Replace `outline-flag-region' with `org-flag-region'
[org-mode/org-tableheadings.git] / lisp / org-feed.el
blobf3c8ab89373368a452fa3819a41fdeaa1b5d79e3
1 ;;; org-feed.el --- Add RSS feed items to Org files -*- lexical-binding: t; -*-
2 ;;
3 ;; Copyright (C) 2009-2018 Free Software Foundation, Inc.
4 ;;
5 ;; Author: Carsten Dominik <carsten at orgmode dot org>
6 ;; Keywords: outlines, hypermedia, calendar, wp
7 ;; Homepage: http://orgmode.org
8 ;;
9 ;; This file is part of GNU Emacs.
11 ;; GNU Emacs is free software: you can redistribute it and/or modify
12 ;; it under the terms of the GNU General Public License as published by
13 ;; the Free Software Foundation, either version 3 of the License, or
14 ;; (at your option) any later version.
16 ;; GNU Emacs is distributed in the hope that it will be useful,
17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
19 ;; GNU General Public License for more details.
21 ;; You should have received a copy of the GNU General Public License
22 ;; along with GNU Emacs. If not, see <https://www.gnu.org/licenses/>.
23 ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
25 ;;; Commentary:
27 ;; This module allows entries to be created and changed in an Org mode
28 ;; file triggered by items in an RSS feed. The basic functionality
29 ;; is geared toward simply adding new items found in a feed as
30 ;; outline nodes to an Org file. Using hooks, arbitrary actions can
31 ;; be triggered for new or changed items.
33 ;; Selecting feeds and target locations
34 ;; ------------------------------------
36 ;; This module is configured through a single variable, `org-feed-alist'.
37 ;; Here is an example, using a notes/tasks feed from reQall.com.
39 ;; (setq org-feed-alist
40 ;; '(("ReQall"
41 ;; "http://www.reqall.com/user/feeds/rss/a1b2c3....."
42 ;; "~/org/feeds.org" "ReQall Entries")
44 ;; With this setup, the command `M-x org-feed-update-all' will
45 ;; collect new entries in the feed at the given URL and create
46 ;; entries as subheadings under the "ReQall Entries" heading in the
47 ;; file "~/org/feeds.org". Each feed should normally have its own
48 ;; heading - however see the `:drawer' parameter.
50 ;; Besides these standard elements that need to be specified for each
51 ;; feed, keyword-value pairs can set additional options. For example,
52 ;; to de-select transitional entries with a title containing
54 ;; "reQall is typing what you said",
56 ;; you could use the `:filter' argument:
58 ;; (setq org-feed-alist
59 ;; '(("ReQall"
60 ;; "http://www.reqall.com/user/feeds/rss/a1b2c3....."
61 ;; "~/org/feeds.org" "ReQall Entries"
62 ;; :filter my-reqall-filter)))
64 ;; (defun my-reqall-filter (e)
65 ;; (if (string-match "reQall is typing what you said"
66 ;; (plist-get e :title))
67 ;; nil
68 ;; e))
70 ;; See the docstring for `org-feed-alist' for more details.
73 ;; Keeping track of previously added entries
74 ;; -----------------------------------------
76 ;; Since Org allows you to delete, archive, or move outline nodes,
77 ;; org-feed.el needs to keep track of which feed items have been handled
78 ;; before, so that they will not be handled again. For this, org-feed.el
79 ;; stores information in a special drawer, FEEDSTATUS, under the heading
80 ;; that received the input of the feed.
83 ;; Acknowledgments
84 ;; ---------------
86 ;; org-feed.el is based on ideas by Brad Bozarth who implemented a
87 ;; similar mechanism using shell and awk scripts.
89 ;;; Code:
91 (require 'org)
92 (require 'sha1)
94 (declare-function url-retrieve-synchronously "url"
95 (url &optional silent inhibit-cookies timeout))
96 (declare-function xml-node-children "xml" (node))
97 (declare-function xml-get-children "xml" (node child-name))
98 (declare-function xml-get-attribute "xml" (node attribute))
99 (declare-function xml-get-attribute-or-nil "xml" (node attribute))
100 (declare-function xml-substitute-special "xml" (string))
102 (declare-function org-capture-escaped-% "org-capture" ())
103 (declare-function org-capture-expand-embedded-elisp "org-capture" (&optional mark))
104 (declare-function org-capture-inside-embedded-elisp-p "org-capture" ())
106 (defgroup org-feed nil
107 "Options concerning RSS feeds as inputs for Org files."
108 :tag "Org Feed"
109 :group 'org)
111 (defcustom org-feed-alist nil
112 "Alist specifying RSS feeds that should create inputs for Org.
113 Each entry in this list specified an RSS feed tat should be queried
114 to create inbox items in Org. Each entry is a list with the following items:
116 name a custom name for this feed
117 URL the Feed URL
118 file the target Org file where entries should be listed, when
119 nil the target becomes the current buffer (may be an
120 indirect buffer) each time the feed update is invoked
121 headline the headline under which entries should be listed
123 Additional arguments can be given using keyword-value pairs. Many of these
124 specify functions that receive one or a list of \"entries\" as their single
125 argument. An entry is a property list that describes a feed item. The
126 property list has properties for each field in the item, for example `:title'
127 for the `<title>' field and `:pubDate' for the publication date. In addition,
128 it contains the following properties:
130 `:item-full-text' the full text in the <item> tag
131 `:guid-permalink' t when the guid property is a permalink
133 Here are the keyword-value pair allows in `org-feed-alist'.
135 :drawer drawer-name
136 The name of the drawer for storing feed information. The default is
137 \"FEEDSTATUS\". Using different drawers for different feeds allows
138 several feeds to target the same inbox heading.
140 :filter filter-function
141 A function to select interesting entries in the feed. It gets a single
142 entry as parameter. It should return the entry if it is relevant, or
143 nil if it is not.
145 :template template-string
146 The default action on new items in the feed is to add them as children
147 under the headline for the feed. The template describes how the entry
148 should be formatted. If not given, it defaults to
149 `org-feed-default-template'.
151 :formatter formatter-function
152 Instead of relying on a template, you may specify a function to format
153 the outline node to be inserted as a child. This function gets passed
154 a property list describing a single feed item, and it should return a
155 string that is a properly formatted Org outline node of level 1.
157 :new-handler function
158 If adding new items as children to the outline is not what you want
159 to do with new items, define a handler function that is called with
160 a list of all new items in the feed, each one represented as a property
161 list. The handler should do what needs to be done, and org-feed will
162 mark all items given to this handler as \"handled\", i.e. they will not
163 be passed to this handler again in future readings of the feed.
164 When the handler is called, point will be at the feed headline.
166 :changed-handler function
167 This function gets passed a list of all entries that have been
168 handled before, but are now still in the feed and have *changed*
169 since last handled (as evidenced by a different sha1 hash).
170 When the handler is called, point will be at the feed headline.
172 :parse-feed function
173 This function gets passed a buffer, and should return a list
174 of entries, each being a property list containing the
175 `:guid' and `:item-full-text' keys. The default is
176 `org-feed-parse-rss-feed'; `org-feed-parse-atom-feed' is an
177 alternative.
179 :parse-entry function
180 This function gets passed an entry as returned by the parse-feed
181 function, and should return the entry with interesting properties added.
182 The default is `org-feed-parse-rss-entry'; `org-feed-parse-atom-entry'
183 is an alternative."
184 :group 'org-feed
185 :type '(repeat
186 (list :value ("" "http://" "" "")
187 (string :tag "Name")
188 (string :tag "Feed URL")
189 (file :tag "File for inbox")
190 (string :tag "Headline for inbox")
191 (repeat :inline t
192 (choice
193 (list :inline t :tag "Filter"
194 (const :filter)
195 (symbol :tag "Filter Function"))
196 (list :inline t :tag "Template"
197 (const :template)
198 (string :tag "Template"))
199 (list :inline t :tag "Formatter"
200 (const :formatter)
201 (symbol :tag "Formatter Function"))
202 (list :inline t :tag "New items handler"
203 (const :new-handler)
204 (symbol :tag "Handler Function"))
205 (list :inline t :tag "Changed items"
206 (const :changed-handler)
207 (symbol :tag "Handler Function"))
208 (list :inline t :tag "Parse Feed"
209 (const :parse-feed)
210 (symbol :tag "Parse Feed Function"))
211 (list :inline t :tag "Parse Entry"
212 (const :parse-entry)
213 (symbol :tag "Parse Entry Function"))
214 )))))
216 (defcustom org-feed-drawer "FEEDSTATUS"
217 "The name of the drawer for feed status information.
218 Each feed may also specify its own drawer name using the `:drawer'
219 parameter in `org-feed-alist'."
220 :group 'org-feed
221 :type '(string :tag "Drawer Name"))
223 (defcustom org-feed-default-template "\n* %h\n %U\n %description\n %a\n"
224 "Template for the Org node created from RSS feed items.
225 This is just the default, each feed can specify its own.
226 Any fields from the feed item can be interpolated into the template with
227 %name, for example %title, %description, %pubDate etc. In addition, the
228 following special escapes are valid as well:
230 %h The title, or the first line of the description
231 %t The date as a stamp, either from <pubDate> (if present), or
232 the current date
233 %T Date and time
234 %u,%U Like %t,%T, but inactive time stamps
235 %a A link, from <guid> if that is a permalink, else from <link>
236 %(sexp) Evaluate elisp `(sexp)' and replace with the result, the simple
237 %-escapes above can be used as arguments, e.g. %(capitalize \\\"%h\\\")"
238 :group 'org-feed
239 :type '(string :tag "Template"))
241 (defcustom org-feed-save-after-adding t
242 "Non-nil means save buffer after adding new feed items."
243 :group 'org-feed
244 :type 'boolean)
246 (defcustom org-feed-retrieve-method 'url-retrieve-synchronously
247 "The method to be used to retrieve a feed URL.
248 This can be `curl' or `wget' to call these external programs, or it can be
249 an Emacs Lisp function that will return a buffer containing the content
250 of the file pointed to by the URL."
251 :group 'org-feed
252 :type '(choice
253 (const :tag "Internally with url.el" url-retrieve-synchronously)
254 (const :tag "Externally with curl" curl)
255 (const :tag "Externally with wget" wget)
256 (function :tag "Function")))
258 (defcustom org-feed-before-adding-hook nil
259 "Hook that is run before adding new feed items to a file.
260 You might want to commit the file in its current state to version control,
261 for example."
262 :group 'org-feed
263 :type 'hook)
265 (defcustom org-feed-after-adding-hook nil
266 "Hook that is run after new items have been added to a file.
267 Depending on `org-feed-save-after-adding', the buffer will already
268 have been saved."
269 :group 'org-feed
270 :type 'hook)
272 (defvar org-feed-buffer "*Org feed*"
273 "The buffer used to retrieve a feed.")
275 ;;;###autoload
276 (defun org-feed-update-all ()
277 "Get inbox items from all feeds in `org-feed-alist'."
278 (interactive)
279 (let ((nfeeds (length org-feed-alist))
280 (nnew (apply '+ (mapcar 'org-feed-update org-feed-alist))))
281 (message "%s from %d %s"
282 (cond ((= nnew 0) "No new entries")
283 ((= nnew 1) "1 new entry")
284 (t (format "%d new entries" nnew)))
285 nfeeds
286 (if (= nfeeds 1) "feed" "feeds"))))
288 ;;;###autoload
289 (defun org-feed-update (feed &optional retrieve-only)
290 "Get inbox items from FEED.
291 FEED can be a string with an association in `org-feed-alist', or
292 it can be a list structured like an entry in `org-feed-alist'."
293 (interactive (list (org-completing-read "Feed name: " org-feed-alist)))
294 (if (stringp feed) (setq feed (assoc feed org-feed-alist)))
295 (unless feed
296 (error "No such feed in `org-feed-alist"))
297 (catch 'exit
298 (let ((name (car feed))
299 (url (nth 1 feed))
300 (file (or (nth 2 feed) (buffer-file-name (or (buffer-base-buffer)
301 (current-buffer)))))
302 (headline (nth 3 feed))
303 (filter (nth 1 (memq :filter feed)))
304 (formatter (nth 1 (memq :formatter feed)))
305 (new-handler (nth 1 (memq :new-handler feed)))
306 (changed-handler (nth 1 (memq :changed-handler feed)))
307 (template (or (nth 1 (memq :template feed))
308 org-feed-default-template))
309 (drawer (or (nth 1 (memq :drawer feed))
310 org-feed-drawer))
311 (parse-feed (or (nth 1 (memq :parse-feed feed))
312 'org-feed-parse-rss-feed))
313 (parse-entry (or (nth 1 (memq :parse-entry feed))
314 'org-feed-parse-rss-entry))
315 feed-buffer inbox-pos new-formatted
316 entries old-status status new changed guid-alist guid olds)
317 (setq feed-buffer (org-feed-get-feed url))
318 (unless (and feed-buffer (bufferp (get-buffer feed-buffer)))
319 (error "Cannot get feed %s" name))
320 (when retrieve-only
321 (throw 'exit feed-buffer))
322 (setq entries (funcall parse-feed feed-buffer))
323 (ignore-errors (kill-buffer feed-buffer))
324 (save-excursion
325 (save-window-excursion
326 (setq inbox-pos (org-feed-goto-inbox-internal file headline))
327 (setq old-status (org-feed-read-previous-status inbox-pos drawer))
328 ;; Add the "handled" status to the appropriate entries
329 (setq entries (mapcar (lambda (e)
330 (setq e
331 (plist-put e :handled
332 (nth 1 (assoc
333 (plist-get e :guid)
334 old-status)))))
335 entries))
336 ;; Find out which entries are new and which are changed
337 (dolist (e entries)
338 (if (not (plist-get e :handled))
339 (push e new)
340 (setq olds (nth 2 (assoc (plist-get e :guid) old-status)))
341 (if (and olds
342 (not (string= (sha1
343 (plist-get e :item-full-text))
344 olds)))
345 (push e changed))))
347 ;; Parse the relevant entries fully
348 (setq new (mapcar parse-entry new)
349 changed (mapcar parse-entry changed))
351 ;; Run the filter
352 (when filter
353 (setq new (delq nil (mapcar filter new))
354 changed (delq nil (mapcar filter new))))
356 (when (not (or new changed))
357 (message "No new items in feed %s" name)
358 (throw 'exit 0))
360 ;; Get alist based on guid, to look up entries
361 (setq guid-alist
362 (append
363 (mapcar (lambda (e) (list (plist-get e :guid) e)) new)
364 (mapcar (lambda (e) (list (plist-get e :guid) e)) changed)))
366 ;; Construct the new status
367 (setq status
368 (mapcar
369 (lambda (e)
370 (setq guid (plist-get e :guid))
371 (list guid
372 ;; things count as handled if we handle them now,
373 ;; or if they were handled previously
374 (if (assoc guid guid-alist) t (plist-get e :handled))
375 ;; A hash, to detect changes
376 (sha1 (plist-get e :item-full-text))))
377 entries))
379 ;; Handle new items in the feed
380 (when new
381 (if new-handler
382 (progn
383 (goto-char inbox-pos)
384 (funcall new-handler new))
385 ;; No custom handler, do the default adding
386 ;; Format the new entries into an alist with GUIDs in the car
387 (setq new-formatted
388 (mapcar
389 (lambda (e) (org-feed-format-entry e template formatter))
390 new)))
392 ;; Insert the new items
393 (org-feed-add-items inbox-pos new-formatted))
395 ;; Handle changed items in the feed
396 (when (and changed-handler changed)
397 (goto-char inbox-pos)
398 (funcall changed-handler changed))
400 ;; Write the new status
401 ;; We do this only now, in case something goes wrong above, so
402 ;; that would would end up with a status that does not reflect
403 ;; which items truely have been handled
404 (org-feed-write-status inbox-pos drawer status)
406 ;; Normalize the visibility of the inbox tree
407 (goto-char inbox-pos)
408 (outline-hide-subtree)
409 (org-show-children)
411 ;; Hooks and messages
412 (when org-feed-save-after-adding (save-buffer))
413 (message "Added %d new item%s from feed %s to file %s, heading %s"
414 (length new) (if (> (length new) 1) "s" "")
415 name
416 (file-name-nondirectory file) headline)
417 (run-hooks 'org-feed-after-adding-hook)
418 (length new))))))
420 ;;;###autoload
421 (defun org-feed-goto-inbox (feed)
422 "Go to the inbox that captures the feed named FEED."
423 (interactive
424 (list (if (= (length org-feed-alist) 1)
425 (car org-feed-alist)
426 (org-completing-read "Feed name: " org-feed-alist))))
427 (if (stringp feed) (setq feed (assoc feed org-feed-alist)))
428 (unless feed
429 (error "No such feed in `org-feed-alist"))
430 (org-feed-goto-inbox-internal (nth 2 feed) (nth 3 feed)))
432 ;;;###autoload
433 (defun org-feed-show-raw-feed (feed)
434 "Show the raw feed buffer of a feed."
435 (interactive
436 (list (if (= (length org-feed-alist) 1)
437 (car org-feed-alist)
438 (org-completing-read "Feed name: " org-feed-alist))))
439 (if (stringp feed) (setq feed (assoc feed org-feed-alist)))
440 (unless feed
441 (error "No such feed in `org-feed-alist"))
442 (pop-to-buffer-same-window
443 (org-feed-update feed 'retrieve-only))
444 (goto-char (point-min)))
446 (defun org-feed-goto-inbox-internal (file heading)
447 "Find or create HEADING in FILE.
448 Switch to that buffer, and return the position of that headline."
449 (find-file file)
450 (widen)
451 (goto-char (point-min))
452 (if (re-search-forward
453 (concat "^\\*+[ \t]+" heading "[ \t]*\\(:.*?:[ \t]*\\)?$")
454 nil t)
455 (goto-char (match-beginning 0))
456 (goto-char (point-max))
457 (insert "\n\n* " heading "\n\n")
458 (org-back-to-heading t))
459 (point))
461 (defun org-feed-read-previous-status (pos drawer)
462 "Get the alist of old GUIDs from the entry at POS.
463 This will find DRAWER and extract the alist."
464 (save-excursion
465 (goto-char pos)
466 (let ((end (save-excursion (org-end-of-subtree t t))))
467 (if (re-search-forward
468 (concat "^[ \t]*:" drawer ":[ \t]*\n\\([^\000]*?\\)\n[ \t]*:END:")
469 end t)
470 (read (match-string 1))
471 nil))))
473 (defun org-feed-write-status (pos drawer status)
474 "Write the feed STATUS to DRAWER in entry at POS."
475 (save-excursion
476 (goto-char pos)
477 (let ((end (save-excursion (org-end-of-subtree t t))))
478 (if (re-search-forward (concat "^[ \t]*:" drawer ":[ \t]*\n")
479 end t)
480 (progn
481 (goto-char (match-end 0))
482 (delete-region (point)
483 (save-excursion
484 (and (re-search-forward "^[ \t]*:END:" nil t)
485 (match-beginning 0)))))
486 (outline-next-heading)
487 (insert " :" drawer ":\n :END:\n")
488 (beginning-of-line 0))
489 (insert (pp-to-string status)))))
491 (defun org-feed-add-items (pos entries)
492 "Add the formatted items to the headline as POS."
493 (let (entry level)
494 (save-excursion
495 (goto-char pos)
496 (unless (looking-at org-complex-heading-regexp)
497 (error "Wrong position"))
498 (setq level (org-get-valid-level (length (match-string 1)) 1))
499 (org-end-of-subtree t t)
500 (skip-chars-backward " \t\n")
501 (beginning-of-line 2)
502 (setq pos (point))
503 (while (setq entry (pop entries))
504 (org-paste-subtree level entry 'yank))
505 (org-mark-ring-push pos))))
507 (defun org-feed-format-entry (entry template formatter)
508 "Format ENTRY so that it can be inserted into an Org file.
509 ENTRY is a property list. This function adds a `:formatted-for-org' property
510 and returns the full property list.
511 If that property is already present, nothing changes."
512 (require 'org-capture)
513 (if formatter (funcall formatter entry)
514 (let* ((dlines
515 (org-split-string (or (plist-get entry :description) "???")
516 "\n"))
517 (time (or (if (plist-get entry :pubDate)
518 (org-read-date t t (plist-get entry :pubDate)))
519 (current-time)))
520 (v-h (or (plist-get entry :title) (car dlines) "???"))
521 (v-t (format-time-string (org-time-stamp-format nil nil) time))
522 (v-T (format-time-string (org-time-stamp-format t nil) time))
523 (v-u (format-time-string (org-time-stamp-format nil t) time))
524 (v-U (format-time-string (org-time-stamp-format t t) time))
525 (v-a (let ((tmp (or (and (plist-get entry :guid-permalink)
526 (plist-get entry :guid))
527 (plist-get entry :link))))
528 (if tmp (format "[[%s]]\n" tmp ) ""))))
529 (with-temp-buffer
530 (insert template)
531 (goto-char (point-min))
533 ;; Mark %() embedded elisp for later evaluation.
534 (org-capture-expand-embedded-elisp 'mark)
536 ;; Simple %-escapes. `org-capture-escaped-%' may modify
537 ;; buffer and cripple match-data. Use markers instead.
538 (while (re-search-forward "%\\([a-zA-Z]+\\)" nil t)
539 (let ((key (match-string 1))
540 (beg (copy-marker (match-beginning 0)))
541 (end (copy-marker (match-end 0))))
542 (unless (org-capture-escaped-%)
543 (delete-region beg end)
544 (set-marker beg nil)
545 (set-marker end nil)
546 (let ((replacement
547 (pcase key
548 ("h" v-h)
549 ("t" v-t)
550 ("T" v-T)
551 ("u" v-u)
552 ("U" v-U)
553 ("a" v-a)
554 (name
555 (let ((v (plist-get entry (intern (concat ":" name)))))
556 (save-excursion
557 (save-match-data
558 (beginning-of-line)
559 (if (looking-at
560 (concat "^\\([ \t]*\\)%" name "[ \t]*$"))
561 (org-feed-make-indented-block
562 v (org-get-indentation))
563 v))))))))
564 (when replacement
565 (insert
566 ;; Escape string delimiters within embedded lisp.
567 (if (org-capture-inside-embedded-elisp-p)
568 (replace-regexp-in-string "\"" "\\\\\"" replacement)
569 replacement)))))))
571 ;; %() embedded elisp
572 (org-capture-expand-embedded-elisp)
574 (decode-coding-string
575 (buffer-string) (detect-coding-region (point-min) (point-max) t))))))
577 (defun org-feed-make-indented-block (s n)
578 "Add indentation of N spaces to a multiline string S."
579 (if (not (string-match "\n" s))
581 (mapconcat 'identity
582 (org-split-string s "\n")
583 (concat "\n" (make-string n ?\ )))))
585 (defun org-feed-skip-http-headers (buffer)
586 "Remove HTTP headers from BUFFER, and return it.
587 Assumes headers are indeed present!"
588 (with-current-buffer buffer
589 (widen)
590 (goto-char (point-min))
591 (search-forward "\n\n")
592 (delete-region (point-min) (point))
593 buffer))
595 (defun org-feed-get-feed (url)
596 "Get the RSS feed file at URL and return the buffer."
597 (cond
598 ((eq org-feed-retrieve-method 'url-retrieve-synchronously)
599 (org-feed-skip-http-headers (url-retrieve-synchronously url)))
600 ((eq org-feed-retrieve-method 'curl)
601 (ignore-errors (kill-buffer org-feed-buffer))
602 (call-process "curl" nil org-feed-buffer nil "--silent" url)
603 org-feed-buffer)
604 ((eq org-feed-retrieve-method 'wget)
605 (ignore-errors (kill-buffer org-feed-buffer))
606 (call-process "wget" nil org-feed-buffer nil "-q" "-O" "-" url)
607 org-feed-buffer)
608 ((functionp org-feed-retrieve-method)
609 (funcall org-feed-retrieve-method url))))
611 (defun org-feed-parse-rss-feed (buffer)
612 "Parse BUFFER for RSS feed entries.
613 Returns a list of entries, with each entry a property list,
614 containing the properties `:guid' and `:item-full-text'."
615 (require 'xml)
616 (let ((case-fold-search t)
617 entries beg end item guid entry)
618 (with-current-buffer buffer
619 (widen)
620 (goto-char (point-min))
621 (while (re-search-forward "<item\\>.*?>" nil t)
622 (setq beg (point)
623 end (and (re-search-forward "</item>" nil t)
624 (match-beginning 0)))
625 (setq item (buffer-substring beg end)
626 guid (if (string-match "<guid\\>.*?>\\([^\000]*?\\)</guid>" item)
627 (xml-substitute-special (match-string-no-properties 1 item))))
628 (setq entry (list :guid guid :item-full-text item))
629 (push entry entries)
630 (widen)
631 (goto-char end))
632 (nreverse entries))))
634 (defun org-feed-parse-rss-entry (entry)
635 "Parse the `:item-full-text' field for xml tags and create new properties."
636 (require 'xml)
637 (with-temp-buffer
638 (insert (plist-get entry :item-full-text))
639 (goto-char (point-min))
640 (while (re-search-forward "<\\([a-zA-Z]+\\>\\).*?>\\([^\000]*?\\)</\\1>"
641 nil t)
642 (setq entry (plist-put entry
643 (intern (concat ":" (match-string 1)))
644 (xml-substitute-special (match-string 2)))))
645 (goto-char (point-min))
646 (unless (re-search-forward "isPermaLink[ \t]*=[ \t]*\"false\"" nil t)
647 (setq entry (plist-put entry :guid-permalink t))))
648 entry)
650 (defun org-feed-parse-atom-feed (buffer)
651 "Parse BUFFER for Atom feed entries.
652 Returns a list of entries, with each entry a property list,
653 containing the properties `:guid' and `:item-full-text'.
655 The `:item-full-text' property actually contains the sexp
656 formatted as a string, not the original XML data."
657 (require 'xml)
658 (with-current-buffer buffer
659 (widen)
660 (let ((feed (car (xml-parse-region (point-min) (point-max)))))
661 (mapcar
662 (lambda (entry)
663 (list
664 :guid (car (xml-node-children (car (xml-get-children entry 'id))))
665 :item-full-text (prin1-to-string entry)))
666 (xml-get-children feed 'entry)))))
668 (defun org-feed-parse-atom-entry (entry)
669 "Parse the `:item-full-text' as a sexp and create new properties."
670 (let ((xml (car (read-from-string (plist-get entry :item-full-text)))))
671 ;; Get first <link href='foo'/>.
672 (setq entry (plist-put entry :link
673 (xml-get-attribute
674 (car (xml-get-children xml 'link))
675 'href)))
676 ;; Add <title/> as :title.
677 (setq entry (plist-put entry :title
678 (xml-substitute-special
679 (car (xml-node-children
680 (car (xml-get-children xml 'title)))))))
681 (let* ((content (car (xml-get-children xml 'content)))
682 (type (xml-get-attribute-or-nil content 'type)))
683 (when content
684 (cond
685 ((string= type "text")
686 ;; We like plain text.
687 (setq entry (plist-put entry :description
688 (xml-substitute-special
689 (car (xml-node-children content))))))
690 ((string= type "html")
691 ;; TODO: convert HTML to Org markup.
692 (setq entry (plist-put entry :description
693 (xml-substitute-special
694 (car (xml-node-children content))))))
695 ((string= type "xhtml")
696 ;; TODO: convert XHTML to Org markup.
697 (setq entry (plist-put entry :description
698 (prin1-to-string
699 (xml-node-children content)))))
701 (setq entry (plist-put entry :description
702 (format-message
703 "Unknown `%s' content." type)))))))
704 entry))
706 (provide 'org-feed)
708 ;; Local variables:
709 ;; generated-autoload-file: "org-loaddefs.el"
710 ;; End:
712 ;;; org-feed.el ends here