Add arch taglines
[gnus] / lisp / nnwarchive.el
index d4f7b21..a9d0d51 100644 (file)
@@ -1,8 +1,8 @@
 ;;; nnwarchive.el --- interfacing with web archives
-;; Copyright (C) 1999 Free Software Foundation, Inc.
+;; Copyright (C) 1999, 2000, 2003 Free Software Foundation, Inc.
 
 ;; Author: Shenghuo Zhu <zsh@cs.rochester.edu>
-;; Keywords: news
+;; Keywords: news egroups mail-archive
 
 ;; This file is part of GNU Emacs.
 
 
 ;;; Commentary:
 
-;; Note: You need to have `url' and `w3' installed for this backend to
-;; work.
+;; Note: You need to have `url' (w3 0.46) or greater version
+;; installed for some functions of this backend to work.
 
-;; A lot of codes stolen from mail-source, nnslashdot, nnweb.
-
-;; Todo: 
+;; Todo:
 ;; 1. To support more web archives.
-;; 2. Support nnwarchive-xover-is-evil.
-
-;; Known bugs: in w3 0.44, there are two copies of url-maybe-relative.
-;; If it is loaded from w3.el, (load-library "url").  Update to w3 0.46 
-;; or greater version.
+;; 2. Generalize webmail to other MHonArc archive.
 
 ;;; Code:
 
 (require 'message)
 (require 'gnus-util)
 (require 'gnus)
+(require 'gnus-bcklg)
 (require 'nnmail)
 (require 'mm-util)
-(require 'mail-source)
-(eval-when-compile
-  (ignore-errors
-    (require 'w3)
-    (require 'url)
-    (require 'w3-forms)
-    (require 'nnweb)))
-;; Report failure to find w3 at load time if appropriate.
-(eval '(progn
-        (require 'w3)
-        (require 'url)
-        (require 'w3-forms)
-        (require 'nnweb)))
+(require 'mm-url)
 
 (nnoo-declare nnwarchive)
 
 (defvar nnwarchive-type-definition
   '((egroups
      (address . "www.egroups.com")
-     (open-url 
-      "http://www.egroups.com/register?method=loginAction&email=%s&password=%s" 
+     (open-url
+      "http://www.egroups.com/login.cgi?&login_email=%s&login_password=%s"
       nnwarchive-login nnwarchive-passwd)
-     (list-url 
-      "http://www.egroups.com/UserGroupsPage?")
+     (list-url
+      "http://www.egroups.com/mygroups")
      (list-dissect . nnwarchive-egroups-list)
      (list-groups . nnwarchive-egroups-list-groups)
-     (xover-url 
-      "http://www.egroups.com/group/%s/?fetchForward=1&start=%d" group aux)
-     (xover-last-url 
-      "http://www.egroups.com/group/%s/?fetchForward=1" group)
+     (xover-url
+      "http://www.egroups.com/messages/%s/%d" group aux)
+     (xover-last-url
+      "http://www.egroups.com/messages/%s/" group)
      (xover-page-size . 13)
      (xover-dissect . nnwarchive-egroups-xover)
-     (article-url 
-      "http://www.egroups.com/group/%s/%d.html?raw=1" group article)
+     (article-url
+      "http://www.egroups.com/message/%s/%d?source=1" group article)
      (article-dissect . nnwarchive-egroups-article)
      (authentication . t)
+     (article-offset . 0)
      (xover-files . nnwarchive-egroups-xover-files))
     (mail-archive
      (address . "www.mail-archive.com")
-     (list-url 
+     (open-url)
+     (list-url
       "http://www.mail-archive.com/lists.html")
      (list-dissect . nnwarchive-mail-archive-list)
      (list-groups . nnwarchive-mail-archive-list-groups)
-     (xover-url 
+     (xover-url
       "http://www.mail-archive.com/%s/mail%d.html" group aux)
-     (xover-last-url 
+     (xover-last-url
       "http://www.mail-archive.com/%s/maillist.html" group)
+     (xover-page-size)
      (xover-dissect . nnwarchive-mail-archive-xover)
-     (article-url 
+     (article-url
       "http://www.mail-archive.com/%s/msg%05d.html" group article1)
      (article-dissect . nnwarchive-mail-archive-article)
      (xover-files . nnwarchive-mail-archive-xover-files)
+     (authentication)
      (article-offset . 1))))
 
 (defvar nnwarchive-default-type 'egroups)
 
 (defvoo nnwarchive-headers-cache nil)
 
-(defvoo nnwarchive-opened nil)
-
 (defvoo nnwarchive-authentication nil)
 
-(defvoo nnwarchive-xover-is-evil nil) ;; not implemented
+(defvoo nnwarchive-nov-is-evil nil)
 
 (defconst nnwarchive-version "nnwarchive 1.0")
 
 
 (defvoo nnwarchive-buffer nil)
 
-(defvar nnwarchive-headers nil)
+(defvoo nnwarchive-keep-backlog 300)
+(defvar nnwarchive-backlog-articles nil)
+(defvar nnwarchive-backlog-hashtb nil)
+
+(defvoo nnwarchive-headers nil)
 
 
 ;;; Interface functions
   (let ((defs (cdr (assq type nnwarchive-type-definition)))
        def)
     (dolist (def defs)
-      (set (intern (concat "nnwarchive-" (symbol-name (car def)))) 
+      (set (intern (concat "nnwarchive-" (symbol-name (car def))))
           (cdr def)))))
 
+(defmacro nnwarchive-backlog (&rest form)
+  `(let ((gnus-keep-backlog nnwarchive-keep-backlog)
+        (gnus-backlog-buffer
+         (format " *nnwarchive backlog %s*" nnwarchive-address))
+        (gnus-backlog-articles nnwarchive-backlog-articles)
+        (gnus-backlog-hashtb nnwarchive-backlog-hashtb))
+     (unwind-protect
+        (progn ,@form)
+       (setq nnwarchive-backlog-articles gnus-backlog-articles
+            nnwarchive-backlog-hashtb gnus-backlog-hashtb))))
+(put 'nnwarchive-backlog 'lisp-indent-function 0)
+(put 'nnwarchive-backlog 'edebug-form-spec '(form body))
+
+(defun nnwarchive-backlog-enter-article (group number buffer)
+  (nnwarchive-backlog
+    (gnus-backlog-enter-article group number buffer)))
+
+(defun nnwarchive-get-article (article &optional group server buffer)
+  (if (numberp article)
+      (if (nnwarchive-backlog
+           (gnus-backlog-request-article group article
+                                         (or buffer nntp-server-buffer)))
+         (cons group article)
+       (let (contents)
+         (save-excursion
+           (set-buffer nnwarchive-buffer)
+           (goto-char (point-min))
+           (let ((article1 (- article nnwarchive-article-offset)))
+             (nnwarchive-url nnwarchive-article-url))
+           (setq contents (funcall nnwarchive-article-dissect group article)))
+         (when contents
+           (save-excursion
+             (set-buffer (or buffer nntp-server-buffer))
+             (erase-buffer)
+             (insert contents)
+             (nnwarchive-backlog-enter-article group article (current-buffer))
+             (nnheader-report 'nnwarchive "Fetched article %s" article)
+             (cons group article)))))
+    nil))
+
 (deffoo nnwarchive-retrieve-headers (articles &optional group server fetch-old)
   (nnwarchive-possibly-change-server group server)
-  (setq nnwarchive-headers (cdr (assoc group nnwarchive-headers-cache)))
-  (save-excursion
-    (set-buffer nnwarchive-buffer)
-    (erase-buffer)
-    (funcall nnwarchive-xover-files group articles))
-  (save-excursion
-    (set-buffer nntp-server-buffer)
-    (erase-buffer)
-    (let (header)
+  (if (or gnus-nov-is-evil nnwarchive-nov-is-evil)
+      (with-temp-buffer
+       (with-current-buffer nntp-server-buffer
+         (erase-buffer))
+       (let ((buf (current-buffer)) b e)
+         (dolist (art articles)
+           (nnwarchive-get-article art group server buf)
+           (setq b (goto-char (point-min)))
+           (if (search-forward "\n\n" nil t)
+               (forward-char -1)
+             (goto-char (point-max)))
+           (setq e (point))
+           (with-current-buffer nntp-server-buffer
+             (insert (format "221 %d Article retrieved.\n" art))
+             (insert-buffer-substring buf b e)
+             (insert ".\n"))))
+       'headers)
+    (setq nnwarchive-headers (cdr (assoc group nnwarchive-headers-cache)))
+    (save-excursion
+      (set-buffer nnwarchive-buffer)
+      (erase-buffer)
+      (funcall nnwarchive-xover-files group articles))
+    (save-excursion
+      (set-buffer nntp-server-buffer)
+      (erase-buffer)
+      (let (header)
       (dolist (art articles)
        (if (setq header (assq art nnwarchive-headers))
            (nnheader-insert-nov (cdr header))))))
-  (let ((elem (assoc group nnwarchive-headers-cache)))
-    (if elem
-       (setcdr elem nnwarchive-headers)
-      (push (cons group nnwarchive-headers) nnwarchive-headers-cache)))
-  'nov)
-
-(deffoo nnwarchive-retrieve-groups (groups &optional server)
-  "Retrieve group info on GROUPS."
-  (nnwarchive-possibly-change-server nil server)
-  (if nnwarchive-list-groups
-      (funcall nnwarchive-list-groups groups))
-  (nnwarchive-write-groups)
-  (nnwarchive-generate-active)
-  'active)
+    (let ((elem (assoc group nnwarchive-headers-cache)))
+      (if elem
+         (setcdr elem nnwarchive-headers)
+       (push (cons group nnwarchive-headers) nnwarchive-headers-cache)))
+    'nov))
 
 (deffoo nnwarchive-request-group (group &optional server dont-check)
   (nnwarchive-possibly-change-server nil server)
-  (if nnwarchive-list-groups
-      (funcall nnwarchive-list-groups (list group)))
-  (nnwarchive-write-groups)
+  (when (and (not dont-check) nnwarchive-list-groups)
+    (funcall nnwarchive-list-groups (list group))
+    (nnwarchive-write-groups))
   (let ((elem (assoc group nnwarchive-groups)))
     (cond
      ((not elem)
        (prin1-to-string group))
       t))))
 
-(deffoo nnwarchive-close-group (group &optional server)
-  (nnwarchive-possibly-change-server group server)
-  (when (gnus-buffer-live-p nnwarchive-buffer)
-    (save-excursion
-      (set-buffer nnwarchive-buffer)
-      (kill-buffer nnwarchive-buffer)))
-  t)
-
 (deffoo nnwarchive-request-article (article &optional group server buffer)
   (nnwarchive-possibly-change-server group server)
-  (let (contents)
-    (save-excursion
-      (set-buffer nnwarchive-buffer)
-      (goto-char (point-min))
-      (let ((article1 (- article nnwarchive-article-offset)))
-       (nnwarchive-url nnwarchive-article-url))
-      (setq contents (funcall nnwarchive-article-dissect group article)))
-    (when contents
-      (save-excursion
-       (set-buffer (or buffer nntp-server-buffer))
-       (erase-buffer)
-       (insert contents)
-       (nnheader-report 'nnwarchive "Fetched article %s" article)
-       (cons group article)))))
+  (nnwarchive-get-article article group server buffer))
 
 (deffoo nnwarchive-close-server (&optional server)
   (when (and (nnwarchive-server-opened server)
     (save-excursion
       (set-buffer nnwarchive-buffer)
       (kill-buffer nnwarchive-buffer)))
+  (nnwarchive-backlog
+    (gnus-backlog-shutdown))
   (nnoo-close-server 'nnwarchive server))
 
 (deffoo nnwarchive-request-list (&optional server)
        (funcall nnwarchive-list-dissect))
     (nnwarchive-write-groups)
     (nnwarchive-generate-active))
-  'active)
-
-(deffoo nnwarchive-request-newgroups (date &optional server)
-  (nnwarchive-possibly-change-server nil server)
-  (nnwarchive-write-groups)
-  (nnwarchive-generate-active)
-  'active)
-
-(deffoo nnwarchive-asynchronous-p ()
-  nil)
-
-(deffoo nnwarchive-server-opened (&optional server)
-  nnwarchive-opened)
+  t)
 
 (deffoo nnwarchive-open-server (server &optional defs connectionless)
+  (nnoo-change-server 'nnwarchive server defs)
   (nnwarchive-init server)
-  (unless (nnwarchive-server-opened server)
-    (when nnwarchive-authentication
-      (setq nnwarchive-login
-           (or nnwarchive-login
-               (read-string
+  (when nnwarchive-authentication
+    (setq nnwarchive-login
+         (or nnwarchive-login
+             (read-string
                 (format "Login at %s: " server)
                 user-mail-address)))
-      (setq nnwarchive-passwd
-           (or nnwarchive-passwd
-               (mail-source-read-passwd
-                (format "Password for %s at %s: " 
-                        nnwarchive-login server)))))
-    (unless nnwarchive-groups
-      (nnwarchive-read-groups))
-    (save-excursion
-      (set-buffer nnwarchive-buffer)
-      (erase-buffer)
-      (if nnwarchive-open-url
-         (nnwarchive-url nnwarchive-open-url))
-      (if nnwarchive-open-dissect
-         (funcall nnwarchive-open-dissect))
-      (setq nnwarchive-opened t)))
+    (setq nnwarchive-passwd
+         (or nnwarchive-passwd
+             (read-passwd
+              (format "Password for %s at %s: "
+                      nnwarchive-login server)))))
+  (unless nnwarchive-groups
+    (nnwarchive-read-groups))
+  (save-excursion
+    (set-buffer nnwarchive-buffer)
+    (erase-buffer)
+    (if nnwarchive-open-url
+       (nnwarchive-url nnwarchive-open-url))
+    (if nnwarchive-open-dissect
+       (funcall nnwarchive-open-dissect)))
   t)
 
 (nnoo-define-skeleton nnwarchive)
     (nnwarchive-open-server server)))
 
 (defun nnwarchive-read-groups ()
-  (let ((file (expand-file-name (concat "groups-" nnwarchive-address) 
+  (let ((file (expand-file-name (concat "groups-" nnwarchive-address)
                                nnwarchive-directory)))
     (when (file-exists-p file)
       (with-temp-buffer
        (setq nnwarchive-groups (read (current-buffer)))))))
 
 (defun nnwarchive-write-groups ()
-  (with-temp-file (expand-file-name (concat "groups-" nnwarchive-address) 
+  (with-temp-file (expand-file-name (concat "groups-" nnwarchive-address)
                                    nnwarchive-directory)
     (prin1 nnwarchive-groups (current-buffer))))
 
 (defun nnwarchive-init (server)
   "Initialize buffers and such."
   (let ((type (intern server)) (defs nnwarchive-type-definition) def)
-    (cond 
+    (cond
      ((equal server "")
       (setq type nnwarchive-default-type))
      ((assq type nnwarchive-type-definition) t)
             (format " *nnwarchive %s %s*" nnwarchive-type server)))))
   (nnwarchive-set-default nnwarchive-type))
 
-(defun nnwarchive-encode-www-form-urlencoded (pairs)
-  "Return PAIRS encoded for forms."
-  (mapconcat
-   (function
-    (lambda (data)
-      (concat (w3-form-encode-xwfu (car data)) "="
-             (w3-form-encode-xwfu (cdr data)))))
-   pairs "&"))
-
-(defun nnwarchive-fetch-form (url pairs)
-  (let ((url-request-data (nnwarchive-encode-www-form-urlencoded pairs))
-       (url-request-method "POST")
-       (url-request-extra-headers
-        '(("Content-type" . "application/x-www-form-urlencoded"))))
-    (nnweb-insert url))
-  t)
-
 (defun nnwarchive-eval (expr)
   (cond
    ((consp expr)
     expr)))
 
 (defun nnwarchive-url (xurl)
-  (let ((url-confirmation-func 'identity))
-    (cond 
-     ((eq (car xurl) 'post)
-      (pop xurl)
-      (nnwarchive-fetch-form (car xurl) (nnwarchive-eval (cdr xurl))))
-     (t
-      (nnweb-insert (apply 'format (nnwarchive-eval xurl)))))))
-
-(defun nnwarchive-decode-entities ()
-  (goto-char (point-min))
-  (while (re-search-forward "&\\(#[0-9]+\\|[a-z]+\\);" nil t)
-    (replace-match (char-to-string 
-                   (if (eq (aref (match-string 1) 0) ?\#)
-                       (string-to-number (substring (match-string 1) 1))
-                     (or (cdr (assq (intern (match-string 1))
-                                    w3-html-entities))
-                         ?#)))
-                  t t)))
-
-(defun nnwarchive-decode-entities-string (str)
-  (with-temp-buffer
-    (insert str)
-    (nnwarchive-decode-entities)
-    (buffer-substring (point-min) (point-max))))
-
-(defun nnwarchive-remove-markup ()
-  (goto-char (point-min))
-  (while (search-forward "<!--" nil t)
-    (delete-region (match-beginning 0)
-                  (or (search-forward "-->" nil t)
-                      (point-max))))
-  (goto-char (point-min))
-  (while (re-search-forward "<[^>]+>" nil t)
-    (replace-match "" t t)))
-
-(defun nnwarchive-date-to-date (sdate)
-  (let ((elem (split-string sdate)))
-    (concat (substring (nth 0 elem) 0 3) " "
-           (substring (nth 1 elem) 0 3) " "
-           (substring (nth 2 elem) 0 2) " "
-           (substring (nth 3 elem) 1 6) " "
-           (format-time-string "%Y") " "
-           (nth 4 elem))))
+  (mm-with-unibyte-current-buffer
+    (let ((url-confirmation-func 'identity) ;; Some hacks.
+         (url-cookie-multiple-line nil))
+      (cond
+       ((eq (car xurl) 'post)
+       (pop xurl)
+       (mm-url-fetch-form (car xurl) (nnwarchive-eval (cdr xurl))))
+       (t
+       (mm-url-insert (apply 'format (nnwarchive-eval xurl))))))))
 
 (defun nnwarchive-generate-active ()
   (save-excursion
   (save-excursion
     (let (articles)
       (set-buffer nnwarchive-buffer)
-      (dolist (group groups) 
+      (dolist (group groups)
        (erase-buffer)
        (nnwarchive-url nnwarchive-xover-last-url)
        (goto-char (point-min))
-       (when (re-search-forward "of \\([0-9]+\\)</title>" nil t)
-         (setq articles (string-to-number (match-string 1)))) 
+       (when (re-search-forward "of \\([0-9]+\\)[ \t\n\r]*</title>" nil t)
+         (setq articles (string-to-number (match-string 1))))
        (let ((elem (assoc group nnwarchive-groups)))
          (if elem
              (setcar (cdr elem) articles)
   (let ((case-fold-search t)
        group description elem articles)
     (goto-char (point-min))
-    (while 
-       (re-search-forward
-        "/group/\\([^/]+\\)/info\\.html[^>]+>[^>]+>[\040\t]*-[\040\t]*\\([^<]+\\)<"
-        nil t)
+    (while
+       (re-search-forward "href=\"/group/\\([^/\"\> ]+\\)" nil t)
       (setq group (match-string 1)
            description (match-string 2))
-      (forward-line 1)
-      (when (re-search-forward ">\\([0-9]+\\)<" nil t)
-       (setq articles (string-to-number (match-string 1)))) 
       (if (setq elem (assoc group nnwarchive-groups))
-         (setcar (cdr elem) articles)
+         (setcar (cdr elem) 0)
        (push (list group articles description) nnwarchive-groups))))
   t)
 
   (let (article subject from date)
     (goto-char (point-min))
     (while (re-search-forward
-           "<a href=\"/group/\\([^/]+\\)/\\([0-9]+\\)\\.html[^>]+>\\([^<]+\\)<"
+           "<a href=\"/group/\\([^/]+\\)/\\([0-9]+\\)[^>]+>\\([^<]+\\)<"
            nil t)
       (setq group  (match-string 1)
            article (string-to-number (match-string 2))
        (push (cons
               article
               (make-full-mail-header
-               article 
-               (nnwarchive-decode-entities-string subject)
-               (nnwarchive-decode-entities-string from)
+               article
+               (mm-url-decode-entities-string subject)
+               (mm-url-decode-entities-string from)
                date
                (concat "<" group "%"
-                       (number-to-string article) 
+                       (number-to-string article)
                        "@egroup.com>")
                ""
                0 0 "")) nnwarchive-headers))))
       (delete-region (point) (point-max)))
   (goto-char (point-min))
   (while (re-search-forward "<a[^>]+>\\([^<]+\\)</a>" nil t)
-    (replace-match "<\\1>"))
-  (nnwarchive-decode-entities)
-  (buffer-substring (point-min) (point-max)))
+    (replace-match "\\1"))
+  (mm-url-decode-entities)
+  (buffer-string))
 
 (defun nnwarchive-egroups-xover-files (group articles)
   (let (aux auxs)
        (let ((elem (assoc group nnwarchive-headers-cache)))
          (if elem
              (setcdr elem nnwarchive-headers)
-           (push (cons group nnwarchive-headers) 
+           (push (cons group nnwarchive-headers)
                  nnwarchive-headers-cache)))))))
 
 (defun nnwarchive-mail-archive-list ()
            subject (match-string 2))
       (forward-line 1)
       (unless (assq article nnwarchive-headers)
-       (if (looking-at "<UL><LI><EM>From</EM>:\\([^&]+\\)&lt;\\([^&]+\\)&gt;")
+       (if (looking-at "<UL><LI><EM>From</EM>: *\\([^<]*[^< ]\\) *&lt;\\([^&]+\\)&gt;")
            (progn
              (setq from (match-string 1)
                    date (identity (match-string 2))))
        (push (cons
               article
               (make-full-mail-header
-               article 
-               (nnwarchive-decode-entities-string subject)
-               (nnwarchive-decode-entities-string from)
+               article
+               (mm-url-decode-entities-string subject)
+               (mm-url-decode-entities-string from)
                date
                (format "<%05d%%%s>\n" (1- article) group)
                ""
       (nnwarchive-mail-archive-xover group)
       (setq min (apply 'min (mapcar 'car nnwarchive-headers))))))
 
+(defvar nnwarchive-caesar-translation-table nil
+  "Modified rot13 table. tr/@A-Z[a-z/N-Z[@A-Mn-za-m/.")
+
+(defun nnwarchive-make-caesar-translation-table ()
+  "Create modified rot13 table. tr/@A-Z[a-z/N-Z[@A-Mn-za-m/."
+  (let ((i -1)
+       (table (make-string 256 0))
+       (a (mm-char-int ?a))
+       (A (mm-char-int ?A)))
+    (while (< (incf i) 256)
+      (aset table i i))
+    (concat
+     (substring table 0 (1- A))
+     (substring table (+ A 13) (+ A 27))
+     (substring table (1- A) (+ A 13))
+     (substring table (+ A 27) a)
+     (substring table (+ a 13) (+ a 26))
+     (substring table a (+ a 13))
+     (substring table (+ a 26) 255))))
+
+(defun nnwarchive-from-r13 (from-r13)
+  (when from-r13
+    (with-temp-buffer
+      (insert from-r13)
+      (let ((message-caesar-translation-table
+            (or nnwarchive-caesar-translation-table
+                (setq nnwarchive-caesar-translation-table
+                      (nnwarchive-make-caesar-translation-table)))))
+       (message-caesar-region (point-min) (point-max))
+       (buffer-string)))))
+
 (defun nnwarchive-mail-archive-article (group article)
-  (let (p refs url mime file)
+  (let (p refs url mime e
+         from subject date id
+         done
+         (case-fold-search t))
     (save-restriction
       (goto-char (point-min))
+      (when (search-forward "X-Head-End" nil t)
+       (beginning-of-line)
+       (narrow-to-region (point-min) (point))
+       (mm-url-decode-entities)
+       (goto-char (point-min))
+       (while (search-forward "<!--X-" nil t)
+         (replace-match ""))
+       (goto-char (point-min))
+       (while (search-forward " -->" nil t)
+         (replace-match ""))
+       (setq from
+             (or (mail-fetch-field "from")
+                 (nnwarchive-from-r13
+                  (mail-fetch-field "from-r13"))))
+       (setq date (mail-fetch-field "date"))
+       (setq id (mail-fetch-field "message-id"))
+       (setq subject (mail-fetch-field "subject"))
+       (goto-char (point-max))
+       (widen))
       (when (search-forward "<ul>" nil t)
        (forward-line)
        (delete-region (point-min) (point))
        (search-forward "</ul>" nil t)
-       (forward-line)
+       (end-of-line)
        (narrow-to-region (point-min) (point))
-       (nnwarchive-remove-markup)
-       (nnwarchive-decode-entities)
+       (mm-url-remove-markup)
+       (mm-url-decode-entities)
        (goto-char (point-min))
        (delete-blank-lines)
+       (when from
+         (message-remove-header "from")
+         (goto-char (point-max))
+         (insert "From: " from "\n"))
+       (when subject
+         (message-remove-header "subject")
+         (goto-char (point-max))
+         (insert "Subject: " subject "\n"))
+       (when id
+         (goto-char (point-max))
+         (insert "X-Message-ID: <" id ">\n"))
+       (when date
+         (message-remove-header "date")
+         (goto-char (point-max))
+         (insert "Date: " date "\n"))
        (goto-char (point-max))
        (widen)
        (insert "\n"))
-      (setq p (point)) 
+      (setq p (point))
       (when (search-forward "X-Body-of-Message" nil t)
-         (forward-line)
-         (delete-region p (point))
-         (search-forward "X-Body-of-Message-End" nil t)
-         (beginning-of-line)
-         (save-restriction
-           (narrow-to-region p (point))
-           (goto-char (point-min))
-           (if (looking-at "<PRE>") 
-               (progn
-                 (delete-char 5)
-                 (setq p (point))
-                 (when (search-forward "</PRE>" nil t)
-                   (goto-char (match-beginning 0))
-                   (delete-char 6)
-                   (save-restriction
-                     (narrow-to-region p (point))
-                     (nnwarchive-remove-markup)
-                     (nnwarchive-decode-entities)
-                     (goto-char (point-max))))
-                 (while (looking-at
-                         "[\040\n\r\t]*<P><A HREF=\"\\([^\"]+\\)[^>]*><[^>]*>\\([^<]+\\)")
-                   (setq url (match-string 1)
-                         file (match-string 2))
-                   (goto-char (match-beginning 1))
-                   (beginning-of-line)
-                   (setq p (point))
-                   (delete-region p (progn (forward-line) (point)))
-                   (insert (format "http://www.mail-archive.com/%s/%s\n"
-                                   group url))))
+       (forward-line)
+       (delete-region p (point))
+       (search-forward "X-Body-of-Message-End" nil t)
+       (beginning-of-line)
+       (save-restriction
+         (narrow-to-region p (point))
+         (goto-char (point-min))
+         (if (> (skip-chars-forward "\040\n\r\t") 0)
+             (delete-region (point-min) (point)))
+         (while (not (eobp))
+           (cond
+            ((looking-at "<PRE>\r?\n?")
+             (delete-region (match-beginning 0) (match-end 0))
+             (setq p (point))
+             (when (search-forward "</PRE>" nil t)
+               (delete-region (match-beginning 0) (match-end 0))
+               (save-restriction
+                 (narrow-to-region p (point))
+                 (mm-url-remove-markup)
+                 (mm-url-decode-entities)
+                 (goto-char (point-max)))))
+            ((looking-at "<P><A HREF=\"\\([^\"]+\\)")
+             (setq url (match-string 1))
+             (delete-region (match-beginning 0)
+                            (progn (forward-line) (point)))
+             ;; I hate to download the url encode it, then immediately
+             ;; decode it.
+             (insert "<#external"
+                     " type="
+                     (or (and url
+                              (string-match "\\.[^\\.]+$" url)
+                              (mailcap-extension-to-mime
+                               (match-string 0 url)))
+                         "application/octet-stream")
+                     (format " url=\"http://www.mail-archive.com/%s/%s\""
+                             group url)
+                     ">\n"
+                     "<#/external>")
              (setq mime t))
-           (goto-char (point-max))))
+            (t
+             (setq p (point))
+             (insert "<#part type=\"text/html\" disposition=inline>")
+             (goto-char
+              (if (re-search-forward
+                   "[\040\n\r\t]*<PRE>\\|[\040\n\r\t]*<P><A HREF=\""
+                   nil t)
+                  (match-beginning 0)
+                (point-max)))
+             (insert "<#/part>")
+             (setq mime t)))
+           (setq p (point))
+           (if (> (skip-chars-forward "\040\n\r\t") 0)
+               (delete-region p (point))))
+         (goto-char (point-max))))
       (setq p (point))
       (when (search-forward "X-References-End" nil t)
-       (narrow-to-region p (point))
-       (goto-char (point-min))
-       (while (re-search-forward "msg\\([0-9]+\\)\\.html" nil t)
-         (push (concat "<" (match-string 1) "%" group ">") refs))
-       (widen))
+       (setq e (point))
+       (beginning-of-line)
+       (search-backward "X-References" p t)
+       (while (re-search-forward "msg\\([0-9]+\\)\\.html" e t)
+         (push (concat "<" (match-string 1) "%" group ">") refs)))
       (delete-region p (point-max))
       (goto-char (point-min))
       (insert (format "Message-ID: <%05d%%%s>\n" (1- article) group))
          (insert " " (pop refs)))
        (insert "\n"))
       (when mime
+       (unless (looking-at "$")
+         (search-forward "\n\n" nil t)
+         (forward-line -1))
+       (narrow-to-region (point) (point-max))
        (insert "MIME-Version: 1.0\n"
-               "Content-Type: text/html\n")))
-    (buffer-string (point-min) (point-max))))
+               (prog1
+                   (mml-generate-mime)
+                 (delete-region (point-min) (point-max))))
+       (widen)))
+    (buffer-string)))
 
 (provide 'nnwarchive)
 
+;;; arch-tag: 1ab7a15c-777a-40e0-95c0-0c41b3963578
 ;;; nnwarchive.el ends here