- (set-syntax-table address-syntax-table))
- ;; commas delimit addresses when outside < > pairs.
- ((and (eq char ?,)
- (or (null <-pos)
- (and >-pos
- ;; handle weird munged addresses
- (> (mail-last-element <-pos) (car >-pos)))))
- (setq multiple-addresses t)
- (delete-char 1)
- (narrow-to-region (point-min) (point)))
- ;; record the position of various interesting chars, determine
- ;; legality later.
- ((setq record-pos-symbol
- (cdr (assq char
- '((?< . <-pos) (?> . >-pos) (?@ . @-pos)
- (?: . :-pos) (?, . ,-pos) (?! . !-pos)
- (?% . %-pos) (?\; . \;-pos)))))
- (set record-pos-symbol
- (cons (point) (symbol-value record-pos-symbol)))
- (forward-char 1))
- ((eq char ?.)
- (forward-char 1))
- ((memq char '(
- ;; comment terminator illegal
- ?\)
- ;; domain literal terminator illegal
- ?\]
- ;; \ allowed only within quoted strings,
- ;; domain literals, and comments
- ?\\
- ))
- (mail-nuke-char-at (point))
- (forward-char 1))
- (t
- (forward-word 1)))
- (or (eq char ?\()
- (setq last-real-pos (point))))
-
- ;; Use only the leftmost <, if any. Replace all others with spaces.
- (while (cdr <-pos)
- (mail-nuke-char-at (car <-pos))
- (setq <-pos (cdr <-pos)))
-
- ;; Use only the rightmost >, if any. Replace all others with spaces.
- (while (cdr >-pos)
- (mail-nuke-char-at (nth 1 >-pos))
- (setcdr >-pos (nthcdr 2 >-pos)))
-
- ;; If multiple @s and a :, but no < and >, insert around buffer.
- ;; This commonly happens on the UUCP "From " line. Ugh.
- (cond ((and (> (length @-pos) 1)
- :-pos ;TODO: check if between @s
- (not <-pos))
- (goto-char (point-min))
- (delete-char 1)
- (setq <-pos (list (point)))
- (insert ?<)))
-
- ;; If < but no >, insert > in rightmost possible position
- (cond ((and <-pos
- (null >-pos))
- (goto-char (point-max))
- (setq >-pos (list (point)))
- (insert ?>)))
-
- ;; If > but no <, replace > with space.
- (cond ((and >-pos
- (null <-pos))
- (mail-nuke-char-at (car >-pos))
- (setq >-pos nil)))
-
- ;; Turn >-pos and <-pos into non-lists
- (setq >-pos (car >-pos)
- <-pos (car <-pos))
-
- ;; Trim other punctuation lists of items outside < > pair to handle
- ;; stupid MTAs.
- (cond (<-pos ; don't need to check >-pos also
- ;; handle bozo software that violates RFC 822 by sticking
- ;; punctuation marks outside of a < > pair
- (mail-nuke-elements-outside-range @-pos <-pos >-pos t)
- ;; RFC 822 says nothing about these two outside < >, but
- ;; remove those positions from the lists to make things
- ;; easier.
- (mail-nuke-elements-outside-range !-pos <-pos >-pos t)
- (mail-nuke-elements-outside-range %-pos <-pos >-pos t)))
-
- ;; Check for : that indicates GROUP list and for : part of
- ;; ROUTE-ADDR spec.
- ;; Can't possibly be more than two :. Nuke any extra.
- (while :-pos
- (setq temp (car :-pos)
- :-pos (cdr :-pos))
- (cond ((and <-pos >-pos
- (> temp <-pos)
- (< temp >-pos))
- (if (or route-addr-:-pos
- (< (length @-pos) 2)
- (> temp (car @-pos))
- (< temp (nth 1 @-pos)))
- (mail-nuke-char-at temp)
- (setq route-addr-:-pos temp)))
- ((or (not <-pos)
- (and <-pos
- (< temp <-pos)))
- (setq group-:-pos temp))))
-
- ;; Nuke any ; that is in or to the left of a < > pair or to the left
- ;; of a GROUP starting :. Also, there may only be one ;.
- (while \;-pos
- (setq temp (car \;-pos)
- \;-pos (cdr \;-pos))
- (cond ((and <-pos >-pos
- (> temp <-pos)
- (< temp >-pos))
- (mail-nuke-char-at temp))
- ((and (or (not group-:-pos)
- (> temp group-:-pos))
- (not group-\;-pos))
- (setq group-\;-pos temp))))
-
- ;; Handle junk like ";@host.company.dom" that sendmail adds.
- ;; **** should I remember comment positions?
- (and group-\;-pos
- ;; this is fine for now
- (mail-nuke-elements-outside-range !-pos group-:-pos group-\;-pos t)
- (mail-nuke-elements-outside-range @-pos group-:-pos group-\;-pos t)
- (mail-nuke-elements-outside-range %-pos group-:-pos group-\;-pos t)
- (mail-nuke-elements-outside-range ,-pos group-:-pos group-\;-pos t)
- (and last-real-pos
- (> last-real-pos (1+ group-\;-pos))
- (setq last-real-pos (1+ group-\;-pos)))
- (and comment-end
- (> comment-end group-\;-pos)
- (setq comment-end nil
- comment-beg nil))
- (and quote-end
- (> quote-end group-\;-pos)
- (setq quote-end nil
- quote-beg nil))
- (narrow-to-region (point-min) group-\;-pos))
-
- ;; Any commas must be between < and : of ROUTE-ADDR. Nuke any
- ;; others.
- ;; Hell, go ahead an nuke all of the commas.
- ;; **** This will cause problems when we start handling commas in
- ;; the PHRASE part .... no it won't ... yes it will ... ?????
- (mail-nuke-elements-outside-range ,-pos 1 1)
-
- ;; can only have multiple @s inside < >. The fact that some MTAs
- ;; put de-bracketed ROUTE-ADDRs in the UUCP-style "From " line is
- ;; handled above.
-
- ;; Locate PHRASE part of ROUTE-ADDR.
- (cond (<-pos
- (goto-char <-pos)
- (skip-chars-backward mail-whitespace)
- (setq phrase-end (point))
- (goto-char (or ;;group-:-pos
- (point-min)))
- (skip-chars-forward mail-whitespace)
- (if (< (point) phrase-end)
- (setq phrase-beg (point))
- (setq phrase-end nil))))
-
- ;; handle ROUTE-ADDRS with real ROUTEs.
- ;; If there are multiple @s, then we assume ROUTE-ADDR syntax, and
- ;; any % or ! must be semantically meaningless.
- ;; TODO: do this processing into canonicalization buffer
- (cond (route-addr-:-pos
- (setq !-pos nil
- %-pos nil
- >-pos (copy-marker >-pos)
- route-addr-:-pos (copy-marker route-addr-:-pos))
- (goto-char >-pos)
- (insert-before-markers ?X)
- (goto-char (car @-pos))
- (while (setq @-pos (cdr @-pos))
- (delete-char 1)
- (setq %-pos (cons (point-marker) %-pos))
- (insert "%")
- (goto-char (1- >-pos))
- (save-excursion
- (insert-buffer-substring extraction-buffer
- (car @-pos) route-addr-:-pos)
- (delete-region (car @-pos) route-addr-:-pos))
- (or (cdr @-pos)
- (setq saved-@-pos (list (point)))))
- (setq @-pos saved-@-pos)
- (goto-char >-pos)
- (delete-char -1)
- (mail-nuke-char-at route-addr-:-pos)
- (mail-demarkerize route-addr-:-pos)
- (setq route-addr-:-pos nil
- >-pos (mail-demarkerize >-pos)
- %-pos (mapcar 'mail-demarkerize %-pos))))
-
- ;; de-listify @-pos
- (setq @-pos (car @-pos))
-
- ;; TODO: remove comments in the middle of an address
-
- (set-buffer canonicalization-buffer)
-
- (buffer-flush-undo canonicalization-buffer)
- (set-syntax-table address-syntax-table)
- (setq case-fold-search nil)
-
- (widen)
- (erase-buffer)
- (insert-buffer-substring extraction-buffer)
-
- (if <-pos
- (narrow-to-region (progn
- (goto-char (1+ <-pos))
- (skip-chars-forward mail-whitespace)
- (point))
- >-pos)
- ;; ****** Oh no! What if the address is completely empty!
- (narrow-to-region first-real-pos last-real-pos))
-
- (and @-pos %-pos
- (mail-nuke-elements-outside-range %-pos (point-min) @-pos))
- (and %-pos !-pos
- (mail-nuke-elements-outside-range !-pos (point-min) (car %-pos)))
- (and @-pos !-pos (not %-pos)
- (mail-nuke-elements-outside-range !-pos (point-min) @-pos))
-
- ;; Error condition:?? (and %-pos (not @-pos))
-
- (cond (!-pos
- ;; **** I don't understand this save-restriction and the
- ;; narrow-to-region inside it. Why did I do that?
- (save-restriction
- (cond ((and @-pos
- mail-@-binds-tighter-than-!)
- (goto-char @-pos)
- (setq %-pos (cons (point) %-pos)
- @-pos nil)
- (delete-char 1)
- (insert "%")
- (setq insert-point (point-max)))
- (mail-@-binds-tighter-than-!
- (setq insert-point (point-max)))
- (%-pos
- (setq insert-point (mail-last-element %-pos)
- saved-%-pos (mapcar 'mail-markerize %-pos)
- %-pos nil
- @-pos (mail-markerize @-pos)))
- (@-pos
- (setq insert-point @-pos)
- (setq @-pos (mail-markerize @-pos)))
- (t
- (setq insert-point (point-max))))
- (narrow-to-region (point-min) insert-point)
- (setq saved-!-pos (car !-pos))
- (while !-pos
+ (t
+ (forward-word 1)))
+ (or (eq char ?\()
+ ;; At the end of first address of a multiple address header.
+ (and (eq char ?,)
+ (eobp))
+ (setq last-real-pos (point))))
+
+ ;; Use only the leftmost <, if any. Replace all others with spaces.
+ (while (cdr <-pos)
+ (mail-extr-nuke-char-at (car <-pos))
+ (setq <-pos (cdr <-pos)))
+
+ ;; Use only the rightmost >, if any. Replace all others with spaces.
+ (while (cdr >-pos)
+ (mail-extr-nuke-char-at (nth 1 >-pos))
+ (setcdr >-pos (nthcdr 2 >-pos)))
+
+ ;; If multiple @s and a :, but no < and >, insert around buffer.
+ ;; Example: @foo.bar.dom,@xxx.yyy.zzz:mailbox@aaa.bbb.ccc
+ ;; This commonly happens on the UUCP "From " line. Ugh.
+ (when (and (> (length @-pos) 1)
+ (eq 1 (length colon-pos)) ;TODO: check if between last two @s
+ (not \;-pos)
+ (not <-pos))
+ (goto-char (point-min))
+ (delete-char 1)
+ (setq <-pos (list (point)))
+ (insert ?<))
+
+ ;; If < but no >, insert > in rightmost possible position
+ (when (and <-pos (null >-pos))
+ (goto-char (point-max))
+ (setq >-pos (list (point)))
+ (insert ?>))
+
+ ;; If > but no <, replace > with space.
+ (when (and >-pos (null <-pos))
+ (mail-extr-nuke-char-at (car >-pos))
+ (setq >-pos nil))
+
+ ;; Turn >-pos and <-pos into non-lists
+ (setq >-pos (car >-pos)
+ <-pos (car <-pos))
+
+ ;; Trim other punctuation lists of items outside < > pair to handle
+ ;; stupid MTAs.
+ (when <-pos ; don't need to check >-pos also
+ ;; handle bozo software that violates RFC 822 by sticking
+ ;; punctuation marks outside of a < > pair
+ (mail-extr-nuke-outside-range @-pos <-pos >-pos t)
+ ;; RFC 822 says nothing about these two outside < >, but
+ ;; remove those positions from the lists to make things
+ ;; easier.
+ (mail-extr-nuke-outside-range !-pos <-pos >-pos t)
+ (mail-extr-nuke-outside-range %-pos <-pos >-pos t))
+
+ ;; Check for : that indicates GROUP list and for : part of
+ ;; ROUTE-ADDR spec.
+ ;; Can't possibly be more than two :. Nuke any extra.
+ (while colon-pos
+ (setq temp (car colon-pos)
+ colon-pos (cdr colon-pos))
+ (cond ((and <-pos >-pos
+ (> temp <-pos)
+ (< temp >-pos))
+ (if (or route-addr-:-pos
+ (< (length @-pos) 2)
+ (> temp (car @-pos))
+ (< temp (nth 1 @-pos)))
+ (mail-extr-nuke-char-at temp)
+ (setq route-addr-:-pos temp)))
+ ((or (not <-pos)
+ (and <-pos
+ (< temp <-pos)))
+ (setq group-:-pos temp))))
+
+ ;; Nuke any ; that is in or to the left of a < > pair or to the left
+ ;; of a GROUP starting :. Also, there may only be one ;.
+ (while \;-pos
+ (setq temp (car \;-pos)
+ \;-pos (cdr \;-pos))
+ (cond ((and <-pos >-pos
+ (> temp <-pos)
+ (< temp >-pos))
+ (mail-extr-nuke-char-at temp))
+ ((and (or (not group-:-pos)
+ (> temp group-:-pos))
+ (not group-\;-pos))
+ (setq group-\;-pos temp))))
+
+ ;; Nuke unmatched GROUP syntax characters.
+ (when (and group-:-pos (not group-\;-pos))
+ ;; *** Do I really need to erase it?
+ (mail-extr-nuke-char-at group-:-pos)
+ (setq group-:-pos nil))
+ (when (and group-\;-pos (not group-:-pos))
+ ;; *** Do I really need to erase it?
+ (mail-extr-nuke-char-at group-\;-pos)
+ (setq group-\;-pos nil))
+
+ ;; Handle junk like ";@host.company.dom" that sendmail adds.
+ ;; **** should I remember comment positions?
+ (when group-\;-pos
+ ;; this is fine for now
+ (mail-extr-nuke-outside-range !-pos group-:-pos group-\;-pos t)
+ (mail-extr-nuke-outside-range @-pos group-:-pos group-\;-pos t)
+ (mail-extr-nuke-outside-range %-pos group-:-pos group-\;-pos t)
+ (mail-extr-nuke-outside-range comma-pos group-:-pos group-\;-pos t)
+ (and last-real-pos
+ (> last-real-pos (1+ group-\;-pos))
+ (setq last-real-pos (1+ group-\;-pos)))
+ ;; *** This may be wrong:
+ (and cend
+ (> cend group-\;-pos)
+ (setq cend nil
+ cbeg nil))
+ (and quote-end
+ (> quote-end group-\;-pos)
+ (setq quote-end nil
+ quote-beg nil))
+ ;; This was both wrong and unnecessary:
+ ;;(narrow-to-region (point-min) group-\;-pos)
+
+ ;; *** The entire handling of GROUP addresses seems rather lame.
+ ;; *** It deserves a complete rethink, except that these addresses
+ ;; *** are hardly ever seen.
+ )
+
+ ;; Any commas must be between < and : of ROUTE-ADDR. Nuke any
+ ;; others.
+ ;; Hell, go ahead and nuke all of the commas.
+ ;; **** This will cause problems when we start handling commas in
+ ;; the PHRASE part .... no it won't ... yes it will ... ?????
+ (mail-extr-nuke-outside-range comma-pos 1 1)
+
+ ;; can only have multiple @s inside < >. The fact that some MTAs
+ ;; put de-bracketed ROUTE-ADDRs in the UUCP-style "From " line is
+ ;; handled above.
+
+ ;; Locate PHRASE part of ROUTE-ADDR.
+ (when <-pos
+ (goto-char <-pos)
+ (mail-extr-skip-whitespace-backward)
+ (setq phrase-end (point))
+ (goto-char (or ;;group-:-pos
+ (point-min)))
+ (mail-extr-skip-whitespace-forward)
+ (if (< (point) phrase-end)
+ (setq phrase-beg (point))
+ (setq phrase-end nil)))
+
+ ;; handle ROUTE-ADDRS with real ROUTEs.
+ ;; If there are multiple @s, then we assume ROUTE-ADDR syntax, and
+ ;; any % or ! must be semantically meaningless.
+ ;; TODO: do this processing into canonicalization buffer
+ (when route-addr-:-pos
+ (setq !-pos nil
+ %-pos nil
+ >-pos (copy-marker >-pos)
+ route-addr-:-pos (copy-marker route-addr-:-pos))
+ (goto-char >-pos)
+ (insert-before-markers ?X)
+ (goto-char (car @-pos))
+ (while (setq @-pos (cdr @-pos))
+ (delete-char 1)
+ (setq %-pos (cons (point-marker) %-pos))
+ (insert "%")
+ (goto-char (1- >-pos))
+ (save-excursion
+ (insert-buffer-substring extraction-buffer
+ (car @-pos) route-addr-:-pos)
+ (delete-region (car @-pos) route-addr-:-pos))
+ (or (cdr @-pos)
+ (setq saved-@-pos (list (point)))))
+ (setq @-pos saved-@-pos)
+ (goto-char >-pos)
+ (delete-char -1)
+ (mail-extr-nuke-char-at route-addr-:-pos)
+ (mail-extr-demarkerize route-addr-:-pos)
+ (setq route-addr-:-pos nil
+ >-pos (mail-extr-demarkerize >-pos)
+ %-pos (mapcar 'mail-extr-demarkerize %-pos)))
+
+ ;; de-listify @-pos
+ (setq @-pos (car @-pos))
+
+ ;; TODO: remove comments in the middle of an address
+
+ (with-current-buffer canonicalization-buffer
+ (widen)
+ (erase-buffer)
+ (insert-buffer-substring extraction-buffer)
+
+ (if <-pos
+ (narrow-to-region (progn
+ (goto-char (1+ <-pos))
+ (mail-extr-skip-whitespace-forward)
+ (point))
+ >-pos)
+ (if (and first-real-pos last-real-pos)
+ (narrow-to-region first-real-pos last-real-pos)
+ ;; ****** Oh no! What if the address is completely empty!
+ ;; *** Is this correct?
+ (narrow-to-region (point-max) (point-max))))
+
+ (and @-pos %-pos
+ (mail-extr-nuke-outside-range %-pos (point-min) @-pos))
+ (and %-pos !-pos
+ (mail-extr-nuke-outside-range !-pos (point-min) (car %-pos)))
+ (and @-pos !-pos (not %-pos)
+ (mail-extr-nuke-outside-range !-pos (point-min) @-pos))
+
+ ;; Error condition:?? (and %-pos (not @-pos))
+
+ ;; WARNING: THIS CODE IS DUPLICATED BELOW.
+ (when (and %-pos (not @-pos))
+ (goto-char (car %-pos))
+ (delete-char 1)
+ (setq @-pos (point))
+ (insert "@")
+ (setq %-pos (cdr %-pos)))
+
+ (when (and mail-extr-mangle-uucp !-pos)
+ ;; **** I don't understand this save-restriction and the
+ ;; narrow-to-region inside it. Why did I do that?
+ (save-restriction
+ (cond ((and @-pos
+ mail-extr-@-binds-tighter-than-!)
+ (goto-char @-pos)
+ (setq %-pos (cons (point) %-pos)
+ @-pos nil)
+ (delete-char 1)
+ (insert "%")
+ (setq insert-point (point-max)))
+ (mail-extr-@-binds-tighter-than-!
+ (setq insert-point (point-max)))
+ (%-pos
+ (setq insert-point (car (last %-pos))
+ saved-%-pos (mapcar 'mail-extr-markerize %-pos)
+ %-pos nil
+ @-pos (mail-extr-markerize @-pos)))
+ (@-pos
+ (setq insert-point @-pos)
+ (setq @-pos (mail-extr-markerize @-pos)))
+ (t
+ (setq insert-point (point-max))))
+ (narrow-to-region (point-min) insert-point)
+ (setq saved-!-pos (car !-pos))
+ (while !-pos
+ (goto-char (point-max))
+ (cond ((and (not @-pos)
+ (not (cdr !-pos)))
+ (setq @-pos (point))
+ (insert-before-markers "@ "))
+ (t
+ (setq %-pos (cons (point) %-pos))
+ (insert-before-markers "% ")))
+ (backward-char 1)
+ (insert-buffer-substring
+ (current-buffer)
+ (if (nth 1 !-pos)
+ (1+ (nth 1 !-pos))
+ (point-min))
+ (car !-pos))
+ (delete-char 1)
+ (or (save-excursion
+ (mail-extr-safe-move-sexp -1)
+ (mail-extr-skip-whitespace-backward)
+ (eq ?. (preceding-char)))
+ (insert-before-markers
+ (if (save-excursion
+ (mail-extr-skip-whitespace-backward)
+ (eq ?. (preceding-char)))
+ ""
+ ".")
+ "uucp"))
+ (setq !-pos (cdr !-pos))))
+ (and saved-%-pos
+ (setq %-pos (append (mapcar 'mail-extr-demarkerize
+ saved-%-pos)
+ %-pos)))
+ (setq @-pos (mail-extr-demarkerize @-pos))
+ (narrow-to-region (1+ saved-!-pos) (point-max)))
+
+ ;; WARNING: THIS CODE IS DUPLICATED ABOVE.
+ (when (and %-pos (not @-pos))
+ (goto-char (car %-pos))
+ (delete-char 1)
+ (setq @-pos (point))
+ (insert "@")
+ (setq %-pos (cdr %-pos)))
+
+ (when (setq %-pos (nreverse %-pos)) ; implies @-pos valid
+ (setq temp %-pos)
+ (catch 'truncated
+ (while temp
+ (goto-char (or (nth 1 temp)
+ @-pos))
+ (mail-extr-skip-whitespace-backward)
+ (save-excursion
+ (mail-extr-safe-move-sexp -1)
+ (setq domain-pos (point))
+ (mail-extr-skip-whitespace-backward)
+ (setq \.-pos (eq ?. (preceding-char))))
+ (when (and \.-pos
+ ;; #### string consing
+ (let ((s (intern-soft
+ (buffer-substring domain-pos (point))
+ mail-extr-all-top-level-domains)))
+ (and s (get s 'domain-name))))
+ (narrow-to-region (point-min) (point))
+ (goto-char (car temp))
+ (delete-char 1)
+ (setq @-pos (point))
+ (setcdr temp nil)
+ (setq %-pos (delq @-pos %-pos))
+ (insert "@")
+ (throw 'truncated t))
+ (setq temp (cdr temp)))))
+ (setq mbox-beg (point-min)
+ mbox-end (if %-pos (car %-pos)
+ (or @-pos
+ (point-max))))
+
+ (when @-pos
+ ;; Make the domain-name part lowercase since it's case
+ ;; insensitive anyway.
+ (downcase-region (1+ @-pos) (point-max))))
+
+ ;; Done canonicalizing address.
+ ;; We are now back in extraction-buffer.
+
+ ;; Decide what part of the address to search to find the full name.
+ (cond (
+ ;; Example: "First M. Last" <fml@foo.bar.dom>
+ (and phrase-beg
+ (eq quote-beg phrase-beg)
+ (<= quote-end phrase-end))
+ (narrow-to-region (1+ quote-beg) (1- quote-end))
+ (mail-extr-undo-backslash-quoting (point-min) (point-max)))
+
+ ;; Example: First Last <fml@foo.bar.dom>
+ (phrase-beg
+ (narrow-to-region phrase-beg phrase-end))
+
+ ;; Example: fml@foo.bar.dom (First M. Last)
+ (cbeg
+ (narrow-to-region (1+ cbeg) (1- cend))
+ (mail-extr-undo-backslash-quoting (point-min) (point-max))
+
+ ;; Deal with spacing problems
+ (goto-char (point-min))
+;;; (cond ((not (search-forward " " nil t))
+;;; (goto-char (point-min))
+;;; (cond ((search-forward "_" nil t)
+;;; ;; Handle the *idiotic* use of underlines as spaces.
+;;; ;; Example: fml@foo.bar.dom (First_M._Last)
+;;; (goto-char (point-min))
+;;; (while (search-forward "_" nil t)
+;;; (replace-match " " t)))
+;;; ((search-forward "." nil t)
+;;; ;; Fix . used as space
+;;; ;; Example: danj1@cb.att.com (daniel.jacobson)
+;;; (goto-char (point-min))
+;;; (while (re-search-forward mail-extr-bad-dot-pattern nil t)
+;;; (replace-match "\\1 \\2" t))))))
+ )
+
+ ;; Otherwise we try to get the name from the mailbox portion
+ ;; of the address.
+ ;; Example: First_M_Last@foo.bar.dom
+ (t
+ ;; *** Work in canon buffer instead? No, can't. Hmm.