1 ;;; std11.el --- STD 11 functions for GNU Emacs
3 ;; Copyright (C) 1995,96,97,98,99,2000,01,02 Free Software Foundation, Inc.
5 ;; Author: MORIOKA Tomohiko <tomo@m17n.org>
6 ;; Keywords: mail, news, RFC 822, STD 11
8 ;; This file is part of FLIM (Faithful Library about Internet Message).
10 ;; This program is free software; you can redistribute it and/or
11 ;; modify it under the terms of the GNU General Public License as
12 ;; published by the Free Software Foundation; either version 2, or (at
13 ;; your option) any later version.
15 ;; This program is distributed in the hope that it will be useful, but
16 ;; WITHOUT ANY WARRANTY; without even the implied warranty of
17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 ;; General Public License for more details.
20 ;; You should have received a copy of the GNU General Public License
21 ;; along with GNU Emacs; see the file COPYING. If not, write to the
22 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
23 ;; Boston, MA 02111-1307, USA.
28 (require 'poem) ; find-non-ascii-charset-string
29 (require 'pcustom) ; std11-lexical-analyzer
35 (defconst std11-field-name-regexp "[!-9;-~]+")
36 (defconst std11-field-head-regexp
37 (concat "^" std11-field-name-regexp ":"))
38 (defconst std11-next-field-head-regexp
39 (concat "\n" std11-field-name-regexp ":"))
41 (defun std11-field-end (&optional bound)
42 "Move to end of field and return this point.
43 The optional argument BOUNDs the search; it is a buffer position."
44 (if (re-search-forward std11-next-field-head-regexp bound t)
45 (goto-char (match-beginning 0))
46 (if (re-search-forward "^$" bound t)
47 (goto-char (1- (match-beginning 0)))
54 (defun std11-fetch-field (name)
55 "Return the value of the header field NAME.
56 The buffer is expected to be narrowed to just the headers of the message."
58 (goto-char (point-min))
59 (let ((case-fold-search t))
60 (if (re-search-forward (concat "^" name ":[ \t]*") nil t)
61 (buffer-substring-no-properties (match-end 0) (std11-field-end))
65 (defun std11-narrow-to-header (&optional boundary)
66 "Narrow to the message header.
67 If BOUNDARY is not nil, it is used as message header separator."
69 (goto-char (point-min))
70 (if (re-search-forward
71 (concat "^\\(" (regexp-quote (or boundary "")) "\\)?$")
78 (defun std11-field-body (name &optional boundary)
79 "Return the value of the header field NAME.
80 If BOUNDARY is not nil, it is used as message header separator."
83 (inline (std11-narrow-to-header boundary)
84 (std11-fetch-field name))
87 (defun std11-find-field-body (field-names &optional boundary)
88 "Return the first found field-body specified by FIELD-NAMES
89 of the message header in current buffer. If BOUNDARY is not nil, it is
90 used as message header separator."
93 (std11-narrow-to-header boundary)
94 (let ((case-fold-search t)
97 (while (setq field-name (car field-names))
98 (goto-char (point-min))
99 (if (re-search-forward (concat "^" field-name ":[ \t]*") nil t)
101 (buffer-substring-no-properties
102 (match-end 0) (std11-field-end)))
104 (setq field-names (cdr field-names))
107 (defun std11-field-bodies (field-names &optional default-value boundary)
108 "Return list of each field-bodies of FIELD-NAMES of the message header
109 in current buffer. If BOUNDARY is not nil, it is used as message
113 (std11-narrow-to-header boundary)
114 (let* ((case-fold-search t)
115 (dest (make-list (length field-names) default-value))
119 (while (setq field-name (car s-rest))
120 (goto-char (point-min))
121 (if (re-search-forward (concat "^" field-name ":[ \t]*") nil t)
123 (buffer-substring-no-properties
124 (match-end 0) (std11-field-end)))
126 (setq s-rest (cdr s-rest)
131 (defun std11-header-string (regexp &optional boundary)
132 "Return string of message header fields matched by REGEXP.
133 If BOUNDARY is not nil, it is used as message header separator."
134 (let ((case-fold-search t))
137 (std11-narrow-to-header boundary)
138 (goto-char (point-min))
140 (while (re-search-forward std11-field-head-regexp nil t)
142 (buffer-substring (match-beginning 0) (std11-field-end)))
143 (if (string-match regexp field)
144 (setq header (concat header field "\n"))
149 (defun std11-header-string-except (regexp &optional boundary)
150 "Return string of message header fields not matched by REGEXP.
151 If BOUNDARY is not nil, it is used as message header separator."
152 (let ((case-fold-search t))
155 (std11-narrow-to-header boundary)
156 (goto-char (point-min))
158 (while (re-search-forward std11-field-head-regexp nil t)
160 (buffer-substring (match-beginning 0) (std11-field-end)))
161 (if (not (string-match regexp field))
162 (setq header (concat header field "\n"))
167 (defun std11-collect-field-names (&optional boundary)
168 "Return list of all field-names of the message header in current buffer.
169 If BOUNDARY is not nil, it is used as message header separator."
172 (std11-narrow-to-header boundary)
173 (goto-char (point-min))
175 (while (re-search-forward std11-field-head-regexp nil t)
176 (setq name (buffer-substring-no-properties
177 (match-beginning 0)(1- (match-end 0))))
178 (or (member name dest)
179 (setq dest (cons name dest))
189 (defun std11-unfold-string (string)
190 "Unfold STRING as message header field."
193 (while (string-match "\n\\([ \t]\\)" string p)
194 (setq dest (concat dest
195 (substring string p (match-beginning 0))
198 (setq p (match-end 0)))
201 (concat dest (substring string p))
208 (defun std11-wrap-as-quoted-pairs (string specials)
212 (len (length string))
215 (let ((chr (aref string i)))
216 (if (memq chr specials)
217 (setq dest (concat dest (substring string b i) "\\")
222 (concat dest (substring string b))
225 (defconst std11-non-qtext-char-list '(?\" ?\\ ?\r ?\n))
227 (defun std11-wrap-as-quoted-string (string)
228 "Wrap STRING as RFC 822 quoted-string."
230 (std11-wrap-as-quoted-pairs string std11-non-qtext-char-list)
233 (defun std11-strip-quoted-pair (string)
234 "Strip quoted-pairs in STRING."
238 (len (length string))
241 (let ((chr (aref string i)))
243 (setq dest (concat dest (substring string b i))
248 (concat dest (substring string b))
251 (defun std11-strip-quoted-string (string)
252 "Strip quoted-string STRING."
253 (let ((len (length string)))
255 (let ((max (1- len)))
256 (and (eq (aref string 0) ?\")
257 (eq (aref string max) ?\")
258 (std11-strip-quoted-pair (substring string 1 max))
263 ;;; @ lexical analyze
266 (defcustom std11-lexical-analyzer
267 '(std11-analyze-quoted-string
268 std11-analyze-domain-literal
269 std11-analyze-comment
271 std11-analyze-special
273 "*List of functions to return result of lexical analyze.
274 Each function must have two arguments: STRING and START.
275 STRING is the target string to be analyzed.
276 START is start position of STRING to analyze.
278 Previous function is preferred to next function. If a function
279 returns nil, next function is used. Otherwise the return value will
283 :type '(repeat function))
286 (defconst std11-space-char-list '(? ?\t ?\n))
287 (defconst std11-special-char-list '(?\] ?\[
292 ;; (defconst std11-spaces-regexp
293 ;; (eval-when-compile (concat "[" std11-space-char-list "]+")))
295 (defconst std11-non-atom-regexp
297 (concat "[" std11-special-char-list std11-space-char-list "]")))
299 (defconst std11-atom-regexp
301 (concat "[^" std11-special-char-list std11-space-char-list "]+")))
303 (defun std11-analyze-spaces (string start)
304 (if (and (string-match (eval-when-compile
305 (concat "[" std11-space-char-list "]+"))
307 (= (match-beginning 0) start))
308 (let ((end (match-end 0)))
309 (cons (cons 'spaces (substring string start end))
310 ;;(substring string end)
314 (defun std11-analyze-special (string start)
315 (if (and (> (length string) start)
316 (memq (aref string start) std11-special-char-list))
317 (cons (cons 'specials (substring string start (1+ start)))
318 ;;(substring string 1)
322 (defun std11-analyze-atom (string start)
323 (if (string-match std11-non-atom-regexp string start)
324 (if (> (match-beginning 0) start)
325 (cons (cons 'atom (substring string start (match-beginning 0)))
328 (cons (cons 'atom (substring string start))
330 ;; (if (and (string-match std11-atom-regexp string start)
331 ;; (= (match-beginning 0) start))
332 ;; (let ((end (match-end 0)))
333 ;; (cons (cons 'atom (substring string start end))
334 ;; ;;(substring string end)
339 (defun std11-check-enclosure (string open close &optional recursive from)
340 (let ((len (length string))
344 (eq (aref string i) open))
349 (setq chr (aref string i))
362 (setq p (std11-check-enclosure
363 string open close recursive i))
373 (defun std11-analyze-quoted-string (string start)
374 (let ((p (std11-check-enclosure string ?\" ?\" nil start)))
376 (cons (cons 'quoted-string (substring string (1+ start) (1- p)))
377 ;;(substring string p))
381 (defun std11-analyze-domain-literal (string start)
382 (let ((p (std11-check-enclosure string ?\[ ?\] nil start)))
384 (cons (cons 'domain-literal (substring string (1+ start) (1- p)))
385 ;;(substring string p))
389 (defun std11-analyze-comment (string start)
390 (let ((p (std11-check-enclosure string ?\( ?\) t start)))
392 (cons (cons 'comment (substring string (1+ start) (1- p)))
393 ;;(substring string p))
398 (defun std11-lexical-analyze (string &optional analyzer start)
399 "Analyze STRING as lexical tokens of STD 11."
401 (setq analyzer std11-lexical-analyzer))
404 (let ((len (length string))
408 (let ((rest analyzer)
410 (while (and (setq func (car rest))
411 (null (setq r (funcall func string start))))
412 (setq rest (cdr rest)))
414 (cons (cons 'error (substring string start)) (1+ len)))
416 (setq dest (cons (car ret) dest)
426 (defun std11-ignored-token-p (token)
427 (let ((type (car token)))
428 (or (eq type 'spaces)(eq type 'comment))
431 (defun std11-parse-token (lal)
435 (setq token (car lal))
436 (std11-ignored-token-p token)
439 (setq itl (cons token itl))
441 (cons (nreverse (cons token itl))
445 (defun std11-parse-ascii-token (lal)
446 (let (token itl parsed token-value)
448 (setq token (car lal))
449 (or (std11-ignored-token-p token)
450 (if (and (setq token-value (cdr token))
451 (find-non-ascii-charset-string token-value))
455 (setq itl (cons token itl))
458 (setq parsed (nreverse (cons token itl)))
460 (cons parsed (cdr lal))
463 (defun std11-parse-token-or-comment (lal)
467 (setq token (car lal))
468 (eq (car token) 'spaces)
471 (setq itl (cons token itl))
473 (cons (nreverse (cons token itl))
477 (defun std11-parse-word (lal)
478 (let ((ret (std11-parse-ascii-token lal)))
480 (let ((elt (car ret))
483 (if (or (assq 'atom elt)
484 (assq 'quoted-string elt))
485 (cons (cons 'word elt) rest)
488 (defun std11-parse-word-or-comment (lal)
489 (let ((ret (std11-parse-token-or-comment lal)))
491 (let ((elt (car ret))
494 (cond ((or (assq 'atom elt)
495 (assq 'quoted-string elt))
496 (cons (cons 'word elt) rest)
499 (cons (cons 'comment-word elt) rest)
503 (defun std11-parse-phrase (lal)
505 (while (setq ret (std11-parse-word-or-comment lal))
506 (setq phrase (append phrase (cdr (car ret))))
510 (cons (cons 'phrase phrase) lal)
513 (defun std11-parse-local-part (lal)
514 (let ((ret (std11-parse-word lal)))
516 (let ((local-part (cdr (car ret))) dot)
518 (while (and (setq ret (std11-parse-ascii-token lal))
520 (string-equal (cdr (assq 'specials dot)) ".")
521 (setq ret (std11-parse-word (cdr ret)))
523 (append local-part dot (cdr (car ret)))
527 (cons (cons 'local-part local-part) lal)
530 (defun std11-parse-sub-domain (lal)
531 (let ((ret (std11-parse-ascii-token lal)))
533 (let ((sub-domain (car ret)))
534 (if (or (assq 'atom sub-domain)
535 (assq 'domain-literal sub-domain)
537 (cons (cons 'sub-domain sub-domain)
542 (defun std11-parse-domain (lal)
543 (let ((ret (std11-parse-sub-domain lal)))
545 (let ((domain (cdr (car ret))) dot)
547 (while (and (setq ret (std11-parse-ascii-token lal))
549 (string-equal (cdr (assq 'specials dot)) ".")
550 (setq ret (std11-parse-sub-domain (cdr ret)))
552 (append domain dot (cdr (car ret)))
556 (cons (cons 'domain domain) lal)
559 (defun std11-parse-at-domain (lal)
560 (let ((ret (std11-parse-ascii-token lal)) at-sign)
562 (setq at-sign (car ret))
563 (string-equal (cdr (assq 'specials at-sign)) "@")
564 (setq ret (std11-parse-domain (cdr ret)))
566 (cons (cons 'at-domain (append at-sign (cdr (car ret))))
570 (defun std11-parse-addr-spec (lal)
571 (let ((ret (std11-parse-local-part lal))
575 (setq addr (cdr (car ret)))
577 (and (setq ret (std11-parse-at-domain lal))
578 (setq addr (append addr (cdr (car ret))))
581 (cons (cons 'addr-spec addr) lal)
584 (defun std11-parse-route (lal)
585 (let ((ret (std11-parse-at-domain lal))
589 (setq route (cdr (car ret)))
591 (while (and (setq ret (std11-parse-ascii-token lal))
592 (setq comma (car ret))
593 (string-equal (cdr (assq 'specials comma)) ",")
594 (setq ret (std11-parse-at-domain (cdr ret)))
596 (setq route (append route comma (cdr (car ret))))
599 (and (setq ret (std11-parse-ascii-token lal))
600 (setq colon (car ret))
601 (string-equal (cdr (assq 'specials colon)) ":")
602 (setq route (append route colon))
605 (cons (cons 'route route)
610 (defun std11-parse-route-addr (lal)
611 (let ((ret (std11-parse-ascii-token lal))
615 (string-equal (cdr (assq 'specials <)) "<")
617 (progn (and (setq ret (std11-parse-route lal))
618 (setq route (cdr (car ret)))
621 (setq ret (std11-parse-addr-spec lal))
623 (setq addr-spec (cdr (car ret)))
625 (setq ret (std11-parse-ascii-token lal))
627 (string-equal (cdr (assq 'specials >)) ">")
629 (cons (cons 'route-addr (append route addr-spec))
634 (defun std11-parse-phrase-route-addr (lal)
635 (let ((ret (std11-parse-phrase lal)) phrase)
638 (setq phrase (cdr (car ret)))
641 (if (setq ret (std11-parse-route-addr lal))
642 (cons (list 'phrase-route-addr
648 (defun std11-parse-mailbox (lal)
649 (let ((ret (or (std11-parse-phrase-route-addr lal)
650 (std11-parse-addr-spec lal)))
654 (setq mbox (car ret))
656 (if (and (setq ret (std11-parse-token-or-comment lal))
657 (setq comment (cdr (assq 'comment (car ret))))
661 (cons (list 'mailbox mbox comment)
665 (defun std11-parse-group (lal)
666 (let ((ret (std11-parse-phrase lal))
667 phrase colon comma mbox semicolon)
669 (setq phrase (cdr (car ret)))
671 (setq ret (std11-parse-ascii-token lal))
672 (setq colon (car ret))
673 (string-equal (cdr (assq 'specials colon)) ":")
676 (and (setq ret (std11-parse-mailbox lal))
677 (setq mbox (list (car ret)))
680 (while (and (setq ret (std11-parse-ascii-token lal))
681 (setq comma (car ret))
683 (cdr (assq 'specials comma)) ",")
685 (setq ret (std11-parse-mailbox lal))
686 (setq mbox (cons (car ret) mbox))
690 (and (setq ret (std11-parse-ascii-token lal))
691 (setq semicolon (car ret))
692 (string-equal (cdr (assq 'specials semicolon)) ";")
694 (cons (list 'group phrase (nreverse mbox))
699 (defun std11-parse-address (lal)
700 (or (std11-parse-group lal)
701 (std11-parse-mailbox lal)
704 (defun std11-parse-addresses (lal)
705 (let ((ret (std11-parse-address lal)))
707 (let ((dest (list (car ret))))
709 (while (and (setq ret (std11-parse-ascii-token lal))
710 (string-equal (cdr (assq 'specials (car ret))) ",")
711 (setq ret (std11-parse-address (cdr ret)))
713 (setq dest (cons (car ret) dest))
719 (defun std11-parse-msg-id (lal)
720 (let ((ret (std11-parse-ascii-token lal))
724 (string-equal (cdr (assq 'specials <)) "<")
726 (setq ret (std11-parse-addr-spec lal))
727 (setq addr-spec (car ret))
729 (setq ret (std11-parse-ascii-token lal))
731 (string-equal (cdr (assq 'specials >)) ">")
733 (cons (cons 'msg-id (cdr addr-spec))
737 (defun std11-parse-msg-ids (tokens)
738 "Parse lexical TOKENS as `*(phrase / msg-id)', and return the result."
739 (let ((ret (or (std11-parse-msg-id tokens)
740 (std11-parse-phrase tokens))))
742 (let ((dest (list (car ret))))
743 (setq tokens (cdr ret))
744 (while (setq ret (or (std11-parse-msg-id tokens)
745 (std11-parse-phrase tokens)))
746 (setq dest (cons (car ret) dest))
747 (setq tokens (cdr ret))
752 (defalias 'std11-parse-in-reply-to 'std11-parse-msg-ids)
753 (make-obsolete 'std11-parse-in-reply-to 'std11-parse-msg-ids)
759 (defun std11-addr-to-string (seq)
760 "Return string from lexical analyzed list SEQ
761 represents addr-spec of RFC 822."
764 (let ((name (car token)))
766 ((eq name 'spaces) "")
767 ((eq name 'comment) "")
768 ((eq name 'quoted-string)
769 (concat "\"" (cdr token) "\""))
776 (defun std11-address-string (address)
777 "Return string of address part from parsed ADDRESS of RFC 822."
778 (cond ((eq (car address) 'group)
779 (mapconcat (function std11-address-string)
783 ((eq (car address) 'mailbox)
784 (let ((addr (nth 1 address)))
785 (std11-addr-to-string
786 (if (eq (car addr) 'phrase-route-addr)
792 (defun std11-comment-value-to-string (value)
794 (std11-strip-quoted-pair value)
799 (if (stringp (car value))
802 (std11-comment-value-to-string
811 (defun std11-full-name-string (address)
812 "Return string of full-name part from parsed ADDRESS of RFC 822."
813 (cond ((eq (car address) 'group)
820 ((eq (car address) 'mailbox)
821 (let ((addr (nth 1 address))
822 (comment (nth 2 address))
824 (if (eq (car addr) 'phrase-route-addr)
829 (let ((type (car token)))
830 (cond ((eq type 'quoted-string)
831 (std11-strip-quoted-pair (cdr token))
835 (std11-comment-value-to-string
844 (cond ((> (length phrase) 0) phrase)
845 (comment (std11-comment-value-to-string comment))
850 (defun std11-msg-id-string (msg-id)
851 "Return string from parsed MSG-ID of RFC 822."
852 (concat "<" (std11-addr-to-string (cdr msg-id)) ">")
856 (defun std11-fill-msg-id-list-string (string &optional column)
857 "Fill list of msg-id in STRING, and return the result."
860 (let ((lal (std11-lexical-analyze string))
862 (let ((ret (std11-parse-msg-id lal)))
864 (let* ((str (std11-msg-id-string (car ret)))
867 (if (> (+ len column) 76)
868 (setq dest (concat dest "\n " str)
871 column (+ column len))
873 (setq dest (concat dest (cdr (car lal)))
877 (let ((ret (std11-parse-msg-id lal)))
879 (let* ((str (std11-msg-id-string (car ret)))
880 (len (1+ (length str))))
882 (if (> (+ len column) 76)
883 (setq dest (concat dest "\n " str)
885 (setq dest (concat dest " " str)
886 column (+ column len))
888 (setq dest (concat dest (cdr (car lal)))
894 ;;; @ parser with lexical analyzer
898 (defun std11-parse-address-string (string)
899 "Parse STRING as mail address."
900 (std11-parse-address (std11-lexical-analyze string))
904 (defun std11-parse-addresses-string (string)
905 "Parse STRING as mail address list."
906 (std11-parse-addresses (std11-lexical-analyze string))
910 (defun std11-parse-msg-id-string (string)
911 "Parse STRING as msg-id."
912 (std11-parse-msg-id (std11-lexical-analyze string))
916 (defun std11-parse-msg-ids-string (string)
917 "Parse STRING as `*(phrase / msg-id)'."
918 (std11-parse-msg-ids (std11-lexical-analyze string))
922 (defun std11-extract-address-components (string)
923 "Extract full name and canonical address from STRING.
924 Returns a list of the form (FULL-NAME CANONICAL-ADDRESS).
925 If no name can be extracted, FULL-NAME will be nil."
926 (let* ((structure (car (std11-parse-address-string
927 (std11-unfold-string string))))
928 (phrase (std11-full-name-string structure))
929 (address (std11-address-string structure))
931 (list phrase address)
940 ;;; std11.el ends here