1 ;;; rfc2047.el --- functions for encoding and decoding rfc2047 messages
3 ;; Copyright (C) 1998-2015 Free Software Foundation, Inc.
5 ;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
6 ;; MORIOKA Tomohiko <morioka@jaist.ac.jp>
7 ;; This file is part of GNU Emacs.
9 ;; GNU Emacs is free software: you can redistribute it and/or modify
10 ;; it under the terms of the GNU General Public License as published by
11 ;; the Free Software Foundation, either version 3 of the License, or
12 ;; (at your option) any later version.
14 ;; GNU Emacs is distributed in the hope that it will be useful,
15 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
16 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
17 ;; GNU General Public License for more details.
19 ;; You should have received a copy of the GNU General Public License
20 ;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>.
24 ;; RFC 2047 is "MIME (Multipurpose Internet Mail Extensions) Part
25 ;; Three: Message Header Extensions for Non-ASCII Text".
31 (defvar message-posting-charset)
35 ;; Fixme: Avoid this (used for mail-parse-charset) mm dependence on gnus.
37 (require 'rfc2045) ;; rfc2045-encode-string
38 (autoload 'mm-body-7-or-8 "mm-bodies")
40 (defvar rfc2047-header-encoding-alist
41 '(("Newsgroups" . nil)
44 ("\\(Resent-\\)?\\(From\\|Cc\\|To\\|Bcc\\|\\(In-\\)?Reply-To\\|Sender\
45 \\|Mail-Followup-To\\|Mail-Copies-To\\|Approved\\)" . address-mime)
47 "*Header/encoding method alist.
48 The list is traversed sequentially. The keys can either be
53 1) nil, in which case no encoding is done;
54 2) `mime', in which case the header will be encoded according to RFC2047;
55 3) `address-mime', like `mime', but takes account of the rules for address
56 fields (where quoted strings and comments must be treated separately);
57 4) a charset, in which case it will be encoded as that charset;
58 5) `default', in which case the field will be encoded as the rest
61 (defvar rfc2047-charset-encoding-alist
87 "Alist of MIME charsets to RFC2047 encodings.
88 Valid encodings are nil, `Q' and `B'. These indicate binary (no) encoding,
89 quoted-printable and base64 respectively.")
91 (defvar rfc2047-encode-function-alist
92 '((Q . rfc2047-q-encode-string)
93 (B . rfc2047-b-encode-string)
95 "Alist of RFC2047 encodings to encoding functions.")
97 (defvar rfc2047-encode-encoded-words t
98 "Whether encoded words should be encoded again.")
100 (defvar rfc2047-allow-irregular-q-encoded-words t
101 "*Whether to decode irregular Q-encoded words.")
103 (eval-and-compile ;; Necessary to hard code them in `rfc2047-decode-region'.
104 (defconst rfc2047-encoded-word-regexp
105 "=\\?\\([^][\000-\040()<>@,\;:*\\\"/?.=]+\\)\\(?:\\*[^?]+\\)?\\?\
106 \\(B\\?[+/0-9A-Za-z]*=*\
109 "Regexp that matches encoded word."
110 ;; The patterns for the B encoding and the Q encoding, i.e. the ones
111 ;; beginning with "B" and "Q" respectively, are restricted into only
112 ;; the characters that those encodings may generally use.
114 (defconst rfc2047-encoded-word-regexp-loose
115 "=\\?\\([^][\000-\040()<>@,\;:*\\\"/?.=]+\\)\\(?:\\*[^?]+\\)?\\?\
116 \\(B\\?[+/0-9A-Za-z]*=*\
117 \\|Q\\?\\(?:\\?+[ -<>@-~]\\)?\\(?:[ ->@-~]+\\?+[ -<>@-~]\\)*[ ->@-~]*\\?*\
119 "Regexp that matches encoded word allowing loose Q encoding."
120 ;; The pattern for the Q encoding, i.e. the one beginning with "Q",
122 ;; "Q\\?\\(\\?+[^\n=?]\\)?\\([^\n?]+\\?+[^\n=?]\\)*[^\n?]*\\?*"
123 ;; <--------1-------><----------2,3----------><--4--><-5->
125 ;; 1. After "Q?", allow "?"s that follow a character other than "=".
126 ;; 2. Allow "=" after "Q?"; it isn't regarded as the terminator.
127 ;; 3. In the middle of an encoded word, allow "?"s that follow a
128 ;; character other than "=".
129 ;; 4. Allow any characters other than "?" in the middle of an
131 ;; 5. At the end, allow "?"s.
135 ;;; Functions for encoding RFC2047 messages
138 (defun rfc2047-qp-or-base64 ()
139 "Return the type with which to encode the buffer.
140 This is either `base64' or `quoted-printable'."
142 (let ((limit (min (point-max) (+ 2000 (point-min))))
144 (goto-char (point-min))
145 (skip-chars-forward "\x20-\x7f\r\n\t" limit)
146 (while (< (point) limit)
149 (skip-chars-forward "\x20-\x7f\r\n\t" limit))
150 (if (or (< (* 6 n8bit) (- limit (point-min)))
151 ;; Don't base64, say, a short line with a single
152 ;; non-ASCII char when splitting parts by charset.
157 (defun rfc2047-narrow-to-field ()
158 "Narrow the buffer to the header on the current line."
164 (if (re-search-forward "^[^ \n\t]" nil t)
167 (goto-char (point-min)))
169 (defun rfc2047-field-value ()
170 "Return the value of the field at point."
173 (rfc2047-narrow-to-field)
174 (re-search-forward ":[ \t\n]*" nil t)
175 (buffer-substring-no-properties (point) (point-max)))))
177 (defun rfc2047-quote-special-characters-in-quoted-strings (&optional
179 "Quote special characters with `\\'s in quoted strings.
180 Quoting will not be done in a quoted string if it contains characters
181 matching ENCODABLE-REGEXP or it is within parentheses."
182 (goto-char (point-min))
183 (let ((tspecials (concat "[" ietf-drums-tspecials "]"))
186 (with-syntax-table (standard-syntax-table)
190 (eq (char-before) ?\)))
192 (goto-char (point-max)))
194 (narrow-to-region start (point))
196 (while (search-forward "\"" nil t)
197 (setq beg (match-beginning 0))
198 (unless (eq (char-before beg) ?\\)
204 (setq end (1- (point)))
206 (if (and encodable-regexp
207 (re-search-forward encodable-regexp end t))
210 (narrow-to-region beg end)
211 (while (re-search-forward tspecials nil 'move)
212 (if (eq (char-before) ?\\)
213 (if (looking-at tspecials) ;; Already quoted.
216 (goto-char (match-beginning 0))
222 (goto-char (point-max)))
224 (setq start (point))))))
226 (defvar rfc2047-encoding-type 'address-mime
227 "The type of encoding done by `rfc2047-encode-region'.
228 This should be dynamically bound around calls to
229 `rfc2047-encode-region' to either `mime' or `address-mime'. See
230 `rfc2047-header-encoding-alist', for definitions.")
232 (defun rfc2047-encode-message-header ()
233 "Encode the message header according to `rfc2047-header-encoding-alist'.
234 Should be called narrowed to the head of the message."
237 (goto-char (point-min))
238 (let (alist elem method charsets)
241 (rfc2047-narrow-to-field)
243 alist rfc2047-header-encoding-alist
244 charsets (mm-find-mime-charset-region (point-min) (point-max)))
245 ;; M$ Outlook boycotts decoding of a header if it consists
246 ;; of two or more encoded words and those charsets differ;
247 ;; it seems to decode all words in a header from a charset
248 ;; found first in the header. So, we unify the charsets into
249 ;; a single one used for encoding the whole text in a header.
250 (let ((mm-coding-system-priorities
251 (if (= (length charsets) 1)
252 (cons (mm-charset-to-coding-system (car charsets))
253 mm-coding-system-priorities)
254 mm-coding-system-priorities)))
255 (while (setq elem (pop alist))
256 (when (or (and (stringp (car elem))
257 (looking-at (car elem)))
261 (if (not (rfc2047-encodable-p))
263 (when (eq method 'address-mime)
264 (rfc2047-quote-special-characters-in-quoted-strings))
265 (if (and (eq (mm-body-7-or-8) '8bit)
268 (car message-posting-charset)))
269 ;; 8 bit must be decoded.
270 (mm-encode-coding-region
271 (point-min) (point-max)
272 (mm-charset-to-coding-system
273 (car message-posting-charset))))
274 ;; No encoding necessary, but folding is nice
278 (goto-char (point-min))
279 (skip-chars-forward "^:")
280 (when (looking-at ": ")
284 ;; We found something that may perhaps be encoded.
285 (re-search-forward "^[^:]+: *" nil t)
287 ((eq method 'address-mime)
288 (rfc2047-encode-region (point) (point-max)))
290 (let ((rfc2047-encoding-type 'mime))
291 (rfc2047-encode-region (point) (point-max))))
292 ((eq method 'default)
293 (if (and (featurep 'mule)
294 (if (boundp 'enable-multibyte-characters)
295 (default-value 'enable-multibyte-characters))
297 (mm-encode-coding-region (point) (point-max)
298 mail-parse-charset)))
299 ;; We get this when CC'ing messages to newsgroups with
300 ;; 8-bit names. The group name mail copy just got
301 ;; unconditionally encoded. Previously, it would ask
302 ;; whether to encode, which was quite confusing for the
303 ;; user. If the new behavior is wrong, tell me. I have
304 ;; left the old code commented out below.
305 ;; -- Per Abrahamsen <abraham@dina.kvl.dk> Date: 2001-10-07.
306 ;; Modified by Dave Love, with the commented-out code changed
307 ;; in accordance with changes elsewhere.
309 (rfc2047-encode-region (point) (point-max)))
311 ;;; (if (or (message-options-get
312 ;;; 'rfc2047-encode-message-header-encode-any)
313 ;;; (message-options-set
314 ;;; 'rfc2047-encode-message-header-encode-any
316 ;;; "Some texts are not encoded. Encode anyway?")))
317 ;;; (rfc2047-encode-region (point-min) (point-max))
318 ;;; (error "Cannot send unencoded text")))
319 ((mm-coding-system-p method)
320 (if (or (and (featurep 'mule)
321 (if (boundp 'enable-multibyte-characters)
322 (default-value 'enable-multibyte-characters)))
323 (featurep 'file-coding))
324 (mm-encode-coding-region (point) (point-max) method)))
327 (goto-char (point-max))))))))
329 ;; Fixme: This, and the require below may not be the Right Thing, but
330 ;; should be safe just before release. -- fx 2001-02-08
332 (defun rfc2047-encodable-p ()
333 "Return non-nil if any characters in current buffer need encoding in headers.
334 The buffer may be narrowed."
335 (require 'message) ; for message-posting-charset
337 (mm-find-mime-charset-region (point-min) (point-max))))
338 (goto-char (point-min))
339 (or (and rfc2047-encode-encoded-words
341 (re-search-forward rfc2047-encoded-word-regexp nil t)
342 (goto-char (point-min))))
344 (not (equal charsets (list (car message-posting-charset))))))))
346 ;; Use this syntax table when parsing into regions that may need
347 ;; encoding. Double quotes are string delimiters, backslash is
348 ;; character quoting, and all other RFC 2822 special characters are
349 ;; treated as punctuation so we can use forward-sexp/forward-word to
350 ;; skip to the end of regions appropriately. Nb. ietf-drums does
351 ;; things differently.
352 (defconst rfc2047-syntax-table
353 ;; (make-char-table 'syntax-table '(2)) only works in Emacs.
354 (let ((table (make-syntax-table)))
355 ;; The following is done to work for setting all elements of the table;
356 ;; it appears to be the cleanest way.
357 ;; Play safe and don't assume the form of the word syntax entry --
359 (if (featurep 'xemacs)
360 (put-char-table t (get-char-table ?a (standard-syntax-table)) table)
361 (set-char-table-range table t (aref (standard-syntax-table) ?a)))
362 (modify-syntax-entry ?\\ "\\" table)
363 (modify-syntax-entry ?\" "\"" table)
364 (modify-syntax-entry ?\( "(" table)
365 (modify-syntax-entry ?\) ")" table)
366 (modify-syntax-entry ?\< "." table)
367 (modify-syntax-entry ?\> "." table)
368 (modify-syntax-entry ?\[ "." table)
369 (modify-syntax-entry ?\] "." table)
370 (modify-syntax-entry ?: "." table)
371 (modify-syntax-entry ?\; "." table)
372 (modify-syntax-entry ?, "." table)
373 (modify-syntax-entry ?@ "." table)
376 (defun rfc2047-encode-region (b e &optional dont-fold)
377 "Encode words in region B to E that need encoding.
378 By default, the region is treated as containing RFC2822 addresses.
379 Dynamically bind `rfc2047-encoding-type' to change that."
381 (narrow-to-region b e)
382 (let ((encodable-regexp (if rfc2047-encode-encoded-words
383 "[^\000-\177]+\\|=\\?"
385 start ; start of current token
387 ;; Whether there's an encoded word before the current token,
388 ;; either immediately or separated by space.
390 (orig-text (buffer-substring-no-properties b e)))
391 (if (eq 'mime rfc2047-encoding-type)
392 ;; Simple case. Continuous words in which all those contain
393 ;; non-ASCII characters are encoded collectively. Encoding
394 ;; ASCII words, including `Re:' used in Subject headers, is
395 ;; avoided for interoperability with non-MIME clients and
396 ;; for making it easy to find keywords.
398 (goto-char (point-min))
399 (while (progn (skip-chars-forward " \t\n")
402 (while (and (looking-at "[ \t\n]*\\([^ \t\n]+\\)")
404 (setq end (match-end 0))
405 (re-search-forward encodable-regexp end t)))
407 (if (> (point) start)
408 (rfc2047-encode start (point))
410 ;; `address-mime' case -- take care of quoted words, comments.
411 (rfc2047-quote-special-characters-in-quoted-strings encodable-regexp)
412 (with-syntax-table rfc2047-syntax-table
413 (goto-char (point-min))
414 (condition-case err ; in case of unbalanced quotes
415 ;; Look for rfc2822-style: sequences of atoms, quoted
416 ;; strings, specials, whitespace. (Specials mustn't be
420 (skip-chars-forward " \t\n")
423 ((not (char-after))) ; eob
425 ((eq ?\" (setq csyntax (char-syntax (char-after))))
429 ;; Does it need encoding?
431 (if (re-search-forward encodable-regexp end 'move)
432 ;; It needs encoding. Strip the quotes first,
433 ;; since encoded words can't occur in quotes.
440 ;; There was a preceding quoted word. We need
441 ;; to include any separating whitespace in this
442 ;; word to avoid it getting lost.
443 (skip-chars-backward " \t")
444 ;; A space is needed between the encoded words.
448 ;; Adjust the end position for the deleted quotes.
449 (rfc2047-encode start (- end 2))
450 (setq last-encoded t)) ; record that it was encoded
451 (setq last-encoded nil)))
453 ;; Skip other delimiters, but record that they've
454 ;; potentially separated quoted words.
456 (setq last-encoded nil))
458 (error "Unbalanced parentheses"))
460 ;; Look for the end of parentheses.
462 ;; Encode text as an unstructured field.
463 (let ((rfc2047-encoding-type 'mime))
464 (rfc2047-encode-region (1+ start) (1- (point))))
465 (skip-chars-forward ")"))
466 (t ; normal token/whitespace sequence
468 ;; Skip one ASCII word, or encode continuous words
469 ;; in which all those contain non-ASCII characters.
471 (while (not (or end (eobp)))
472 (when (looking-at "[\000-\177]+")
476 (while (and (or (re-search-forward
477 "[ \t\n]\\|\\Sw" end 'move)
479 (eq ?\\ (char-syntax (char-before))))
480 ;; Skip backslash-quoted characters.
483 (setq end (match-beginning 0))
484 (if rfc2047-encode-encoded-words
487 (when (search-forward "=?" end 'move)
488 (goto-char (match-beginning 0))
491 ;; Where the value nil of `end' means there may be
492 ;; text to have to be encoded following the point.
493 ;; Otherwise, the point reached to the end of ASCII
494 ;; words separated by whitespace or a special char.
496 (when (looking-at encodable-regexp)
497 (goto-char (setq begin (match-end 0)))
498 (while (and (looking-at "[ \t\n]+\\([^ \t\n]+\\)")
499 (setq end (match-end 0))
501 (while (re-search-forward
502 encodable-regexp end t))
505 (or (not (re-search-forward "\\Sw" end t))
507 (goto-char (match-beginning 0))
510 (when (looking-at "[^ \t\n]+")
511 (setq end (match-end 0))
512 (if (re-search-forward "\\Sw+" end t)
513 ;; There are special characters better
514 ;; to be encoded so that MTAs may parse
516 (cond ((= end (point)))
517 ((looking-at (concat "\\sw*\\("
522 (goto-char (1- (match-end 0)))
523 (unless (= (point) (match-beginning 0))
524 ;; Separate encodable text and
528 (skip-chars-forward " \t\n")
529 (if (and (looking-at "[^ \t\n]+")
530 (string-match encodable-regexp
533 (goto-char end)))))))
534 (skip-chars-backward " \t\n")
537 (if (re-search-forward encodable-regexp end 'move)
539 (unless (memq (char-before start) '(nil ?\t ? ))
542 (skip-chars-backward "^ \t\n")
543 (and (looking-at "\\Sw+")
544 (= (match-end 0) start)))
545 ;; Also encode bogus delimiters.
547 ;; Separate encodable text and delimiter.
550 (setq start (1+ start)
552 (rfc2047-encode start end)
553 (setq last-encoded t))
554 (setq last-encoded nil)))))
556 (if (or debug-on-quit debug-on-error)
557 (signal (car err) (cdr err))
558 (error "Invalid data for rfc2047 encoding: %s"
559 (mm-replace-in-string orig-text "[ \t\n]+" " "))))))))
561 (rfc2047-fold-region b (point)))
562 (goto-char (point-max))))
564 (defun rfc2047-encode-string (string &optional dont-fold)
565 "Encode words in STRING.
566 By default, the string is treated as containing addresses (see
567 `rfc2047-encoding-type')."
568 (mm-with-multibyte-buffer
570 (rfc2047-encode-region (point-min) (point-max) dont-fold)
574 ;; 2. Syntax of encoded-words
576 ;; While there is no limit to the length of a multiple-line header
577 ;; field, each line of a header field that contains one or more
578 ;; 'encoded-word's is limited to 76 characters.
580 ;; In `rfc2047-encode-parameter' it is bound to nil, so don't defconst it.
581 (defvar rfc2047-encode-max-chars 76
582 "Maximum characters of each header line that contain encoded-words.
583 According to RFC 2047, it is 76. If it is nil, encoded-words
584 will not be folded. Too small value may cause an error. You
585 should not change this value.")
587 (defun rfc2047-encode-1 (column string cs encoder start crest tail
589 "Subroutine used by `rfc2047-encode'."
590 (cond ((string-equal string "")
592 ((not rfc2047-encode-max-chars)
594 (funcall encoder (if cs
595 (mm-encode-coding-string string cs)
598 ((>= column rfc2047-encode-max-chars)
600 (cond ((string-match "\n[ \t]+\\'" eword)
601 ;; Remove a superfluous empty line.
602 (setq eword (substring eword 0 (match-beginning 0))))
603 ((string-match "(+\\'" eword)
604 ;; Break the line before the open parenthesis.
605 (setq crest (concat crest (match-string 0 eword))
606 eword (substring eword 0 (match-beginning 0))))))
607 (rfc2047-encode-1 (length crest) string cs encoder start " " tail
608 (concat eword "\n" crest)))
611 (limit (1- (length string)))
616 (setq next (concat start
619 (mm-encode-coding-string
620 (substring string 0 (1+ index))
622 (substring string 0 (1+ index))))
624 len (+ column (length next)))
625 (if (> len rfc2047-encode-max-chars)
628 (if (or (< index limit)
629 (<= (+ len (or (string-match "\n" tail)
631 rfc2047-encode-max-chars))
634 (if (string-match "\\`)+" tail)
635 ;; Break the line after the close parenthesis.
636 (setq tail (concat (substring tail 0 (match-end 0))
638 (substring tail (match-end 0)))
644 (concat eword next tail)
647 (string-match "(+\\'" eword))
648 (setq crest (concat crest (match-string 0 eword))
649 eword (substring eword 0 (match-beginning 0)))
650 (setq eword (concat eword next)))
652 eword (concat eword next)))
653 (when (string-match "\n[ \t]+\\'" eword)
654 ;; Remove a superfluous empty line.
655 (setq eword (substring eword 0 (match-beginning 0))))
656 (rfc2047-encode-1 (length crest) (substring string index)
657 cs encoder start " " tail
658 (concat eword "\n" crest)))))))
660 (defun rfc2047-encode (b e)
661 "Encode the word(s) in the region B to E.
662 Point moves to the end of the region."
663 (let ((mime-charset (or (mm-find-mime-charset-region b e) (list 'us-ascii)))
664 cs encoding tail crest eword)
665 ;; Use utf-8 as a last resort if determining charset of text fails.
666 (if (memq nil mime-charset)
667 (setq mime-charset (list 'utf-8)))
668 (cond ((> (length mime-charset) 1)
669 (error "Can't rfc2047-encode `%s'"
670 (buffer-substring-no-properties b e)))
671 ((= (length mime-charset) 1)
672 (setq mime-charset (car mime-charset)
673 cs (mm-charset-to-coding-system mime-charset))
674 (unless (and (mm-multibyte-p)
675 (mm-coding-system-p cs))
678 (narrow-to-region b e)
680 (or (cdr (assq mime-charset
681 rfc2047-charset-encoding-alist))
682 ;; For the charsets that don't have a preferred
683 ;; encoding, choose the one that's shorter.
684 (if (eq (rfc2047-qp-or-base64) 'base64)
689 (skip-chars-forward "^ \t\n")
690 ;; `tail' may contain a close parenthesis.
691 (setq tail (buffer-substring-no-properties e (point)))
693 (setq b (point-marker)
694 e (set-marker (make-marker) e))
695 (rfc2047-fold-region (point-at-bol) b)
697 (skip-chars-backward "^ \t\n")
698 (unless (= 0 (skip-chars-backward " \t"))
699 ;; `crest' may contain whitespace and an open parenthesis.
700 (setq crest (buffer-substring-no-properties (point) b)))
701 (setq eword (rfc2047-encode-1
703 (mm-replace-in-string
704 (buffer-substring-no-properties b e)
705 "\n\\([ \t]?\\)" "\\1")
707 (or (cdr (assq encoding
708 rfc2047-encode-function-alist))
710 (concat "=?" (downcase (symbol-name mime-charset))
711 "?" (upcase (symbol-name encoding)) "?")
714 (delete-region (if (eq (aref eword 0) ?\n)
716 ;; The line was folded before encoding.
721 ;; `eword' contains `crest' and `tail'.
725 (unless (or (/= 0 (length tail))
727 (looking-at "[ \t\n)]"))
732 (defun rfc2047-fold-field ()
733 "Fold the current header field."
736 (rfc2047-narrow-to-field)
737 (rfc2047-fold-region (point-min) (point-max)))))
739 (defun rfc2047-fold-region (b e)
740 "Fold long lines in region B to E."
742 (narrow-to-region b e)
743 (goto-char (point-min))
747 (bol (save-restriction
751 (when (and (or break qword-break)
752 (> (- (point) bol) 76))
753 (goto-char (or break qword-break))
756 (skip-chars-backward " \t")
757 (if (looking-at "[ \t]")
760 (setq bol (1- (point)))
761 ;; Don't break before the first non-LWSP characters.
762 (skip-chars-forward " \t")
766 ((eq (char-after) ?\n)
771 (skip-chars-forward " \t")
772 (unless (or (eobp) (eq (char-after) ?\n))
774 ((eq (char-after) ?\r)
776 ((memq (char-after) '(? ?\t))
777 (skip-chars-forward " \t")
778 (unless first ;; Don't break just after the header name.
779 (setq break (point))))
781 (if (not (looking-at "=\\?[^=]"))
782 (if (eq (char-after) ?=)
784 (skip-chars-forward "^ \t\n\r="))
785 ;; Don't break at the start of the field.
786 (unless (= (point) b)
787 (setq qword-break (point)))
788 (skip-chars-forward "^ \t\n\r")))
790 (skip-chars-forward "^ \t\n\r")))
792 (when (and (or break qword-break)
793 (> (- (point) bol) 76))
794 (goto-char (or break qword-break))
797 (if (or (> 0 (skip-chars-backward " \t"))
798 (looking-at "[ \t]"))
801 (setq bol (1- (point)))
802 ;; Don't break before the first non-LWSP characters.
803 (skip-chars-forward " \t")
805 (forward-char 1))))))
807 (defun rfc2047-unfold-field ()
808 "Fold the current line."
811 (rfc2047-narrow-to-field)
812 (rfc2047-unfold-region (point-min) (point-max)))))
814 (defun rfc2047-unfold-region (b e)
815 "Unfold lines in region B to E."
817 (narrow-to-region b e)
818 (goto-char (point-min))
819 (let ((bol (save-restriction
822 (eol (point-at-eol)))
825 (if (and (looking-at "[ \t]")
826 (< (- (point-at-eol) bol) 76))
827 (delete-region eol (progn
829 (skip-chars-forward "\r\n")
831 (setq bol (point-at-bol)))
832 (setq eol (point-at-eol))
835 (defun rfc2047-b-encode-string (string)
836 "Base64-encode the header contained in STRING."
837 (base64-encode-string string t))
839 (autoload 'quoted-printable-encode-region "qp")
841 (defun rfc2047-q-encode-string (string)
842 "Quoted-printable-encode the header in STRING."
843 (mm-with-unibyte-buffer
845 (quoted-printable-encode-region
846 (point-min) (point-max) nil
847 ;; = (\075), _ (\137), ? (\077) are used in the encoded word.
848 ;; Avoid using 8bit characters.
849 ;; This list excludes `especials' (see the RFC2047 syntax),
850 ;; meaning that some characters in non-structured fields will
851 ;; get encoded when they con't need to be. The following is
852 ;; what it used to be.
853 ;;; ;; Equivalent to "^\000-\007\011\013\015-\037\200-\377=_?"
854 ;;; "\010\012\014\040-\074\076\100-\136\140-\177")
855 "-\b\n\f !#-'*+0-9A-Z\\^`-~\d")
856 (subst-char-in-region (point-min) (point-max) ? ?_)
859 (defun rfc2047-encode-parameter (param value)
860 "Return and PARAM=VALUE string encoded in the RFC2047-like style.
861 This is a substitution for the `rfc2231-encode-string' function, that
862 is the standard but many mailers don't support it."
863 (let ((rfc2047-encoding-type 'mime)
864 (rfc2047-encode-max-chars nil))
865 (rfc2045-encode-string param (rfc2047-encode-string value t))))
868 ;;; Functions for decoding RFC2047 messages
871 (defvar rfc2047-quote-decoded-words-containing-tspecials nil
872 "If non-nil, quote decoded words containing special characters.")
874 (defvar rfc2047-allow-incomplete-encoded-text t
875 "*Non-nil means allow incomplete encoded-text in successive encoded-words.
876 Dividing of encoded-text in the place other than character boundaries
877 violates RFC2047 section 5, while we have a capability to decode it.
878 If it is non-nil, the decoder will decode B- or Q-encoding in each
879 encoded-word, concatenate them, and decode it by charset. Otherwise,
880 the decoder will fully decode each encoded-word before concatenating
883 (defun rfc2047-strip-backslashes-in-quoted-strings ()
884 "Strip backslashes in quoted strings. `\\\"' remains."
885 (goto-char (point-min))
887 (with-syntax-table (standard-syntax-table)
888 (while (search-forward "\"" nil t)
889 (unless (eq (char-before) ?\\)
890 (setq beg (match-end 0))
891 (goto-char (match-beginning 0))
896 (narrow-to-region beg (1- (point)))
898 (while (search-forward "\\" nil 'move)
899 (unless (memq (char-after) '(?\"))
904 (goto-char beg))))))))
906 (defun rfc2047-charset-to-coding-system (charset &optional allow-override)
907 "Return coding-system corresponding to MIME CHARSET.
908 If your Emacs implementation can't decode CHARSET, return nil.
910 If allow-override is given, use `mm-charset-override-alist' to
911 map undesired charset names to their replacement. This should
912 only be used for decoding, not for encoding."
913 (when (stringp charset)
914 (setq charset (intern (downcase charset))))
915 (when (or (not charset)
916 (eq 'gnus-all mail-parse-ignored-charsets)
917 (memq 'gnus-all mail-parse-ignored-charsets)
918 (memq charset mail-parse-ignored-charsets))
919 (setq charset mail-parse-charset))
920 (let ((cs (mm-charset-to-coding-system charset nil allow-override)))
921 (cond ((eq cs 'ascii)
922 (setq cs (or (mm-charset-to-coding-system mail-parse-charset)
924 ((mm-coding-system-p cs))
926 (listp mail-parse-ignored-charsets)
927 (memq 'gnus-unknown mail-parse-ignored-charsets))
928 (setq cs (mm-charset-to-coding-system mail-parse-charset))))
933 (autoload 'quoted-printable-decode-string "qp")
935 (defun rfc2047-decode-encoded-words (words)
936 "Decode successive encoded-words in WORDS and return a decoded string.
937 Each element of WORDS looks like (CHARSET ENCODING ENCODED-TEXT
939 (let (word charset cs encoding text rest)
941 (setq word (pop words))
942 (if (and (setq cs (rfc2047-charset-to-coding-system
943 (setq charset (car word)) t))
945 (cond ((char-equal ?B (nth 1 word))
946 (setq text (base64-decode-string
947 (rfc2047-pad-base64 (nth 2 word)))))
948 ((char-equal ?Q (nth 1 word))
949 (setq text (quoted-printable-decode-string
950 (mm-subst-char-in-string
951 ?_ ? (nth 2 word) t)))))
953 (message "%s" (error-message-string code))
955 (if (and rfc2047-allow-incomplete-encoded-text
957 ;; Concatenate text of which the charset is the same.
958 (setcdr (car rest) (concat (cdar rest) text))
959 (push (cons cs text) rest))
960 ;; Don't decode encoded-word.
961 (push (cons nil (nth 3 word)) rest)))
964 (or (and (setq cs (caar rest))
966 (mm-decode-coding-string (cdar rest) cs)
968 (message "%s" (error-message-string code))
970 (concat (when (cdr rest) " ")
973 (not (eq (string-to-char words) ? )))
979 ;; Fixme: This should decode in place, not cons intermediate strings.
980 ;; Also check whether it needs to worry about delimiting fields like
983 ;; In fact it's reported that (invalid) encoding of mailboxes in
984 ;; addr-specs is in use, so delimiting fields might help. Probably
985 ;; not decoding a word which isn't properly delimited is good enough
986 ;; and worthwhile (is it more correct or not?), e.g. something like
987 ;; `=?iso-8859-1?q?foo?=@'.
989 (defun rfc2047-decode-region (start end &optional address-mime)
990 "Decode MIME-encoded words in region between START and END.
991 If ADDRESS-MIME is non-nil, strip backslashes which precede characters
992 other than `\"' and `\\' in quoted strings."
994 (let ((case-fold-search t)
996 (if rfc2047-allow-irregular-q-encoded-words
998 (concat "[\n\t ]*\\(" rfc2047-encoded-word-regexp-loose "\\)"))
1000 (concat "[\n\t ]*\\(" rfc2047-encoded-word-regexp "\\)"))))
1004 (narrow-to-region start end)
1006 (rfc2047-strip-backslashes-in-quoted-strings))
1007 (goto-char (setq b start))
1008 ;; Look for the encoded-words.
1009 (while (setq match (re-search-forward eword-regexp nil t))
1010 (setq e (match-beginning 1)
1014 (push (list (match-string 2) ;; charset
1015 (char-after (match-beginning 3)) ;; encoding
1016 (substring (match-string 3) 2) ;; encoded-text
1017 (match-string 1)) ;; encoded-word
1019 ;; Look for the subsequent encoded-words.
1020 (when (setq match (looking-at eword-regexp))
1021 (goto-char (setq end (match-end 0)))))
1022 ;; Replace the encoded-words with the decoded one.
1023 (delete-region e end)
1024 (insert (rfc2047-decode-encoded-words (nreverse words)))
1026 (narrow-to-region e (point))
1028 ;; Remove newlines between decoded words, though such
1029 ;; things essentially must not be there.
1030 (while (re-search-forward "[\n\r]+" nil t)
1031 (replace-match " "))
1032 (setq end (point-max))
1033 ;; Quote decoded words if there are special characters
1034 ;; which might violate RFC2822.
1035 (when (and rfc2047-quote-decoded-words-containing-tspecials
1036 (let ((regexp (car (rassq
1038 rfc2047-header-encoding-alist))))
1043 ;; Don't quote words if already quoted.
1044 (not (and (eq (char-before e) ?\")
1045 (eq (char-after end) ?\")))
1048 (while (and (memq (char-after) '(? ?\t))
1049 (zerop (forward-line -1))))
1050 (looking-at regexp)))))))
1053 (skip-chars-forward " \t")
1054 (setq start (point))
1055 (setq quoted (eq (char-after) ?\"))
1056 (goto-char (point-max))
1057 (skip-chars-backward " \t" start)
1058 (if (setq quoted (and quoted
1059 (> (point) (1+ start))
1060 (eq (char-before) ?\")))
1063 (setq start (1+ start)
1064 end (point-marker)))
1065 (setq end (point-marker)))
1067 (while (search-forward "\"" end t)
1070 (zerop (% (skip-chars-backward "\\\\") 2))
1071 (goto-char (match-beginning 0)))
1074 (when (and (not quoted)
1078 (concat "[" ietf-drums-tspecials "]")
1084 (set-marker end nil)))
1085 (goto-char (point-max)))
1086 (when (and (mm-multibyte-p)
1088 (not (eq mail-parse-charset 'us-ascii))
1089 (not (eq mail-parse-charset 'gnus-decoded)))
1090 (mm-decode-coding-region b e mail-parse-charset))
1092 (when (and (mm-multibyte-p)
1094 (not (eq mail-parse-charset 'us-ascii))
1095 (not (eq mail-parse-charset 'gnus-decoded)))
1096 (mm-decode-coding-region b (point-max) mail-parse-charset))))))
1098 (defun rfc2047-decode-address-region (start end)
1099 "Decode MIME-encoded words in region between START and END.
1100 Backslashes which precede characters other than `\"' and `\\' in quoted
1101 strings are stripped."
1102 (rfc2047-decode-region start end t))
1104 (defun rfc2047-decode-string (string &optional address-mime)
1105 "Decode MIME-encoded STRING and return the result.
1106 If ADDRESS-MIME is non-nil, strip backslashes which precede characters
1107 other than `\"' and `\\' in quoted strings."
1108 ;; (let ((m (mm-multibyte-p)))
1109 (if (string-match "=\\?" string)
1111 ;; We used to only call mm-enable-multibyte if `m' is non-nil,
1112 ;; but this can't be the right criterion. Don't just revert this
1113 ;; change if it encounters a bug. Please help me fix it
1114 ;; right instead. --Stef
1115 ;; The string returned should always be multibyte in a multibyte
1116 ;; session, i.e. the buffer should be multibyte before
1117 ;; `buffer-string' is called.
1118 (mm-enable-multibyte)
1121 (rfc2047-decode-region (point-min) (point-max) address-mime))
1126 (when (mm-multibyte-string-p string)
1127 (mm-enable-multibyte))
1129 (rfc2047-strip-backslashes-in-quoted-strings)
1131 ;; Fixme: As above, `m' here is inappropriate.
1134 (not (eq mail-parse-charset 'us-ascii))
1135 (not (eq mail-parse-charset 'gnus-decoded)))
1136 ;; `decode-coding-string' in Emacs offers a third optional
1137 ;; arg NOCOPY to avoid consing a new string if the decoding
1138 ;; is "trivial". Unfortunately it currently doesn't
1139 ;; consider anything else than a nil coding system
1141 ;; `rfc2047-decode-string' is called multiple times for each
1142 ;; article during summary buffer generation, and we really
1143 ;; want to avoid unnecessary consing. So we bypass
1144 ;; `decode-coding-string' if the string is purely ASCII.
1145 (if (and (fboundp 'detect-coding-string)
1146 ;; string is purely ASCII
1147 (eq (detect-coding-string string t) 'undecided))
1149 (mm-decode-coding-string string mail-parse-charset))
1150 (mm-string-to-multibyte string)))) ;; )
1152 (defun rfc2047-decode-address-string (string)
1153 "Decode MIME-encoded STRING and return the result.
1154 Backslashes which precede characters other than `\"' and `\\' in quoted
1155 strings are stripped."
1156 (rfc2047-decode-string string t))
1158 (defun rfc2047-pad-base64 (string)
1159 "Pad STRING to quartets."
1160 ;; Be more liberal to accept buggy base64 strings. If
1161 ;; base64-decode-string accepts buggy strings, this function could
1162 ;; be aliased to identity.
1163 (if (= 0 (mod (length string) 4))
1165 (when (string-match "=+$" string)
1166 (setq string (substring string 0 (match-beginning 0))))
1167 (case (mod (length string) 4)
1169 (1 string) ;; Error, don't pad it.
1170 (2 (concat string "=="))
1171 (3 (concat string "=")))))
1175 ;;; rfc2047.el ends here