file, and insert a newline if missing.
2444156Seric 		*/
2454156Seric 
24655012Seric 		e->e_msgsize += strlen(bp) + 1;
2474551Seric 		fputs(bp, tf);
2487852Seric 		fputs("\n", tf);
2491392Seric 		if (ferror(tf))
25055012Seric 			tferror(tf, e);
25164718Seric 		if (sfgets(buf, MAXLINE, InChannel, TimeOuts.to_datablock,
25264718Seric 				"message body read") == NULL)
25364718Seric 			goto readerr;
25464718Seric 	}
25540965Sbostic 
25664718Seric 	if (feof(InChannel) || ferror(InChannel))
25764718Seric 	{
25840965Sbostic readerr:
25964916Seric 		if (tTd(30, 1))
26064916Seric 			printf("collect: read error\n");
26164718Seric 		inputerr = TRUE;
26264718Seric 	}
26364718Seric 
26411544Seric 	if (fflush(tf) != 0)
26555012Seric 		tferror(tf, e);
26664762Seric 	if (fsync(fileno(tf)) < 0 || fclose(tf) < 0)
26764762Seric 	{
26864762Seric 		syserr("cannot sync message data to disk (%s)", e->e_df);
26964762Seric 		finis();
27064762Seric 	}
2712900Seric 
27211145Seric 	/* An EOF when running SMTP is an error */
273*65580Seric 	if (inputerr && (OpMode == MD_SMTP || OpMode == MD_DAEMON))
27416136Seric 	{
27558308Seric 		char *host;
27664718Seric 		char *problem;
27758082Seric 
27858308Seric 		host = RealHostName;
27958308Seric 		if (host == NULL)
28058308Seric 			host = "localhost";
28158308Seric 
28264718Seric 		if (feof(InChannel))
28364718Seric 			problem = "unexpected close";
28464718Seric 		else if (ferror(InChannel))
28564718Seric 			problem = "I/O error";
28664718Seric 		else
28764718Seric 			problem = "read timeout";
28836233Skarels # ifdef LOG
28958308Seric 		if (LogLevel > 0 && feof(InChannel))
29036230Skarels 			syslog(LOG_NOTICE,
29164718Seric 			    "collect: %s on connection from %s, sender=%s: %m\n",
29264718Seric 			    problem, host, e->e_from.q_paddr);
29336233Skarels # endif
29458082Seric 		(feof(InChannel) ? usrerr : syserr)
29564718Seric 			("451 collect: %s on connection from %s, from=%s",
29664718Seric 				problem, host, e->e_from.q_paddr);
29711145Seric 
29816136Seric 		/* don't return an error indication */
29955012Seric 		e->e_to = NULL;
30055012Seric 		e->e_flags &= ~EF_FATALERRS;
30164124Seric 		e->e_flags |= EF_CLRQUEUE;
30216136Seric 
30316136Seric 		/* and don't try to deliver the partial message either */
30464718Seric 		if (InChild)
30564718Seric 			ExitStat = EX_QUIT;
30616136Seric 		finis();
30716136Seric 	}
30816136Seric 
3092900Seric 	/*
3102900Seric 	**  Find out some information from the headers.
3113386Seric 	**	Examples are who is the from person & the date.
3122900Seric 	*/
3132900Seric 
31458929Seric 	eatheader(e, !requeueflag);
3157673Seric 
31664068Seric 	/* collect statistics */
31764068Seric 	if (OpMode != MD_VERIFY)
31864068Seric 		markstats(e, (ADDRESS *) NULL);
31964068Seric 
3207782Seric 	/*
3217782Seric 	**  Add an Apparently-To: line if we have no recipient lines.
3227782Seric 	*/
3234622Seric 
32455012Seric 	if (hvalue("to", e) == NULL && hvalue("cc", e) == NULL &&
32555012Seric 	    hvalue("bcc", e) == NULL && hvalue("apparently-to", e) == NULL)
3267367Seric 	{
3277367Seric 		register ADDRESS *q;
3287367Seric 
3297367Seric 		/* create an Apparently-To: field */
3307367Seric 		/*    that or reject the message.... */
33155012Seric 		for (q = e->e_sendqueue; q != NULL; q = q->q_next)
3327367Seric 		{
3337389Seric 			if (q->q_alias != NULL)
3347389Seric 				continue;
3357673Seric 			if (tTd(30, 3))
3367367Seric 				printf("Adding Apparently-To: %s\n", q->q_paddr);
33759579Seric 			addheader("Apparently-To", q->q_paddr, e);
3387367Seric 		}
3397367Seric 	}
3407367Seric 
34159320Seric 	/* check for message too large */
34259320Seric 	if (MaxMessageSize > 0 && e->e_msgsize > MaxMessageSize)
34359320Seric 	{
34459320Seric 		usrerr("552 Message exceeds maximum fixed size (%ld)",
34559320Seric 			MaxMessageSize);
34659320Seric 	}
34759320Seric 
34855012Seric 	if ((e->e_dfp = fopen(e->e_df, "r")) == NULL)
34958690Seric 	{
35058690Seric 		/* we haven't acked receipt yet, so just chuck this */
35155012Seric 		syserr("Cannot reopen %s", e->e_df);
35258690Seric 		finis();
35358690Seric 	}
3541392Seric }
3551392Seric /*
35640965Sbostic **  FLUSHEOL -- if not at EOL, throw away rest of input line.
35740965Sbostic **
35840965Sbostic **	Parameters:
35940965Sbostic **		buf -- last line read in (checked for '\n'),
36040965Sbostic **		fp -- file to be read from.
36140965Sbostic **
36240965Sbostic **	Returns:
36340965Sbostic **		FALSE on error from sfgets(), TRUE otherwise.
36440965Sbostic **
36540965Sbostic **	Side Effects:
36640965Sbostic **		none.
36740965Sbostic */
36840965Sbostic 
36940965Sbostic bool
37040965Sbostic flusheol(buf, fp)
37140965Sbostic 	char *buf;
37240965Sbostic 	FILE *fp;
37340965Sbostic {
37440965Sbostic 	register char *p = buf;
37557134Seric 	bool printmsg = TRUE;
37657134Seric 	char junkbuf[MAXLINE];
37740965Sbostic 
37857134Seric 	while (strchr(p, '\n') == NULL)
37957134Seric 	{
38057134Seric 		if (printmsg)
38158151Seric 			usrerr("553 header line too long");
38257134Seric 		printmsg = FALSE;
38361093Seric 		if (sfgets(junkbuf, MAXLINE, fp, TimeOuts.to_datablock,
38461093Seric 				"long line flush") == NULL)
38557134Seric 			return (FALSE);
38640965Sbostic 		p = junkbuf;
38740965Sbostic 	}
38840965Sbostic 
38957134Seric 	return (TRUE);
39040965Sbostic }
39140965Sbostic /*
39211544Seric **  TFERROR -- signal error on writing the temporary file.
39311544Seric **
39411544Seric **	Parameters:
39511544Seric **		tf -- the file pointer for the temporary file.
39611544Seric **
39711544Seric **	Returns:
39811544Seric **		none.
39911544Seric **
40011544Seric **	Side Effects:
40111544Seric **		Gives an error message.
40211544Seric **		Arranges for following output to go elsewhere.
40311544Seric */
40411544Seric 
40555012Seric tferror(tf, e)
40611544Seric 	FILE *tf;
40755012Seric 	register ENVELOPE *e;
40811544Seric {
40911544Seric 	if (errno == ENOSPC)
41011544Seric 	{
41155012Seric 		(void) freopen(e->e_df, "w", tf);
41211544Seric 		fputs("\nMAIL DELETED BECAUSE OF LACK OF DISK SPACE\n\n", tf);
41311544Seric 		usrerr("452 Out of disk space for temp file");
41411544Seric 	}
41511544Seric 	else
41655012Seric 		syserr("collect: Cannot write %s", e->e_df);
41711544Seric 	(void) freopen("/dev/null", "w", tf);
41811544Seric }
41911544Seric /*
4202900Seric **  EATFROM -- chew up a UNIX style from line and process
4212900Seric **
4222900Seric **	This does indeed make some assumptions about the format
4232900Seric **	of UNIX messages.
4242900Seric **
4252900Seric **	Parameters:
4262900Seric **		fm -- the from line.
4272900Seric **
4282900Seric **	Returns:
4292900Seric **		none.
4302900Seric **
4312900Seric **	Side Effects:
4322900Seric **		extracts what information it can from the header,
4333386Seric **		such as the date.
4342900Seric */
4352900Seric 
4364321Seric # ifndef NOTUNIX
4374321Seric 
4384203Seric char	*DowList[] =
4394203Seric {
4404203Seric 	"Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat", NULL
4414203Seric };
4424203Seric 
4432900Seric char	*MonthList[] =
4442900Seric {
4452900Seric 	"Jan", "Feb", "Mar", "Apr", "May", "Jun",
4462900Seric 	"Jul", "Aug", "Sep", "Oct", "Nov", "Dec",
4472900Seric 	NULL
4482900Seric };
4492900Seric 
45055012Seric eatfrom(fm, e)
4512900Seric 	char *fm;
45255012Seric 	register ENVELOPE *e;
4532900Seric {
4542900Seric 	register char *p;
4552900Seric 	register char **dt;
4562900Seric 
4577673Seric 	if (tTd(30, 2))
4584203Seric 		printf("eatfrom(%s)\n", fm);
4594203Seric 
4602900Seric 	/* find the date part */
4612900Seric 	p = fm;
4622900Seric 	while (*p != '\0')
4632900Seric 	{
4642900Seric 		/* skip a word */
4652900Seric 		while (*p != '\0' && *p != ' ')
46616896Seric 			p++;
4672900Seric 		while (*p == ' ')
46816896Seric 			p++;
46958050Seric 		if (!(isascii(*p) && isupper(*p)) ||
47058050Seric 		    p[3] != ' ' || p[13] != ':' || p[16] != ':')
4712900Seric 			continue;
4722900Seric 
4732900Seric 		/* we have a possible date */
4744203Seric 		for (dt = DowList; *dt != NULL; dt++)
4752900Seric 			if (strncmp(*dt, p, 3) == 0)
4762900Seric 				break;
4774203Seric 		if (*dt == NULL)
4784203Seric 			continue;
4792900Seric 
4804203Seric 		for (dt = MonthList; *dt != NULL; dt++)
4814203Seric 			if (strncmp(*dt, &p[4], 3) == 0)
4824203Seric 				break;
4832900Seric 		if (*dt != NULL)
4842900Seric 			break;
4852900Seric 	}
4862900Seric 
48760502Seric 	if (*p != '\0')
4882900Seric 	{
4893386Seric 		char *q;
4905366Seric 		extern char *arpadate();
4913386Seric 
4922900Seric 		/* we have found a date */
4933386Seric 		q = xalloc(25);
49423103Seric 		(void) strncpy(q, p, 25);
4953386Seric 		q[24] = '\0';
4965366Seric 		q = arpadate(q);
49755012Seric 		define('a', newstr(q), e);
4982900Seric 	}
4992900Seric }
5004321Seric 
50156795Seric # endif /* NOTUNIX */
502