[Buzilla 376] Preliminary DKIM support
[exim.git] / src / src / spool_in.c
CommitLineData
f7572e5a 1/* $Cambridge: exim/src/src/spool_in.c,v 1.21 2007/09/28 12:21:57 tom Exp $ */
059ec3d9
PH
2
3/*************************************************
4* Exim - an Internet mail transport agent *
5*************************************************/
6
184e8823 7/* Copyright (c) University of Cambridge 1995 - 2007 */
059ec3d9
PH
8/* See the file NOTICE for conditions of use and distribution. */
9
10/* Functions for reading spool files. When compiling for a utility (eximon),
11not all are needed, and some functionality can be cut out. */
12
13
14#include "exim.h"
15
16
17
18#ifndef COMPILE_UTILITY
19/*************************************************
20* Open and lock data file *
21*************************************************/
22
23/* The data file is the one that is used for locking, because the header file
24can get replaced during delivery because of header rewriting. The file has
25to opened with write access so that we can get an exclusive lock, but in
26fact it won't be written to. Just in case there's a major disaster (e.g.
27overwriting some other file descriptor with the value of this one), open it
28with append.
29
30Argument: the id of the message
31Returns: TRUE if file successfully opened and locked
32
33Side effect: deliver_datafile is set to the fd of the open file.
34*/
35
36BOOL
37spool_open_datafile(uschar *id)
38{
39int i;
40struct stat statbuf;
41flock_t lock_data;
42uschar spoolname[256];
43
44/* If split_spool_directory is set, first look for the file in the appropriate
45sub-directory of the input directory. If it is not found there, try the input
46directory itself, to pick up leftovers from before the splitting. If split_
47spool_directory is not set, first look in the main input directory. If it is
48not found there, try the split sub-directory, in case it is left over from a
49splitting state. */
50
51for (i = 0; i < 2; i++)
52 {
53 int save_errno;
54 message_subdir[0] = (split_spool_directory == (i == 0))? id[5] : 0;
55 sprintf(CS spoolname, "%s/input/%s/%s-D", spool_directory, message_subdir, id);
56 deliver_datafile = Uopen(spoolname, O_RDWR | O_APPEND, 0);
57 if (deliver_datafile >= 0) break;
58 save_errno = errno;
59 if (errno == ENOENT)
60 {
61 if (i == 0) continue;
62 if (!queue_running)
63 log_write(0, LOG_MAIN, "Spool file %s-D not found", id);
64 }
65 else log_write(0, LOG_MAIN, "Spool error for %s: %s", spoolname,
66 strerror(errno));
67 errno = save_errno;
68 return FALSE;
69 }
70
71/* File is open and message_subdir is set. Set the close-on-exec flag, and lock
72the file. We lock only the first line of the file (containing the message ID)
73because this apparently is needed for running Exim under Cygwin. If the entire
74file is locked in one process, a sub-process cannot access it, even when passed
75an open file descriptor (at least, I think that's the Cygwin story). On real
76Unix systems it doesn't make any difference as long as Exim is consistent in
77what it locks. */
78
ff790e47 79(void)fcntl(deliver_datafile, F_SETFD, fcntl(deliver_datafile, F_GETFD) |
059ec3d9
PH
80 FD_CLOEXEC);
81
82lock_data.l_type = F_WRLCK;
83lock_data.l_whence = SEEK_SET;
84lock_data.l_start = 0;
85lock_data.l_len = SPOOL_DATA_START_OFFSET;
86
87if (fcntl(deliver_datafile, F_SETLK, &lock_data) < 0)
88 {
89 log_write(L_skip_delivery,
90 LOG_MAIN,
91 "Spool file is locked (another process is handling this message)");
f1e894f3 92 (void)close(deliver_datafile);
059ec3d9
PH
93 deliver_datafile = -1;
94 errno = 0;
95 return FALSE;
96 }
97
98/* Get the size of the data; don't include the leading filename line
99in the count, but add one for the newline before the data. */
100
101if (fstat(deliver_datafile, &statbuf) == 0)
102 {
103 message_body_size = statbuf.st_size - SPOOL_DATA_START_OFFSET;
104 message_size = message_body_size + 1;
105 }
106
107return TRUE;
108}
109#endif /* COMPILE_UTILITY */
110
111
112
113/*************************************************
114* Read non-recipients tree from spool file *
115*************************************************/
116
117/* The tree of non-recipients is written to the spool file in a form that
118makes it easy to read back into a tree. The format is as follows:
119
120 . Each node is preceded by two letter(Y/N) indicating whether it has left
121 or right children. There's one space after the two flags, before the name.
122
123 . The left subtree (if any) then follows, then the right subtree (if any).
124
125This function is entered with the next input line in the buffer. Note we must
126save the right flag before recursing with the same buffer.
127
128Once the tree is read, we re-construct the balance fields by scanning the tree.
129I forgot to write them out originally, and the compatible fix is to do it this
130way. This initial local recursing function does the necessary.
131
132Arguments:
133 node tree node
134
135Returns: maximum depth below the node, including the node itself
136*/
137
138static int
139count_below(tree_node *node)
140{
141int nleft, nright;
142if (node == NULL) return 0;
143nleft = count_below(node->left);
144nright = count_below(node->right);
145node->balance = (nleft > nright)? 1 : ((nright > nleft)? 2 : 0);
146return 1 + ((nleft > nright)? nleft : nright);
147}
148
149/* This is the real function...
150
151Arguments:
152 connect pointer to the root of the tree
153 f FILE to read data from
154 buffer contains next input line; further lines read into it
155 buffer_size size of the buffer
156
157Returns: FALSE on format error
158*/
159
160static BOOL
161read_nonrecipients_tree(tree_node **connect, FILE *f, uschar *buffer,
162 int buffer_size)
163{
164tree_node *node;
165int n = Ustrlen(buffer);
166BOOL right = buffer[1] == 'Y';
167
168if (n < 5) return FALSE; /* malformed line */
169buffer[n-1] = 0; /* Remove \n */
170node = store_get(sizeof(tree_node) + n - 3);
171*connect = node;
172Ustrcpy(node->name, buffer + 3);
173node->data.ptr = NULL;
174
175if (buffer[0] == 'Y')
176 {
177 if (Ufgets(buffer, buffer_size, f) == NULL ||
178 !read_nonrecipients_tree(&node->left, f, buffer, buffer_size))
179 return FALSE;
180 }
181else node->left = NULL;
182
183if (right)
184 {
185 if (Ufgets(buffer, buffer_size, f) == NULL ||
186 !read_nonrecipients_tree(&node->right, f, buffer, buffer_size))
187 return FALSE;
188 }
189else node->right = NULL;
190
191(void) count_below(*connect);
192return TRUE;
193}
194
195
196
197
198/*************************************************
199* Read spool header file *
200*************************************************/
201
202/* This function reads a spool header file and places the data into the
203appropriate global variables. The header portion is always read, but header
204structures are built only if read_headers is set true. It isn't, for example,
205while generating -bp output.
206
207It may be possible for blocks of nulls (binary zeroes) to get written on the
208end of a file if there is a system crash during writing. It was observed on an
209earlier version of Exim that omitted to fsync() the files - this is thought to
210have been the cause of that incident, but in any case, this code must be robust
211against such an event, and if such a file is encountered, it must be treated as
212malformed.
213
214Arguments:
215 name name of the header file, including the -H
216 read_headers TRUE if in-store header structures are to be built
217 subdir_set TRUE is message_subdir is already set
218
219Returns: spool_read_OK success
220 spool_read_notopen open failed
221 spool_read_enverror error in the envelope portion
222 spool_read_hdrdrror error in the header portion
223*/
224
225int
226spool_read_header(uschar *name, BOOL read_headers, BOOL subdir_set)
227{
228FILE *f = NULL;
229int n;
230int rcount = 0;
231long int uid, gid;
232BOOL inheader = FALSE;
1e70f85b 233uschar *p;
059ec3d9
PH
234
235/* Reset all the global variables to their default values. However, there is
236one exception. DO NOT change the default value of dont_deliver, because it may
237be forced by an external setting. */
238
38a0a95f 239acl_var_c = acl_var_m = NULL;
059ec3d9
PH
240authenticated_id = NULL;
241authenticated_sender = NULL;
242allow_unqualified_recipient = FALSE;
243allow_unqualified_sender = FALSE;
244body_linecount = 0;
245body_zerocount = 0;
246deliver_firsttime = FALSE;
247deliver_freeze = FALSE;
248deliver_frozen_at = 0;
249deliver_manual_thaw = FALSE;
250/* dont_deliver must NOT be reset */
251header_list = header_last = NULL;
b08b24c8 252host_lookup_deferred = FALSE;
059ec3d9
PH
253host_lookup_failed = FALSE;
254interface_address = NULL;
255interface_port = 0;
256local_error_message = FALSE;
257local_scan_data = NULL;
d677b2f2 258max_received_linelength = 0;
059ec3d9
PH
259message_linecount = 0;
260received_protocol = NULL;
261received_count = 0;
262recipients_list = NULL;
263sender_address = NULL;
264sender_fullhost = NULL;
265sender_helo_name = NULL;
266sender_host_address = NULL;
267sender_host_name = NULL;
268sender_host_port = 0;
269sender_host_authenticated = NULL;
270sender_ident = NULL;
271sender_local = FALSE;
272sender_set_untrusted = FALSE;
1f5b4c3d 273smtp_active_hostname = primary_hostname;
059ec3d9
PH
274tree_nonrecipients = NULL;
275
8523533c
TK
276#ifdef EXPERIMENTAL_BRIGHTMAIL
277bmi_run = 0;
278bmi_verdicts = NULL;
279#endif
280
fb2274d4
TK
281#ifdef EXPERIMENTAL_DOMAINKEYS
282dk_do_verify = 0;
283#endif
284
f7572e5a
TK
285#ifdef EXPERIMENTAL_DKIM
286dkim_do_verify = 0;
287#endif
288
059ec3d9
PH
289#ifdef SUPPORT_TLS
290tls_certificate_verified = FALSE;
291tls_cipher = NULL;
292tls_peerdn = NULL;
293#endif
294
8523533c 295#ifdef WITH_CONTENT_SCAN
8523533c
TK
296spam_score_int = NULL;
297#endif
298
059ec3d9
PH
299/* Generate the full name and open the file. If message_subdir is already
300set, just look in the given directory. Otherwise, look in both the split
301and unsplit directories, as for the data file above. */
302
303for (n = 0; n < 2; n++)
304 {
305 if (!subdir_set)
306 message_subdir[0] = (split_spool_directory == (n == 0))? name[5] : 0;
307 sprintf(CS big_buffer, "%s/input/%s/%s", spool_directory, message_subdir,
308 name);
309 f = Ufopen(big_buffer, "rb");
310 if (f != NULL) break;
311 if (n != 0 || subdir_set || errno != ENOENT) return spool_read_notopen;
312 }
313
314errno = 0;
315
316#ifndef COMPILE_UTILITY
317DEBUG(D_deliver) debug_printf("reading spool file %s\n", name);
318#endif /* COMPILE_UTILITY */
319
320/* The first line of a spool file contains the message id followed by -H (i.e.
321the file name), in order to make the file self-identifying. */
322
323if (Ufgets(big_buffer, big_buffer_size, f) == NULL) goto SPOOL_READ_ERROR;
324if (Ustrlen(big_buffer) != MESSAGE_ID_LENGTH + 3 ||
325 Ustrncmp(big_buffer, name, MESSAGE_ID_LENGTH + 2) != 0)
326 goto SPOOL_FORMAT_ERROR;
327
328/* The next three lines in the header file are in a fixed format. The first
329contains the login, uid, and gid of the user who caused the file to be written.
ebb6e6d5
PH
330There are known cases where a negative gid is used, so we allow for both
331negative uids and gids. The second contains the mail address of the message's
332sender, enclosed in <>. The third contains the time the message was received,
333and the number of warning messages for delivery delays that have been sent. */
059ec3d9
PH
334
335if (Ufgets(big_buffer, big_buffer_size, f) == NULL) goto SPOOL_READ_ERROR;
336
1e70f85b
PH
337p = big_buffer + Ustrlen(big_buffer);
338while (p > big_buffer && isspace(p[-1])) p--;
339*p = 0;
340if (!isdigit(p[-1])) goto SPOOL_FORMAT_ERROR;
ebb6e6d5 341while (p > big_buffer && (isdigit(p[-1]) || '-' == p[-1])) p--;
1e70f85b
PH
342gid = Uatoi(p);
343if (p <= big_buffer || *(--p) != ' ') goto SPOOL_FORMAT_ERROR;
344*p = 0;
345if (!isdigit(p[-1])) goto SPOOL_FORMAT_ERROR;
ebb6e6d5 346while (p > big_buffer && (isdigit(p[-1]) || '-' == p[-1])) p--;
1e70f85b
PH
347uid = Uatoi(p);
348if (p <= big_buffer || *(--p) != ' ') goto SPOOL_FORMAT_ERROR;
349*p = 0;
8e669ac1 350
1e70f85b 351originator_login = string_copy(big_buffer);
059ec3d9
PH
352originator_uid = (uid_t)uid;
353originator_gid = (gid_t)gid;
354
355if (Ufgets(big_buffer, big_buffer_size, f) == NULL) goto SPOOL_READ_ERROR;
356n = Ustrlen(big_buffer);
357if (n < 3 || big_buffer[0] != '<' || big_buffer[n-2] != '>')
358 goto SPOOL_FORMAT_ERROR;
359
360sender_address = store_get(n-2);
361Ustrncpy(sender_address, big_buffer+1, n-3);
362sender_address[n-3] = 0;
363
364if (Ufgets(big_buffer, big_buffer_size, f) == NULL) goto SPOOL_READ_ERROR;
365if (sscanf(CS big_buffer, "%d %d", &received_time, &warning_count) != 2)
366 goto SPOOL_FORMAT_ERROR;
367
368message_age = time(NULL) - received_time;
369
370#ifndef COMPILE_UTILITY
371DEBUG(D_deliver) debug_printf("user=%s uid=%ld gid=%ld sender=%s\n",
372 originator_login, (long int)originator_uid, (long int)originator_gid,
373 sender_address);
374#endif /* COMPILE_UTILITY */
375
08955dd3
PH
376/* Now there may be a number of optional lines, each starting with "-". If you
377add a new setting here, make sure you set the default above.
059ec3d9 378
08955dd3
PH
379Because there are now quite a number of different possibilities, we use a
380switch on the first character to avoid too many failing tests. Thanks to Nico
381Erfurth for the patch that implemented this. I have made it even more efficient
382by not re-scanning the first two characters.
383
384To allow new versions of Exim that add additional flags to interwork with older
385versions that do not understand them, just ignore any lines starting with "-"
386that we don't recognize. Otherwise it wouldn't be possible to back off a new
387version that left new-style flags written on the spool. */
388
389p = big_buffer + 2;
059ec3d9
PH
390for (;;)
391 {
392 if (Ufgets(big_buffer, big_buffer_size, f) == NULL) goto SPOOL_READ_ERROR;
393 if (big_buffer[0] != '-') break;
059ec3d9 394 big_buffer[Ustrlen(big_buffer) - 1] = 0;
47ca6d6c 395
08955dd3 396 switch(big_buffer[1])
059ec3d9 397 {
08955dd3
PH
398 case 'a':
399
400 /* Nowadays we use "-aclc" and "-aclm" for the different types of ACL
401 variable, because Exim allows any number of them, with arbitrary names.
402 The line in the spool file is "-acl[cm] <name> <length>". The name excludes
403 the c or m. */
404
405 if (Ustrncmp(p, "clc ", 4) == 0 ||
406 Ustrncmp(p, "clm ", 4) == 0)
407 {
408 uschar *name, *endptr;
409 int count;
410 tree_node *node;
411 endptr = Ustrchr(big_buffer + 6, ' ');
412 if (endptr == NULL) goto SPOOL_FORMAT_ERROR;
413 name = string_sprintf("%c%.*s", big_buffer[4], endptr - big_buffer - 6,
414 big_buffer + 6);
415 if (sscanf(CS endptr, " %d", &count) != 1) goto SPOOL_FORMAT_ERROR;
416 node = acl_var_create(name);
417 node->data.ptr = store_get(count + 1);
418 if (fread(node->data.ptr, 1, count+1, f) < count) goto SPOOL_READ_ERROR;
419 ((uschar*)node->data.ptr)[count] = 0;
420 }
421
422 else if (Ustrcmp(p, "llow_unqualified_recipient") == 0)
423 allow_unqualified_recipient = TRUE;
424 else if (Ustrcmp(p, "llow_unqualified_sender") == 0)
425 allow_unqualified_sender = TRUE;
426
427 else if (Ustrncmp(p, "uth_id", 6) == 0)
428 authenticated_id = string_copy(big_buffer + 9);
429 else if (Ustrncmp(p, "uth_sender", 10) == 0)
430 authenticated_sender = string_copy(big_buffer + 13);
431 else if (Ustrncmp(p, "ctive_hostname", 14) == 0)
432 smtp_active_hostname = string_copy(big_buffer + 17);
433
434 /* For long-term backward compatibility, we recognize "-acl", which was
435 used before the number of ACL variables changed from 10 to 20. This was
436 before the subsequent change to an arbitrary number of named variables.
437 This code is retained so that upgrades from very old versions can still
438 handle old-format spool files. The value given after "-acl" is a number
439 that is 0-9 for connection variables, and 10-19 for message variables. */
440
441 else if (Ustrncmp(p, "cl ", 3) == 0)
442 {
443 int index, count;
8dce1a6f 444 uschar name[20]; /* Need plenty of space for %d format */
08955dd3
PH
445 tree_node *node;
446 if (sscanf(CS big_buffer + 5, "%d %d", &index, &count) != 2)
447 goto SPOOL_FORMAT_ERROR;
8dce1a6f
PH
448 if (index < 10)
449 (void) string_format(name, sizeof(name), "%c%d", 'c', index);
450 else if (index < 20) /* ignore out-of-range index */
451 (void) string_format(name, sizeof(name), "%c%d", 'm', index - 10);
08955dd3
PH
452 node = acl_var_create(name);
453 node->data.ptr = store_get(count + 1);
454 if (fread(node->data.ptr, 1, count+1, f) < count) goto SPOOL_READ_ERROR;
455 ((uschar*)node->data.ptr)[count] = 0;
456 }
457 break;
458
459 case 'b':
460 if (Ustrncmp(p, "ody_linecount", 13) == 0)
461 body_linecount = Uatoi(big_buffer + 15);
462 else if (Ustrncmp(p, "ody_zerocount", 13) == 0)
463 body_zerocount = Uatoi(big_buffer + 15);
464 #ifdef EXPERIMENTAL_BRIGHTMAIL
465 else if (Ustrncmp(p, "mi_verdicts ", 12) == 0)
466 bmi_verdicts = string_copy(big_buffer + 14);
467 #endif
468 break;
469
470 case 'd':
471 if (Ustrcmp(p, "eliver_firsttime") == 0)
472 deliver_firsttime = TRUE;
473 break;
474
475 case 'f':
476 if (Ustrncmp(p, "rozen", 5) == 0)
477 {
478 deliver_freeze = TRUE;
479 deliver_frozen_at = Uatoi(big_buffer + 7);
480 }
481 break;
482
483 case 'h':
484 if (Ustrcmp(p, "ost_lookup_deferred") == 0)
485 host_lookup_deferred = TRUE;
486 else if (Ustrcmp(p, "ost_lookup_failed") == 0)
487 host_lookup_failed = TRUE;
488 else if (Ustrncmp(p, "ost_auth", 8) == 0)
489 sender_host_authenticated = string_copy(big_buffer + 11);
490 else if (Ustrncmp(p, "ost_name", 8) == 0)
491 sender_host_name = string_copy(big_buffer + 11);
492 else if (Ustrncmp(p, "elo_name", 8) == 0)
493 sender_helo_name = string_copy(big_buffer + 11);
494
495 /* We now record the port number after the address, separated by a
496 dot. For compatibility during upgrading, do nothing if there
497 isn't a value (it gets left at zero). */
498
499 else if (Ustrncmp(p, "ost_address", 11) == 0)
500 {
501 sender_host_port = host_address_extract_port(big_buffer + 14);
502 sender_host_address = string_copy(big_buffer + 14);
503 }
504 break;
505
506 case 'i':
507 if (Ustrncmp(p, "nterface_address", 16) == 0)
508 {
509 interface_port = host_address_extract_port(big_buffer + 19);
510 interface_address = string_copy(big_buffer + 19);
511 }
512 else if (Ustrncmp(p, "dent", 4) == 0)
513 sender_ident = string_copy(big_buffer + 7);
514 break;
515
516 case 'l':
517 if (Ustrcmp(p, "ocal") == 0) sender_local = TRUE;
518 else if (Ustrcmp(big_buffer, "-localerror") == 0)
519 local_error_message = TRUE;
520 else if (Ustrncmp(p, "ocal_scan ", 10) == 0)
521 local_scan_data = string_copy(big_buffer + 12);
522 break;
523
524 case 'm':
525 if (Ustrcmp(p, "anual_thaw") == 0) deliver_manual_thaw = TRUE;
d677b2f2
PH
526 else if (Ustrncmp(p, "ax_received_linelength", 22) == 0)
527 max_received_linelength = Uatoi(big_buffer + 24);
08955dd3
PH
528 break;
529
530 case 'N':
531 if (*p == 0) dont_deliver = TRUE; /* -N */
532 break;
533
534 case 'r':
535 if (Ustrncmp(p, "eceived_protocol", 16) == 0)
536 received_protocol = string_copy(big_buffer + 19);
537 break;
538
539 case 's':
540 if (Ustrncmp(p, "ender_set_untrusted", 19) == 0)
541 sender_set_untrusted = TRUE;
542 #ifdef WITH_CONTENT_SCAN
543 else if (Ustrncmp(p, "pam_score_int ", 14) == 0)
544 spam_score_int = string_copy(big_buffer + 16);
545 #endif
546 break;
547
548 #ifdef SUPPORT_TLS
549 case 't':
550 if (Ustrncmp(p, "ls_certificate_verified", 23) == 0)
551 tls_certificate_verified = TRUE;
552 else if (Ustrncmp(p, "ls_cipher", 9) == 0)
553 tls_cipher = string_copy(big_buffer + 12);
554 else if (Ustrncmp(p, "ls_peerdn", 9) == 0)
555 tls_peerdn = string_copy(big_buffer + 12);
556 break;
557 #endif
558
559 default: /* Present because some compilers complain if all */
560 break; /* possibilities are not covered. */
059ec3d9 561 }
059ec3d9
PH
562 }
563
564/* Build sender_fullhost if required */
565
566#ifndef COMPILE_UTILITY
567host_build_sender_fullhost();
568#endif /* COMPILE_UTILITY */
569
570#ifndef COMPILE_UTILITY
571DEBUG(D_deliver)
572 debug_printf("sender_local=%d ident=%s\n", sender_local,
573 (sender_ident == NULL)? US"unset" : sender_ident);
574#endif /* COMPILE_UTILITY */
575
576/* We now have the tree of addresses NOT to deliver to, or a line
577containing "XX", indicating no tree. */
578
579if (Ustrncmp(big_buffer, "XX\n", 3) != 0 &&
580 !read_nonrecipients_tree(&tree_nonrecipients, f, big_buffer, big_buffer_size))
581 goto SPOOL_FORMAT_ERROR;
582
583#ifndef COMPILE_UTILITY
584DEBUG(D_deliver)
585 {
586 debug_printf("Non-recipients:\n");
587 debug_print_tree(tree_nonrecipients);
588 }
589#endif /* COMPILE_UTILITY */
590
591/* After reading the tree, the next line has not yet been read into the
592buffer. It contains the count of recipients which follow on separate lines. */
593
594if (Ufgets(big_buffer, big_buffer_size, f) == NULL) goto SPOOL_READ_ERROR;
595if (sscanf(CS big_buffer, "%d", &rcount) != 1) goto SPOOL_FORMAT_ERROR;
596
597#ifndef COMPILE_UTILITY
598DEBUG(D_deliver) debug_printf("recipients_count=%d\n", rcount);
599#endif /* COMPILE_UTILITY */
600
601recipients_list_max = rcount;
602recipients_list = store_get(rcount * sizeof(recipient_item));
603
604for (recipients_count = 0; recipients_count < rcount; recipients_count++)
605 {
606 int nn;
607 int pno = -1;
608 uschar *errors_to = NULL;
609 uschar *p;
610
611 if (Ufgets(big_buffer, big_buffer_size, f) == NULL) goto SPOOL_READ_ERROR;
612 nn = Ustrlen(big_buffer);
613 if (nn < 2) goto SPOOL_FORMAT_ERROR;
614
615 /* Remove the newline; this terminates the address if there is no additional
616 data on the line. */
617
618 p = big_buffer + nn - 1;
619 *p-- = 0;
620
621 /* Look back from the end of the line for digits and special terminators.
622 Since an address must end with a domain, we can tell that extra data is
623 present by the presence of the terminator, which is always some character
624 that cannot exist in a domain. (If I'd thought of the need for additional
625 data early on, I'd have put it at the start, with the address at the end. As
626 it is, we have to operate backwards. Addresses are permitted to contain
627 spaces, you see.)
628
629 This code has to cope with various versions of this data that have evolved
630 over time. In all cases, the line might just contain an address, with no
631 additional data. Otherwise, the possibilities are as follows:
632
633 Exim 3 type: <address><space><digits>,<digits>,<digits>
634
635 The second set of digits is the parent number for one_time addresses. The
636 other values were remnants of earlier experiments that were abandoned.
637
638 Exim 4 first type: <address><space><digits>
639
640 The digits are the parent number for one_time addresses.
641
642 Exim 4 new type: <address><space><data>#<type bits>
643
644 The type bits indicate what the contents of the data are.
645
646 Bit 01 indicates that, reading from right to left, the data
647 ends with <errors_to address><space><len>,<pno> where pno is
648 the parent number for one_time addresses, and len is the length
649 of the errors_to address (zero meaning none).
650 */
651
652 while (isdigit(*p)) p--;
653
654 /* Handle Exim 3 spool files */
655
656 if (*p == ',')
657 {
658 int dummy;
659 while (isdigit(*(--p)) || *p == ',');
660 if (*p == ' ')
661 {
662 *p++ = 0;
ff790e47 663 (void)sscanf(CS p, "%d,%d", &dummy, &pno);
059ec3d9
PH
664 }
665 }
666
667 /* Handle early Exim 4 spool files */
668
669 else if (*p == ' ')
670 {
671 *p++ = 0;
ff790e47 672 (void)sscanf(CS p, "%d", &pno);
059ec3d9
PH
673 }
674
675 /* Handle current format Exim 4 spool files */
676
677 else if (*p == '#')
678 {
679 int flags;
ff790e47 680 (void)sscanf(CS p+1, "%d", &flags);
059ec3d9
PH
681
682 if ((flags & 0x01) != 0) /* one_time data exists */
683 {
684 int len;
685 while (isdigit(*(--p)) || *p == ',' || *p == '-');
ff790e47 686 (void)sscanf(CS p+1, "%d,%d", &len, &pno);
059ec3d9
PH
687 *p = 0;
688 if (len > 0)
689 {
690 p -= len;
691 errors_to = string_copy(p);
692 }
693 }
694
695 *(--p) = 0; /* Terminate address */
696 }
697
698 recipients_list[recipients_count].address = string_copy(big_buffer);
699 recipients_list[recipients_count].pno = pno;
700 recipients_list[recipients_count].errors_to = errors_to;
701 }
702
703/* The remainder of the spool header file contains the headers for the message,
704separated off from the previous data by a blank line. Each header is preceded
705by a count of its length and either a certain letter (for various identified
706headers), space (for a miscellaneous live header) or an asterisk (for a header
707that has been rewritten). Count the Received: headers. We read the headers
708always, in order to check on the format of the file, but only create a header
709list if requested to do so. */
710
711inheader = TRUE;
712if (Ufgets(big_buffer, big_buffer_size, f) == NULL) goto SPOOL_READ_ERROR;
713if (big_buffer[0] != '\n') goto SPOOL_FORMAT_ERROR;
714
715while ((n = fgetc(f)) != EOF)
716 {
717 header_line *h;
718 uschar flag[4];
719 int i;
720
721 if (!isdigit(n)) goto SPOOL_FORMAT_ERROR;
ff790e47
PH
722 (void)ungetc(n, f);
723 (void)fscanf(f, "%d%c ", &n, flag);
059ec3d9
PH
724 if (flag[0] != '*') message_size += n; /* Omit non-transmitted headers */
725
726 if (read_headers)
727 {
728 h = store_get(sizeof(header_line));
729 h->next = NULL;
730 h->type = flag[0];
731 h->slen = n;
732 h->text = store_get(n+1);
733
734 if (h->type == htype_received) received_count++;
735
736 if (header_list == NULL) header_list = h;
737 else header_last->next = h;
738 header_last = h;
739
740 for (i = 0; i < n; i++)
741 {
742 int c = fgetc(f);
743 if (c == 0 || c == EOF) goto SPOOL_FORMAT_ERROR;
744 if (c == '\n' && h->type != htype_old) message_linecount++;
745 h->text[i] = c;
746 }
747 h->text[i] = 0;
748 }
749
750 /* Not requiring header data, just skip through the bytes */
751
752 else for (i = 0; i < n; i++)
753 {
754 int c = fgetc(f);
755 if (c == 0 || c == EOF) goto SPOOL_FORMAT_ERROR;
756 }
757 }
758
759/* We have successfully read the data in the header file. Update the message
760line count by adding the body linecount to the header linecount. Close the file
761and give a positive response. */
762
763#ifndef COMPILE_UTILITY
764DEBUG(D_deliver) debug_printf("body_linecount=%d message_linecount=%d\n",
765 body_linecount, message_linecount);
766#endif /* COMPILE_UTILITY */
767
768message_linecount += body_linecount;
769
770fclose(f);
771return spool_read_OK;
772
773
774/* There was an error reading the spool or there was missing data,
775or there was a format error. A "read error" with no errno means an
776unexpected EOF, which we treat as a format error. */
777
778SPOOL_READ_ERROR:
779if (errno != 0)
780 {
781 n = errno;
782
783 #ifndef COMPILE_UTILITY
784 DEBUG(D_any) debug_printf("Error while reading spool file %s\n", name);
785 #endif /* COMPILE_UTILITY */
786
787 fclose(f);
788 errno = n;
789 return inheader? spool_read_hdrerror : spool_read_enverror;
790 }
791
792SPOOL_FORMAT_ERROR:
793
794#ifndef COMPILE_UTILITY
795DEBUG(D_any) debug_printf("Format error in spool file %s\n", name);
796#endif /* COMPILE_UTILITY */
797
798fclose(f);
799errno = ERRNO_SPOOLFORMAT;
800return inheader? spool_read_hdrerror : spool_read_enverror;
801}
802
803/* End of spool_in.c */