;; A lot of codes stolen from mail-source, nnslashdot, nnweb.
-;; Todo: To support more web archives.
+;; 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"). w3 0.45 should
-;; work.
+;; If it is loaded from w3.el, (load-library "url"). Update to w3 0.46
+;; or greater version.
;;; Code:
(nnoo-declare nnwarchive)
-(eval-and-compile
- (defvar nnwarchive-type-definition
- '((egroups
- (open-url
- "http://www.egroups.com/register?method=loginAction&email=%s&password=%s"
- login passwd)
- (list-url
- "http://www.egroups.com/UserGroupsPage?")
- (list-dissect . nnwarchive-egroups-list)
- (list-groups . nnwarchive-egroups-list-groups)
- (xover-url
- "http://www.egroups.com/group/%s/?fetchForward=1&start=%d" group start)
- (xover-last-url
- "http://www.egroups.com/group/%s/?fetchForward=1" 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-dissect . nnwarchive-egroups-article)))))
-
-(eval-and-compile
- (defvar nnwarchive-short-names
- '(login passwd)))
+(defvar nnwarchive-type-definition
+ '((egroups
+ (address . "www.egroups.com")
+ (open-url
+ "http://www.egroups.com/register?method=loginAction&email=%s&password=%s"
+ nnwarchive-login nnwarchive-passwd)
+ (list-url
+ "http://www.egroups.com/UserGroupsPage?")
+ (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-page-size . 13)
+ (xover-dissect . nnwarchive-egroups-xover)
+ (article-url
+ "http://www.egroups.com/group/%s/%d.html?raw=1" group article)
+ (article-dissect . nnwarchive-egroups-article)
+ (authentication . t)
+ (xover-files . nnwarchive-egroups-xover-files))
+ (mail-archive
+ (address . "www.mail-archive.com")
+ (list-url
+ "http://www.mail-archive.com/lists.html")
+ (list-dissect . nnwarchive-mail-archive-list)
+ (list-groups . nnwarchive-mail-archive-list-groups)
+ (xover-url
+ "http://www.mail-archive.com/%s/mail%d.html" group aux)
+ (xover-last-url
+ "http://www.mail-archive.com/%s/maillist.html" group)
+ (xover-dissect . nnwarchive-mail-archive-xover)
+ (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)
+ (article-offset . 1))))
+
+(defvar nnwarchive-default-type 'egroups)
(defvoo nnwarchive-directory (nnheader-concat gnus-directory "warchive/")
"Where nnwarchive will save its files.")
(eval-and-compile
- (defvoo nnwarchive-type 'egroups
+ (defvoo nnwarchive-type nil
"The type of nnwarchive."))
-(defvoo nnwarchive-address "egroups.com"
+(defvoo nnwarchive-address ""
"The address of nnwarchive.")
(defvoo nnwarchive-login nil
(defvoo nnwarchive-opened nil)
+(defvoo nnwarchive-authentication nil)
+
+(defvoo nnwarchive-xover-is-evil nil) ;; not implemented
+
(defconst nnwarchive-version "nnwarchive 1.0")
;;; Internal variables
-(defvar nnwarchive-open-url nil)
-(defvar nnwarchive-open-dissect nil)
+(defvoo nnwarchive-open-url nil)
+(defvoo nnwarchive-open-dissect nil)
-(defvar nnwarchive-list-url nil)
-(defvar nnwarchive-list-dissect nil)
-(defvar nnwarchive-list-groups nil)
+(defvoo nnwarchive-list-url nil)
+(defvoo nnwarchive-list-dissect nil)
+(defvoo nnwarchive-list-groups nil)
-(defvar nnwarchive-xover-url nil)
-(defvar nnwarchive-xover-last-url nil)
-(defvar nnwarchive-xover-dissect nil)
-(defvar nnwarchive-xover-page-size nil)
+(defvoo nnwarchive-xover-files nil)
+(defvoo nnwarchive-xover-url nil)
+(defvoo nnwarchive-xover-last-url nil)
+(defvoo nnwarchive-xover-dissect nil)
+(defvoo nnwarchive-xover-page-size nil)
-(defvar nnwarchive-article-url nil)
-(defvar nnwarchive-article-dissect nil)
+(defvoo nnwarchive-article-url nil)
+(defvoo nnwarchive-article-dissect nil)
+(defvoo nnwarchive-xover-files nil)
+(defvoo nnwarchive-article-offset 0)
-(defvar nnwarchive-buffer nil)
+(defvoo nnwarchive-buffer nil)
(defvar nnwarchive-headers nil)
+
;;; Interface functions
(nnoo-define-basics nnwarchive)
-(eval-and-compile
- (defun nnwarchive-bind-1 ()
- (let ((defaults (cdr (assq nnwarchive-type nnwarchive-type-definition)))
- (short-names nnwarchive-short-names)
- default bind)
- (while (setq default (pop defaults))
- (push (list (intern (concat "nnwarchive-" (symbol-name (car default))))
- (list 'quote (cdr default))) bind))
- (while (setq default (pop short-names))
- (push (list default
- (intern (concat "nnwarchive-"
- (symbol-name default))))
- bind))
- bind)))
-
-(defmacro nnwarchive-bind (&rest body)
- "Return a `let' form that binds all variables in TYPE.
-Read `mail-source-bind' for details."
- `(let ,(nnwarchive-bind-1)
- ,@body))
-
-(put 'nnwarchive-bind 'lisp-indent-function 0)
-(put 'nnwarchive-bind 'edebug-form-spec '(form body))
+(defun nnwarchive-set-default (type)
+ (let ((defs (cdr (assq type nnwarchive-type-definition)))
+ def)
+ (dolist (def defs)
+ (set (intern (concat "nnwarchive-" (symbol-name (car def))))
+ (cdr def)))))
(deffoo nnwarchive-retrieve-headers (articles &optional group server fetch-old)
(nnwarchive-possibly-change-server group server)
- (nnwarchive-bind
- (setq nnwarchive-headers (cdr (assoc group nnwarchive-headers-cache)))
- (save-excursion
- (set-buffer nnwarchive-buffer)
- (erase-buffer)
- (let (point start starts)
- (setq starts (nnwarchive-paged (sort articles '<)))
- (while (setq start (pop starts))
- (goto-char (point-max))
- (nnwarchive-url nnwarchive-xover-url))
- (if nnwarchive-xover-dissect
- (funcall nnwarchive-xover-dissect))))
- (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))
+ (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)
- (nnwarchive-bind
- (if nnwarchive-list-groups
- (funcall nnwarchive-list-groups groups))
- (nnwarchive-write-groups)
- (nnwarchive-generate-active)
- 'active))
+ (if nnwarchive-list-groups
+ (funcall nnwarchive-list-groups groups))
+ (nnwarchive-write-groups)
+ (nnwarchive-generate-active)
+ 'active)
(deffoo nnwarchive-request-group (group &optional server dont-check)
(nnwarchive-possibly-change-server nil server)
- (nnwarchive-bind
- (if nnwarchive-list-groups
- (funcall nnwarchive-list-groups (list group)))
- (nnwarchive-write-groups)
- (let ((elem (assoc group nnwarchive-groups)))
- (cond
- ((not elem)
- (nnheader-report 'nnwarchive "Group does not exist"))
- (t
- (nnheader-report 'nnwarchive "Opened group %s" group)
- (nnheader-insert
- "211 %d %d %d %s\n" (or (cadr elem) 0) 1 (or (cadr elem) 0)
- (prin1-to-string group))
- t)))))
+ (if nnwarchive-list-groups
+ (funcall nnwarchive-list-groups (list group)))
+ (nnwarchive-write-groups)
+ (let ((elem (assoc group nnwarchive-groups)))
+ (cond
+ ((not elem)
+ (nnheader-report 'nnwarchive "Group does not exist"))
+ (t
+ (nnheader-report 'nnwarchive "Opened group %s" group)
+ (nnheader-insert
+ "211 %d %d %d %s\n" (or (cadr elem) 0) 1 (or (cadr elem) 0)
+ (prin1-to-string group))
+ t))))
(deffoo nnwarchive-close-group (group &optional server)
(nnwarchive-possibly-change-server group server)
- (nnwarchive-bind
- (when (gnus-buffer-live-p nnwarchive-buffer)
- (save-excursion
- (set-buffer nnwarchive-buffer)
- (kill-buffer nnwarchive-buffer)))
- t))
+ (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)
- (nnwarchive-bind
- (let (contents)
- (save-excursion
- (set-buffer nnwarchive-buffer)
- (goto-char (point-min))
- (nnwarchive-url nnwarchive-article-url)
- (setq contents (funcall nnwarchive-article-dissect)))
- (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))))))
+ (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)))))
(deffoo nnwarchive-close-server (&optional server)
(when (and (nnwarchive-server-opened server)
(deffoo nnwarchive-request-list (&optional server)
(nnwarchive-possibly-change-server nil server)
- (nnwarchive-bind
- (save-excursion
- (set-buffer nnwarchive-buffer)
- (erase-buffer)
- (if nnwarchive-list-url
- (nnwarchive-url nnwarchive-list-url))
- (if nnwarchive-list-dissect
- (funcall nnwarchive-list-dissect))
- (nnwarchive-write-groups)
- (nnwarchive-generate-active)))
+ (save-excursion
+ (set-buffer nnwarchive-buffer)
+ (erase-buffer)
+ (if nnwarchive-list-url
+ (nnwarchive-url nnwarchive-list-url))
+ (if nnwarchive-list-dissect
+ (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-bind
- (nnwarchive-write-groups)
- (nnwarchive-generate-active))
+ (nnwarchive-write-groups)
+ (nnwarchive-generate-active)
'active)
(deffoo nnwarchive-asynchronous-p ()
(deffoo nnwarchive-open-server (server &optional defs connectionless)
(nnwarchive-init server)
- (if (nnwarchive-server-opened server)
- t
- (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))))
- (nnwarchive-bind
- (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)))
- t))
+ (unless (nnwarchive-server-opened server)
+ (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)))
+ t)
(nnoo-define-skeleton nnwarchive)
(defun nnwarchive-init (server)
"Initialize buffers and such."
+ (let ((type (intern server)) (defs nnwarchive-type-definition) def)
+ (cond
+ ((equal server "")
+ (setq type nnwarchive-default-type))
+ ((assq type nnwarchive-type-definition) t)
+ (t
+ (setq type nil)
+ (while (setq def (pop defs))
+ (when (equal (cdr (assq 'address (cdr def))) server)
+ (setq defs nil)
+ (setq type (car def))))
+ (unless type
+ (error "Undefined server %s" server))))
+ (setq nnwarchive-type type))
(unless (file-exists-p nnwarchive-directory)
(gnus-make-directory nnwarchive-directory))
(unless (gnus-buffer-live-p nnwarchive-buffer)
(setq nnwarchive-buffer
(save-excursion
(nnheader-set-temp-buffer
- (format " *nnwarchive %s %s*" nnwarchive-type server))))))
+ (format " *nnwarchive %s %s*" nnwarchive-type server)))))
+ (nnwarchive-set-default nnwarchive-type))
(defun nnwarchive-encode-www-form-urlencoded (pairs)
"Return PAIRS encoded for forms."
(setcar (cdr elem) articles)
(push (list group articles "") nnwarchive-groups)))
(setq nnwarchive-headers (cdr (assoc group nnwarchive-headers-cache)))
- (nnwarchive-egroups-xover)
+ (nnwarchive-egroups-xover group)
(let ((elem (assoc group nnwarchive-headers-cache)))
(if elem
(setcdr elem nnwarchive-headers)
(nnwarchive-egroups-list-groups (mapcar 'identity nnwarchive-groups)))
t)
-(defun nnwarchive-egroups-xover()
- (let (article subject from date group)
+(defun nnwarchive-egroups-xover (group)
+ (let (article subject from date)
(goto-char (point-min))
(while (re-search-forward
"<a href=\"/group/\\([^/]+\\)/\\([0-9]+\\)\\.html[^>]+>\\([^<]+\\)<"
0 0 "")) nnwarchive-headers))))
nnwarchive-headers)
-(defun nnwarchive-egroups-article ()
+(defun nnwarchive-egroups-article (group articles)
(goto-char (point-min))
(if (search-forward "<pre>" nil t)
(delete-region (point-min) (point)))
(nnwarchive-decode-entities)
(buffer-substring (point-min) (point-max)))
+(defun nnwarchive-egroups-xover-files (group articles)
+ (let (aux auxs)
+ (setq auxs (nnwarchive-paged (sort articles '<)))
+ (while (setq aux (pop auxs))
+ (goto-char (point-max))
+ (nnwarchive-url nnwarchive-xover-url))
+ (if nnwarchive-xover-dissect
+ (nnwarchive-egroups-xover group))))
+
+;; mail-archive
+
+(defun nnwarchive-mail-archive-list-groups (groups)
+ (save-excursion
+ (let (articles)
+ (set-buffer nnwarchive-buffer)
+ (dolist (group groups)
+ (erase-buffer)
+ (nnwarchive-url nnwarchive-xover-last-url)
+ (goto-char (point-min))
+ (when (re-search-forward "msg\\([0-9]+\\)\\.html" nil t)
+ (setq articles (1+ (string-to-number (match-string 1)))))
+ (let ((elem (assoc group nnwarchive-groups)))
+ (if elem
+ (setcar (cdr elem) articles)
+ (push (list group articles "") nnwarchive-groups)))
+ (setq nnwarchive-headers (cdr (assoc group nnwarchive-headers-cache)))
+ (nnwarchive-mail-archive-xover group)
+ (let ((elem (assoc group nnwarchive-headers-cache)))
+ (if elem
+ (setcdr elem nnwarchive-headers)
+ (push (cons group nnwarchive-headers)
+ nnwarchive-headers-cache)))))))
+
+(defun nnwarchive-mail-archive-list ()
+ (let ((case-fold-search t)
+ group description elem articles)
+ (goto-char (point-min))
+ (while (re-search-forward "<a href=\"\\([^/]+\\)/\">\\([^>]+\\)<" nil t)
+ (setq group (match-string 1)
+ description (match-string 2))
+ (forward-line 1)
+ (setq articles 0)
+ (if (setq elem (assoc group nnwarchive-groups))
+ (setcar (cdr elem) articles)
+ (push (list group articles description) nnwarchive-groups))))
+ t)
+
+(defun nnwarchive-mail-archive-xover (group)
+ (let (article subject from date)
+ (goto-char (point-min))
+ (while (re-search-forward
+ "<A[^>]*HREF=\"msg\\([0-9]+\\)\\.html[^>]+>\\([^<]+\\)<"
+ nil t)
+ (setq article (1+ (string-to-number (match-string 1)))
+ subject (match-string 2))
+ (forward-line 1)
+ (unless (assq article nnwarchive-headers)
+ (if (looking-at "<UL><LI><EM>From</EM>:\\([^&]+\\)<\\([^&]+\\)>")
+ (progn
+ (setq from (match-string 1)
+ date (identity (match-string 2))))
+ (setq from "" date ""))
+ (push (cons
+ article
+ (make-full-mail-header
+ article
+ (nnwarchive-decode-entities-string subject)
+ (nnwarchive-decode-entities-string from)
+ date
+ (format "<%05d%%%s>\n" (1- article) group)
+ ""
+ 0 0 "")) nnwarchive-headers))))
+ nnwarchive-headers)
+
+(defun nnwarchive-mail-archive-xover-files (group articles)
+ (unless nnwarchive-headers
+ (erase-buffer)
+ (nnwarchive-url nnwarchive-xover-last-url)
+ (goto-char (point-min))
+ (nnwarchive-mail-archive-xover group))
+ (let ((minart (apply 'min articles))
+ (min (apply 'min (mapcar 'car nnwarchive-headers)))
+ (aux 2))
+ (while (> min minart)
+ (erase-buffer)
+ (nnwarchive-url nnwarchive-xover-url)
+ (nnwarchive-mail-archive-xover group)
+ (setq min (apply 'min (mapcar 'car nnwarchive-headers))))))
+
+(defun nnwarchive-mail-archive-article (group article)
+ (let (p refs url mime file)
+ (save-restriction
+ (goto-char (point-min))
+ (when (search-forward "<ul>" nil t)
+ (forward-line)
+ (delete-region (point-min) (point))
+ (search-forward "</ul>" nil t)
+ (forward-line)
+ (narrow-to-region (point-min) (point))
+ (nnwarchive-remove-markup)
+ (nnwarchive-decode-entities)
+ (goto-char (point-min))
+ (delete-blank-lines)
+ (goto-char (point-max))
+ (widen)
+ (insert "\n"))
+ (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))))
+ (setq mime t))
+ (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))
+ (delete-region p (point-max))
+ (goto-char (point-min))
+ (insert (format "Message-ID: <%05d%%%s>\n" (1- article) group))
+ (when refs
+ (insert "References:")
+ (while refs
+ (insert " " (pop refs)))
+ (insert "\n"))
+ (when mime
+ (insert "MIME-Version: 1.0\n"
+ "Content-Type: text/html\n")))
+ (buffer-string (point-min) (point-max))))
+
(provide 'nnwarchive)
;;; nnwarchive.el ends here