comparison lisp/mail/feedmail.el @ 22801:e71025b5e000

Avoid using all-caps for non-arguments.

22800:125816f37462

;;        there is also a queue for draft messages
;;
;;    --- you can get one last look at the prepped outbound message and
;;        be prompted for confirmation
;;
;;    --- removes BCC:/RESENT-BCC: headers after getting address info
;;
;;    --- does smart filling of address headers
;;
;;    --- calls a routine to process FCC: lines and removes them
;;
;;    --- empty headers are removed
;;
;;    --- can force FROM: or SENDER: line
;;
;;    --- can generate a MESSAGE-ID: line
;;
;;    --- can generate a DATE: line; the date can be the time the
;;        message was written or the time it is being sent
;;
;;    --- strips comments from address info (both "()" and "<>" are
;;        handled via a call to mail-strip-quoted-names); the
;;        comments are stripped in the simplified address list given

;;        list
;;
;;    --- you can specify something other than /bin/mail for the
;;        subprocess
;;
;;    --- you can generate/modify an X-MAILER: message header
;;
;; After a long list of options below, you will find the function
;; feedmail-send-it. Hers's the best way to use the stuff in this
;; file:
;;

;; patchlevel 4, issued by someone else
;; patchlevel 5, issued by someone else
;; patchlevel 6, not issued as far as I know
;; patchlevel 7,  20 May 1997
;;         abandon futile support of Emacs 18 (sorry if that hurts you)
;;         provide a DATE: header by default
;;         provide a default for generating MESSAGE-ID: header contents
;;            and use it by default (slightly changed API)
;;         return value from feedmail-run-the-queue
;;         new wrapper function feedmail-run-the-queue-no-prompts
;;         user-mail-address as default for FROM:
;;         properly deal with RESENT-{TO,CC,BCC}
;;         BCC and RESENT-* now included in smart filling
;;         limited support for a "drafts" directory
;;         user-configurable default message action
;;         allow timeout for confirmation prompt (where available)
;;         move FCC handling to as late as possible to get max
;;            header munging in the saved file
;;         work around sendmail.el's prompts when working from queue
;;         more reliably detect voluntary user bailouts
;;         offer to save modified buffers visiting queue files
;;         offer to delete old file copies of messages being queued

;;         eliminate reliance on directory-sep-char and feedmail-sep-thing
;;         tweak smart filling (reminded of comma problem by [e-mail])
;;         option to control writing in text vs binary mode
;; patchlevel 8, 15 June 1998
;;         reliable re-editing of text-mode (vs binary) queued messages
;;         user option to keep BCC: in FCC: copy (keep by default)
;;         user option to delete body from FCC: copy (keep by default)
;;         feedmail-deduce-bcc-where for envelope (API change for
;;           feedmail-deduce-address list)
;;         feedmail-queue-alternative-mail-header-separator
;;         at message action prompt, "I"/"S" bypass message confirmation prompt
;;         feedmail-mail-send-hook-splitter, feedmail-mail-send-hook,

;;         feedmail-queue-reminder, feedmail-queue-reminder-alist (after suggestions
;;           and/or code fragments from [e-mail] (Tony Lam) and
;;           [e-mail] (Shane Burgess); bumped up the default value of
;;           feedmail-queue-chatty-sit-for since info is more complex sometimes
;;         feedmail-enable-spray (individual transmissions, crude mailmerge)
;;         blank SUBJECT: no longer a special case; see feedmail-nuke-empty-headers
;;         fiddle-plexes data structure used lots of places; see feedmail-fiddle-plex-blurb
;;         feedmail-fiddle-plex-user-list
;;         feedmail-is-a-resend
;;         honor mail-from-style in constructing default for feedmail-from-line
;;         re-implement feedmail-from-line and feedmail-sender-line with

  :type '(choice (const nil) integer)
  )


(defcustom feedmail-nuke-bcc t
  "*If non-nil remove BCC: lines from the message headers.
In any case, the BCC: lines do participate in the composed address
list.  You may want to leave them in if you're using sendmail
\(see feedmail-buffer-eating-function\)."
  :group 'feedmail-headers
  :type 'boolean
  )


(defcustom feedmail-nuke-resent-bcc t
  "*If non-nil remove RESENT-BCC: lines from the message headers.
In any case, the RESENT-BCC: lines do participate in the composed
address list.  You may want to leave them in if you're using sendmail
\(see feedmail-buffer-eating-function\)."
  :group 'feedmail-headers
  :type 'boolean
  )

or, it can be set to the symbol 'last, in which case they will appear
at the end of the list.

Why should you care?  Well, maybe you don't, and certainly the same
things could be accomplished by affecting the order of message headers
in the outgoing message.  Some people use BCC: as a way of getting
their own \"come back\" copy of each message they send.  If BCC:
addresses are not handled first, there can be substantial delays in
seeing the message again.  Some configurations of sendmail, for example,
seem to try to deliver to each addressee at least once, immediately
and serially, so slow SMTP conversations can add up to a delay.  There
is an option for either 'first or 'last because you might have a

(defcustom feedmail-fill-to-cc t
  "*If non-nil do smart filling of addressee header lines.
Smart filling means breaking long lines at appropriate points and
making continuation lines.  Despite the function name, it includes
TO:, CC:, BCC: (and their RESENT-* forms), as well as FROM: and
REPLY-TO: (though they seldom need it).  If nil, the lines are left
as-is.  The filling is done after mail address alias expansion."
  :group 'feedmail-headers
  :type 'boolean
  )

  :type 'integer
  )


(defcustom feedmail-nuke-bcc-in-fcc nil
  "*If non-nil remove [RESENT-]BCC: lines in message copies saved via FCC:.
This is independent of whether the BCC: header lines are actually sent
with the message (see feedmail-nuke-bcc).  Though not implied in the name,
the same FCC: treatment applies to both BCC: and RESENT-BCC: lines."
  :group 'feedmail-headers
  :type 'boolean
  )


(defcustom feedmail-nuke-body-in-fcc nil
  "*If non-nil remove body of message in copies saved via FCC:.
If an positive integer value, leave (up to) that many lines of the
beginning of the body intact.  The result is that the FCC: copy will
consist only of the message headers, serving as a sort of an outgoing
message log."
  :group 'feedmail-headers
  ;;:type 'boolean
  :type '(choice (const nil) (const t) integer)

  )


(defcustom feedmail-nuke-empty-headers t
  "*If non-nil, remove header lines which have no contents.
A completely empty SUBJECT: header is always removed, regardless of
the setting of this variable.  The only time you would want them left
in would be if you used some headers whose presence indicated
something rather than their contents.  This is rare in Internet email
but common in some proprietary systems."
  :group 'feedmail-headers
  :type 'boolean
  )

;; wjc sez:  I think the use of the SENDER: line is pretty pointless,
;; but I left it in to be compatible with sendmail.el and because
;; maybe some distant mail system needs it.  Really, though, if you
;; want a sender line in your mail, just put one in there and don't
;; wait for feedmail to do it for you.  (Yes, I know all about
;; RFC-822 and RFC-1123, but are you *really* one of those cases
;; they're talking about?  I doubt it.)
(defcustom feedmail-sender-line nil
  "*If non-nil and the email has no SENDER: header, use this value.
May be nil, in which case nothing in particular is done with respect
to SENDER: lines.  By design, will not replace an existing SENDER:
line, but you can achieve that with a fiddle-plex 'replace action.
NB: it makes no sense to use the value t since there is no sensible
default for SENDER:.

If not nil, it may be a string, a fiddle-plex, or a function which
returns either nil, t, a string, or a fiddle-plex (or, in fact,
another function, but let's not be ridiculous).  If a string, it
should be just the contents of the header, not the name of the header

cases the name element of the fiddle-plex is ignored and is hardwired
by feedmail to either \"X-Sender\" or \"X-Resent-Sender\".

You can probably leave this nil, but if you feel like using it, a good
value would be a string of a fully-qualified domain name form of your
address.  For example, \"[e-mail] (WJCarpenter)\".  The SENDER:
header is fiddled after the FROM: header is fiddled."
  :group 'feedmail-headers
  :type '(choice (const nil) string)
  )


(defcustom feedmail-force-binary-write t
  "*If non-nil, force writing file as binary (this applies to queues and FCC:).
On systems where there is a difference between binary and text files,
feedmail will temporarily manipulate the values of `buffer-file-type'
and/or default-buffer-file-type to make the writing as binary.  If
nil, writing will be in text mode.  On systems where there is no
distinction or where it is controlled by other variables or other

  :type 'boolean
  )


(defcustom feedmail-from-line t
  "*If non-nil and the email has no FROM: header, use this value.
May be t, in which case a default is computed (and you probably won't
be happy with it).  May be nil, in which case nothing in particular is
done with respect to FROM: lines.  By design, will not replace an
existing FROM: line, but you can achieve that with a fiddle-plex 'replace
action.

If neither nil nor t, it may be a string, a fiddle-plex, or a function
which returns either nil, t, a string, or a fiddle-plex (or, in fact,
another function, but let's not be ridiculous).  If a string, it

your address.  For example, \"[e-mail] (WJCarpenter)\".  The
default value of this variable uses the standard elisp variable
`user-mail-address' which should be set on every system but has a decent
chance of being wrong.  It also honors `mail-from-style'.  Better to set
this variable explicitly to the string you want or find some other way
to arrange for the message to get a FROM: line."
  :group 'feedmail-headers
  :type '(choice (const nil) string)
  )


(defcustom feedmail-deduce-envelope-from t
  "*If non-nil, deduce message envelope \"from\" from header FROM: or SENDER:.
In other words, if there is a SENDER: header in the message, temporarily
change the value of `user-mail-address' to be the same while the message
is being sent.  If there is no SENDER: header, use the FROM: header,
if any.  Address values are taken from the actual message just before
it is sent, and the process is independent of the values of
feedmail-from-line and/or feedmail-sender-line.

There are many and good reasons for having the message header
FROM:/SENDER: be different from the message envelope \"from\"
information.  However, for most people and for most circumstances, it
is usual for them to be the same (this is probably especially true for
the case where the user doesn't understand the difference between the
two in the first place).

The idea behind this feature is that you can have everything set up
some normal way for yourself.  If for some reason you want to send a
message with another FROM: line, you can just type it at the top of
the message, and feedmail will take care of \"fixing up\" the envelope
\"from\".  This only works for mail senders which make use of
`user-mail-address' as the envelope \"from\" value.  For some mail
senders (e.g., feedmail-buffer-to-bin-mail), there is no simple way to
influence what they will use as the envelope."

  :type '(choice (const nil) string)
  )


(defcustom feedmail-x-mailer-line t
  "*Control the form of an X-MAILER: header in an outgoing message.
Moderately useful for debugging, keeping track of your correspondents'
mailer preferences, or just wearing your MUA on your sleeve.  You
should probably know that some people are fairly emotional about the
presence of X-MAILER: lines in email.

If nil, nothing is done about X-MAILER:.

If t, an X-MAILER: header of a predetermined format is produced,
combining its efforts with any existing X-MAILER: header.  If you want
to take the default construct and just add a little blob of your own
at the end, define the variable feedmail-x-mailer-line-user-appendage
as that blob string.  A value of t is equivalent to using the function
feedmail-default-x-mailer-generator.

  :type '(choice (const t) (const nil) string function)
  )


(defcustom feedmail-message-id-generator t
  "*Specifies the creation of a MESSAGE-ID: header field.

If nil, nothing is done about MESSAGE-ID:.

If t, a MESSAGE-ID: header of a predetermined format is produced, but
only if there is not already a MESSAGE-ID: in the message.  A value of
t is equivalent to using the function feedmail-default-message-id-generator.

If neither nil nor t, it may be a string, a fiddle-plex, or a function
which returns either nil, t, a string, or a fiddle-plex (or, in fact,
another function, but let's not be ridiculous).  If a string, it

itself nor the trailing newline.  If a function, it will be called
with one argument: the possibly-nil name of the file associated with
the message buffer.  For an explanation of fiddle-plexes, see the
documentation for the variable feedmail-fiddle-plex-blurb.  In all
cases the name element of the fiddle-plex is ignored and is hardwired
by feedmail to either \"Message-ID\" or \"Resent-Message-ID\".

You should let feedmail generate a MESSAGE-ID: for you unless you are sure
that whatever you give your messages to will do it for you (e.g., most
configurations of sendmail).  Even if the latter case is true, it
probably won't hurt you to generate your own, and it will then show up
in the saved message if you use FCC:."
  :group 'feedmail-headers
  :type '(choice (const nil) function)
  )

  )

;; this was suggested in various forms by several people; first was
;; Tony DeSimone in Oct 1992; sorry to be so tardy
(defcustom feedmail-date-generator t
  "*Specifies the creation of a DATE: header field.

If nil, nothing is done about DATE:.

If t, a DATE: header of a predetermined format is produced, but only
if there is not already a DATE: in the message.  A value of t is
equivalent to using the function feedmail-default-date-generator.

If neither nil nor t, it may be a string, a fiddle-plex, or a function
which returns either nil, t, a string, or a fiddle-plex (or, in fact,
another function, but let's not be ridiculous).  If a string, it

If you decide to format your own date field, do us all a favor and know
what you're doing.  Study the relevant parts of RFC-822 and RFC-1123.
Don't make me come up there!

You should let feedmail generate a DATE: for you unless you are sure
that whatever you give your messages to will do it for you (e.g., most
configurations of sendmail).  Even if the latter case is true, it
probably won't hurt you to generate your own, and it will then show up
in the saved message if you use FCC:."
  :group 'feedmail-headers
  :type '(choice (const nil) function)
  )

message headers except as noted.

Spray mode is usually pointless, and if you can't think of a good reason for
it, you should avoid it since it is inherently less efficient than normal
multiple delivery.  One reason to use it is to overcome mis-featured mail
transports which betray your trust by revealing BCC: addressees in the
headers of a message.  Another use is to do a crude form of mailmerge, for
which see feedmail-spray-address-fiddle-plex-list.

If one of the calls to the buffer-eating function results in an error,
what happens next is carelessly defined, so beware."

`feedmail-fiddle-plex-blurb', for an overview of fiddle-plex data structures.

May be nil, in which case nothing in particular is done about message
headers for specific addresses.

May be t, in which case a \"TO:\" header is added to the message with
the stripped address as the header contents.  The fiddle-plex operator
is 'supplement.

May be a string, in which case the string is assumed to be the name of
a message header field with the stripped address serving as the value.

called and will consult feedmail-spray-this-address to find the
stripped envelope email address (no comments or angle brackets).  The
function should return an embellished form of the address.

The recipe for sending form letters is:  (1) create a message with all
addressees on BCC: headers; (2) tell feedmail to remove BCC: headers
before sending the message; (3) create a function which will embellish
stripped addresses, if desired; (4) define feedmail-spray-address-fiddle-plex-list
appropriately; (5) send the message with feedmail-enable-spray set
non-nil; (6) stand back and watch co-workers wonder at how efficient
you are at accomplishing inherently inefficient things."

(defcustom feedmail-queue-draft-directory
  (if (memq system-type '(axp-vms vax-vms))
      (expand-file-name (concat (getenv "HOME") "[.MAIL.DRAFT]"))
    (concat (getenv "HOME") "/mail/draft"))
  "*Name of an directory where DRAFT messages will be queued.
Directory will be created if necessary.  Should be a string that
doesn't end with a slash.  Default, except on VMS, is \"$HOME/mail/draft\"."
  :group 'feedmail-queue
  :type 'string
  )

  :type '(choice (const nil) function)
  )


(defcustom feedmail-queue-use-send-time-for-date nil
  "*If non-nil, use send time for the DATE: header value.
This variable is used by the default date generating function,
feedmail-default-date-generator.  If nil, the default, the
last-modified timestamp of the queue file is used to create the
message DATE: header; if there is no queue file, the current time is
used."
  :group 'feedmail-queue
  :type 'boolean
  )


(defcustom feedmail-queue-use-send-time-for-message-id nil
  "*If non-nil, use send time for the MESSAGE-ID: header value.
This variable is used by the default MESSAGE-ID: generating function,
feedmail-default-message-id-generator.  If nil, the default, the
last-modified timestamp of the queue file is used to create the
message MESSAGE-ID: header; if there is no queue file, the current time is
used."
  :group 'feedmail-queue
  :type 'boolean
  )

  :type 'hook
  )


(defcustom feedmail-before-fcc-hook nil
  "*User's last opportunity to modify the message before FCC action.
It has already had all the header prepping from the standard package.
The next step after running the hook will be to save the message via
FCC: processing. The hook might be interested in these: (1)
feedmail-prepped-text-buffer contains the header and body of the
message, ready to go; (2) feedmail-address-list contains a list of
simplified recipients of addressees to whom the message was sent (3)
feedmail-error-buffer is an empty buffer intended to soak up errors
for display to the user.  If the hook allows interactive activity, the

for different scenarios.  Users shouldn't set or change this
variable, but may depend on its value as described here.")


(defvar feedmail-is-a-resend nil
  "*Non-nil means the the message is a RESEND (in the RFC-822 sense).
This affects the composition of certain headers.  feedmail sets this
variable as soon as it starts prepping the message text buffer, so any
user-supplied functions can rely on it.  Users shouldn't set or change this
variable, but may depend on its value as described here.")

(defcustom feedmail-binmail-template (if mail-interactive "/bin/mail %s" "/bin/rmail %s")
  "*Command template for the subprocess which will get rid of the mail.
It can result in any command understandable by /bin/sh.  Might not
work at all in non-UNIX environments.  The single '%s', if present,
gets replaced by the space-separated, simplified list of addressees.
Used in feedmail-buffer-to-binmail to form the shell command which
will receive the contents of the prepped buffer as stdin.  If you'd
like your errors to come back as mail instead of immediately in a
buffer, try /bin/rmail instead of /bin/mail (this can be accomplished

			 (mapconcat 'identity addr-listoid " "))))))


(defun feedmail-buffer-to-sendmail (prepped errors-to addr-listoid)
  "Function which actually calls sendmail as a subprocess.
Feeds the buffer to it.  Probably has some flaws for RESENT-* and other
complicated cases."
  (set-buffer prepped)
  (apply 'call-process-region
	 (append (list (point-min) (point-max)
		       (if (boundp 'sendmail-program) sendmail-program "/usr/lib/sendmail")

  )

(defun feedmail-queue-subject-slug-maker (&optional queue-directory)
  "Create a name for storing the message in the queue.
Optional argument QUEUE-DIRECTORY specifies into which directory the
file will be placed.  The name is based on the SUBJECT: header (if
there is one).  If there is no subject,
feedmail-queue-default-file-slug is consulted Special characters are
mapped to mostly alphanumerics for safety."
  (let ((eoh-marker) (case-fold-search t) (subject "") (s-point))
    (setq eoh-marker (feedmail-find-eoh))
    (goto-char (point-min))
    ;; get raw subject value (first line, anyhow)
    (if (re-search-forward "^SUBJECT:" eoh-marker t)
	(progn (setq s-point (point))
	       (end-of-line)
	       (setq subject (buffer-substring s-point (point)))))
    (setq subject (feedmail-tidy-up-slug subject))
    (if (zerop (length subject))

	 (feedmail-raw-text-buffer (current-buffer))
	 (feedmail-address-list)
	 (eoh-marker)
	 (bcc-holder)
	 (resent-bcc-holder)
	 (a-re-rtcb  "^RESENT-\\(TO\\|CC\\|BCC\\):")
	 (a-re-rtc   "^RESENT-\\(TO\\|CC\\):")
	 (a-re-rb    "^RESENT-BCC:")
	 (a-re-dtcb  "^\\(TO\\|CC\\|BCC\\):")
	 (a-re-dtc   "^\\(TO\\|CC\\):")
	 (a-re-db    "^BCC:")
	 ;; to get a temporary changable copy
	 (mail-header-separator mail-header-separator)
	 )
    (unwind-protect
	(save-excursion

	    ;; use resent-* stuff only if there is at least one non-empty one
	    (setq feedmail-is-a-resend
		  (re-search-forward
		   ;; header name, followed by optional whitespace, followed by
		   ;; non-whitespace, followed by anything, followed by newline;
		   ;; the idea is empty RESENT-* headers are ignored
		   "^\\(RESENT-TO:\\|RESENT-CC:\\|RESENT-BCC:\\)\\s-*\\S-+.*$"
		   eoh-marker t))
	    ;; if we say so, gather the BCC stuff before the main course
	    (if (eq feedmail-deduce-bcc-where 'first)
		(progn (if feedmail-is-a-resend (setq addr-regexp a-re-rb) (setq addr-regexp a-re-db))
		       (setq feedmail-address-list (feedmail-deduce-address-list feedmail-prepped-text-buffer (point-min) eoh-marker addr-regexp feedmail-address-list))))
	    ;; the main course
	    (if (or (eq feedmail-deduce-bcc-where 'first) (eq feedmail-deduce-bcc-where 'last))
		;; handled by first or last cases, so don't get BCC stuff
		(progn (if feedmail-is-a-resend (setq addr-regexp a-re-rtc) (setq addr-regexp a-re-dtc))
		       (setq feedmail-address-list (feedmail-deduce-address-list feedmail-prepped-text-buffer (point-min) eoh-marker addr-regexp feedmail-address-list)))
	      ;; not handled by first or last cases, so also get BCC stuff
	      (progn (if feedmail-is-a-resend (setq addr-regexp a-re-rtcb) (setq addr-regexp a-re-dtcb))
		     (setq feedmail-address-list (feedmail-deduce-address-list feedmail-prepped-text-buffer (point-min) eoh-marker addr-regexp feedmail-address-list))))
	    ;; if we say so, gather the BCC stuff after the main course
	    (if (eq feedmail-deduce-bcc-where 'last)
		(progn (if feedmail-is-a-resend (setq addr-regexp a-re-rb) (setq addr-regexp a-re-db))
		       (setq feedmail-address-list (feedmail-deduce-address-list feedmail-prepped-text-buffer (point-min) eoh-marker addr-regexp feedmail-address-list))))
	    (if (not feedmail-address-list) (error "FQM: Sending...abandoned, no addressees"))
	    ;; not needed, but meets user expectations
	    (setq feedmail-address-list (nreverse feedmail-address-list))
	    ;; Find and handle any BCC fields.
	    (setq bcc-holder (feedmail-accume-n-nuke-header eoh-marker "^BCC:"))
	    (setq resent-bcc-holder (feedmail-accume-n-nuke-header eoh-marker "^RESENT-BCC:"))
	    (if (and bcc-holder (not feedmail-nuke-bcc))
		(progn (goto-char (point-min))
		       (insert bcc-holder)))
	    (if (and resent-bcc-holder (not feedmail-nuke-resent-bcc))
		(progn (goto-char (point-min))

		 (while (re-search-forward "^[A-Za-z0-9-]+:[ \t]*\n" eoh-marker t)
		   (replace-match ""))))

	  (run-hooks 'feedmail-last-chance-hook)

	  (let ((fcc (feedmail-accume-n-nuke-header eoh-marker "^FCC:"))
		(also-file)
		(confirm (cond
			  ((eq feedmail-confirm-outgoing 'immediate)
			   (not feedmail-queue-runner-is-active))
			  ((eq feedmail-confirm-outgoing 'queued) feedmail-queue-runner-is-active)

			      (setq buffer-offer-save nil)
			      (setq buffer-file-name nil)
			      )
			  )))
		  (goto-char (point-min))
		  ;; re-insert and handle any FCC fields (and, optionally, any BCC).
		  (if fcc (let ((default-buffer-file-type feedmail-force-binary-write))
			    (insert fcc)
			    (if (not feedmail-nuke-bcc-in-fcc)
				(progn (if bcc-holder (insert bcc-holder))
				       (if resent-bcc-holder (insert resent-bcc-holder))))

		(erase-buffer)
		;; not life's most efficient methodology, but spraying isn't
		;; an every-5-minutes event either
		(insert-buffer feedmail-prepped-text-buffer)
		;; There's a good case to me made that each separate transmission of
		;; a message in the spray should have a distinct MESSAGE-ID:.  There
		;; is also a less compelling argument in the other direction.  I think
		;; they technically should have distinct MESSAGE-ID:s, but I doubt that
		;; anyone cares, practically.  If someone complains about it, I'll add
		;; it.
		(feedmail-fiddle-list-of-spray-fiddle-plexes feedmail-spray-address-fiddle-plex-list)
		;; this (let ) is just in case some buffer eater
		;; is cheating and using the global variable name instead

	       feedmail-address-list))))


(defun feedmail-envelope-deducer (eoh-marker)
  "If feedmail-deduce-envelope-from is false, simply return `user-mail-address'.
Else, look for SENDER: or FROM: (or RESENT-*) and
return that value."
  (if (not feedmail-deduce-envelope-from)
      user-mail-address
    (let ((from-list))
      (setq from-list
	    (feedmail-deduce-address-list
	     (current-buffer) (point-min) eoh-marker (if feedmail-is-a-resend "^RESENT-SENDER:" "^SENDER:")
	     from-list))
      (if (not from-list)
	  (setq from-list
		(feedmail-deduce-address-list
		 (current-buffer) (point-min) eoh-marker (if feedmail-is-a-resend "^RESENT-FROM:" "^FROM:")
		 from-list)))
      (if (and from-list (car from-list)) (car from-list) user-mail-address))))


(defun feedmail-fiddle-from ()
  "Fiddle FROM:."
  ;; default is to fall off the end of the list and do nothing
  (cond
   ;; nil means do nothing
   ((eq nil feedmail-from-line) nil)
   ;; t is the same a using the default computation, so compute it and recurse

     (nth 2 feedmail-from-line)		; action
     (nth 3 feedmail-from-line)))))	; folding


(defun feedmail-fiddle-sender ()
  "Fiddle SENDER:."
  ;; default is to fall off the end of the list and do nothing
  (cond
   ;; nil means do nothing
   ((eq nil feedmail-sender-line) nil)
   ;; t is not allowed, but handled it just to avoid bugs later

     (nth 2 feedmail-sender-line)	; action
     (nth 3 feedmail-sender-line)))))	; folding


(defun feedmail-default-date-generator (maybe-file)
  "Default function for generating DATE: header contents."
  (let ((date-time))
    (if (and (not feedmail-queue-use-send-time-for-date) maybe-file)
	(setq date-time (nth 5 (file-attributes maybe-file))))
    (feedmail-rfc822-date date-time))
  )


(defun feedmail-fiddle-date (maybe-file)
  "Fiddle DATE:.  See documentation of feedmail-date-generator."
  ;; default is to fall off the end of the list and do nothing
  (cond
   ;; nil means do nothing
   ((eq nil feedmail-date-generator) nil)
   ;; t is the same a using the function feedmail-default-date-generator, so let it and recurse

     (nth 2 feedmail-date-generator)	; action
     (nth 3 feedmail-date-generator))))) ; folding


(defun feedmail-default-message-id-generator (maybe-file)
  "Default function for generating MESSAGE-ID: header contents.
Based on a date and a sort of random number for tie breaking.  Unless
feedmail-message-id-suffix is defined, uses `user-mail-address', so be
sure it's set."
  (let ((date-time)
	(end-stuff (if feedmail-message-id-suffix feedmail-message-id-suffix user-mail-address)))

	    (feedmail-rfc822-time-zone date-time)
	    end-stuff))
  )

(defun feedmail-fiddle-message-id (maybe-file)
  "Fiddle MESSAGE-ID:.  See documentation of feedmail-message-id-generator."
  ;; default is to fall off the end of the list and do nothing
  (cond
   ;; nil means do nothing
   ((eq nil feedmail-message-id-generator) nil)
   ;; t is the same a using the function feedmail-default-message-id-generator, so let it and recurse

      (feedmail-fiddle-message-id maybe-file)))

   ;; if it's a list, it must be a fiddle-plex -- so fiddle, man, fiddle
   ((listp feedmail-message-id-generator)
    (feedmail-fiddle-header
     (if feedmail-is-a-resend "Resent-Message-ID" "Message-ID")
     (nth 1 feedmail-message-id-generator) ; value
     (nth 2 feedmail-message-id-generator) ; action
     (nth 3 feedmail-message-id-generator))))) ; folding


(defun feedmail-default-x-mailer-generator ()
  "Default function for generating X-MAILER: header contents."
  (concat
   (let ((case-fold-search t)) (if (string-match "emacs" emacs-version) "" "emacs "))
   emacs-version " (via feedmail " feedmail-patch-level
   (if feedmail-queue-runner-is-active " Q" " I")
   (if feedmail-enable-spray "S" "")
   (if feedmail-x-mailer-line-user-appendage ") " ")")
   feedmail-x-mailer-line-user-appendage))


(defun feedmail-fiddle-x-mailer ()
  "Fiddle X-MAILER:.  See documentation of feedmail-x-mailer-line."
  ;; default is to fall off the end of the list and do nothing
  (cond
   ;; t is the same a using the function feedmail-default-x-mailer-generator, so let it and recurse
   ((eq t feedmail-x-mailer-line)
    (let ((feedmail-x-mailer-line (feedmail-default-x-mailer-generator)))

  "Fiddle header for single spray address.  Uses feedmail-spray-this-address."
  ;; default is to fall off the end of the list and do nothing
  (cond
   ;; nil means do nothing
   ((eq nil addy-plex) nil)
   ;; t means the same as using "TO: and unembellished addy
   ((eq t addy-plex)
    (let ((addy-plex (list "To" feedmail-spray-this-address)))
      (feedmail-fiddle-spray-address addy-plex)))

   ;; if it's a string, simply make a fiddle-plex out of it and recurse, assuming
   ;; the string names a header field (e.g., "TO")
   ((stringp addy-plex)
    (let ((addy-plex (list addy-plex feedmail-spray-this-address)))
      (feedmail-fiddle-spray-address addy-plex)))

   ;; if it's a function, call it and recurse with the resulting value

(defun feedmail-fill-to-cc-function (header-end)
  "Smart filling of address headers (don't be fooled by the name).
The filling tries to avoid splitting lines except at commas.  This
avoids, in particular, splitting within parenthesized comments in
addresses.  Headers filled include FROM:, REPLY-TO:, TO:, CC:, BCC:,
RESENT-TO:, RESENT-CC:, and RESENT-BCC:."
  (let ((case-fold-search t)
	this-line
	this-line-end)
    (save-excursion
      (goto-char (point-min))
      ;; iterate over all TO:/CC:, etc, lines
      (while
	  (re-search-forward
	   "^\\(FROM:\\|REPLY-TO:\\|TO:\\|CC:\\|BCC:\\|RESENT-TO:\\|RESENT-CC:\\|RESENT-BCC:\\)"
	   header-end t)
	(setq this-line (match-beginning 0))
	;; replace 0 or more leading spaces with a single space
	(and (looking-at "[ \t]*") (replace-match " "))
	(forward-line 1)

22801:e71025b5e000

;;        there is also a queue for draft messages
;;
;;    --- you can get one last look at the prepped outbound message and
;;        be prompted for confirmation
;;
;;    --- removes Bcc:/Resent-Bcc: headers after getting address info
;;
;;    --- does smart filling of address headers
;;
;;    --- calls a routine to process Fcc: lines and removes them
;;
;;    --- empty headers are removed
;;
;;    --- can force From: or Sender: line
;;
;;    --- can generate a Message-Id: line
;;
;;    --- can generate a Date: line; the date can be the time the
;;        message was written or the time it is being sent
;;
;;    --- strips comments from address info (both "()" and "<>" are
;;        handled via a call to mail-strip-quoted-names); the
;;        comments are stripped in the simplified address list given

;;        list
;;
;;    --- you can specify something other than /bin/mail for the
;;        subprocess
;;
;;    --- you can generate/modify an X-Mailer: message header
;;
;; After a long list of options below, you will find the function
;; feedmail-send-it. Hers's the best way to use the stuff in this
;; file:
;;

;; patchlevel 4, issued by someone else
;; patchlevel 5, issued by someone else
;; patchlevel 6, not issued as far as I know
;; patchlevel 7,  20 May 1997
;;         abandon futile support of Emacs 18 (sorry if that hurts you)
;;         provide a Date: header by default
;;         provide a default for generating Message-Id: header contents
;;            and use it by default (slightly changed API)
;;         return value from feedmail-run-the-queue
;;         new wrapper function feedmail-run-the-queue-no-prompts
;;         user-mail-address as default for From:
;;         properly deal with Resent-{To,Cc,Bcc}
;;         Bcc and Resent-* now included in smart filling
;;         limited support for a "drafts" directory
;;         user-configurable default message action
;;         allow timeout for confirmation prompt (where available)
;;         move Fcc handling to as late as possible to get max
;;            header munging in the saved file
;;         work around sendmail.el's prompts when working from queue
;;         more reliably detect voluntary user bailouts
;;         offer to save modified buffers visiting queue files
;;         offer to delete old file copies of messages being queued

;;         eliminate reliance on directory-sep-char and feedmail-sep-thing
;;         tweak smart filling (reminded of comma problem by [e-mail])
;;         option to control writing in text vs binary mode
;; patchlevel 8, 15 June 1998
;;         reliable re-editing of text-mode (vs binary) queued messages
;;         user option to keep Bcc: in Fcc: copy (keep by default)
;;         user option to delete body from Fcc: copy (keep by default)
;;         feedmail-deduce-bcc-where for envelope (API change for
;;           feedmail-deduce-address list)
;;         feedmail-queue-alternative-mail-header-separator
;;         at message action prompt, "I"/"S" bypass message confirmation prompt
;;         feedmail-mail-send-hook-splitter, feedmail-mail-send-hook,

;;         feedmail-queue-reminder, feedmail-queue-reminder-alist (after suggestions
;;           and/or code fragments from [e-mail] (Tony Lam) and
;;           [e-mail] (Shane Burgess); bumped up the default value of
;;           feedmail-queue-chatty-sit-for since info is more complex sometimes
;;         feedmail-enable-spray (individual transmissions, crude mailmerge)
;;         blank Subject: no longer a special case; see feedmail-nuke-empty-headers
;;         fiddle-plexes data structure used lots of places; see feedmail-fiddle-plex-blurb
;;         feedmail-fiddle-plex-user-list
;;         feedmail-is-a-resend
;;         honor mail-from-style in constructing default for feedmail-from-line
;;         re-implement feedmail-from-line and feedmail-sender-line with

  :type '(choice (const nil) integer)
  )


(defcustom feedmail-nuke-bcc t
  "*If non-nil remove Bcc: lines from the message headers.
In any case, the Bcc: lines do participate in the composed address
list.  You may want to leave them in if you're using sendmail
\(see feedmail-buffer-eating-function\)."
  :group 'feedmail-headers
  :type 'boolean
  )


(defcustom feedmail-nuke-resent-bcc t
  "*If non-nil remove Resent-Bcc: lines from the message headers.
In any case, the Resent-Bcc: lines do participate in the composed
address list.  You may want to leave them in if you're using sendmail
\(see feedmail-buffer-eating-function\)."
  :group 'feedmail-headers
  :type 'boolean
  )

or, it can be set to the symbol 'last, in which case they will appear
at the end of the list.

Why should you care?  Well, maybe you don't, and certainly the same
things could be accomplished by affecting the order of message headers
in the outgoing message.  Some people use Bcc: as a way of getting
their own \"come back\" copy of each message they send.  If Bcc:
addresses are not handled first, there can be substantial delays in
seeing the message again.  Some configurations of sendmail, for example,
seem to try to deliver to each addressee at least once, immediately
and serially, so slow SMTP conversations can add up to a delay.  There
is an option for either 'first or 'last because you might have a

(defcustom feedmail-fill-to-cc t
  "*If non-nil do smart filling of addressee header lines.
Smart filling means breaking long lines at appropriate points and
making continuation lines.  Despite the function name, it includes
To:, Cc:, Bcc: (and their Resent-* forms), as well as From: and
Reply-To: (though they seldom need it).  If nil, the lines are left
as-is.  The filling is done after mail address alias expansion."
  :group 'feedmail-headers
  :type 'boolean
  )

  :type 'integer
  )


(defcustom feedmail-nuke-bcc-in-fcc nil
  "*If non-nil remove [Resent-]Bcc: lines in message copies saved via Fcc:.
This is independent of whether the Bcc: header lines are actually sent
with the message (see feedmail-nuke-bcc).  Though not implied in the name,
the same Fcc: treatment applies to both Bcc: and Resent-Bcc: lines."
  :group 'feedmail-headers
  :type 'boolean
  )


(defcustom feedmail-nuke-body-in-fcc nil
  "*If non-nil remove body of message in copies saved via Fcc:.
If an positive integer value, leave (up to) that many lines of the
beginning of the body intact.  The result is that the Fcc: copy will
consist only of the message headers, serving as a sort of an outgoing
message log."
  :group 'feedmail-headers
  ;;:type 'boolean
  :type '(choice (const nil) (const t) integer)

  )


(defcustom feedmail-nuke-empty-headers t
  "*If non-nil, remove header lines which have no contents.
A completely empty Subject: header is always removed, regardless of
the setting of this variable.  The only time you would want them left
in would be if you used some headers whose presence indicated
something rather than their contents.  This is rare in Internet email
but common in some proprietary systems."
  :group 'feedmail-headers
  :type 'boolean
  )

;; wjc sez:  I think the use of the Sender: line is pretty pointless,
;; but I left it in to be compatible with sendmail.el and because
;; maybe some distant mail system needs it.  Really, though, if you
;; want a sender line in your mail, just put one in there and don't
;; wait for feedmail to do it for you.  (Yes, I know all about
;; RFC-822 and RFC-1123, but are you *really* one of those cases
;; they're talking about?  I doubt it.)
(defcustom feedmail-sender-line nil
  "*If non-nil and the email has no Sender: header, use this value.
May be nil, in which case nothing in particular is done with respect
to Sender: lines.  By design, will not replace an existing Sender:
line, but you can achieve that with a fiddle-plex 'replace action.
NB: it makes no sense to use the value t since there is no sensible
default for Sender:.

If not nil, it may be a string, a fiddle-plex, or a function which
returns either nil, t, a string, or a fiddle-plex (or, in fact,
another function, but let's not be ridiculous).  If a string, it
should be just the contents of the header, not the name of the header

cases the name element of the fiddle-plex is ignored and is hardwired
by feedmail to either \"X-Sender\" or \"X-Resent-Sender\".

You can probably leave this nil, but if you feel like using it, a good
value would be a string of a fully-qualified domain name form of your
address.  For example, \"[e-mail] (WJCarpenter)\".  The Sender:
header is fiddled after the From: header is fiddled."
  :group 'feedmail-headers
  :type '(choice (const nil) string)
  )


(defcustom feedmail-force-binary-write t
  "*If non-nil, force writing file as binary (this applies to queues and Fcc:).
On systems where there is a difference between binary and text files,
feedmail will temporarily manipulate the values of `buffer-file-type'
and/or default-buffer-file-type to make the writing as binary.  If
nil, writing will be in text mode.  On systems where there is no
distinction or where it is controlled by other variables or other

  :type 'boolean
  )


(defcustom feedmail-from-line t
  "*If non-nil and the email has no From: header, use this value.
May be t, in which case a default is computed (and you probably won't
be happy with it).  May be nil, in which case nothing in particular is
done with respect to From: lines.  By design, will not replace an
existing From: line, but you can achieve that with a fiddle-plex 'replace
action.

If neither nil nor t, it may be a string, a fiddle-plex, or a function
which returns either nil, t, a string, or a fiddle-plex (or, in fact,
another function, but let's not be ridiculous).  If a string, it

your address.  For example, \"[e-mail] (WJCarpenter)\".  The
default value of this variable uses the standard elisp variable
`user-mail-address' which should be set on every system but has a decent
chance of being wrong.  It also honors `mail-from-style'.  Better to set
this variable explicitly to the string you want or find some other way
to arrange for the message to get a From: line."
  :group 'feedmail-headers
  :type '(choice (const nil) string)
  )


(defcustom feedmail-deduce-envelope-from t
  "*If non-nil, deduce message envelope \"from\" from header From: or Sender:.
In other words, if there is a Sender: header in the message, temporarily
change the value of `user-mail-address' to be the same while the message
is being sent.  If there is no Sender: header, use the From: header,
if any.  Address values are taken from the actual message just before
it is sent, and the process is independent of the values of
feedmail-from-line and/or feedmail-sender-line.

There are many and good reasons for having the message header
From:/Sender: be different from the message envelope \"from\"
information.  However, for most people and for most circumstances, it
is usual for them to be the same (this is probably especially true for
the case where the user doesn't understand the difference between the
two in the first place).

The idea behind this feature is that you can have everything set up
some normal way for yourself.  If for some reason you want to send a
message with another From: line, you can just type it at the top of
the message, and feedmail will take care of \"fixing up\" the envelope
\"from\".  This only works for mail senders which make use of
`user-mail-address' as the envelope \"from\" value.  For some mail
senders (e.g., feedmail-buffer-to-bin-mail), there is no simple way to
influence what they will use as the envelope."

  :type '(choice (const nil) string)
  )


(defcustom feedmail-x-mailer-line t
  "*Control the form of an X-Mailer: header in an outgoing message.
Moderately useful for debugging, keeping track of your correspondents'
mailer preferences, or just wearing your MUA on your sleeve.  You
should probably know that some people are fairly emotional about the
presence of X-Mailer: lines in email.

If nil, nothing is done about X-Mailer:.

If t, an X-Mailer: header of a predetermined format is produced,
combining its efforts with any existing X-Mailer: header.  If you want
to take the default construct and just add a little blob of your own
at the end, define the variable feedmail-x-mailer-line-user-appendage
as that blob string.  A value of t is equivalent to using the function
feedmail-default-x-mailer-generator.

  :type '(choice (const t) (const nil) string function)
  )


(defcustom feedmail-message-id-generator t
  "*Specifies the creation of a Message-Id: header field.

If nil, nothing is done about Message-Id:.

If t, a Message-Id: header of a predetermined format is produced, but
only if there is not already a Message-Id: in the message.  A value of
t is equivalent to using the function feedmail-default-message-id-generator.

If neither nil nor t, it may be a string, a fiddle-plex, or a function
which returns either nil, t, a string, or a fiddle-plex (or, in fact,
another function, but let's not be ridiculous).  If a string, it

itself nor the trailing newline.  If a function, it will be called
with one argument: the possibly-nil name of the file associated with
the message buffer.  For an explanation of fiddle-plexes, see the
documentation for the variable feedmail-fiddle-plex-blurb.  In all
cases the name element of the fiddle-plex is ignored and is hardwired
by feedmail to either \"Message-Id\" or \"Resent-Message-Id\".

You should let feedmail generate a Message-Id: for you unless you are sure
that whatever you give your messages to will do it for you (e.g., most
configurations of sendmail).  Even if the latter case is true, it
probably won't hurt you to generate your own, and it will then show up
in the saved message if you use Fcc:."
  :group 'feedmail-headers
  :type '(choice (const nil) function)
  )

  )

;; this was suggested in various forms by several people; first was
;; Tony DeSimone in Oct 1992; sorry to be so tardy
(defcustom feedmail-date-generator t
  "*Specifies the creation of a Date: header field.

If nil, nothing is done about Date:.

If t, a Date: header of a predetermined format is produced, but only
if there is not already a Date: in the message.  A value of t is
equivalent to using the function feedmail-default-date-generator.

If neither nil nor t, it may be a string, a fiddle-plex, or a function
which returns either nil, t, a string, or a fiddle-plex (or, in fact,
another function, but let's not be ridiculous).  If a string, it

If you decide to format your own date field, do us all a favor and know
what you're doing.  Study the relevant parts of RFC-822 and RFC-1123.
Don't make me come up there!

You should let feedmail generate a Date: for you unless you are sure
that whatever you give your messages to will do it for you (e.g., most
configurations of sendmail).  Even if the latter case is true, it
probably won't hurt you to generate your own, and it will then show up
in the saved message if you use Fcc:."
  :group 'feedmail-headers
  :type '(choice (const nil) function)
  )

message headers except as noted.

Spray mode is usually pointless, and if you can't think of a good reason for
it, you should avoid it since it is inherently less efficient than normal
multiple delivery.  One reason to use it is to overcome mis-featured mail
transports which betray your trust by revealing Bcc: addressees in the
headers of a message.  Another use is to do a crude form of mailmerge, for
which see feedmail-spray-address-fiddle-plex-list.

If one of the calls to the buffer-eating function results in an error,
what happens next is carelessly defined, so beware."

`feedmail-fiddle-plex-blurb', for an overview of fiddle-plex data structures.

May be nil, in which case nothing in particular is done about message
headers for specific addresses.

May be t, in which case a \"To:\" header is added to the message with
the stripped address as the header contents.  The fiddle-plex operator
is 'supplement.

May be a string, in which case the string is assumed to be the name of
a message header field with the stripped address serving as the value.

called and will consult feedmail-spray-this-address to find the
stripped envelope email address (no comments or angle brackets).  The
function should return an embellished form of the address.

The recipe for sending form letters is:  (1) create a message with all
addressees on Bcc: headers; (2) tell feedmail to remove Bcc: headers
before sending the message; (3) create a function which will embellish
stripped addresses, if desired; (4) define feedmail-spray-address-fiddle-plex-list
appropriately; (5) send the message with feedmail-enable-spray set
non-nil; (6) stand back and watch co-workers wonder at how efficient
you are at accomplishing inherently inefficient things."

(defcustom feedmail-queue-draft-directory
  (if (memq system-type '(axp-vms vax-vms))
      (expand-file-name (concat (getenv "HOME") "[.MAIL.DRAFT]"))
    (concat (getenv "HOME") "/mail/draft"))
  "*Name of an directory where draft messages will be queued.
Directory will be created if necessary.  Should be a string that
doesn't end with a slash.  Default, except on VMS, is \"$HOME/mail/draft\"."
  :group 'feedmail-queue
  :type 'string
  )

  :type '(choice (const nil) function)
  )


(defcustom feedmail-queue-use-send-time-for-date nil
  "*If non-nil, use send time for the Date: header value.
This variable is used by the default date generating function,
feedmail-default-date-generator.  If nil, the default, the
last-modified timestamp of the queue file is used to create the
message Date: header; if there is no queue file, the current time is
used."
  :group 'feedmail-queue
  :type 'boolean
  )


(defcustom feedmail-queue-use-send-time-for-message-id nil
  "*If non-nil, use send time for the Message-Id: header value.
This variable is used by the default Message-Id: generating function,
feedmail-default-message-id-generator.  If nil, the default, the
last-modified timestamp of the queue file is used to create the
message Message-Id: header; if there is no queue file, the current time is
used."
  :group 'feedmail-queue
  :type 'boolean
  )

  :type 'hook
  )


(defcustom feedmail-before-fcc-hook nil
  "*User's last opportunity to modify the message before Fcc action.
It has already had all the header prepping from the standard package.
The next step after running the hook will be to save the message via
Fcc: processing. The hook might be interested in these: (1)
feedmail-prepped-text-buffer contains the header and body of the
message, ready to go; (2) feedmail-address-list contains a list of
simplified recipients of addressees to whom the message was sent (3)
feedmail-error-buffer is an empty buffer intended to soak up errors
for display to the user.  If the hook allows interactive activity, the

for different scenarios.  Users shouldn't set or change this
variable, but may depend on its value as described here.")


(defvar feedmail-is-a-resend nil
  "*Non-nil means the the message is a Resend (in the RFC-822 sense).
This affects the composition of certain headers.  feedmail sets this
variable as soon as it starts prepping the message text buffer, so any
user-supplied functions can rely on it.  Users shouldn't set or change this
variable, but may depend on its value as described here.")

(defcustom feedmail-binmail-template (if mail-interactive "/bin/mail %s" "/bin/rmail %s")
  "*Command template for the subprocess which will get rid of the mail.
It can result in any command understandable by /bin/sh.  Might not
work at all in non-Unix environments.  The single '%s', if present,
gets replaced by the space-separated, simplified list of addressees.
Used in feedmail-buffer-to-binmail to form the shell command which
will receive the contents of the prepped buffer as stdin.  If you'd
like your errors to come back as mail instead of immediately in a
buffer, try /bin/rmail instead of /bin/mail (this can be accomplished

			 (mapconcat 'identity addr-listoid " "))))))


(defun feedmail-buffer-to-sendmail (prepped errors-to addr-listoid)
  "Function which actually calls sendmail as a subprocess.
Feeds the buffer to it.  Probably has some flaws for Resent-* and other
complicated cases."
  (set-buffer prepped)
  (apply 'call-process-region
	 (append (list (point-min) (point-max)
		       (if (boundp 'sendmail-program) sendmail-program "/usr/lib/sendmail")

  )

(defun feedmail-queue-subject-slug-maker (&optional queue-directory)
  "Create a name for storing the message in the queue.
Optional argument QUEUE-DIRECTORY specifies into which directory the
file will be placed.  The name is based on the Subject: header (if
there is one).  If there is no subject,
feedmail-queue-default-file-slug is consulted Special characters are
mapped to mostly alphanumerics for safety."
  (let ((eoh-marker) (case-fold-search t) (subject "") (s-point))
    (setq eoh-marker (feedmail-find-eoh))
    (goto-char (point-min))
    ;; get raw subject value (first line, anyhow)
    (if (re-search-forward "^Subject:" eoh-marker t)
	(progn (setq s-point (point))
	       (end-of-line)
	       (setq subject (buffer-substring s-point (point)))))
    (setq subject (feedmail-tidy-up-slug subject))
    (if (zerop (length subject))

	 (feedmail-raw-text-buffer (current-buffer))
	 (feedmail-address-list)
	 (eoh-marker)
	 (bcc-holder)
	 (resent-bcc-holder)
	 (a-re-rtcb  "^Resent-\\(To\\|Cc\\|Bcc\\):")
	 (a-re-rtc   "^Resent-\\(To\\|Cc\\):")
	 (a-re-rb    "^Resent-Bcc:")
	 (a-re-dtcb  "^\\(To\\|Cc\\|Bcc\\):")
	 (a-re-dtc   "^\\(To\\|Cc\\):")
	 (a-re-db    "^Bcc:")
	 ;; to get a temporary changable copy
	 (mail-header-separator mail-header-separator)
	 )
    (unwind-protect
	(save-excursion

	    ;; use resent-* stuff only if there is at least one non-empty one
	    (setq feedmail-is-a-resend
		  (re-search-forward
		   ;; header name, followed by optional whitespace, followed by
		   ;; non-whitespace, followed by anything, followed by newline;
		   ;; the idea is empty Resent-* headers are ignored
		   "^\\(Resent-To:\\|Resent-Cc:\\|Resent-Bcc:\\)\\s-*\\S-+.*$"
		   eoh-marker t))
	    ;; if we say so, gather the Bcc stuff before the main course
	    (if (eq feedmail-deduce-bcc-where 'first)
		(progn (if feedmail-is-a-resend (setq addr-regexp a-re-rb) (setq addr-regexp a-re-db))
		       (setq feedmail-address-list (feedmail-deduce-address-list feedmail-prepped-text-buffer (point-min) eoh-marker addr-regexp feedmail-address-list))))
	    ;; the main course
	    (if (or (eq feedmail-deduce-bcc-where 'first) (eq feedmail-deduce-bcc-where 'last))
		;; handled by first or last cases, so don't get Bcc stuff
		(progn (if feedmail-is-a-resend (setq addr-regexp a-re-rtc) (setq addr-regexp a-re-dtc))
		       (setq feedmail-address-list (feedmail-deduce-address-list feedmail-prepped-text-buffer (point-min) eoh-marker addr-regexp feedmail-address-list)))
	      ;; not handled by first or last cases, so also get Bcc stuff
	      (progn (if feedmail-is-a-resend (setq addr-regexp a-re-rtcb) (setq addr-regexp a-re-dtcb))
		     (setq feedmail-address-list (feedmail-deduce-address-list feedmail-prepped-text-buffer (point-min) eoh-marker addr-regexp feedmail-address-list))))
	    ;; if we say so, gather the Bcc stuff after the main course
	    (if (eq feedmail-deduce-bcc-where 'last)
		(progn (if feedmail-is-a-resend (setq addr-regexp a-re-rb) (setq addr-regexp a-re-db))
		       (setq feedmail-address-list (feedmail-deduce-address-list feedmail-prepped-text-buffer (point-min) eoh-marker addr-regexp feedmail-address-list))))
	    (if (not feedmail-address-list) (error "FQM: Sending...abandoned, no addressees"))
	    ;; not needed, but meets user expectations
	    (setq feedmail-address-list (nreverse feedmail-address-list))
	    ;; Find and handle any Bcc fields.
	    (setq bcc-holder (feedmail-accume-n-nuke-header eoh-marker "^Bcc:"))
	    (setq resent-bcc-holder (feedmail-accume-n-nuke-header eoh-marker "^Resent-Bcc:"))
	    (if (and bcc-holder (not feedmail-nuke-bcc))
		(progn (goto-char (point-min))
		       (insert bcc-holder)))
	    (if (and resent-bcc-holder (not feedmail-nuke-resent-bcc))
		(progn (goto-char (point-min))

		 (while (re-search-forward "^[A-Za-z0-9-]+:[ \t]*\n" eoh-marker t)
		   (replace-match ""))))

	  (run-hooks 'feedmail-last-chance-hook)

	  (let ((fcc (feedmail-accume-n-nuke-header eoh-marker "^Fcc:"))
		(also-file)
		(confirm (cond
			  ((eq feedmail-confirm-outgoing 'immediate)
			   (not feedmail-queue-runner-is-active))
			  ((eq feedmail-confirm-outgoing 'queued) feedmail-queue-runner-is-active)

			      (setq buffer-offer-save nil)
			      (setq buffer-file-name nil)
			      )
			  )))
		  (goto-char (point-min))
		  ;; re-insert and handle any Fcc fields (and, optionally, any Bcc).
		  (if fcc (let ((default-buffer-file-type feedmail-force-binary-write))
			    (insert fcc)
			    (if (not feedmail-nuke-bcc-in-fcc)
				(progn (if bcc-holder (insert bcc-holder))
				       (if resent-bcc-holder (insert resent-bcc-holder))))

		(erase-buffer)
		;; not life's most efficient methodology, but spraying isn't
		;; an every-5-minutes event either
		(insert-buffer feedmail-prepped-text-buffer)
		;; There's a good case to me made that each separate transmission of
		;; a message in the spray should have a distinct Message-Id:.  There
		;; is also a less compelling argument in the other direction.  I think
		;; they technically should have distinct Message-Id:s, but I doubt that
		;; anyone cares, practically.  If someone complains about it, I'll add
		;; it.
		(feedmail-fiddle-list-of-spray-fiddle-plexes feedmail-spray-address-fiddle-plex-list)
		;; this (let ) is just in case some buffer eater
		;; is cheating and using the global variable name instead

	       feedmail-address-list))))


(defun feedmail-envelope-deducer (eoh-marker)
  "If feedmail-deduce-envelope-from is false, simply return `user-mail-address'.
Else, look for Sender: or From: (or Resent-*) and
return that value."
  (if (not feedmail-deduce-envelope-from)
      user-mail-address
    (let ((from-list))
      (setq from-list
	    (feedmail-deduce-address-list
	     (current-buffer) (point-min) eoh-marker (if feedmail-is-a-resend "^Resent-Sender:" "^Sender:")
	     from-list))
      (if (not from-list)
	  (setq from-list
		(feedmail-deduce-address-list
		 (current-buffer) (point-min) eoh-marker (if feedmail-is-a-resend "^Resent-From:" "^From:")
		 from-list)))
      (if (and from-list (car from-list)) (car from-list) user-mail-address))))


(defun feedmail-fiddle-from ()
  "Fiddle From:."
  ;; default is to fall off the end of the list and do nothing
  (cond
   ;; nil means do nothing
   ((eq nil feedmail-from-line) nil)
   ;; t is the same a using the default computation, so compute it and recurse

     (nth 2 feedmail-from-line)		; action
     (nth 3 feedmail-from-line)))))	; folding


(defun feedmail-fiddle-sender ()
  "Fiddle Sender:."
  ;; default is to fall off the end of the list and do nothing
  (cond
   ;; nil means do nothing
   ((eq nil feedmail-sender-line) nil)
   ;; t is not allowed, but handled it just to avoid bugs later

     (nth 2 feedmail-sender-line)	; action
     (nth 3 feedmail-sender-line)))))	; folding


(defun feedmail-default-date-generator (maybe-file)
  "Default function for generating Date: header contents."
  (let ((date-time))
    (if (and (not feedmail-queue-use-send-time-for-date) maybe-file)
	(setq date-time (nth 5 (file-attributes maybe-file))))
    (feedmail-rfc822-date date-time))
  )


(defun feedmail-fiddle-date (maybe-file)
  "Fiddle Date:.  See documentation of feedmail-date-generator."
  ;; default is to fall off the end of the list and do nothing
  (cond
   ;; nil means do nothing
   ((eq nil feedmail-date-generator) nil)
   ;; t is the same a using the function feedmail-default-date-generator, so let it and recurse

     (nth 2 feedmail-date-generator)	; action
     (nth 3 feedmail-date-generator))))) ; folding


(defun feedmail-default-message-id-generator (maybe-file)
  "Default function for generating Message-Id: header contents.
Based on a date and a sort of random number for tie breaking.  Unless
feedmail-message-id-suffix is defined, uses `user-mail-address', so be
sure it's set."
  (let ((date-time)
	(end-stuff (if feedmail-message-id-suffix feedmail-message-id-suffix user-mail-address)))

	    (feedmail-rfc822-time-zone date-time)
	    end-stuff))
  )

(defun feedmail-fiddle-message-id (maybe-file)
  "Fiddle Message-Id:.  See documentation of feedmail-message-id-generator."
  ;; default is to fall off the end of the list and do nothing
  (cond
   ;; nil means do nothing
   ((eq nil feedmail-message-id-generator) nil)
   ;; t is the same a using the function feedmail-default-message-id-generator, so let it and recurse

      (feedmail-fiddle-message-id maybe-file)))

   ;; if it's a list, it must be a fiddle-plex -- so fiddle, man, fiddle
   ((listp feedmail-message-id-generator)
    (feedmail-fiddle-header
     (if feedmail-is-a-resend "Resent-Message-Id" "Message-Id")
     (nth 1 feedmail-message-id-generator) ; value
     (nth 2 feedmail-message-id-generator) ; action
     (nth 3 feedmail-message-id-generator))))) ; folding


(defun feedmail-default-x-mailer-generator ()
  "Default function for generating X-Mailer: header contents."
  (concat
   (let ((case-fold-search t)) (if (string-match "emacs" emacs-version) "" "emacs "))
   emacs-version " (via feedmail " feedmail-patch-level
   (if feedmail-queue-runner-is-active " Q" " I")
   (if feedmail-enable-spray "S" "")
   (if feedmail-x-mailer-line-user-appendage ") " ")")
   feedmail-x-mailer-line-user-appendage))


(defun feedmail-fiddle-x-mailer ()
  "Fiddle X-Mailer:.  See documentation of feedmail-x-mailer-line."
  ;; default is to fall off the end of the list and do nothing
  (cond
   ;; t is the same a using the function feedmail-default-x-mailer-generator, so let it and recurse
   ((eq t feedmail-x-mailer-line)
    (let ((feedmail-x-mailer-line (feedmail-default-x-mailer-generator)))

  "Fiddle header for single spray address.  Uses feedmail-spray-this-address."
  ;; default is to fall off the end of the list and do nothing
  (cond
   ;; nil means do nothing
   ((eq nil addy-plex) nil)
   ;; t means the same as using "To:" and unembellished addy
   ((eq t addy-plex)
    (let ((addy-plex (list "To" feedmail-spray-this-address)))
      (feedmail-fiddle-spray-address addy-plex)))

   ;; if it's a string, simply make a fiddle-plex out of it and recurse, assuming
   ;; the string names a header field (e.g., "To")
   ((stringp addy-plex)
    (let ((addy-plex (list addy-plex feedmail-spray-this-address)))
      (feedmail-fiddle-spray-address addy-plex)))

   ;; if it's a function, call it and recurse with the resulting value

(defun feedmail-fill-to-cc-function (header-end)
  "Smart filling of address headers (don't be fooled by the name).
The filling tries to avoid splitting lines except at commas.  This
avoids, in particular, splitting within parenthesized comments in
addresses.  Headers filled include From:, Reply-To:, To:, Cc:, Bcc:,
Resent-To:, Resent-Cc:, and Resent-Bcc:."
  (let ((case-fold-search t)
	this-line
	this-line-end)
    (save-excursion
      (goto-char (point-min))
      ;; iterate over all To:/Cc:, etc, lines
      (while
	  (re-search-forward
	   "^\\(From:\\|Reply-To:\\|To:\\|Cc:\\|Bcc:\\|Resent-To:\\|Resent-Cc:\\|Resent-Bcc:\\)"
	   header-end t)
	(setq this-line (match-beginning 0))
	;; replace 0 or more leading spaces with a single space
	(and (looking-at "[ \t]*") (replace-match " "))
	(forward-line 1)