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*66864Seric static char sccsid[] = "@(#)collect.c	8.13 (Berkeley) 04/18/94";
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 
4066796Seric char	*CollectErrorMessage;
4166796Seric bool	CollectErrno;
4266796Seric 
4358929Seric collect(smtpmode, requeueflag, e)
4452105Seric 	bool smtpmode;
4558929Seric 	bool requeueflag;
4655012Seric 	register ENVELOPE *e;
471392Seric {
481392Seric 	register FILE *tf;
4952105Seric 	bool ignrdot = smtpmode ? FALSE : IgnrDot;
5057135Seric 	char buf[MAXLINE], buf2[MAXLINE];
5140965Sbostic 	register char *workbuf, *freebuf;
5264718Seric 	bool inputerr = FALSE;
532900Seric 	extern char *hvalue();
5440965Sbostic 	extern bool isheader(), flusheol();
551392Seric 
5666796Seric 	CollectErrorMessage = NULL;
5766796Seric 	CollectErrno = 0;
5866796Seric 
591392Seric 	/*
601392Seric 	**  Create the temp file name and create the file.
611392Seric 	*/
621392Seric 
6364086Seric 	e->e_df = queuename(e, 'd');
6464086Seric 	e->e_df = newstr(e->e_df);
6559745Seric 	if ((tf = dfopen(e->e_df, O_WRONLY|O_CREAT, FileMode)) == NULL)
661392Seric 	{
6755012Seric 		syserr("Cannot create %s", e->e_df);
685366Seric 		NoReturn = TRUE;
695366Seric 		finis();
701392Seric 	}
711392Seric 
724316Seric 	/*
734322Seric 	**  Tell ARPANET to go ahead.
744322Seric 	*/
754322Seric 
7652105Seric 	if (smtpmode)
7758151Seric 		message("354 Enter mail, end with \".\" on a line by itself");
784322Seric 
7966765Seric 	/* set global timer to monitor progress */
8066765Seric 	sfgetset(TimeOuts.to_datablock);
8166765Seric 
824322Seric 	/*
834316Seric 	**  Try to read a UNIX-style From line
844316Seric 	*/
854316Seric 
8661093Seric 	if (sfgets(buf, MAXLINE, InChannel, TimeOuts.to_datablock,
8761093Seric 			"initial message read") == NULL)
8840965Sbostic 		goto readerr;
894557Seric 	fixcrlf(buf, FALSE);
904321Seric # ifndef NOTUNIX
914322Seric 	if (!SaveFrom && strncmp(buf, "From ", 5) == 0)
922900Seric 	{
9340965Sbostic 		if (!flusheol(buf, InChannel))
9440965Sbostic 			goto readerr;
9555012Seric 		eatfrom(buf, e);
9661093Seric 		if (sfgets(buf, MAXLINE, InChannel, TimeOuts.to_datablock,
9761093Seric 				"message header read") == NULL)
9840965Sbostic 			goto readerr;
994557Seric 		fixcrlf(buf, FALSE);
1002900Seric 	}
10156795Seric # endif /* NOTUNIX */
1022900Seric 
1031392Seric 	/*
1045975Seric 	**  Copy InChannel to temp file & do message editing.
1051392Seric 	**	To keep certain mailers from getting confused,
1061392Seric 	**	and to keep the output clean, lines that look
10713932Seric 	**	like UNIX "From" lines are deleted in the header.
1081392Seric 	*/
1091392Seric 
11040965Sbostic 	workbuf = buf;		/* `workbuf' contains a header field */
11140965Sbostic 	freebuf = buf2;		/* `freebuf' can be used for read-ahead */
11240965Sbostic 	for (;;)
1131392Seric 	{
11457135Seric 		char *curbuf;
11557135Seric 		int curbuffree;
11657135Seric 		register int curbuflen;
11757135Seric 		char *p;
11857135Seric 
11940965Sbostic 		/* first, see if the header is over */
12040965Sbostic 		if (!isheader(workbuf))
12140965Sbostic 		{
12240965Sbostic 			fixcrlf(workbuf, TRUE);
12319036Seric 			break;
12440965Sbostic 		}
12519036Seric 
1267681Seric 		/* if the line is too long, throw the rest away */
12740965Sbostic 		if (!flusheol(workbuf, InChannel))
12840965Sbostic 			goto readerr;
1297681Seric 
13040965Sbostic 		/* it's okay to toss '\n' now (flusheol() needed it) */
13140965Sbostic 		fixcrlf(workbuf, TRUE);
1324557Seric 
13357135Seric 		curbuf = workbuf;
13457135Seric 		curbuflen = strlen(curbuf);
13557135Seric 		curbuffree = MAXLINE - curbuflen;
13657135Seric 		p = curbuf + curbuflen;
1372900Seric 
1382900Seric 		/* get the rest of this field */
13940965Sbostic 		for (;;)
1401392Seric 		{
14157135Seric 			int clen;
14257135Seric 
14361093Seric 			if (sfgets(freebuf, MAXLINE, InChannel,
14461093Seric 					TimeOuts.to_datablock,
14561093Seric 					"message header read") == NULL)
14664916Seric 			{
14764916Seric 				freebuf[0] = '\0';
14864916Seric 				break;
14964916Seric 			}
15040965Sbostic 
15140965Sbostic 			/* is this a continuation line? */
15240965Sbostic 			if (*freebuf != ' ' && *freebuf != '\t')
1532900Seric 				break;
15440965Sbostic 
15540965Sbostic 			if (!flusheol(freebuf, InChannel))
15640965Sbostic 				goto readerr;
15740965Sbostic 
15857135Seric 			fixcrlf(freebuf, TRUE);
15957135Seric 			clen = strlen(freebuf) + 1;
16057135Seric 
16157135Seric 			/* if insufficient room, dynamically allocate buffer */
16257135Seric 			if (clen >= curbuffree)
16340965Sbostic 			{
16457135Seric 				/* reallocate buffer */
16557135Seric 				int nbuflen = ((p - curbuf) + clen) * 2;
16657135Seric 				char *nbuf = xalloc(nbuflen);
16740965Sbostic 
16857135Seric 				p = nbuf + curbuflen;
16957135Seric 				curbuffree = nbuflen - curbuflen;
17057135Seric 				bcopy(curbuf, nbuf, curbuflen);
17157135Seric 				if (curbuf != buf && curbuf != buf2)
17257135Seric 					free(curbuf);
17357135Seric 				curbuf = nbuf;
17440965Sbostic 			}
17557135Seric 			*p++ = '\n';
17657135Seric 			bcopy(freebuf, p, clen - 1);
17757135Seric 			p += clen - 1;
17857135Seric 			curbuffree -= clen;
17957135Seric 			curbuflen += clen;
1801392Seric 		}
18157135Seric 		*p++ = '\0';
1821392Seric 
18357135Seric 		e->e_msgsize += curbuflen;
1841392Seric 
1852900Seric 		/*
18640965Sbostic 		**  The working buffer now becomes the free buffer, since
18740965Sbostic 		**  the free buffer contains a new header field.
18840965Sbostic 		**
18940965Sbostic 		**  This is premature, since we still havent called
19040965Sbostic 		**  chompheader() to process the field we just created
19140965Sbostic 		**  (so the call to chompheader() will use `freebuf').
19240965Sbostic 		**  This convolution is necessary so that if we break out
19340965Sbostic 		**  of the loop due to H_EOH, `workbuf' will always be
19440965Sbostic 		**  the next unprocessed buffer.
19540965Sbostic 		*/
19640965Sbostic 
19740965Sbostic 		{
19840965Sbostic 			register char *tmp = workbuf;
19940965Sbostic 			workbuf = freebuf;
20040965Sbostic 			freebuf = tmp;
20140965Sbostic 		}
20240965Sbostic 
20340965Sbostic 		/*
2042900Seric 		**  Snarf header away.
2052900Seric 		*/
2062900Seric 
20757135Seric 		if (bitset(H_EOH, chompheader(curbuf, FALSE, e)))
2083058Seric 			break;
20957135Seric 
21057135Seric 		/*
21157135Seric 		**  If the buffer was dynamically allocated, free it.
21257135Seric 		*/
21357135Seric 
21457135Seric 		if (curbuf != buf && curbuf != buf2)
21557135Seric 			free(curbuf);
21640965Sbostic 	}
2171392Seric 
2187673Seric 	if (tTd(30, 1))
2192900Seric 		printf("EOH\n");
2202900Seric 
22140965Sbostic 	if (*workbuf == '\0')
22240965Sbostic 	{
22340965Sbostic 		/* throw away a blank line */
22461093Seric 		if (sfgets(buf, MAXLINE, InChannel, TimeOuts.to_datablock,
22561093Seric 				"message separator read") == NULL)
22640965Sbostic 			goto readerr;
22740965Sbostic 	}
22840965Sbostic 	else if (workbuf == buf2)	/* guarantee `buf' contains data */
22940965Sbostic 		(void) strcpy(buf, buf2);
2302900Seric 
2312900Seric 	/*
2322900Seric 	**  Collect the body of the message.
2332900Seric 	*/
2342900Seric 
23564718Seric 	for (;;)
2362900Seric 	{
2374551Seric 		register char *bp = buf;
2384156Seric 
2397852Seric 		fixcrlf(buf, TRUE);
2404557Seric 
2412900Seric 		/* check for end-of-message */
24252105Seric 		if (!ignrdot && buf[0] == '.' && (buf[1] == '\n' || buf[1] == '\0'))
2432900Seric 			break;
2442900Seric 
2454551Seric 		/* check for transparent dot */
24665580Seric 		if ((OpMode == MD_SMTP || OpMode == MD_DAEMON) &&
24765580Seric 		    bp[0] == '.' && bp[1] == '.')
2484551Seric 			bp++;
2494551Seric 
2504156Seric 		/*
2514156Seric 		**  Figure message length, output the line to the temp
2524156Seric 		**  file, and insert a newline if missing.
2534156Seric 		*/
2544156Seric 
25555012Seric 		e->e_msgsize += strlen(bp) + 1;
2564551Seric 		fputs(bp, tf);
2577852Seric 		fputs("\n", tf);
2581392Seric 		if (ferror(tf))
25955012Seric 			tferror(tf, e);
26064718Seric 		if (sfgets(buf, MAXLINE, InChannel, TimeOuts.to_datablock,
26164718Seric 				"message body read") == NULL)
26264718Seric 			goto readerr;
26364718Seric 	}
26440965Sbostic 
26564718Seric 	if (feof(InChannel) || ferror(InChannel))
26664718Seric 	{
26740965Sbostic readerr:
26864916Seric 		if (tTd(30, 1))
26964916Seric 			printf("collect: read error\n");
27064718Seric 		inputerr = TRUE;
27164718Seric 	}
27264718Seric 
27366765Seric 	/* reset global timer */
27466765Seric 	sfgetset((time_t) 0);
27566765Seric 
27611544Seric 	if (fflush(tf) != 0)
27755012Seric 		tferror(tf, e);
27864762Seric 	if (fsync(fileno(tf)) < 0 || fclose(tf) < 0)
27964762Seric 	{
28066796Seric 		tferror(tf, e);
28164762Seric 		finis();
28264762Seric 	}
2832900Seric 
28466796Seric 	if (CollectErrorMessage != NULL && Errors <= 0)
28516136Seric 	{
28666796Seric 		if (CollectErrno != 0)
28766796Seric 		{
28866796Seric 			errno = CollectErrno;
28966796Seric 			syserr(CollectErrorMessage, e->e_df);
29066796Seric 			finis();
29166796Seric 		}
29266796Seric 		usrerr(CollectErrorMessage);
29366796Seric 	}
29466796Seric 	else if (inputerr && (OpMode == MD_SMTP || OpMode == MD_DAEMON))
29566796Seric 	{
29666796Seric 		/* An EOF when running SMTP is an error */
29758308Seric 		char *host;
29864718Seric 		char *problem;
29958082Seric 
30058308Seric 		host = RealHostName;
30158308Seric 		if (host == NULL)
30258308Seric 			host = "localhost";
30358308Seric 
30464718Seric 		if (feof(InChannel))
30564718Seric 			problem = "unexpected close";
30664718Seric 		else if (ferror(InChannel))
30764718Seric 			problem = "I/O error";
30864718Seric 		else
30964718Seric 			problem = "read timeout";
31036233Skarels # ifdef LOG
31158308Seric 		if (LogLevel > 0 && feof(InChannel))
31236230Skarels 			syslog(LOG_NOTICE,
313*66864Seric 			    "collect: %s on connection from %s, sender=%s: %s\n",
314*66864Seric 			    problem, host, e->e_from.q_paddr, errstring(errno));
31536233Skarels # endif
31665951Seric 		if (feof(InChannel))
31765951Seric 			usrerr("451 collect: %s on connection from %s, from=%s",
31864718Seric 				problem, host, e->e_from.q_paddr);
31965951Seric 		else
32065951Seric 			syserr("451 collect: %s on connection from %s, from=%s",
32165951Seric 				problem, host, e->e_from.q_paddr);
32211145Seric 
32316136Seric 		/* don't return an error indication */
32455012Seric 		e->e_to = NULL;
32555012Seric 		e->e_flags &= ~EF_FATALERRS;
32664124Seric 		e->e_flags |= EF_CLRQUEUE;
32716136Seric 
32816136Seric 		/* and don't try to deliver the partial message either */
32964718Seric 		if (InChild)
33064718Seric 			ExitStat = EX_QUIT;
33116136Seric 		finis();
33216136Seric 	}
33316136Seric 
3342900Seric 	/*
3352900Seric 	**  Find out some information from the headers.
3363386Seric 	**	Examples are who is the from person & the date.
3372900Seric 	*/
3382900Seric 
33958929Seric 	eatheader(e, !requeueflag);
3407673Seric 
34164068Seric 	/* collect statistics */
34264068Seric 	if (OpMode != MD_VERIFY)
34364068Seric 		markstats(e, (ADDRESS *) NULL);
34464068Seric 
3457782Seric 	/*
3467782Seric 	**  Add an Apparently-To: line if we have no recipient lines.
3477782Seric 	*/
3484622Seric 
34955012Seric 	if (hvalue("to", e) == NULL && hvalue("cc", e) == NULL &&
35055012Seric 	    hvalue("bcc", e) == NULL && hvalue("apparently-to", e) == NULL)
3517367Seric 	{
3527367Seric 		register ADDRESS *q;
3537367Seric 
3547367Seric 		/* create an Apparently-To: field */
3557367Seric 		/*    that or reject the message.... */
35655012Seric 		for (q = e->e_sendqueue; q != NULL; q = q->q_next)
3577367Seric 		{
3587389Seric 			if (q->q_alias != NULL)
3597389Seric 				continue;
3607673Seric 			if (tTd(30, 3))
3617367Seric 				printf("Adding Apparently-To: %s\n", q->q_paddr);
36259579Seric 			addheader("Apparently-To", q->q_paddr, e);
3637367Seric 		}
3647367Seric 	}
3657367Seric 
36659320Seric 	/* check for message too large */
36759320Seric 	if (MaxMessageSize > 0 && e->e_msgsize > MaxMessageSize)
36859320Seric 	{
36959320Seric 		usrerr("552 Message exceeds maximum fixed size (%ld)",
37059320Seric 			MaxMessageSize);
37159320Seric 	}
37259320Seric 
37355012Seric 	if ((e->e_dfp = fopen(e->e_df, "r")) == NULL)
37458690Seric 	{
37558690Seric 		/* we haven't acked receipt yet, so just chuck this */
37655012Seric 		syserr("Cannot reopen %s", e->e_df);
37758690Seric 		finis();
37858690Seric 	}
3791392Seric }
3801392Seric /*
38140965Sbostic **  FLUSHEOL -- if not at EOL, throw away rest of input line.
38240965Sbostic **
38340965Sbostic **	Parameters:
38440965Sbostic **		buf -- last line read in (checked for '\n'),
38540965Sbostic **		fp -- file to be read from.
38640965Sbostic **
38740965Sbostic **	Returns:
38840965Sbostic **		FALSE on error from sfgets(), TRUE otherwise.
38940965Sbostic **
39040965Sbostic **	Side Effects:
39140965Sbostic **		none.
39240965Sbostic */
39340965Sbostic 
39440965Sbostic bool
39540965Sbostic flusheol(buf, fp)
39640965Sbostic 	char *buf;
39740965Sbostic 	FILE *fp;
39840965Sbostic {
39940965Sbostic 	register char *p = buf;
40057134Seric 	char junkbuf[MAXLINE];
40140965Sbostic 
40257134Seric 	while (strchr(p, '\n') == NULL)
40357134Seric 	{
40466796Seric 		CollectErrorMessage = "553 header line too long";
40566796Seric 		CollectErrno = 0;
40661093Seric 		if (sfgets(junkbuf, MAXLINE, fp, TimeOuts.to_datablock,
40761093Seric 				"long line flush") == NULL)
40857134Seric 			return (FALSE);
40940965Sbostic 		p = junkbuf;
41040965Sbostic 	}
41140965Sbostic 
41257134Seric 	return (TRUE);
41340965Sbostic }
41440965Sbostic /*
41511544Seric **  TFERROR -- signal error on writing the temporary file.
41611544Seric **
41711544Seric **	Parameters:
41811544Seric **		tf -- the file pointer for the temporary file.
41911544Seric **
42011544Seric **	Returns:
42111544Seric **		none.
42211544Seric **
42311544Seric **	Side Effects:
42411544Seric **		Gives an error message.
42511544Seric **		Arranges for following output to go elsewhere.
42611544Seric */
42711544Seric 
42855012Seric tferror(tf, e)
42911544Seric 	FILE *tf;
43055012Seric 	register ENVELOPE *e;
43111544Seric {
43266796Seric 	CollectErrno = errno;
43311544Seric 	if (errno == ENOSPC)
43411544Seric 	{
43566782Seric 		struct stat st;
43666782Seric 		long avail;
43766782Seric 		long bsize;
43866782Seric 
43966782Seric 		NoReturn = TRUE;
44066782Seric 		if (fstat(fileno(tf), &st) < 0)
44166782Seric 			st.st_size = 0;
44255012Seric 		(void) freopen(e->e_df, "w", tf);
44366782Seric 		if (st.st_size <= 0)
44466782Seric 			fprintf(tf, "\n*** Mail could not be accepted");
44566782Seric 		else if (sizeof st.st_size > sizeof (long))
44666782Seric 			fprintf(tf, "\n*** Mail of at least %qd bytes could not be accepted\n",
44766782Seric 				st.st_size);
44866782Seric 		else
44966782Seric 			fprintf(tf, "\n*** Mail of at least %ld bytes could not be accepted\n",
45066782Seric 				st.st_size);
45166782Seric 		fprintf(tf, "*** at %s due to lack of disk space for temp file.\n",
45266782Seric 			MyHostName);
45366782Seric 		avail = freespace(QueueDir, &bsize);
45466782Seric 		if (avail > 0)
45566782Seric 		{
45666782Seric 			if (bsize > 1024)
45766782Seric 				avail *= bsize / 1024;
45866782Seric 			else if (bsize < 1024)
45966782Seric 				avail /= 1024 / bsize;
46066782Seric 			fprintf(tf, "*** Currently, %ld kilobytes are available for mail temp files.\n",
46166782Seric 				avail);
46266782Seric 		}
46366796Seric 		CollectErrorMessage = "452 Out of disk space for temp file";
46411544Seric 	}
46511544Seric 	else
46666796Seric 	{
46766796Seric 		CollectErrorMessage = "cannot write message body to disk (%s)";
46866796Seric 	}
46911544Seric 	(void) freopen("/dev/null", "w", tf);
47011544Seric }
47111544Seric /*
4722900Seric **  EATFROM -- chew up a UNIX style from line and process
4732900Seric **
4742900Seric **	This does indeed make some assumptions about the format
4752900Seric **	of UNIX messages.
4762900Seric **
4772900Seric **	Parameters:
4782900Seric **		fm -- the from line.
4792900Seric **
4802900Seric **	Returns:
4812900Seric **		none.
4822900Seric **
4832900Seric **	Side Effects:
4842900Seric **		extracts what information it can from the header,
4853386Seric **		such as the date.
4862900Seric */
4872900Seric 
4884321Seric # ifndef NOTUNIX
4894321Seric 
4904203Seric char	*DowList[] =
4914203Seric {
4924203Seric 	"Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat", NULL
4934203Seric };
4944203Seric 
4952900Seric char	*MonthList[] =
4962900Seric {
4972900Seric 	"Jan", "Feb", "Mar", "Apr", "May", "Jun",
4982900Seric 	"Jul", "Aug", "Sep", "Oct", "Nov", "Dec",
4992900Seric 	NULL
5002900Seric };
5012900Seric 
50255012Seric eatfrom(fm, e)
5032900Seric 	char *fm;
50455012Seric 	register ENVELOPE *e;
5052900Seric {
5062900Seric 	register char *p;
5072900Seric 	register char **dt;
5082900Seric 
5097673Seric 	if (tTd(30, 2))
5104203Seric 		printf("eatfrom(%s)\n", fm);
5114203Seric 
5122900Seric 	/* find the date part */
5132900Seric 	p = fm;
5142900Seric 	while (*p != '\0')
5152900Seric 	{
5162900Seric 		/* skip a word */
5172900Seric 		while (*p != '\0' && *p != ' ')
51816896Seric 			p++;
5192900Seric 		while (*p == ' ')
52016896Seric 			p++;
52158050Seric 		if (!(isascii(*p) && isupper(*p)) ||
52258050Seric 		    p[3] != ' ' || p[13] != ':' || p[16] != ':')
5232900Seric 			continue;
5242900Seric 
5252900Seric 		/* we have a possible date */
5264203Seric 		for (dt = DowList; *dt != NULL; dt++)
5272900Seric 			if (strncmp(*dt, p, 3) == 0)
5282900Seric 				break;
5294203Seric 		if (*dt == NULL)
5304203Seric 			continue;
5312900Seric 
5324203Seric 		for (dt = MonthList; *dt != NULL; dt++)
5334203Seric 			if (strncmp(*dt, &p[4], 3) == 0)
5344203Seric 				break;
5352900Seric 		if (*dt != NULL)
5362900Seric 			break;
5372900Seric 	}
5382900Seric 
53960502Seric 	if (*p != '\0')
5402900Seric 	{
5413386Seric 		char *q;
5425366Seric 		extern char *arpadate();
5433386Seric 
5442900Seric 		/* we have found a date */
5453386Seric 		q = xalloc(25);
54623103Seric 		(void) strncpy(q, p, 25);
5473386Seric 		q[24] = '\0';
5485366Seric 		q = arpadate(q);
54955012Seric 		define('a', newstr(q), e);
5502900Seric 	}
5512900Seric }
5524321Seric 
55356795Seric # endif /* NOTUNIX */
554