1 ;;; nnrss.el --- interfacing with RSS
3 ;; Copyright (C) 2001, 2002, 2003, 2004, 2005 Free Software Foundation, Inc.
5 ;; Author: Shenghuo Zhu <zsh@cs.rochester.edu>
8 ;; This file is part of GNU Emacs.
10 ;; GNU Emacs is free software; you can redistribute it and/or modify
11 ;; it under the terms of the GNU General Public License as published
12 ;; by the Free Software Foundation; either version 2, or (at your
13 ;; option) any later version.
15 ;; GNU Emacs is distributed in the hope that it will be useful, but
16 ;; WITHOUT ANY WARRANTY; without even the implied warranty of
17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
18 ;; General Public License for more details.
20 ;; You should have received a copy of the GNU General Public License
21 ;; along with GNU Emacs; see the file COPYING. If not, write to the
22 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
23 ;; Boston, MA 02110-1301, USA.
29 (eval-when-compile (require 'cl))
45 (eval '(require 'xml))
49 (defvoo nnrss-directory (nnheader-concat gnus-directory "rss/")
50 "Where nnrss will save its files.")
52 ;; (group max rss-url)
53 (defvoo nnrss-server-data nil)
55 ;; (num timestamp url subject author date extra)
56 (defvoo nnrss-group-data nil)
57 (defvoo nnrss-group-max 0)
58 (defvoo nnrss-group-min 1)
59 (defvoo nnrss-group nil)
60 (defvoo nnrss-group-hashtb (make-hash-table :test 'equal))
61 (defvoo nnrss-status-string "")
63 (defconst nnrss-version "nnrss 1.0")
65 (defvar nnrss-group-alist '()
66 "List of RSS addresses.")
68 (defvar nnrss-use-local nil)
70 (defvar nnrss-description-field 'X-Gnus-Description
71 "Field name used for DESCRIPTION.
72 To use the description in headers, put this name into `nnmail-extra-headers'.")
74 (defvar nnrss-url-field 'X-Gnus-Url
75 "Field name used for URL.
76 To use the description in headers, put this name into `nnmail-extra-headers'.")
78 (defvar nnrss-content-function nil
79 "A function which is called in `nnrss-request-article'.
80 The arguments are (ENTRY GROUP ARTICLE).
81 ENTRY is the record of the current headline. GROUP is the group name.
82 ARTICLE is the article number of the current headline.")
84 (defvar nnrss-file-coding-system mm-universal-coding-system
85 "Coding system used when reading and writing files.")
87 (defvar nnrss-compatible-encoding-alist '((iso-8859-1 . windows-1252))
88 "Alist of encodings and those supersets.
89 The cdr of each element is used to decode data if it is available when
90 the car is what the data specify as the encoding. Or, the car is used
91 for decoding when the cdr that the data specify is not available.")
93 (defvar nnrss-wash-html-in-text-plain-parts nil
94 "*Non-nil means render text in text/plain parts as HTML.
95 The function specified by the `mm-text-html-renderer' variable will be
96 used to render text. If it is nil, text will simply be folded.")
98 (nnoo-define-basics nnrss)
100 ;;; Interface functions
102 (defsubst nnrss-format-string (string)
103 (gnus-replace-in-string string " *\n *" " "))
105 (defun nnrss-decode-group-name (group)
106 (if (and group (mm-coding-system-p 'utf-8))
107 (setq group (mm-decode-coding-string group 'utf-8))
110 (deffoo nnrss-retrieve-headers (articles &optional group server fetch-old)
111 (setq group (nnrss-decode-group-name group))
112 (nnrss-possibly-change-group group server)
115 (set-buffer nntp-server-buffer)
117 (dolist (article articles)
118 (if (setq e (assq article nnrss-group-data))
119 (insert (number-to-string (car e)) "\t" ;; number
124 (or (nth 4 e) "(nobody)")
130 (format "<%d@%s.nnrss>" (car e) group)
141 (memq nnrss-description-field
142 nnmail-extra-headers))
143 (concat (symbol-name nnrss-description-field)
145 (nnrss-format-string (nth 6 e))
149 (memq nnrss-url-field
150 nnmail-extra-headers))
151 (concat (symbol-name nnrss-url-field)
153 (nnrss-format-string (nth 2 e))
159 (deffoo nnrss-request-group (group &optional server dont-check)
160 (setq group (nnrss-decode-group-name group))
161 (nnheader-message 6 "nnrss: Requesting %s..." group)
162 (nnrss-possibly-change-group group server)
166 (nnrss-check-group group server)
167 (nnheader-report 'nnrss "Opened group %s" group)
169 "211 %d %d %d %s\n" nnrss-group-max nnrss-group-min nnrss-group-max
170 (prin1-to-string group)
172 (nnheader-message 6 "nnrss: Requesting %s...done" group)))
174 (deffoo nnrss-close-group (group &optional server)
178 (defvar mm-text-html-renderer)
179 (defvar mm-text-html-washer-alist))
181 (deffoo nnrss-request-article (article &optional group server buffer)
182 (setq group (nnrss-decode-group-name group))
183 (when (stringp article)
184 (setq article (if (string-match "\\`<\\([0-9]+\\)@" article)
185 (string-to-number (match-string 1 article))
187 (nnrss-possibly-change-group group server)
188 (let ((e (assq article nnrss-group-data))
189 (nntp-server-buffer (or buffer nntp-server-buffer))
192 (with-current-buffer nntp-server-buffer
195 (insert "Newsgroups: " group "\n"))
197 (insert "Subject: " (nth 3 e) "\n"))
199 (insert "From: " (nth 4 e) "\n"))
201 (insert "Date: " (nnrss-format-string (nth 5 e)) "\n"))
202 (let ((header (buffer-string))
205 (enclosure (nth 7 e))
207 ;; Enable encoding of Newsgroups header in XEmacs.
208 (default-enable-multibyte-characters t)
209 (rfc2047-header-encoding-alist
210 (if (mm-coding-system-p 'utf-8)
211 (cons '("Newsgroups" . utf-8)
212 rfc2047-header-encoding-alist)
213 rfc2047-header-encoding-alist))
214 rfc2047-encode-encoded-words body fn)
215 (when (or text link enclosure comments)
217 (insert "<#multipart type=alternative>\n"
218 "<#part type=\"text/plain\">\n")
223 (if (and nnrss-wash-html-in-text-plain-parts
226 (setq fn (or (cdr (assq mm-text-html-renderer
227 mm-text-html-washer-alist))
228 mm-text-html-renderer))))
230 (narrow-to-region body (point-max))
233 (apply (car fn) (cdr fn)))
236 (re-search-forward "[^\t\n ]" nil t)
238 (delete-region body (point))
239 (goto-char (point-max))
240 (skip-chars-backward "\t\n ")
242 (delete-region (point) (point-max))
244 (while (re-search-forward "\n+" nil t)
247 ;; See `nnrss-check-group', which inserts "<br /><br />".
248 (when (search-forward "<br /><br />" nil t)
251 (replace-match "\n\n")))
253 (let ((fill-column default-fill-column)
254 (window (get-buffer-window nntp-server-buffer)))
257 (max 1 (/ (* (window-width window) 7) 8))))
258 (fill-region (point) (point-max))
259 (goto-char (point-max))
260 ;; XEmacs version of `fill-region' inserts newline.
263 (when (or link enclosure)
268 (insert (car enclosure) " "
269 (nth 2 enclosure) " "
270 (nth 3 enclosure) "\n"))
272 (insert comments "\n"))
273 (setq body (buffer-substring body (point)))
275 "<#part type=\"text/html\">\n"
276 "<html><head></head><body>\n")
280 (insert "<p><a href=\"" link "\">link</a></p>\n"))
282 (insert "<p><a href=\"" (car enclosure) "\">"
283 (cadr enclosure) "</a> " (nth 2 enclosure)
284 " " (nth 3 enclosure) "</p>\n"))
286 (insert "<p><a href=\"" comments "\">comments</a></p>\n"))
287 (insert "</body></html>\n"
295 "Content-Type: text/plain; charset=gnus-decoded\n"
296 "Content-Transfer-Encoding: 8bit\n\n"
299 3 "Warning - there might be invalid characters"))))
300 (goto-char (point-min))
301 (search-forward "\n\n")
303 (insert (format "Message-ID: <%d@%s.nnrss>\n"
305 (let ((rfc2047-encoding-type 'mime)
306 rfc2047-encode-max-chars)
307 (rfc2047-encode-string
308 (gnus-replace-in-string group "[\t\n ]+" "_")))))
309 (when nnrss-content-function
310 (funcall nnrss-content-function e group article))))
313 (nnheader-report 'nnrss err))
315 (nnheader-report 'nnrss "no such id: %d" article))
317 (nnheader-report 'nnrss "article %s retrieved" (car e))
318 ;; we return the article number.
319 (cons nnrss-group (car e))))))
321 (deffoo nnrss-request-list (&optional server)
322 (nnrss-possibly-change-group nil server)
323 (nnrss-generate-active)
326 (deffoo nnrss-open-server (server &optional defs connectionless)
327 (nnrss-read-server-data server)
328 (nnoo-change-server 'nnrss server defs)
331 (deffoo nnrss-request-expire-articles
332 (articles group &optional server force)
333 (setq group (nnrss-decode-group-name group))
334 (nnrss-possibly-change-group group server)
335 (let (e days not-expirable changed)
336 (dolist (art articles)
337 (if (and (setq e (assq art nnrss-group-data))
338 (nnmail-expired-article-p
340 (if (listp (setq days (nth 1 e))) days
341 (days-to-time (- days (time-to-days '(0 0)))))
343 (setq nnrss-group-data (delq e nnrss-group-data)
345 (push art not-expirable)))
347 (nnrss-save-group-data group server))
350 (deffoo nnrss-request-delete-group (group &optional force server)
351 (setq group (nnrss-decode-group-name group))
352 (nnrss-possibly-change-group group server)
354 ;; There may be two or more entries in `nnrss-group-alist' since
355 ;; this function didn't delete them formerly.
356 (while (setq elem (assoc group nnrss-group-alist))
357 (setq nnrss-group-alist (delq elem nnrss-group-alist))))
358 (setq nnrss-server-data
359 (delq (assoc group nnrss-server-data) nnrss-server-data))
360 (nnrss-save-server-data server)
362 (delete-file (nnrss-make-filename group server)))
365 (deffoo nnrss-request-list-newsgroups (&optional server)
366 (nnrss-possibly-change-group nil server)
368 (set-buffer nntp-server-buffer)
370 (dolist (elem nnrss-group-alist)
372 (insert (car elem) "\t" (third elem) "\n"))))
375 (nnoo-define-skeleton nnrss)
377 ;;; Internal functions
378 (eval-when-compile (defun xml-rpc-method-call (&rest args)))
380 (defun nnrss-get-encoding ()
381 "Return an encoding attribute specified in the current xml contents.
382 If `nnrss-compatible-encoding-alist' specifies the compatible encoding,
383 it is used instead. If the xml contents doesn't specify the encoding,
384 return `utf-8' which is the default encoding for xml if it is available,
385 otherwise return nil."
386 (goto-char (point-min))
387 (if (re-search-forward
388 "<\\?[^>]*encoding=\\(?:\"\\([^\">]+\\)\"\\|'\\([^'>]+\\)'\\)"
390 (let ((encoding (intern (downcase (or (match-string 1)
391 (match-string 2))))))
393 (mm-coding-system-p (cdr (assq encoding
394 nnrss-compatible-encoding-alist)))
395 (mm-coding-system-p encoding)
396 (mm-coding-system-p (car (rassq encoding
397 nnrss-compatible-encoding-alist)))))
398 (mm-coding-system-p 'utf-8)))
400 (defun nnrss-fetch (url &optional local)
401 "Fetch URL and put it in a the expected Lisp structure."
402 (mm-with-unibyte-buffer
403 ;;some CVS versions of url.el need this to close the connection quickly
404 (let (cs xmlform htmlform)
405 ;; bit o' work necessary for w3 pre-cvs and post-cvs
407 (let ((coding-system-for-read 'binary))
408 (insert-file-contents url))
409 ;; FIXME: shouldn't binding `coding-system-for-read' be moved
410 ;; to `mm-url-insert'?
411 (let ((coding-system-for-read 'binary))
414 (error (if (or debug-on-quit debug-on-error)
415 (signal (car err) (cdr err))
416 (message "nnrss: Failed to fetch %s" url))))))
417 (nnheader-remove-cr-followed-by-lf)
418 ;; Decode text according to the encoding attribute.
419 (when (setq cs (nnrss-get-encoding))
420 (mm-decode-coding-region (point-min) (point-max) cs)
421 (mm-enable-multibyte))
422 (goto-char (point-min))
424 ;; Because xml-parse-region can't deal with anything that isn't
425 ;; xml and w3-parse-buffer can't deal with some xml, we have to
426 ;; parse with xml-parse-region first and, if that fails, parse
427 ;; with w3-parse-buffer. Yuck. Eventually, someone should find out
428 ;; why w3-parse-buffer fails to parse some well-formed xml and
432 (setq xmlform (xml-parse-region (point-min) (point-max)))
435 (setq htmlform (caddar (w3-parse-buffer
439 nnrss: %s: Not valid XML %s and w3-parse doesn't work %s"
445 (defun nnrss-possibly-change-group (&optional group server)
447 (not (nnrss-server-opened server)))
448 (nnrss-open-server server))
449 (when (and group (not (equal group nnrss-group)))
450 (nnrss-read-group-data group server)
451 (setq nnrss-group group)))
453 (defvar nnrss-extra-categories '(nnrss-snarf-moreover-categories))
455 (defun nnrss-generate-active ()
456 (when (y-or-n-p "Fetch extra categories? ")
457 (mapc 'funcall nnrss-extra-categories))
459 (set-buffer nntp-server-buffer)
461 (dolist (elem nnrss-group-alist)
462 (insert (prin1-to-string (car elem)) " 0 1 y\n"))
463 (dolist (elem nnrss-server-data)
464 (unless (assoc (car elem) nnrss-group-alist)
465 (insert (prin1-to-string (car elem)) " 0 1 y\n")))))
467 (eval-and-compile (autoload 'timezone-parse-date "timezone"))
469 (defun nnrss-normalize-date (date)
470 "Return a date string of DATE in the RFC822 style.
471 This function handles the ISO 8601 date format described in
472 <URL:http://www.w3.org/TR/NOTE-datetime>, and also the RFC822 style
473 which RSS 2.0 allows."
474 (let (case-fold-search vector year month day time zone cts)
477 ((string-match " [0-9]+ " date)
478 (setq vector (timezone-parse-date date)
479 year (string-to-number (aref vector 0)))
481 (setq month (string-to-number (aref vector 1))
482 day (string-to-number (aref vector 2)))
483 (unless (>= (length (setq time (aref vector 3))) 3)
484 (setq time "00:00:00"))
485 (when (and (setq zone (aref vector 4))
486 (not (string-match "\\`[A-Z+-]" zone)))
493 "\\(199[0-9]\\|20[0-9][0-9]\\)"
502 "\\([012][0-9]:[0-5][0-9]\\)"
506 "\\(?:\\.[0-9]+\\)?\\)?\\)?"
508 "\\(?:\\(?:\\([+-][012][0-9]\\):\\([0-5][0-9]\\)\\)"
509 "\\|\\([+-][012][0-9][0-5][0-9]\\)"
512 (setq year (string-to-number (match-string 1 date))
513 month (string-to-number (or (match-string 2 date) "1"))
514 day (string-to-number (or (match-string 3 date) "1"))
515 time (if (match-beginning 5)
516 (substring date (match-beginning 4) (match-end 5))
517 (concat (or (match-string 4 date) "00:00") ":00"))
518 zone (cond ((match-beginning 6)
519 (concat (match-string 6 date)
520 (match-string 7 date)))
521 ((match-beginning 9) ;; Z
523 (t ;; nil if zone is not provided.
524 (match-string 8 date))))))
527 (setq cts (current-time-string (encode-time 0 0 0 day month year)))
528 (format "%s, %02d %s %04d %s%s"
529 (substring cts 0 3) day (substring cts 4 7) year time
533 (message-make-date))))
537 (defun nnrss-read-server-data (server)
538 (setq nnrss-server-data nil)
539 (let ((file (nnrss-make-filename "nnrss" server)))
540 (when (file-exists-p file)
541 ;; In Emacs 21.3 and earlier, `load' doesn't support non-ASCII
542 ;; file names. So, we use `insert-file-contents' instead.
543 (mm-with-multibyte-buffer
544 (let ((coding-system-for-read nnrss-file-coding-system)
545 (file-name-coding-system nnmail-pathname-coding-system))
546 (insert-file-contents file)
547 (eval-region (point-min) (point-max)))))))
549 (defun nnrss-save-server-data (server)
550 (gnus-make-directory nnrss-directory)
551 (let ((coding-system-for-write nnrss-file-coding-system)
552 (file-name-coding-system nnmail-pathname-coding-system))
553 (with-temp-file (nnrss-make-filename "nnrss" server)
554 (insert (format ";; -*- coding: %s; -*-\n"
555 nnrss-file-coding-system))
556 (gnus-prin1 `(setq nnrss-group-alist ',nnrss-group-alist))
558 (gnus-prin1 `(setq nnrss-server-data ',nnrss-server-data)))))
560 (defun nnrss-read-group-data (group server)
561 (setq nnrss-group-data nil)
562 (if (hash-table-p nnrss-group-hashtb)
563 (clrhash nnrss-group-hashtb)
564 (setq nnrss-group-hashtb (make-hash-table :test 'equal)))
565 (let ((pair (assoc group nnrss-server-data)))
566 (setq nnrss-group-max (or (cadr pair) 0))
567 (setq nnrss-group-min (+ nnrss-group-max 1)))
568 (let ((file (nnrss-make-filename group server)))
569 (when (file-exists-p file)
570 ;; In Emacs 21.3 and earlier, `load' doesn't support non-ASCII
571 ;; file names. So, we use `insert-file-contents' instead.
572 (mm-with-multibyte-buffer
573 (let ((coding-system-for-read nnrss-file-coding-system)
574 (file-name-coding-system nnmail-pathname-coding-system))
575 (insert-file-contents file)
576 (eval-region (point-min) (point-max))))
577 (dolist (e nnrss-group-data)
578 (puthash (or (nth 2 e) (nth 6 e)) t nnrss-group-hashtb)
579 (when (and (car e) (> nnrss-group-min (car e)))
580 (setq nnrss-group-min (car e)))
581 (when (and (car e) (< nnrss-group-max (car e)))
582 (setq nnrss-group-max (car e)))))))
584 (defun nnrss-save-group-data (group server)
585 (gnus-make-directory nnrss-directory)
586 (let ((coding-system-for-write nnrss-file-coding-system)
587 (file-name-coding-system nnmail-pathname-coding-system))
588 (with-temp-file (nnrss-make-filename group server)
589 (insert (format ";; -*- coding: %s; -*-\n"
590 nnrss-file-coding-system))
591 (gnus-prin1 `(setq nnrss-group-data ',nnrss-group-data)))))
593 (defun nnrss-make-filename (name server)
595 (nnrss-translate-file-chars
598 (not (equal server ""))
604 (gnus-add-shutdown 'nnrss-close 'gnus)
606 (defun nnrss-close ()
607 "Clear internal nnrss variables."
608 (setq nnrss-group-data nil
609 nnrss-server-data nil
610 nnrss-group-hashtb nil
611 nnrss-group-alist nil))
615 (defun nnrss-no-cache (url)
618 (defun nnrss-insert-w3 (url)
619 (mm-with-unibyte-current-buffer
622 (error (if (or debug-on-quit debug-on-error)
623 (signal (car err) (cdr err))
624 (message "nnrss: Failed to fetch %s" url))))))
626 (defun nnrss-decode-entities-string (string)
628 (mm-with-multibyte-buffer
630 (mm-url-decode-entities-nbsp)
633 (defalias 'nnrss-insert 'nnrss-insert-w3)
635 (defun nnrss-mime-encode-string (string)
636 (mm-with-multibyte-buffer
638 (mm-url-decode-entities-nbsp)
639 (goto-char (point-min))
640 (while (re-search-forward "[\t\n ]+" nil t)
642 (goto-char (point-min))
643 (skip-chars-forward " ")
644 (delete-region (point-min) (point))
645 (goto-char (point-max))
646 (skip-chars-forward " ")
647 (delete-region (point) (point-max))
648 (let ((rfc2047-encoding-type 'mime)
649 rfc2047-encode-max-chars)
650 (rfc2047-encode-region (point-min) (point-max)))
651 (goto-char (point-min))
652 (while (search-forward "\n" nil t)
653 (delete-backward-char 1))
658 (defun nnrss-check-group (group server)
659 (let (file xml subject url extra changed author date feed-subject
660 enclosure comments rss-ns rdf-ns content-ns dc-ns)
661 (if (and nnrss-use-local
662 (file-exists-p (setq file (expand-file-name
663 (nnrss-translate-file-chars
664 (concat group ".xml"))
666 (setq xml (nnrss-fetch file t))
667 (setq url (or (nth 2 (assoc group nnrss-server-data))
668 (second (assoc group nnrss-group-alist))))
675 (format "URL to search for %s: " group) "http://")))))
676 (let ((pair (assoc group nnrss-server-data)))
678 (setcdr (cdr pair) (list url))
679 (push (list group nnrss-group-max url) nnrss-server-data)))
681 (setq xml (nnrss-fetch url)))
683 ;; http://feeds.archive.org/validator/docs/howto/declare_namespaces.html
684 ;; for more RSS namespaces.
685 (setq dc-ns (nnrss-get-namespace-prefix xml "http://purl.org/dc/elements/1.1/")
686 rdf-ns (nnrss-get-namespace-prefix xml "http://www.w3.org/1999/02/22-rdf-syntax-ns#")
687 rss-ns (nnrss-get-namespace-prefix xml "http://purl.org/rss/1.0/")
688 content-ns (nnrss-get-namespace-prefix xml "http://purl.org/rss/1.0/modules/content/"))
689 (dolist (item (nreverse (nnrss-find-el (intern (concat rss-ns "item")) xml)))
690 (when (and (listp item)
691 (string= (concat rss-ns "item") (car item))
692 (if (setq url (nnrss-decode-entities-string
693 (nnrss-node-text rss-ns 'link (cddr item))))
694 (not (gethash url nnrss-group-hashtb))
695 (setq extra (or (nnrss-node-text content-ns 'encoded item)
696 (nnrss-node-text rss-ns 'description item)))
697 (not (gethash extra nnrss-group-hashtb))))
698 (setq subject (nnrss-node-text rss-ns 'title item))
699 (setq extra (or extra
700 (nnrss-node-text content-ns 'encoded item)
701 (nnrss-node-text rss-ns 'description item)))
702 (if (setq feed-subject (nnrss-node-text dc-ns 'subject item))
703 (setq extra (concat feed-subject "<br /><br />" extra)))
704 (setq author (or (nnrss-node-text rss-ns 'author item)
705 (nnrss-node-text dc-ns 'creator item)
706 (nnrss-node-text dc-ns 'contributor item)))
707 (setq date (nnrss-normalize-date
708 (or (nnrss-node-text dc-ns 'date item)
709 (nnrss-node-text rss-ns 'pubDate item))))
710 (setq comments (nnrss-node-text rss-ns 'comments item))
711 (when (setq enclosure (cadr (assq (intern (concat rss-ns "enclosure")) item)))
712 (let ((url (cdr (assq 'url enclosure)))
713 (len (cdr (assq 'length enclosure)))
714 (type (cdr (assq 'type enclosure)))
717 (if (and len (integerp (setq len (string-to-number len))))
718 ;; actually already in `ls-lisp-format-file-size' but
719 ;; probably not worth to require it for one function
720 (do ((size (/ len 1.0) (/ size 1024.0))
721 (post-fixes (list "" "k" "M" "G" "T" "P" "E")
724 (format "%.1f%s" size (car post-fixes))))
726 (setq url (or url ""))
727 (setq name (if (string-match "/\\([^/]*\\)$" url)
730 (setq type (or type ""))
731 (setq enclosure (list url name len type))))
734 (incf nnrss-group-max)
737 (and subject (nnrss-mime-encode-string subject))
738 (and author (nnrss-mime-encode-string author))
740 (and extra (nnrss-decode-entities-string extra))
744 (puthash (or url extra) t nnrss-group-hashtb)
748 (nnrss-save-group-data group server)
749 (let ((pair (assoc group nnrss-server-data)))
751 (setcar (cdr pair) nnrss-group-max)
752 (push (list group nnrss-group-max) nnrss-server-data)))
753 (nnrss-save-server-data server))))
755 (defun nnrss-opml-import (opml-file)
756 "OPML subscriptions import.
757 Read the file and attempt to subscribe to each Feed in the file."
758 (interactive "fImport file: ")
760 (lambda (node) (gnus-group-make-rss-group
761 (cdr (assq 'xmlUrl (cadr node)))))
762 (nnrss-find-el 'outline
764 (find-file opml-file)
765 (xml-parse-region (point-min)
768 (defun nnrss-opml-export ()
769 "OPML subscription export.
770 Export subscriptions to a buffer in OPML Format."
772 (with-current-buffer (get-buffer-create "*OPML Export*")
773 (mm-set-buffer-file-coding-system 'utf-8)
774 (insert "<?xml version=\"1.0\" encoding=\"utf-8\"?>\n"
775 "<!-- OPML generated by Emacs Gnus' nnrss.el -->\n"
776 "<opml version=\"1.1\">\n"
778 " <title>mySubscriptions</title>\n"
779 " <dateCreated>" (format-time-string "%a, %d %b %Y %T %z")
781 " <ownerEmail>" user-mail-address "</ownerEmail>\n"
782 " <ownerName>" (user-full-name) "</ownerName>\n"
785 (dolist (sub nnrss-group-alist)
786 (insert " <outline text=\"" (car sub)
787 "\" xmlUrl=\"" (cadr sub) "\"/>\n"))
790 (pop-to-buffer "*OPML Export*")
791 (when (fboundp 'sgml-mode)
794 (defun nnrss-generate-download-script ()
795 "Generate a download script in the current buffer.
796 It is useful when `(setq nnrss-use-local t)'."
798 (insert "#!/bin/sh\n")
799 (insert "WGET=wget\n")
800 (insert "RSSDIR='" (expand-file-name nnrss-directory) "'\n")
801 (dolist (elem nnrss-server-data)
802 (let ((url (or (nth 2 elem)
803 (second (assoc (car elem) nnrss-group-alist)))))
804 (insert "$WGET -q -O \"$RSSDIR\"/'"
805 (nnrss-translate-file-chars (concat (car elem) ".xml"))
808 (defun nnrss-translate-file-chars (name)
809 (let ((nnheader-file-name-translation-alist
810 (append nnheader-file-name-translation-alist '((?' . ?_)))))
811 (nnheader-translate-file-chars name)))
813 (defvar nnrss-moreover-url
814 "http://w.moreover.com/categories/category_list_rss.html"
815 "The url of moreover.com categories.")
817 (defun nnrss-snarf-moreover-categories ()
818 "Snarf RSS links from moreover.com."
820 (let (category name url changed)
822 (nnrss-insert nnrss-moreover-url)
823 (goto-char (point-min))
824 (while (re-search-forward
825 "<a name=\"\\([^\"]+\\)\">\\|<a href=\"\\(http://[^\"]*moreover\\.com[^\"]+page\\?c=\\([^\"&]+\\)&o=rss\\)" nil t)
827 (setq category (match-string 1))
828 (setq url (match-string 2)
829 name (mm-url-decode-entities-string
830 (rfc2231-decode-encoded-string
833 (setq name (concat category "." name)))
834 (unless (assoc name nnrss-server-data)
836 (push (list name 0 url) nnrss-server-data)))))
838 (nnrss-save-server-data ""))))
840 (defun nnrss-node-text (namespace local-name element)
841 (let* ((node (assq (intern (concat namespace (symbol-name local-name)))
843 (text (if (and node (listp node))
844 (nnrss-node-just-text node)
846 (cleaned-text (if text
847 (gnus-replace-in-string
848 (gnus-replace-in-string
849 text "^[\000-\037\177]+\\|^ +\\| +$" "")
851 (if (string-equal "" cleaned-text)
855 (defun nnrss-node-just-text (node)
856 (if (and node (listp node))
857 (mapconcat 'nnrss-node-just-text (cddr node) " ")
860 (defun nnrss-find-el (tag data &optional found-list)
861 "Find the all matching elements in the data.
862 Careful with this on large documents!"
866 (when (equal tag (car bit))
867 ;; Old xml.el may return a list of string.
868 (when (and (consp (caddr bit))
869 (stringp (caaddr bit)))
870 (setcar (cddr bit) (caaddr bit)))
874 (if (and (consp (car-safe (caddr bit)))
875 (not (stringp (caddr bit))))
883 tag (cddr bit))))))))
886 (defun nnrss-rsslink-p (el)
887 "Test if the element we are handed is an RSS autodiscovery link."
888 (and (eq (car-safe el) 'link)
889 (string-equal (cdr (assoc 'rel (cadr el))) "alternate")
890 (or (string-equal (cdr (assoc 'type (cadr el)))
891 "application/rss+xml")
892 (string-equal (cdr (assoc 'type (cadr el))) "text/xml"))))
894 (defun nnrss-get-rsslinks (data)
895 "Extract the <link> elements that are links to RSS from the parsed data."
898 (if (nnrss-rsslink-p el) el))
899 (nnrss-find-el 'link data))))
901 (defun nnrss-extract-hrefs (data)
902 "Recursively extract hrefs from a page's source.
903 DATA should be the output of `xml-parse-region' or
905 (mapcar (lambda (ahref)
906 (cdr (assoc 'href (cadr ahref))))
907 (nnrss-find-el 'a data)))
909 (defmacro nnrss-match-macro (base-uri item onsite-list offsite-list)
910 `(cond ((or (string-match (concat "^" ,base-uri) ,item)
911 (not (string-match "://" ,item)))
912 (setq ,onsite-list (append ,onsite-list (list ,item))))
913 (t (setq ,offsite-list (append ,offsite-list (list ,item))))))
915 (defun nnrss-order-hrefs (base-uri hrefs)
916 "Given a list of hrefs, sort them using the following priorities:
917 1. links ending in .rss
918 2. links ending in .rdf
919 3. links ending in .xml
920 4. links containing the above
923 BASE-URI is used to determine the location of the links and
924 whether they are `offsite' or `onsite'."
925 (let (rss-onsite-end rdf-onsite-end xml-onsite-end
926 rss-onsite-in rdf-onsite-in xml-onsite-in
927 rss-offsite-end rdf-offsite-end xml-offsite-end
928 rss-offsite-in rdf-offsite-in xml-offsite-in)
931 ((string-match "\\.rss$" href)
933 base-uri href rss-onsite-end rss-offsite-end))
934 ((string-match "\\.rdf$" href)
936 base-uri href rdf-onsite-end rdf-offsite-end))
937 ((string-match "\\.xml$" href)
939 base-uri href xml-onsite-end xml-offsite-end))
940 ((string-match "rss" href)
942 base-uri href rss-onsite-in rss-offsite-in))
943 ((string-match "rdf" href)
945 base-uri href rdf-onsite-in rdf-offsite-in))
946 ((string-match "xml" href)
948 base-uri href xml-onsite-in xml-offsite-in))))
950 rss-onsite-end rdf-onsite-end xml-onsite-end
951 rss-onsite-in rdf-onsite-in xml-onsite-in
952 rss-offsite-end rdf-offsite-end xml-offsite-end
953 rss-offsite-in rdf-offsite-in xml-offsite-in)))
955 (defun nnrss-discover-feed (url)
956 "Given a page, find an RSS feed using Mark Pilgrim's
957 `ultra-liberal rss locator' (http://diveintomark.org/2002/08/15.html)."
959 (let ((parsed-page (nnrss-fetch url)))
961 ;; 1. if this url is the rss, use it.
962 (if (nnrss-rss-p parsed-page)
963 (let ((rss-ns (nnrss-get-namespace-prefix parsed-page "http://purl.org/rss/1.0/")))
964 (nnrss-rss-title-description rss-ns parsed-page url))
966 ;; 2. look for the <link rel="alternate"
967 ;; type="application/rss+xml" and use that if it is there.
968 (let ((links (nnrss-get-rsslinks parsed-page)))
970 (let* ((xml (nnrss-fetch
971 (cdr (assoc 'href (cadar links)))))
972 (rss-ns (nnrss-get-namespace-prefix xml "http://purl.org/rss/1.0/")))
973 (nnrss-rss-title-description rss-ns xml (cdr (assoc 'href (cadar links)))))
975 ;; 3. look for links on the site in the following order:
976 ;; - onsite links ending in .rss, .rdf, or .xml
977 ;; - onsite links containing any of the above
978 ;; - offsite links ending in .rss, .rdf, or .xml
979 ;; - offsite links containing any of the above
980 (let* ((base-uri (progn (string-match ".*://[^/]+/?" url)
981 (match-string 0 url)))
982 (hrefs (nnrss-order-hrefs
983 base-uri (nnrss-extract-hrefs parsed-page)))
985 (while (and (eq rss-link nil) (not (eq hrefs nil)))
986 (let ((href-data (nnrss-fetch (car hrefs))))
987 (if (nnrss-rss-p href-data)
988 (let* ((rss-ns (nnrss-get-namespace-prefix href-data "http://purl.org/rss/1.0/")))
989 (setq rss-link (nnrss-rss-title-description
990 rss-ns href-data (car hrefs))))
991 (setq hrefs (cdr hrefs)))))
992 (if rss-link rss-link
995 (nnrss-find-rss-via-syndic8 url))))))))
997 (defun nnrss-find-rss-via-syndic8 (url)
998 "Query syndic8 for the rss feeds it has for URL."
999 (if (not (locate-library "xml-rpc"))
1001 (message "XML-RPC is not available... not checking Syndic8.")
1004 (let ((feedid (xml-rpc-method-call
1005 "http://www.syndic8.com/xmlrpc.php"
1009 (let* ((feedinfo (xml-rpc-method-call
1010 "http://www.syndic8.com/xmlrpc.php"
1011 'syndic8.GetFeedInfo
1018 (cdr (assoc "status" listinfo))
1021 (cdr (assoc "sitename" listinfo))
1025 "sitename" listinfo)))
1028 "dataurl" listinfo)))))))
1030 (if (not (> (length urllist) 1))
1032 (let ((completion-ignore-case t)
1034 (mapcar (lambda (listinfo)
1035 (cons (cdr (assoc "sitename" listinfo))
1037 (cdr (assoc "feedid" listinfo)))))
1041 "Multiple feeds found. Select one: "
1042 selection nil t) urllist)))))))))
1044 (defun nnrss-rss-p (data)
1045 "Test if DATA is an RSS feed.
1046 Simply ensures that the first element is rss or rdf."
1047 (or (eq (caar data) 'rss)
1048 (eq (caar data) 'rdf:RDF)))
1050 (defun nnrss-rss-title-description (rss-namespace data url)
1051 "Return the title of an RSS feed."
1052 (if (nnrss-rss-p data)
1053 (let ((description (intern (concat rss-namespace "description")))
1054 (title (intern (concat rss-namespace "title")))
1055 (channel (nnrss-find-el (intern (concat rss-namespace "channel"))
1058 (cons 'description (caddr (nth 0 (nnrss-find-el description channel))))
1059 (cons 'title (caddr (nth 0 (nnrss-find-el title channel))))
1060 (cons 'href url)))))
1062 (defun nnrss-get-namespace-prefix (el uri)
1063 "Given EL (containing a parsed element) and URI (containing a string
1064 that gives the URI for which you want to retrieve the namespace
1065 prefix), return the prefix."
1066 (let* ((prefix (car (rassoc uri (cadar el))))
1068 (split-string (symbol-name prefix) ":")))
1069 (ns (cond ((eq (length nslist) 1) ; no prefix given
1071 ((eq (length nslist) 2) ; extract prefix
1073 (if (and ns (not (string= ns "")))
1080 ;;; nnrss.el ends here
1082 ;;; arch-tag: 12910c07-0cdf-44fb-8d2c-416ded64c267