This makes this file more suitable for import into SCFG->Message Areas
(something I'll work on soon), i.e. as an alternative to importing badareas.lst

to before the actual export, so that if an area has no downlinks, you won't
actually see this errorneous but harmless log message. Also include the
"export-to" address in the message.

Areafix requests to unlink a node from an area would corrupt the list of
linked nodes: the *last* listed node would always be removed. If this was
not the node that submitted the areafix request, then 2 nodes would be
removed from the list of linked-nodes for an echo.
To simplify this, we're just going to not write the removed node back to
the area file, but leave it in the in-memory list. So technically, the node
won't be unlinked until the next run of SBBSecho when the area file is
re-parsed. If that's a problem, we can always add run-time removal from
the in-memory list later. Reported by Alterego (ALTERANT).

*.msg).

(e.g. for Hold status) even if/when it's a point off of us.

Don't ever route netmail when the destination zone/net/node matches a local
AKA (ignoring the point value).

If a packed messages contains no PATH or SEEN-BY lines, we can still detect
and prevent a message loop by comparing the origin address in the packet header
against the downlink's address and if it's a match, skip that downlink.
It is still possible that a packed message header contains a different origin
address than the packet header, and we're actually over-writing the packed
messge header variable with the parsed Origin: line address (if there is one),
so perhaps we'll want to compare the (actual) packed message header origin
address too at some point in the future, if loops continue to be a problem.

write_to_pkts():
If for some weird reason a downlink's address is in the PATH, but not in the
SEEN-BYs, detect as a loop and don't send to (add to an outbound pkt for) the
downlink.

This is an experimental change to see if it addresses the issue reported by
Richard Williamson and Mark Lewis with regards to dupes in the COOKING
echo from point nodes off 1:396/45.

in case something (e.g. Argus mailer) deleted it while we were waiting our
turn to use it. As reported by Alterego (ALTERANT)

aux/net attributes to FidoNet msg header attributes:
- CRASH
- FILEREQUEST (FREQ)
- RECEIPTREQ (RRREQ)

This should fix the FREQ netmail issue reported by Richard Williamson
(RICHARDF). Thanks for the bug report!

in-place or by also copying to a new string in the process.

Don't string "soft-CRs" (0x8d) from UTF-8 encoded FTN messages, ever.

Default strip_soft_cr SBBSecho setting to true/on.

If an incoming message contains no CHRS/CHARSET control line *and* the message
text contains valid UTF-8 character encodings, set the FTN charset value to
UTF-8 so the message will be displayed/handled accordingly.
I did not add checks for header fields (to/from/subject) - we should probably
auto-detect UTF-8 in those as well, but for now, I don't see messages coming
into FidoNet echoes with UTF-8 in the header fields.
Incremented SBBSecho/EchoCfg version to 3.10.

was not a recoverable message header (e.g. the toUserName field was *not*
at offset 0x22):

xx 02 00 01 00 01 00 6A - 02 FE 01 00 00 00 00 30   : .......j.......0
34 20 46 65 62 20 31 31 - 39 20 20 32 30 3A 32 36   : 4 Feb 119  20:26
3A 33 32 04 00 4B 75 72 - 74 20 57 65 69 73 6B 65   : :32..Kurt Weiske

So restore the DateTime field validation: the 20th byte must be null.

messages is a null (0x00) byte. Some broken FidoNet software may include
a full 20 usable characters in their DateTime header field, like so:
"04 Feb 119  20:26:32" - representing February 4th, 2019.  Y2K. <sigh>

The DateTime won't be parsed fully correct, but at least the packet won't be
rejected outright because it is "Grunged". - for Alterego (ALTERANT)

This requires that the AutoAddSubs=true in sbbsecho.ini (the default) and if
you don't want an area file created/updated with new subs, just set new option
AutoAddToAreaFile=false.
Both of these options default to true which is the same as the previous
behavior: if you have a link that configured to hub for a msg group, any subs
not already in the area list/file will be considered new and added dynamically.
So what's new here is that you can set SBBSecho not to create or update an
area file and SBBSecho will run just fine if no area file exists or contains no
area definitions.
Incremented version to 3.09.

Alterego reported, would cause a crash if the netmail message did *not* have
a FLAGS kludge line.

standard time-zone (e.g. -0300 == EST) as this is not accurate when daylight
saving time is in effect (-0400 == EDT or AST?).

Always log the "exiting" line when exiting with a non-zero error level.

exit), just log the error and continue processing. Reported by wkitty42.

- Don't ever try to unpack a 0-byte file.
- Don't log anything or try delete if the file is < 24 hours old.
- Try to delete the file if it's >= 24 hours old (and log a msg, INFO-level).

contain UTF-8 characters.

file-attachment deletion in write_flofile() - not the KILLSENT attribute flag.
Export the SMB MSG_KILLSENT auxattr from SMB mail to FTN netmail \1FLAGS KFS.

"UTF-8", set the msg's auxattr MSG_HFIELDS_UTF8 flag because FTS-5003 states:
  "The character set identifier applies to all parts of the message,
   including the header information and the control lines like origin
   and tear line."

When writing non-bundle file paths/names to BSO/FLO files, don't use the
delete (^) prefix unless the message has the "Kill Sent" attribute set. This
seems kind of wrong to me. The KFS "Kill File Sent" flag has been defined in
FSC-0053 since 1992, that seems more likely the appropriate flag to determine
if a message attachment should be deleted (or not) after being sent. But
parsing/using the "Flags" control line flags isn't already in SBBSecho, so
I'll just punt for now and do what Mark asked for. <shrug>

the Ctrl-A codes: L (CLS), < (backspace), [ (CR), or ] (LF).

start of a valid control paragraph (kludge line), then convert the Ctrl-A char
to an '@'. This handles the situation where someone quotes control paragraphs
with Ctrl-A chars, for example, we don't want those Ctrl-A chars being
misinterpreted as Ctrl-A codes causing weird colors or pauses or whatever.

FIDOCHARSET
Export CHRS: control line with value determined by sub-board settings (e.g.
ASCII only) and the contents of the message (UTF-8 or CP437).
Import "NOTE:" control line values into the SMB_EDITOR hfield and export
from SMB_EDITOR -> NOTE:.
Increment SBBSecho version to 3.08.

linked nodes: UseOutboxes (default: true)

BinkIT will continue to outboxes even when this option is set to false, but
SBBSecho won't place any mail files in the outboxes when this option is set to
true.

For PSI-Jack who was surprised that SBBSecho put mail files into outboxes.
Since it appears BinkD supports both outboxes and normal outbound directories
for linked nodes and BinkIT does as well, this shouldn't really make any
difference - just a sysop preference.

doesn't actually *use* the password for anything, EchoCfg uses the nodecfg_t
definition from here (which includes the 'ticpwd') and the SBBSecho AreaManager
allows remote changes of the TicPwd by linked-nodes. For PSI-Jack.

When processing area manager netmail messages, SBBSecho can modify
(add/change) keys in the [node:*] section of the sbbsecho.ini file. Although
SBBSecho would read node sections that had a domain (e.g. 5D address) just fine
when processing area manager changes and writing those changes back to the
sbbsecho.ini, only 4D addresses (no "@domain") were used in the section naming.
This could result in duplicate node sections being created in the sbbsecho.ini
file (one with a "@domain", one without) and then "bad things" could happen
(e.g. BinkP authentication failures, packet authentication failures, etc.).

Now, when SBBSecho writes linked-node setting to the sbbsecho.ini, it first
checks for the full 5D address of the node and if that section exists, modify
it, otherwise fall-through to the 3D/4D address of the node for the sectoin
naming.

Log free disk space errors to sbbsecho.log (as well as print to stdout).

(when set to Yes).