1 ;;; rfc2047.el --- functions for encoding and decoding rfc2047 messages
3 ;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004,
4 ;; 2005, 2006, 2007 Free Software Foundation, Inc.
6 ;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
7 ;; MORIOKA Tomohiko <morioka@jaist.ac.jp>
8 ;; This file is part of GNU Emacs.
10 ;; GNU Emacs is free software; you can redistribute it and/or modify
11 ;; it under the terms of the GNU General Public License as published by
12 ;; the Free Software Foundation; either version 3, or (at your option)
15 ;; GNU Emacs is distributed in the hope that it will be useful,
16 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
18 ;; GNU 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., 51 Franklin Street, Fifth Floor,
23 ;; Boston, MA 02110-1301, USA.
27 ;; RFC 2047 is "MIME (Multipurpose Internet Mail Extensions) Part
28 ;; Three: Message Header Extensions for Non-ASCII Text".
34 (defvar message-posting-charset
)
39 ;; Fixme: Avoid this (used for mail-parse-charset) mm dependence on gnus.
41 (require 'rfc2045
) ;; rfc2045-encode-string
42 (autoload 'mm-body-7-or-8
"mm-bodies")
44 (defvar rfc2047-header-encoding-alist
45 '(("Newsgroups" . nil
)
48 ("\\(Resent-\\)?\\(From\\|Cc\\|To\\|Bcc\\|\\(In-\\)?Reply-To\\|Sender\
49 \\|Mail-Followup-To\\|Mail-Copies-To\\|Approved\\)" . address-mime
)
51 "*Header/encoding method alist.
52 The list is traversed sequentially. The keys can either be
57 1) nil, in which case no encoding is done;
58 2) `mime', in which case the header will be encoded according to RFC2047;
59 3) `address-mime', like `mime', but takes account of the rules for address
60 fields (where quoted strings and comments must be treated separately);
61 4) a charset, in which case it will be encoded as that charset;
62 5) `default', in which case the field will be encoded as the rest
65 (defvar rfc2047-charset-encoding-alist
89 "Alist of MIME charsets to RFC2047 encodings.
90 Valid encodings are nil, `Q' and `B'. These indicate binary (no) encoding,
91 quoted-printable and base64 respectively.")
93 (defvar rfc2047-encode-function-alist
94 '((Q . rfc2047-q-encode-string
)
95 (B . rfc2047-b-encode-string
)
97 "Alist of RFC2047 encodings to encoding functions.")
99 (defvar rfc2047-encode-encoded-words t
100 "Whether encoded words should be encoded again.")
103 ;;; Functions for encoding RFC2047 messages
106 (defun rfc2047-qp-or-base64 ()
107 "Return the type with which to encode the buffer.
108 This is either `base64' or `quoted-printable'."
110 (let ((limit (min (point-max) (+ 2000 (point-min))))
112 (goto-char (point-min))
113 (skip-chars-forward "\x20-\x7f\r\n\t" limit
)
114 (while (< (point) limit
)
117 (skip-chars-forward "\x20-\x7f\r\n\t" limit
))
118 (if (or (< (* 6 n8bit
) (- limit
(point-min)))
119 ;; Don't base64, say, a short line with a single
120 ;; non-ASCII char when splitting parts by charset.
125 (defun rfc2047-narrow-to-field ()
126 "Narrow the buffer to the header on the current line."
132 (if (re-search-forward "^[^ \n\t]" nil t
)
135 (goto-char (point-min)))
137 (defun rfc2047-field-value ()
138 "Return the value of the field at point."
141 (rfc2047-narrow-to-field)
142 (re-search-forward ":[ \t\n]*" nil t
)
143 (buffer-substring-no-properties (point) (point-max)))))
145 (defun rfc2047-quote-special-characters-in-quoted-strings (&optional
147 "Quote special characters with `\\'s in quoted strings.
148 Quoting will not be done in a quoted string if it contains characters
149 matching ENCODABLE-REGEXP or it is within parentheses."
150 (goto-char (point-min))
151 (let ((tspecials (concat "[" ietf-drums-tspecials
"]"))
154 (with-syntax-table (standard-syntax-table)
158 (eq (char-before) ?\
)))
160 (goto-char (point-max)))
162 (narrow-to-region start
(point))
164 (while (search-forward "\"" nil t
)
165 (setq beg
(match-beginning 0))
166 (unless (eq (char-before beg
) ?
\\)
172 (setq end
(1- (point)))
174 (if (and encodable-regexp
175 (re-search-forward encodable-regexp end t
))
178 (narrow-to-region beg end
)
179 (while (re-search-forward tspecials nil
'move
)
180 (if (eq (char-before) ?
\\)
181 (if (looking-at tspecials
) ;; Already quoted.
184 (goto-char (match-beginning 0))
190 (goto-char (point-max)))
192 (setq start
(point))))))
194 (defvar rfc2047-encoding-type
'address-mime
195 "The type of encoding done by `rfc2047-encode-region'.
196 This should be dynamically bound around calls to
197 `rfc2047-encode-region' to either `mime' or `address-mime'. See
198 `rfc2047-header-encoding-alist', for definitions.")
200 (defun rfc2047-encode-message-header ()
201 "Encode the message header according to `rfc2047-header-encoding-alist'.
202 Should be called narrowed to the head of the message."
205 (goto-char (point-min))
206 (let (alist elem method
)
209 (rfc2047-narrow-to-field)
211 alist rfc2047-header-encoding-alist
)
212 (while (setq elem
(pop alist
))
213 (when (or (and (stringp (car elem
))
214 (looking-at (car elem
)))
218 (if (not (rfc2047-encodable-p))
220 (when (eq method
'address-mime
)
221 (rfc2047-quote-special-characters-in-quoted-strings))
222 (if (and (eq (mm-body-7-or-8) '8bit
)
225 (car message-posting-charset
)))
226 ;; 8 bit must be decoded.
227 (mm-encode-coding-region
228 (point-min) (point-max)
229 (mm-charset-to-coding-system
230 (car message-posting-charset
))))
231 ;; No encoding necessary, but folding is nice
235 (goto-char (point-min))
236 (skip-chars-forward "^:")
237 (when (looking-at ": ")
241 ;; We found something that may perhaps be encoded.
242 (re-search-forward "^[^:]+: *" nil t
)
244 ((eq method
'address-mime
)
245 (rfc2047-encode-region (point) (point-max)))
247 (let ((rfc2047-encoding-type 'mime
))
248 (rfc2047-encode-region (point) (point-max))))
249 ((eq method
'default
)
250 (if (and (featurep 'mule
)
251 (if (boundp 'default-enable-multibyte-characters
)
252 default-enable-multibyte-characters
)
254 (mm-encode-coding-region (point) (point-max)
255 mail-parse-charset
)))
256 ;; We get this when CC'ing messsages to newsgroups with
257 ;; 8-bit names. The group name mail copy just got
258 ;; unconditionally encoded. Previously, it would ask
259 ;; whether to encode, which was quite confusing for the
260 ;; user. If the new behaviour is wrong, tell me. I have
261 ;; left the old code commented out below.
262 ;; -- Per Abrahamsen <abraham@dina.kvl.dk> Date: 2001-10-07.
263 ;; Modified by Dave Love, with the commented-out code changed
264 ;; in accordance with changes elsewhere.
266 (rfc2047-encode-region (point) (point-max)))
268 ;;; (if (or (message-options-get
269 ;;; 'rfc2047-encode-message-header-encode-any)
270 ;;; (message-options-set
271 ;;; 'rfc2047-encode-message-header-encode-any
273 ;;; "Some texts are not encoded. Encode anyway?")))
274 ;;; (rfc2047-encode-region (point-min) (point-max))
275 ;;; (error "Cannot send unencoded text")))
276 ((mm-coding-system-p method
)
277 (if (or (and (featurep 'mule
)
278 (if (boundp 'default-enable-multibyte-characters
)
279 default-enable-multibyte-characters
))
280 (featurep 'file-coding
))
281 (mm-encode-coding-region (point) (point-max) method
)))
284 (goto-char (point-max)))))))
286 ;; Fixme: This, and the require below may not be the Right Thing, but
287 ;; should be safe just before release. -- fx 2001-02-08
289 (defun rfc2047-encodable-p ()
290 "Return non-nil if any characters in current buffer need encoding in headers.
291 The buffer may be narrowed."
292 (require 'message
) ; for message-posting-charset
294 (mm-find-mime-charset-region (point-min) (point-max))))
295 (goto-char (point-min))
296 (or (and rfc2047-encode-encoded-words
298 (search-forward "=?" nil t
)
299 (goto-char (point-min))))
301 (not (equal charsets
(list (car message-posting-charset
))))))))
303 ;; Use this syntax table when parsing into regions that may need
304 ;; encoding. Double quotes are string delimiters, backslash is
305 ;; character quoting, and all other RFC 2822 special characters are
306 ;; treated as punctuation so we can use forward-sexp/forward-word to
307 ;; skip to the end of regions appropriately. Nb. ietf-drums does
308 ;; things differently.
309 (defconst rfc2047-syntax-table
310 ;; (make-char-table 'syntax-table '(2)) only works in Emacs.
311 (let ((table (make-syntax-table)))
312 ;; The following is done to work for setting all elements of the table
313 ;; in Emacs 21 and 22 and XEmacs; it appears to be the cleanest way.
314 ;; Play safe and don't assume the form of the word syntax entry --
316 (if (fboundp 'set-char-table-range
) ; Emacs
317 (funcall (intern "set-char-table-range")
318 table t
(aref (standard-syntax-table) ?a
))
319 (if (fboundp 'put-char-table
)
320 (if (fboundp 'get-char-table
) ; warning avoidance
321 (put-char-table t
(get-char-table ?a
(standard-syntax-table))
323 (modify-syntax-entry ?
\\ "\\" table
)
324 (modify-syntax-entry ?
\" "\"" table
)
325 (modify-syntax-entry ?\
( "(" table
)
326 (modify-syntax-entry ?\
) ")" table
)
327 (modify-syntax-entry ?\
< "." table
)
328 (modify-syntax-entry ?\
> "." table
)
329 (modify-syntax-entry ?\
[ "." table
)
330 (modify-syntax-entry ?\
] "." table
)
331 (modify-syntax-entry ?
: "." table
)
332 (modify-syntax-entry ?\
; "." table)
333 (modify-syntax-entry ?
, "." table
)
334 (modify-syntax-entry ?
@ "." table
)
337 (defun rfc2047-encode-region (b e
)
338 "Encode words in region B to E that need encoding.
339 By default, the region is treated as containing RFC2822 addresses.
340 Dynamically bind `rfc2047-encoding-type' to change that."
342 (narrow-to-region b e
)
343 (let ((encodable-regexp (if rfc2047-encode-encoded-words
344 "[^\000-\177]+\\|=\\?"
346 start
; start of current token
348 ;; Whether there's an encoded word before the current token,
349 ;; either immediately or separated by space.
351 (orig-text (buffer-substring-no-properties b e
)))
352 (if (eq 'mime rfc2047-encoding-type
)
353 ;; Simple case. Continuous words in which all those contain
354 ;; non-ASCII characters are encoded collectively. Encoding
355 ;; ASCII words, including `Re:' used in Subject headers, is
356 ;; avoided for interoperability with non-MIME clients and
357 ;; for making it easy to find keywords.
359 (goto-char (point-min))
360 (while (progn (skip-chars-forward " \t\n")
363 (while (and (looking-at "[ \t\n]*\\([^ \t\n]+\\)")
365 (setq end
(match-end 0))
366 (re-search-forward encodable-regexp end t
)))
368 (if (> (point) start
)
369 (rfc2047-encode start
(point))
371 ;; `address-mime' case -- take care of quoted words, comments.
372 (rfc2047-quote-special-characters-in-quoted-strings encodable-regexp
)
373 (with-syntax-table rfc2047-syntax-table
374 (goto-char (point-min))
375 (condition-case err
; in case of unbalanced quotes
376 ;; Look for rfc2822-style: sequences of atoms, quoted
377 ;; strings, specials, whitespace. (Specials mustn't be
381 (skip-chars-forward " \t\n")
384 ((not (char-after))) ; eob
386 ((eq ?
\" (setq csyntax
(char-syntax (char-after))))
390 ;; Does it need encoding?
392 (if (re-search-forward encodable-regexp end
'move
)
393 ;; It needs encoding. Strip the quotes first,
394 ;; since encoded words can't occur in quotes.
397 (delete-backward-char 1)
401 ;; There was a preceding quoted word. We need
402 ;; to include any separating whitespace in this
403 ;; word to avoid it getting lost.
404 (skip-chars-backward " \t")
405 ;; A space is needed between the encoded words.
409 ;; Adjust the end position for the deleted quotes.
410 (rfc2047-encode start
(- end
2))
411 (setq last-encoded t
)) ; record that it was encoded
412 (setq last-encoded nil
)))
414 ;; Skip other delimiters, but record that they've
415 ;; potentially separated quoted words.
417 (setq last-encoded nil
))
419 (error "Unbalanced parentheses"))
421 ;; Look for the end of parentheses.
423 ;; Encode text as an unstructured field.
424 (let ((rfc2047-encoding-type 'mime
))
425 (rfc2047-encode-region (1+ start
) (1- (point))))
426 (skip-chars-forward ")"))
427 (t ; normal token/whitespace sequence
429 ;; Skip one ASCII word, or encode continuous words
430 ;; in which all those contain non-ASCII characters.
432 (while (not (or end
(eobp)))
433 (when (looking-at "[\000-\177]+")
437 (while (and (or (re-search-forward
438 "[ \t\n]\\|\\Sw" end
'move
)
440 (eq ?
\\ (char-syntax (char-before))))
441 ;; Skip backslash-quoted characters.
444 (setq end
(match-beginning 0))
445 (if rfc2047-encode-encoded-words
448 (when (search-forward "=?" end
'move
)
449 (goto-char (match-beginning 0))
452 ;; Where the value nil of `end' means there may be
453 ;; text to have to be encoded following the point.
454 ;; Otherwise, the point reached to the end of ASCII
455 ;; words separated by whitespace or a special char.
457 (when (looking-at encodable-regexp
)
458 (goto-char (setq begin
(match-end 0)))
459 (while (and (looking-at "[ \t\n]+\\([^ \t\n]+\\)")
460 (setq end
(match-end 0))
462 (while (re-search-forward
463 encodable-regexp end t
))
466 (or (not (re-search-forward "\\Sw" end t
))
468 (goto-char (match-beginning 0))
471 (when (looking-at "[^ \t\n]+")
472 (setq end
(match-end 0))
473 (if (re-search-forward "\\Sw+" end t
)
474 ;; There are special characters better
475 ;; to be encoded so that MTAs may parse
477 (cond ((= end
(point)))
478 ((looking-at (concat "\\sw*\\("
483 (goto-char (1- (match-end 0)))
484 (unless (= (point) (match-beginning 0))
485 ;; Separate encodable text and
489 (skip-chars-forward " \t\n")
490 (if (and (looking-at "[^ \t\n]+")
491 (string-match encodable-regexp
494 (goto-char end
)))))))
495 (skip-chars-backward " \t\n")
498 (if (re-search-forward encodable-regexp end
'move
)
500 (unless (memq (char-before start
) '(nil ?
\t ?
))
503 (skip-chars-backward "^ \t\n")
504 (and (looking-at "\\Sw+")
505 (= (match-end 0) start
)))
506 ;; Also encode bogus delimiters.
508 ;; Separate encodable text and delimiter.
511 (setq start
(1+ start
)
513 (rfc2047-encode start end
)
514 (setq last-encoded t
))
515 (setq last-encoded nil
)))))
517 (if (or debug-on-quit debug-on-error
)
518 (signal (car err
) (cdr err
))
519 (error "Invalid data for rfc2047 encoding: %s"
520 (mm-replace-in-string orig-text
"[ \t\n]+" " "))))))))
521 (rfc2047-fold-region b
(point))
522 (goto-char (point-max))))
524 (defun rfc2047-encode-string (string)
525 "Encode words in STRING.
526 By default, the string is treated as containing addresses (see
527 `rfc2047-encoding-type')."
528 (mm-with-multibyte-buffer
530 (rfc2047-encode-region (point-min) (point-max))
533 (defvar rfc2047-encode-max-chars
76
534 "Maximum characters of each header line that contain encoded-words.
535 If it is nil, encoded-words will not be folded. Too small value may
536 cause an error. Don't change this for no particular reason.")
538 (defun rfc2047-encode-1 (column string cs encoder start crest tail
540 "Subroutine used by `rfc2047-encode'."
541 (cond ((string-equal string
"")
543 ((not rfc2047-encode-max-chars
)
545 (funcall encoder
(if cs
546 (mm-encode-coding-string string cs
)
549 ((>= column rfc2047-encode-max-chars
)
551 (cond ((string-match "\n[ \t]+\\'" eword
)
552 ;; Reomove a superfluous empty line.
553 (setq eword
(substring eword
0 (match-beginning 0))))
554 ((string-match "(+\\'" eword
)
555 ;; Break the line before the open parenthesis.
556 (setq crest
(concat crest
(match-string 0 eword
))
557 eword
(substring eword
0 (match-beginning 0))))))
558 (rfc2047-encode-1 (length crest
) string cs encoder start
" " tail
559 (concat eword
"\n" crest
)))
562 (limit (1- (length string
)))
567 (setq next
(concat start
570 (mm-encode-coding-string
571 (substring string
0 (1+ index
))
573 (substring string
0 (1+ index
))))
575 len
(+ column
(length next
)))
576 (if (> len rfc2047-encode-max-chars
)
579 (if (or (< index limit
)
580 (<= (+ len
(or (string-match "\n" tail
)
582 rfc2047-encode-max-chars
))
585 (if (string-match "\\`)+" tail
)
586 ;; Break the line after the close parenthesis.
587 (setq tail
(concat (substring tail
0 (match-end 0))
589 (substring tail
(match-end 0)))
595 (concat eword next tail
)
598 (string-match "(+\\'" eword
))
599 (setq crest
(concat crest
(match-string 0 eword
))
600 eword
(substring eword
0 (match-beginning 0)))
601 (setq eword
(concat eword next
)))
603 eword
(concat eword next
)))
604 (when (string-match "\n[ \t]+\\'" eword
)
605 ;; Reomove a superfluous empty line.
606 (setq eword
(substring eword
0 (match-beginning 0))))
607 (rfc2047-encode-1 (length crest
) (substring string index
)
608 cs encoder start
" " tail
609 (concat eword
"\n" crest
)))))))
611 (defun rfc2047-encode (b e
)
612 "Encode the word(s) in the region B to E.
613 Point moves to the end of the region."
614 (let ((mime-charset (or (mm-find-mime-charset-region b e
) (list 'us-ascii
)))
615 cs encoding tail crest eword
)
616 (cond ((> (length mime-charset
) 1)
617 (error "Can't rfc2047-encode `%s'"
618 (buffer-substring-no-properties b e
)))
619 ((= (length mime-charset
) 1)
620 (setq mime-charset
(car mime-charset
)
621 cs
(mm-charset-to-coding-system mime-charset
))
622 (unless (and (mm-multibyte-p)
623 (mm-coding-system-p cs
))
626 (narrow-to-region b e
)
628 (or (cdr (assq mime-charset
629 rfc2047-charset-encoding-alist
))
630 ;; For the charsets that don't have a preferred
631 ;; encoding, choose the one that's shorter.
632 (if (eq (rfc2047-qp-or-base64) 'base64
)
637 (skip-chars-forward "^ \t\n")
638 ;; `tail' may contain a close parenthesis.
639 (setq tail
(buffer-substring-no-properties e
(point)))
641 (setq b
(point-marker)
642 e
(set-marker (make-marker) e
))
643 (rfc2047-fold-region (point-at-bol) b
)
645 (skip-chars-backward "^ \t\n")
646 (unless (= 0 (skip-chars-backward " \t"))
647 ;; `crest' may contain whitespace and an open parenthesis.
648 (setq crest
(buffer-substring-no-properties (point) b
)))
649 (setq eword
(rfc2047-encode-1
651 (mm-replace-in-string
652 (buffer-substring-no-properties b e
)
653 "\n\\([ \t]?\\)" "\\1")
655 (or (cdr (assq encoding
656 rfc2047-encode-function-alist
))
658 (concat "=?" (downcase (symbol-name mime-charset
))
659 "?" (upcase (symbol-name encoding
)) "?")
662 (delete-region (if (eq (aref eword
0) ?
\n)
664 ;; The line was folded before encoding.
669 ;; `eword' contains `crest' and `tail'.
673 (unless (or (/= 0 (length tail
))
675 (looking-at "[ \t\n)]"))
680 (defun rfc2047-fold-field ()
681 "Fold the current header field."
684 (rfc2047-narrow-to-field)
685 (rfc2047-fold-region (point-min) (point-max)))))
687 (defun rfc2047-fold-region (b e
)
688 "Fold long lines in region B to E."
690 (narrow-to-region b e
)
691 (goto-char (point-min))
695 (bol (save-restriction
699 (when (and (or break qword-break
)
700 (> (- (point) bol
) 76))
701 (goto-char (or break qword-break
))
704 (skip-chars-backward " \t")
705 (if (looking-at "[ \t]")
708 (setq bol
(1- (point)))
709 ;; Don't break before the first non-LWSP characters.
710 (skip-chars-forward " \t")
714 ((eq (char-after) ?
\n)
719 (skip-chars-forward " \t")
720 (unless (or (eobp) (eq (char-after) ?
\n))
722 ((eq (char-after) ?
\r)
724 ((memq (char-after) '(? ?
\t))
725 (skip-chars-forward " \t")
726 (unless first
;; Don't break just after the header name.
727 (setq break
(point))))
729 (if (not (looking-at "=\\?[^=]"))
730 (if (eq (char-after) ?
=)
732 (skip-chars-forward "^ \t\n\r="))
733 ;; Don't break at the start of the field.
734 (unless (= (point) b
)
735 (setq qword-break
(point)))
736 (skip-chars-forward "^ \t\n\r")))
738 (skip-chars-forward "^ \t\n\r")))
740 (when (and (or break qword-break
)
741 (> (- (point) bol
) 76))
742 (goto-char (or break qword-break
))
745 (if (or (> 0 (skip-chars-backward " \t"))
746 (looking-at "[ \t]"))
749 (setq bol
(1- (point)))
750 ;; Don't break before the first non-LWSP characters.
751 (skip-chars-forward " \t")
753 (forward-char 1))))))
755 (defun rfc2047-unfold-field ()
756 "Fold the current line."
759 (rfc2047-narrow-to-field)
760 (rfc2047-unfold-region (point-min) (point-max)))))
762 (defun rfc2047-unfold-region (b e
)
763 "Unfold lines in region B to E."
765 (narrow-to-region b e
)
766 (goto-char (point-min))
767 (let ((bol (save-restriction
770 (eol (point-at-eol)))
773 (if (and (looking-at "[ \t]")
774 (< (- (point-at-eol) bol
) 76))
775 (delete-region eol
(progn
777 (skip-chars-forward "\r\n")
779 (setq bol
(point-at-bol)))
780 (setq eol
(point-at-eol))
783 (defun rfc2047-b-encode-string (string)
784 "Base64-encode the header contained in STRING."
785 (base64-encode-string string t
))
787 (defun rfc2047-q-encode-string (string)
788 "Quoted-printable-encode the header in STRING."
789 (mm-with-unibyte-buffer
791 (quoted-printable-encode-region
792 (point-min) (point-max) nil
793 ;; = (\075), _ (\137), ? (\077) are used in the encoded word.
794 ;; Avoid using 8bit characters.
795 ;; This list excludes `especials' (see the RFC2047 syntax),
796 ;; meaning that some characters in non-structured fields will
797 ;; get encoded when they con't need to be. The following is
798 ;; what it used to be.
799 ;;; ;; Equivalent to "^\000-\007\011\013\015-\037\200-\377=_?"
800 ;;; "\010\012\014\040-\074\076\100-\136\140-\177")
801 "-\b\n\f !#-'*+0-9A-Z\\^`-~\d")
802 (subst-char-in-region (point-min) (point-max) ? ?_
)
805 (defun rfc2047-encode-parameter (param value
)
806 "Return and PARAM=VALUE string encoded in the RFC2047-like style.
807 This is a replacement for the `rfc2231-encode-string' function.
809 When attaching files as MIME parts, we should use the RFC2231 encoding
810 to specify the file names containing non-ASCII characters. However,
811 many mail softwares don't support it in practice and recipients won't
812 be able to extract files with correct names. Instead, the RFC2047-like
813 encoding is acceptable generally. This function provides the very
814 RFC2047-like encoding, resigning to such a regrettable trend. To use
815 it, put the following line in your ~/.gnus.el file:
817 \(defalias 'mail-header-encode-parameter 'rfc2047-encode-parameter)
819 (let ((rfc2047-encoding-type 'mime
)
820 (rfc2047-encode-max-chars nil
))
821 (rfc2045-encode-string param
(rfc2047-encode-string value
))))
824 ;;; Functions for decoding RFC2047 messages
828 (defconst rfc2047-encoded-word-regexp
829 "=\\?\\([^][\000-\040()<>@,\;:*\\\"/?.=]+\\)\\(?:\\*[^?]+\\)?\
830 \\?\\(B\\|Q\\)\\?\\([!->@-~ ]*\\)\\?="))
832 (defvar rfc2047-quote-decoded-words-containing-tspecials nil
833 "If non-nil, quote decoded words containing special characters.")
835 (defvar rfc2047-allow-incomplete-encoded-text t
836 "*Non-nil means allow incomplete encoded-text in successive encoded-words.
837 Dividing of encoded-text in the place other than character boundaries
838 violates RFC2047 section 5, while we have a capability to decode it.
839 If it is non-nil, the decoder will decode B- or Q-encoding in each
840 encoded-word, concatenate them, and decode it by charset. Otherwise,
841 the decoder will fully decode each encoded-word before concatenating
844 (defun rfc2047-strip-backslashes-in-quoted-strings ()
845 "Strip backslashes in quoted strings. `\\\"' remains."
846 (goto-char (point-min))
848 (with-syntax-table (standard-syntax-table)
849 (while (search-forward "\"" nil t
)
850 (unless (eq (char-before) ?
\\)
851 (setq beg
(match-end 0))
852 (goto-char (match-beginning 0))
857 (narrow-to-region beg
(1- (point)))
859 (while (search-forward "\\" nil
'move
)
860 (unless (memq (char-after) '(?
\"))
861 (delete-backward-char 1))
865 (goto-char beg
))))))))
867 (defun rfc2047-charset-to-coding-system (charset)
868 "Return coding-system corresponding to MIME CHARSET.
869 If your Emacs implementation can't decode CHARSET, return nil."
870 (when (stringp charset
)
871 (setq charset
(intern (downcase charset
))))
872 (when (or (not charset
)
873 (eq 'gnus-all mail-parse-ignored-charsets
)
874 (memq 'gnus-all mail-parse-ignored-charsets
)
875 (memq charset mail-parse-ignored-charsets
))
876 (setq charset mail-parse-charset
))
877 (let ((cs (mm-charset-to-coding-system charset
)))
878 (cond ((eq cs
'ascii
)
879 (setq cs
(or (mm-charset-to-coding-system mail-parse-charset
)
881 ((mm-coding-system-p cs
))
883 (listp mail-parse-ignored-charsets
)
884 (memq 'gnus-unknown mail-parse-ignored-charsets
))
885 (setq cs
(mm-charset-to-coding-system mail-parse-charset
))))
890 (defun rfc2047-decode-encoded-words (words)
891 "Decode successive encoded-words in WORDS and return a decoded string.
892 Each element of WORDS looks like (CHARSET ENCODING ENCODED-TEXT
894 (let (word charset cs encoding text rest
)
896 (setq word
(pop words
))
897 (if (and (setq cs
(rfc2047-charset-to-coding-system
898 (setq charset
(car word
))))
900 (cond ((char-equal ?B
(nth 1 word
))
901 (setq text
(base64-decode-string
902 (rfc2047-pad-base64 (nth 2 word
)))))
903 ((char-equal ?Q
(nth 1 word
))
904 (setq text
(quoted-printable-decode-string
905 (mm-subst-char-in-string
906 ?_ ?
(nth 2 word
) t
)))))
908 (message "%s" (error-message-string code
))
910 (if (and rfc2047-allow-incomplete-encoded-text
912 ;; Concatenate text of which the charset is the same.
913 (setcdr (car rest
) (concat (cdar rest
) text
))
914 (push (cons cs text
) rest
))
915 ;; Don't decode encoded-word.
916 (push (cons nil
(nth 3 word
)) rest
)))
919 (or (and (setq cs
(caar rest
))
921 (mm-decode-coding-string (cdar rest
) cs
)
923 (message "%s" (error-message-string code
))
925 (concat (when (cdr rest
) " ")
928 (not (eq (string-to-char words
) ?
)))
934 ;; Fixme: This should decode in place, not cons intermediate strings.
935 ;; Also check whether it needs to worry about delimiting fields like
938 ;; In fact it's reported that (invalid) encoding of mailboxes in
939 ;; addr-specs is in use, so delimiting fields might help. Probably
940 ;; not decoding a word which isn't properly delimited is good enough
941 ;; and worthwhile (is it more correct or not?), e.g. something like
942 ;; `=?iso-8859-1?q?foo?=@'.
944 (defun rfc2047-decode-region (start end
&optional address-mime
)
945 "Decode MIME-encoded words in region between START and END.
946 If ADDRESS-MIME is non-nil, strip backslashes which precede characters
947 other than `\"' and `\\' in quoted strings."
949 (let ((case-fold-search t
)
950 (eword-regexp (eval-when-compile
951 ;; Ignore whitespace between encoded-words.
952 (concat "[\n\t ]*\\(" rfc2047-encoded-word-regexp
957 (narrow-to-region start end
)
959 (rfc2047-strip-backslashes-in-quoted-strings))
960 (goto-char (setq b start
))
961 ;; Look for the encoded-words.
962 (while (setq match
(re-search-forward eword-regexp nil t
))
963 (setq e
(match-beginning 1)
967 (push (list (match-string 2) ;; charset
968 (char-after (match-beginning 3)) ;; encoding
969 (match-string 4) ;; encoded-text
970 (match-string 1)) ;; encoded-word
972 ;; Look for the subsequent encoded-words.
973 (when (setq match
(looking-at eword-regexp
))
974 (goto-char (setq end
(match-end 0)))))
975 ;; Replace the encoded-words with the decoded one.
976 (delete-region e end
)
977 (insert (rfc2047-decode-encoded-words (nreverse words
)))
979 (narrow-to-region e
(point))
981 ;; Remove newlines between decoded words, though such
982 ;; things essentially must not be there.
983 (while (re-search-forward "[\n\r]+" nil t
)
985 ;; Quote decoded words if there are special characters
986 ;; which might violate RFC2822.
987 (when (and rfc2047-quote-decoded-words-containing-tspecials
988 (let ((regexp (car (rassq
990 rfc2047-header-encoding-alist
))))
995 (while (and (memq (char-after) '(? ?
\t))
996 (zerop (forward-line -
1))))
997 (looking-at regexp
)))))
1000 (skip-chars-forward " \t")
1001 (setq start
(point))
1002 (setq quoted
(eq (char-after) ?
\"))
1003 (goto-char (point-max))
1004 (skip-chars-backward " \t")
1005 (if (setq quoted
(and quoted
1006 (> (point) (1+ start
))
1007 (eq (char-before) ?
\")))
1010 (setq start
(1+ start
)
1011 end
(point-marker)))
1012 (setq end
(point-marker)))
1014 (while (search-forward "\"" end t
)
1017 (zerop (%
(skip-chars-backward "\\\\") 2))
1018 (goto-char (match-beginning 0)))
1021 (when (and (not quoted
)
1025 (concat "[" ietf-drums-tspecials
"]")
1031 (set-marker end nil
)))
1032 (goto-char (point-max)))
1033 (when (and (mm-multibyte-p)
1035 (not (eq mail-parse-charset
'us-ascii
))
1036 (not (eq mail-parse-charset
'gnus-decoded
)))
1037 (mm-decode-coding-region b e mail-parse-charset
))
1039 (when (and (mm-multibyte-p)
1041 (not (eq mail-parse-charset
'us-ascii
))
1042 (not (eq mail-parse-charset
'gnus-decoded
)))
1043 (mm-decode-coding-region b
(point-max) mail-parse-charset
))))))
1045 (defun rfc2047-decode-address-region (start end
)
1046 "Decode MIME-encoded words in region between START and END.
1047 Backslashes which precede characters other than `\"' and `\\' in quoted
1048 strings are stripped."
1049 (rfc2047-decode-region start end t
))
1051 (defun rfc2047-decode-string (string &optional address-mime
)
1052 "Decode MIME-encoded STRING and return the result.
1053 If ADDRESS-MIME is non-nil, strip backslashes which precede characters
1054 other than `\"' and `\\' in quoted strings."
1055 (let ((m (mm-multibyte-p)))
1056 (if (string-match "=\\?" string
)
1058 ;; Fixme: This logic is wrong, but seems to be required by
1059 ;; Gnus summary buffer generation. The value of `m' depends
1060 ;; on the current buffer, not global multibyteness or that
1061 ;; of the string. Also the string returned should always be
1062 ;; multibyte in a multibyte session, i.e. the buffer should
1063 ;; be multibyte before `buffer-string' is called.
1065 (mm-enable-multibyte))
1068 (rfc2047-decode-region (point-min) (point-max) address-mime
))
1073 (when (mm-multibyte-string-p string
)
1074 (mm-enable-multibyte))
1076 (rfc2047-strip-backslashes-in-quoted-strings)
1078 ;; Fixme: As above, `m' here is inappropriate.
1081 (not (eq mail-parse-charset
'us-ascii
))
1082 (not (eq mail-parse-charset
'gnus-decoded
)))
1083 ;; `decode-coding-string' in Emacs offers a third optional
1084 ;; arg NOCOPY to avoid consing a new string if the decoding
1085 ;; is "trivial". Unfortunately it currently doesn't
1086 ;; consider anything else than a `nil' coding system
1088 ;; `rfc2047-decode-string' is called multiple times for each
1089 ;; article during summary buffer generation, and we really
1090 ;; want to avoid unnecessary consing. So we bypass
1091 ;; `decode-coding-string' if the string is purely ASCII.
1092 (if (and (fboundp 'detect-coding-string
)
1093 ;; string is purely ASCII
1094 (eq (detect-coding-string string t
) 'undecided
))
1096 (mm-decode-coding-string string mail-parse-charset
))
1097 (mm-string-as-multibyte string
)))))
1099 (defun rfc2047-decode-address-string (string)
1100 "Decode MIME-encoded STRING and return the result.
1101 Backslashes which precede characters other than `\"' and `\\' in quoted
1102 strings are stripped."
1103 (rfc2047-decode-string string t
))
1105 (defun rfc2047-pad-base64 (string)
1106 "Pad STRING to quartets."
1107 ;; Be more liberal to accept buggy base64 strings. If
1108 ;; base64-decode-string accepts buggy strings, this function could
1109 ;; be aliased to identity.
1110 (if (= 0 (mod (length string
) 4))
1112 (when (string-match "=+$" string
)
1113 (setq string
(substring string
0 (match-beginning 0))))
1114 (case (mod (length string
) 4)
1116 (1 string
) ;; Error, don't pad it.
1117 (2 (concat string
"=="))
1118 (3 (concat string
"=")))))
1122 ;;; arch-tag: a07fe3d4-22b5-4c4a-bd89-b1f82d5d36f6
1123 ;;; rfc2047.el ends here