2003-02-13 Michael Shields <shields@msrl.com>

	* gnus-cite.el
	(gnus-cite-attribution-suffix, gnus-cite-parse):
	Better handling for Microsoft citation styles.
	(gnus-unsightly-citation-regexp): New.

diff --git a/lisp/ChangeLog b/lisp/ChangeLog
index fe20bb0..121ed5f 100644 (file)
--- a/lisp/ChangeLog
+++ b/lisp/ChangeLog
@@ -1,3 +1,10 @@
+2003-02-13  Michael Shields  <shields@msrl.com>
+
+       * gnus-cite.el
+       (gnus-cite-attribution-suffix, gnus-cite-parse):
+       Better handling for Microsoft citation styles.
+       (gnus-unsightly-citation-regexp): New.
+
 2003-02-12  Michael Shields  <shields@msrl.com>
 
        * gnus-art.el (article-strip-banner): Strip both per-group and

diff --git a/lisp/gnus-cite.el b/lisp/gnus-cite.el
index 1db19cc..cbeb5d6 100644 (file)
--- a/lisp/gnus-cite.el
+++ b/lisp/gnus-cite.el
@@ -90,19 +90,42 @@ The first regexp group should match the Supercite attribution."
   :group 'gnus-cite
   :type 'integer)
 
+;; Some Microsoft products put in a citation that extends to the
+;; remainder of the message:
+;;
+;;     -----Original Message-----
+;;     From: ...
+;;     To: ...
+;;     Sent: ...   [date, in non-RFC-2822 format]
+;;     Subject: ...
+;;
+;;     Cited message, with no prefixes
+;;
+;; The four headers are always the same.  But note they are prone to
+;; folding without additional indentation.
+;;
+;; Others use "----- Original Message -----" instead, and properly quote
+;; the body using "> ".  This style is handled without special cases.
+
 (defcustom gnus-cite-attribution-prefix
-  "In article\\|in <\\|On \\(Mon\\|Tue\\|Wed\\|Thu\\|Fri\\|Sat\\|Sun\\),\\|-----Original Message-----"
+  "In article\\|in <\\|On \\(Mon\\|Tue\\|Wed\\|Thu\\|Fri\\|Sat\\|Sun\\),\\|----- ?Original Message ?-----"
   "*Regexp matching the beginning of an attribution line."
   :group 'gnus-cite
   :type 'regexp)
 
 (defcustom gnus-cite-attribution-suffix
-  "\\(\\(wrote\\|writes\\|said\\|says\\|>\\)\\(:\\|\\.\\.\\.\\)\\|-----Original Message-----\\)[ \t]*$"
+  "\\(\\(wrote\\|writes\\|said\\|says\\|>\\)\\(:\\|\\.\\.\\.\\)\\|----- ?Original Message ?-----\\)[ \t]*$"
   "*Regexp matching the end of an attribution line.
 The text matching the first grouping will be used as a button."
   :group 'gnus-cite
   :type 'regexp)
 
+(defcustom gnus-unsightly-citation-regexp
+  "^-----Original Message-----\nFrom: \\(.+\n\\)+\n"
+  "Regexp matching Microsoft-type rest-of-message citations."
+  :group 'gnus-cite
+  :type 'regexp)
+
 (defface gnus-cite-attribution-face '((t
                                       (:italic t)))
   "Face used for attribution lines.")
@@ -724,9 +747,19 @@ See also the documentation for `gnus-article-highlight-citation'."
        (goto-char begin))
       (goto-char start)
       (setq line (1+ line)))
+    ;; Horrible special case for some Microsoft mailers.
+    (goto-char (point-min))
+    (when (re-search-forward gnus-unsightly-citation-regexp max t)
+      (setq begin (count-lines (point-min) (point)))
+      (setq end (count-lines (point-min) max))
+      (setq entry nil)
+      (while (< begin end)
+       (push begin entry)
+       (setq begin (1+ begin)))
+      (push (cons "" entry) alist))
     ;; We got all the potential prefixes.  Now create
     ;; `gnus-cite-prefix-alist' containing the oldest prefix for each
-    ;; line that appears at least gnus-cite-minimum-match-count
+    ;; line that appears at least `gnus-cite-minimum-match-count'
     ;; times.  First sort them by length.  Longer is older.
     (setq alist (sort alist (lambda (a b)
                              (> (length (car a)) (length (car b))))))