summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorEli Zaretskii <eliz@gnu.org>2015-02-28 14:03:34 +0200
committerEli Zaretskii <eliz@gnu.org>2015-02-28 14:03:34 +0200
commit31ecbf8d513540855aa07588f6746942aed453ba (patch)
tree0b2389ed3bfa8928f8f65473a0dae1f987f3cdbd
parent0537943561a37b54467bec19d1b8afbeba8e1e58 (diff)
downloademacs-31ecbf8d513540855aa07588f6746942aed453ba.tar.gz
Better decoding of HTML payload in Rmail
lisp/mail/rmailmm.el (rmail-mime-insert-html): Decode HTML payload when the charset is only given by the HTML <head>, and allow to specify the encoding with "C-x RET c".
-rw-r--r--lisp/ChangeLog6
-rw-r--r--lisp/mail/rmailmm.el17
2 files changed, 23 insertions, 0 deletions
diff --git a/lisp/ChangeLog b/lisp/ChangeLog
index b9681d35cf0..1cfd08fc95d 100644
--- a/lisp/ChangeLog
+++ b/lisp/ChangeLog
@@ -1,3 +1,9 @@
+2015-02-28 Eli Zaretskii <eliz@gnu.org>
+
+ * mail/rmailmm.el (rmail-mime-insert-html): Decode HTML payload
+ when the charset is only given by the HTML <head>, and allow to
+ specify the encoding with "C-x RET c".
+
2015-02-27 Mark Laws <mdl@60hz.org>
Support daemon mode on MS-Windows (bug#19688)
diff --git a/lisp/mail/rmailmm.el b/lisp/mail/rmailmm.el
index 120d517f55c..00fc25dd440 100644
--- a/lisp/mail/rmailmm.el
+++ b/lisp/mail/rmailmm.el
@@ -661,6 +661,7 @@ HEADER is a header component of a MIME-entity object (see
(transfer-encoding (rmail-mime-entity-transfer-encoding entity))
(charset (cdr (assq 'charset (cdr (rmail-mime-entity-type entity)))))
(buffer (current-buffer))
+ (case-fold-search t)
coding-system)
(if charset (setq coding-system (coding-system-from-name charset)))
(or (and coding-system (coding-system-p coding-system))
@@ -674,6 +675,22 @@ HEADER is a header component of a MIME-entity object (see
(ignore-errors (base64-decode-region (point-min) (point-max))))
((string= transfer-encoding "quoted-printable")
(quoted-printable-decode-region (point-min) (point-max))))
+ ;; Some broken MUAs state the charset only in the HTML <head>,
+ ;; so if we don't have a non-trivial coding-system at this
+ ;; point, make one last attempt to find it there.
+ (if (eq coding-system 'undecided)
+ (save-excursion
+ (goto-char (point-min))
+ (when (re-search-forward
+ "^<html><head><meta[^;]*; charset=\\([-a-zA-Z0-9]+\\)"
+ nil t)
+ (setq coding-system (coding-system-from-name (match-string 1)))
+ (or (and coding-system (coding-system-p coding-system))
+ (setq coding-system 'undecided)))
+ ;; Finally, let them manually force decoding if they know it.
+ (if (and (eq coding-system 'undecided)
+ (not (null coding-system-for-read)))
+ (setq coding-system coding-system-for-read))))
(decode-coding-region (point-min) (point) coding-system)
(if (and
(or (not rmail-mime-coding-system) (consp rmail-mime-coding-system))