1 /*
2  * Copyright (c) 1983 Eric P. Allman
3  * Copyright (c) 1988, 1993
4  *	The Regents of the University of California.  All rights reserved.
5  *
6  * %sccs.include.redist.c%
7  */
8 
9 #ifndef lint
10 static char sccsid[] = "@(#)collect.c	8.24 (Berkeley) 08/18/94";
11 #endif /* not lint */
12 
13 # include <errno.h>
14 # include "sendmail.h"
15 
16 /*
17 **  COLLECT -- read & parse message header & make temp file.
18 **
19 **	Creates a temporary file name and copies the standard
20 **	input to that file.  Leading UNIX-style "From" lines are
21 **	stripped off (after important information is extracted).
22 **
23 **	Parameters:
24 **		fp -- file to read.
25 **		smtpmode -- if set, we are running SMTP: give an RFC821
26 **			style message to say we are ready to collect
27 **			input, and never ignore a single dot to mean
28 **			end of message.
29 **		requeueflag -- this message will be requeued later, so
30 **			don't do final processing on it.
31 **		hdrp -- the location to stash the header.
32 **		e -- the current envelope.
33 **
34 **	Returns:
35 **		none.
36 **
37 **	Side Effects:
38 **		Temp file is created and filled.
39 **		The from person may be set.
40 */
41 
42 char	*CollectErrorMessage;
43 bool	CollectErrno;
44 
45 static jmp_buf	CtxCollectTimeout;
46 static int	collecttimeout();
47 static bool	CollectProgress;
48 static EVENT	*CollectTimeout;
49 
50 /* values for input state machine */
51 #define IS_NORM		0	/* middle of line */
52 #define IS_BOL		1	/* beginning of line */
53 #define IS_DOT		2	/* read a dot at beginning of line */
54 #define IS_DOTCR	3	/* read ".\r" at beginning of line */
55 #define IS_CR		4	/* read a carriage return */
56 
57 /* values for message state machine */
58 #define MS_UFROM	0	/* reading Unix from line */
59 #define MS_HEADER	1	/* reading message header */
60 #define MS_BODY		2	/* reading message body */
61 
62 
63 collect(fp, smtpmode, requeueflag, hdrp, e)
64 	FILE *fp;
65 	bool smtpmode;
66 	bool requeueflag;
67 	HDR **hdrp;
68 	register ENVELOPE *e;
69 {
70 	register FILE *tf;
71 	bool ignrdot = smtpmode ? FALSE : IgnrDot;
72 	time_t dbto = smtpmode ? TimeOuts.to_datablock : 0;
73 	register char *bp;
74 	register int c;
75 	bool inputerr = FALSE;
76 	bool headeronly = FALSE;
77 	char *buf;
78 	int buflen;
79 	int istate;
80 	int mstate;
81 	char *pbp;
82 	char peekbuf[8];
83 	char bufbuf[MAXLINE];
84 	extern bool isheader();
85 
86 	CollectErrorMessage = NULL;
87 	CollectErrno = 0;
88 	if (hdrp == NULL)
89 		hdrp = &e->e_header;
90 	else
91 		headeronly = TRUE;
92 
93 	/*
94 	**  Create the temp file name and create the file.
95 	*/
96 
97 	if (!headeronly)
98 	{
99 		struct stat stbuf;
100 
101 		e->e_df = queuename(e, 'd');
102 		e->e_df = newstr(e->e_df);
103 		if ((tf = dfopen(e->e_df, O_WRONLY|O_CREAT|O_TRUNC, FileMode)) == NULL)
104 		{
105 			syserr("Cannot create %s", e->e_df);
106 			e->e_flags |= EF_NORETURN;
107 			finis();
108 		}
109 		if (fstat(fileno(tf), &stbuf) < 0)
110 			e->e_dfino = -1;
111 		else
112 		{
113 			e->e_dfdev = stbuf.st_dev;
114 			e->e_dfino = stbuf.st_ino;
115 		}
116 		HasEightBits = FALSE;
117 	}
118 
119 	/*
120 	**  Tell ARPANET to go ahead.
121 	*/
122 
123 	if (smtpmode)
124 		message("354 Enter mail, end with \".\" on a line by itself");
125 
126 	/*
127 	**  Read the message.
128 	**
129 	**	This is done using two interleaved state machines.
130 	**	The input state machine is looking for things like
131 	**	hidden dots; the message state machine is handling
132 	**	the larger picture (e.g., header versus body).
133 	*/
134 
135 	buf = bp = bufbuf;
136 	buflen = sizeof bufbuf;
137 	pbp = peekbuf;
138 	istate = IS_BOL;
139 	mstate = SaveFrom ? MS_HEADER : MS_UFROM;
140 	CollectProgress = FALSE;
141 
142 	/* if transmitting binary, don't map NL to EOL */
143 	if (e->e_bodytype != NULL && strcasecmp(e->e_bodytype, "8BITMIME") == 0)
144 		e->e_flags |= EF_NL_NOT_EOL;
145 
146 	if (dbto != 0)
147 	{
148 		/* handle possible input timeout */
149 		if (setjmp(CtxCollectTimeout) != 0)
150 		{
151 #ifdef LOG
152 			syslog(LOG_NOTICE,
153 			    "timeout waiting for input from %s during message collect",
154 			    CurHostName ? CurHostName : "<local machine>");
155 #endif
156 			errno = 0;
157 			usrerr("451 timeout waiting for input during message collect");
158 			goto readerr;
159 		}
160 		CollectTimeout = setevent(dbto, collecttimeout, dbto);
161 	}
162 
163 	for (;;)
164 	{
165 		if (tTd(30, 35))
166 			printf("top, istate=%d, mstate=%d\n", istate, mstate);
167 		for (;;)
168 		{
169 			if (pbp > peekbuf)
170 				c = *--pbp;
171 			else
172 			{
173 				while (!feof(InChannel) && !ferror(InChannel))
174 				{
175 					errno = 0;
176 					c = fgetc(InChannel);
177 					if (errno != EINTR)
178 						break;
179 					clearerr(InChannel);
180 				}
181 				CollectProgress = TRUE;
182 				if (TrafficLogFile != NULL)
183 				{
184 					if (istate == IS_BOL)
185 						fprintf(TrafficLogFile, "%05d <<< ",
186 							getpid());
187 					if (c == EOF)
188 						fprintf(TrafficLogFile, "[EOF]\n");
189 					else
190 						fputc(c, TrafficLogFile);
191 				}
192 				if (c == EOF)
193 					goto readerr;
194 				if (SevenBitInput)
195 					c &= 0x7f;
196 				else
197 					HasEightBits |= bitset(0x80, c);
198 				e->e_msgsize++;
199 			}
200 			if (tTd(30, 94))
201 				printf("istate=%d, c=%c (0x%x)\n",
202 					istate, c, c);
203 			switch (istate)
204 			{
205 			  case IS_BOL:
206 				if (c == '.')
207 				{
208 					istate = IS_DOT;
209 					continue;
210 				}
211 				break;
212 
213 			  case IS_DOT:
214 				if (c == '\n' && !ignrdot &&
215 				    !bitset(EF_NL_NOT_EOL, e->e_flags))
216 					goto readerr;
217 				else if (c == '\r' &&
218 					 !bitset(EF_CRLF_NOT_EOL, e->e_flags))
219 				{
220 					istate = IS_DOTCR;
221 					continue;
222 				}
223 				else if (c != '.' ||
224 					 (OpMode != MD_SMTP &&
225 					  OpMode != MD_DAEMON &&
226 					  OpMode != MD_ARPAFTP))
227 				{
228 					*pbp++ = c;
229 					c = '.';
230 				}
231 				break;
232 
233 			  case IS_DOTCR:
234 				if (c == '\n')
235 					goto readerr;
236 				else
237 				{
238 					/* push back the ".\rx" */
239 					*pbp++ = c;
240 					*pbp++ = '\r';
241 					c = '.';
242 				}
243 				break;
244 
245 			  case IS_CR:
246 				if (c == '\n')
247 					istate = IS_BOL;
248 				else
249 				{
250 					ungetc(c, InChannel);
251 					c = '\r';
252 					istate = IS_NORM;
253 				}
254 				goto bufferchar;
255 			}
256 
257 			if (c == '\r' && !bitset(EF_CRLF_NOT_EOL, e->e_flags))
258 			{
259 				istate = IS_CR;
260 				continue;
261 			}
262 			else if (c == '\n' && !bitset(EF_NL_NOT_EOL, e->e_flags))
263 				istate = IS_BOL;
264 			else
265 				istate = IS_NORM;
266 
267 bufferchar:
268 			if (mstate == MS_BODY)
269 			{
270 				/* just put the character out */
271 				fputc(c, tf);
272 				continue;
273 			}
274 
275 			/* header -- buffer up */
276 			if (bp >= &buf[buflen - 2])
277 			{
278 				char *obuf;
279 
280 				if (mstate != MS_HEADER)
281 					break;
282 
283 				/* out of space for header */
284 				obuf = buf;
285 				if (buflen < MEMCHUNKSIZE)
286 					buflen *= 2;
287 				else
288 					buflen += MEMCHUNKSIZE;
289 				buf = xalloc(buflen);
290 				bcopy(obuf, buf, bp - obuf);
291 				bp = &buf[bp - obuf];
292 				if (obuf != bufbuf)
293 					free(obuf);
294 			}
295 			*bp++ = c;
296 			if (istate == IS_BOL)
297 				break;
298 		}
299 		*bp = '\0';
300 
301 nextstate:
302 		if (tTd(30, 35))
303 			printf("nextstate, istate=%d, mstate=%d, line = \"%s\"\n",
304 				istate, mstate, buf);
305 		switch (mstate)
306 		{
307 		  case MS_UFROM:
308 			mstate = MS_HEADER;
309 			if (strncmp(buf, "From ", 5) == 0)
310 			{
311 				bp = buf;
312 				eatfrom(buf, e);
313 				continue;
314 			}
315 			/* fall through */
316 
317 		  case MS_HEADER:
318 			if (!isheader(buf))
319 			{
320 				mstate = MS_BODY;
321 				goto nextstate;
322 			}
323 
324 			/* check for possible continuation line */
325 			do
326 			{
327 				clearerr(InChannel);
328 				errno = 0;
329 				c = fgetc(InChannel);
330 			} while (errno == EINTR);
331 			if (c != EOF)
332 				ungetc(c, InChannel);
333 			if (c == ' ' || c == '\t')
334 			{
335 				/* yep -- defer this */
336 				continue;
337 			}
338 
339 			/* trim off trailing CRLF or NL */
340 			if (*--bp != '\n' || *--bp != '\r')
341 				bp++;
342 			*bp = '\0';
343 			if (bitset(H_EOH, chompheader(buf, FALSE, e)))
344 				mstate = MS_BODY;
345 			break;
346 
347 		  case MS_BODY:
348 			if (tTd(30, 1))
349 				printf("EOH\n");
350 			if (headeronly)
351 				goto readerr;
352 			bp = buf;
353 
354 			/* toss blank line */
355 			if ((!bitset(EF_CRLF_NOT_EOL, e->e_flags) &&
356 				bp[0] == '\r' && bp[1] == '\n') ||
357 			    (!bitset(EF_NL_NOT_EOL, e->e_flags) &&
358 				bp[0] == '\n'))
359 			{
360 				break;
361 			}
362 
363 			/* if not a blank separator, write it out */
364 			while (*bp != '\0')
365 				fputc(*bp++, tf);
366 			break;
367 		}
368 		bp = buf;
369 	}
370 
371 readerr:
372 	if ((feof(fp) && smtpmode) || ferror(fp))
373 	{
374 		if (tTd(30, 1))
375 			printf("collect: read error\n");
376 		inputerr = TRUE;
377 	}
378 
379 	/* reset global timer */
380 	clrevent(CollectTimeout);
381 
382 	if (headeronly)
383 		return;
384 
385 	if (tf != NULL)
386 	{
387 		if (fflush(tf) != 0)
388 			tferror(tf, e);
389 		if (fsync(fileno(tf)) < 0 || fclose(tf) < 0)
390 		{
391 			tferror(tf, e);
392 			finis();
393 		}
394 	}
395 
396 	if (CollectErrorMessage != NULL && Errors <= 0)
397 	{
398 		if (CollectErrno != 0)
399 		{
400 			errno = CollectErrno;
401 			syserr(CollectErrorMessage, e->e_df);
402 			finis();
403 		}
404 		usrerr(CollectErrorMessage);
405 	}
406 	else if (inputerr && (OpMode == MD_SMTP || OpMode == MD_DAEMON))
407 	{
408 		/* An EOF when running SMTP is an error */
409 		char *host;
410 		char *problem;
411 
412 		host = RealHostName;
413 		if (host == NULL)
414 			host = "localhost";
415 
416 		if (feof(fp))
417 			problem = "unexpected close";
418 		else if (ferror(fp))
419 			problem = "I/O error";
420 		else
421 			problem = "read timeout";
422 # ifdef LOG
423 		if (LogLevel > 0 && feof(fp))
424 			syslog(LOG_NOTICE,
425 			    "collect: %s on connection from %s, sender=%s: %s\n",
426 			    problem, host, e->e_from.q_paddr, errstring(errno));
427 # endif
428 		if (feof(fp))
429 			usrerr("451 collect: %s on connection from %s, from=%s",
430 				problem, host, e->e_from.q_paddr);
431 		else
432 			syserr("451 collect: %s on connection from %s, from=%s",
433 				problem, host, e->e_from.q_paddr);
434 
435 		/* don't return an error indication */
436 		e->e_to = NULL;
437 		e->e_flags &= ~EF_FATALERRS;
438 		e->e_flags |= EF_CLRQUEUE;
439 
440 		/* and don't try to deliver the partial message either */
441 		if (InChild)
442 			ExitStat = EX_QUIT;
443 		finis();
444 	}
445 
446 	/*
447 	**  Find out some information from the headers.
448 	**	Examples are who is the from person & the date.
449 	*/
450 
451 	eatheader(e, !requeueflag);
452 
453 	/* collect statistics */
454 	if (OpMode != MD_VERIFY)
455 		markstats(e, (ADDRESS *) NULL);
456 
457 	/*
458 	**  Add an Apparently-To: line if we have no recipient lines.
459 	*/
460 
461 	if (hvalue("to", e->e_header) == NULL &&
462 	    hvalue("cc", e->e_header) == NULL &&
463 	    hvalue("bcc", e->e_header) == NULL &&
464 	    hvalue("apparently-to", e->e_header) == NULL)
465 	{
466 		register ADDRESS *q;
467 
468 		/* create an Apparently-To: field */
469 		/*    that or reject the message.... */
470 		for (q = e->e_sendqueue; q != NULL; q = q->q_next)
471 		{
472 			if (q->q_alias != NULL)
473 				continue;
474 			if (tTd(30, 3))
475 				printf("Adding Apparently-To: %s\n", q->q_paddr);
476 			addheader("Apparently-To", q->q_paddr, &e->e_header);
477 		}
478 	}
479 
480 	/* check for message too large */
481 	if (MaxMessageSize > 0 && e->e_msgsize > MaxMessageSize)
482 	{
483 		usrerr("552 Message exceeds maximum fixed size (%ld)",
484 			MaxMessageSize);
485 	}
486 
487 	/* check for illegal 8-bit data */
488 	if (HasEightBits)
489 	{
490 		e->e_flags |= EF_HAS8BIT;
491 		if (bitset(MM_MIME8BIT, MimeMode))
492 		{
493 			/* convert it to MIME */
494 			if (hvalue("MIME-Version", e->e_header) == NULL)
495 			{
496 				char mimebuf[20];
497 
498 				strcpy(mimebuf, "MIME-Version: 1.0");
499 				chompheader(mimebuf, FALSE, e);
500 			}
501 			if (e->e_bodytype == NULL)
502 				e->e_bodytype = "8BITMIME";
503 		}
504 		else if (!bitset(MM_PASS8BIT, MimeMode))
505 			usrerr("554 Eight bit data not allowed");
506 	}
507 
508 	if ((e->e_dfp = fopen(e->e_df, "r")) == NULL)
509 	{
510 		/* we haven't acked receipt yet, so just chuck this */
511 		syserr("Cannot reopen %s", e->e_df);
512 		finis();
513 	}
514 }
515 
516 
517 static
518 collecttimeout(timeout)
519 	time_t timeout;
520 {
521 	/* if no progress was made, die now */
522 	if (!CollectProgress)
523 		longjmp(CtxCollectTimeout, 1);
524 
525 	/* otherwise reset the timeout */
526 	CollectTimeout = setevent(timeout, collecttimeout, timeout);
527 	CollectProgress = FALSE;
528 }
529 /*
530 **  TFERROR -- signal error on writing the temporary file.
531 **
532 **	Parameters:
533 **		tf -- the file pointer for the temporary file.
534 **
535 **	Returns:
536 **		none.
537 **
538 **	Side Effects:
539 **		Gives an error message.
540 **		Arranges for following output to go elsewhere.
541 */
542 
543 tferror(tf, e)
544 	FILE *tf;
545 	register ENVELOPE *e;
546 {
547 	CollectErrno = errno;
548 	if (errno == ENOSPC)
549 	{
550 		struct stat st;
551 		long avail;
552 		long bsize;
553 
554 		e->e_flags |= EF_NORETURN;
555 		if (fstat(fileno(tf), &st) < 0)
556 			st.st_size = 0;
557 		(void) freopen(e->e_df, "w", tf);
558 		if (st.st_size <= 0)
559 			fprintf(tf, "\n*** Mail could not be accepted");
560 		else if (sizeof st.st_size > sizeof (long))
561 			fprintf(tf, "\n*** Mail of at least %qd bytes could not be accepted\n",
562 				st.st_size);
563 		else
564 			fprintf(tf, "\n*** Mail of at least %ld bytes could not be accepted\n",
565 				st.st_size);
566 		fprintf(tf, "*** at %s due to lack of disk space for temp file.\n",
567 			MyHostName);
568 		avail = freespace(QueueDir, &bsize);
569 		if (avail > 0)
570 		{
571 			if (bsize > 1024)
572 				avail *= bsize / 1024;
573 			else if (bsize < 1024)
574 				avail /= 1024 / bsize;
575 			fprintf(tf, "*** Currently, %ld kilobytes are available for mail temp files.\n",
576 				avail);
577 		}
578 		CollectErrorMessage = "452 Out of disk space for temp file";
579 	}
580 	else
581 	{
582 		CollectErrorMessage = "cannot write message body to disk (%s)";
583 	}
584 	(void) freopen("/dev/null", "w", tf);
585 }
586 /*
587 **  EATFROM -- chew up a UNIX style from line and process
588 **
589 **	This does indeed make some assumptions about the format
590 **	of UNIX messages.
591 **
592 **	Parameters:
593 **		fm -- the from line.
594 **
595 **	Returns:
596 **		none.
597 **
598 **	Side Effects:
599 **		extracts what information it can from the header,
600 **		such as the date.
601 */
602 
603 # ifndef NOTUNIX
604 
605 char	*DowList[] =
606 {
607 	"Sun", "Mon", "Tue", "Wed", "Thu", "Fri", "Sat", NULL
608 };
609 
610 char	*MonthList[] =
611 {
612 	"Jan", "Feb", "Mar", "Apr", "May", "Jun",
613 	"Jul", "Aug", "Sep", "Oct", "Nov", "Dec",
614 	NULL
615 };
616 
617 eatfrom(fm, e)
618 	char *fm;
619 	register ENVELOPE *e;
620 {
621 	register char *p;
622 	register char **dt;
623 
624 	if (tTd(30, 2))
625 		printf("eatfrom(%s)\n", fm);
626 
627 	/* find the date part */
628 	p = fm;
629 	while (*p != '\0')
630 	{
631 		/* skip a word */
632 		while (*p != '\0' && *p != ' ')
633 			p++;
634 		while (*p == ' ')
635 			p++;
636 		if (!(isascii(*p) && isupper(*p)) ||
637 		    p[3] != ' ' || p[13] != ':' || p[16] != ':')
638 			continue;
639 
640 		/* we have a possible date */
641 		for (dt = DowList; *dt != NULL; dt++)
642 			if (strncmp(*dt, p, 3) == 0)
643 				break;
644 		if (*dt == NULL)
645 			continue;
646 
647 		for (dt = MonthList; *dt != NULL; dt++)
648 			if (strncmp(*dt, &p[4], 3) == 0)
649 				break;
650 		if (*dt != NULL)
651 			break;
652 	}
653 
654 	if (*p != '\0')
655 	{
656 		char *q;
657 		extern char *arpadate();
658 
659 		/* we have found a date */
660 		q = xalloc(25);
661 		(void) strncpy(q, p, 25);
662 		q[24] = '\0';
663 		q = arpadate(q);
664 		define('a', newstr(q), e);
665 	}
666 }
667 
668 # endif /* NOTUNIX */
669