org-feed.el 25 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512513514515516517518519520521522523524525526527528529530531532533534535536537538539540541542543544545546547548549550551552553554555556557558559560561562563564565566567568569570571572573574575576577578579580581582583584585586587588589590591592593594595596597598599600601602603604605606607608609610611612613614615616617618619620621622623624625626627628629630631632633634635636637638639640641642643644645646647648649650651652653654655656657658659660661662663664665666667668669670671672673674675676677678679680681682683684685686687688689690691692693694695696697698699
  1. ;;; org-feed.el --- Add RSS feed items to Org files
  2. ;;
  3. ;; Copyright (C) 2009-2014 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.
  10. ;;
  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.
  15. ;; GNU Emacs is distributed in the hope that it will be useful,
  16. ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  17. ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  18. ;; GNU General Public License for more details.
  19. ;; You should have received a copy of the GNU General Public License
  20. ;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>.
  21. ;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
  22. ;;
  23. ;;; Commentary:
  24. ;;
  25. ;; This module allows to create and change entries in an Org-mode
  26. ;; file triggered by items in an RSS feed. The basic functionality is
  27. ;; geared toward simply adding new items found in a feed as outline nodes
  28. ;; to an Org file. Using hooks, arbitrary actions can be triggered for
  29. ;; new or changed items.
  30. ;;
  31. ;; Selecting feeds and target locations
  32. ;; ------------------------------------
  33. ;;
  34. ;; This module is configured through a single variable, `org-feed-alist'.
  35. ;; Here is an example, using a notes/tasks feed from reQall.com.
  36. ;;
  37. ;; (setq org-feed-alist
  38. ;; '(("ReQall"
  39. ;; "http://www.reqall.com/user/feeds/rss/a1b2c3....."
  40. ;; "~/org/feeds.org" "ReQall Entries")
  41. ;;
  42. ;; With this setup, the command `M-x org-feed-update-all' will
  43. ;; collect new entries in the feed at the given URL and create
  44. ;; entries as subheadings under the "ReQall Entries" heading in the
  45. ;; file "~/org/feeds.org". Each feed should normally have its own
  46. ;; heading - however see the `:drawer' parameter.
  47. ;;
  48. ;; Besides these standard elements that need to be specified for each
  49. ;; feed, keyword-value pairs can set additional options. For example,
  50. ;; to de-select transitional entries with a title containing
  51. ;;
  52. ;; "reQall is typing what you said",
  53. ;;
  54. ;; you could use the `:filter' argument:
  55. ;;
  56. ;; (setq org-feed-alist
  57. ;; '(("ReQall"
  58. ;; "http://www.reqall.com/user/feeds/rss/a1b2c3....."
  59. ;; "~/org/feeds.org" "ReQall Entries"
  60. ;; :filter my-reqall-filter)))
  61. ;;
  62. ;; (defun my-reqall-filter (e)
  63. ;; (if (string-match "reQall is typing what you said"
  64. ;; (plist-get e :title))
  65. ;; nil
  66. ;; e))
  67. ;;
  68. ;; See the docstring for `org-feed-alist' for more details.
  69. ;;
  70. ;;
  71. ;; Keeping track of previously added entries
  72. ;; -----------------------------------------
  73. ;;
  74. ;; Since Org allows you to delete, archive, or move outline nodes,
  75. ;; org-feed.el needs to keep track of which feed items have been handled
  76. ;; before, so that they will not be handled again. For this, org-feed.el
  77. ;; stores information in a special drawer, FEEDSTATUS, under the heading
  78. ;; that received the input of the feed. You should add FEEDSTATUS
  79. ;; to your list of drawers in the files that receive feed input:
  80. ;;
  81. ;; #+DRAWERS: PROPERTIES CLOCK LOGBOOK RESULTS FEEDSTATUS
  82. ;;
  83. ;; Acknowledgments
  84. ;; ---------------
  85. ;;
  86. ;; org-feed.el is based on ideas by Brad Bozarth who implemented a
  87. ;; similar mechanism using shell and awk scripts.
  88. ;;; Code:
  89. (require 'org)
  90. (require 'sha1)
  91. (declare-function url-retrieve-synchronously "url" (url))
  92. (declare-function xml-node-children "xml" (node))
  93. (declare-function xml-get-children "xml" (node child-name))
  94. (declare-function xml-get-attribute "xml" (node attribute))
  95. (declare-function xml-get-attribute-or-nil "xml" (node attribute))
  96. (declare-function xml-substitute-special "xml" (string))
  97. (declare-function org-capture-escaped-% "org-capture" ())
  98. (declare-function org-capture-inside-embedded-elisp-p "org-capture" ())
  99. (declare-function org-capture-expand-embedded-elisp "org-capture" ())
  100. (defgroup org-feed nil
  101. "Options concerning RSS feeds as inputs for Org files."
  102. :tag "Org Feed"
  103. :group 'org)
  104. (defcustom org-feed-alist nil
  105. "Alist specifying RSS feeds that should create inputs for Org.
  106. Each entry in this list specified an RSS feed tat should be queried
  107. to create inbox items in Org. Each entry is a list with the following items:
  108. name a custom name for this feed
  109. URL the Feed URL
  110. file the target Org file where entries should be listed
  111. headline the headline under which entries should be listed
  112. Additional arguments can be given using keyword-value pairs. Many of these
  113. specify functions that receive one or a list of \"entries\" as their single
  114. argument. An entry is a property list that describes a feed item. The
  115. property list has properties for each field in the item, for example `:title'
  116. for the `<title>' field and `:pubDate' for the publication date. In addition,
  117. it contains the following properties:
  118. `:item-full-text' the full text in the <item> tag
  119. `:guid-permalink' t when the guid property is a permalink
  120. Here are the keyword-value pair allows in `org-feed-alist'.
  121. :drawer drawer-name
  122. The name of the drawer for storing feed information. The default is
  123. \"FEEDSTATUS\". Using different drawers for different feeds allows
  124. several feeds to target the same inbox heading.
  125. :filter filter-function
  126. A function to select interesting entries in the feed. It gets a single
  127. entry as parameter. It should return the entry if it is relevant, or
  128. nil if it is not.
  129. :template template-string
  130. The default action on new items in the feed is to add them as children
  131. under the headline for the feed. The template describes how the entry
  132. should be formatted. If not given, it defaults to
  133. `org-feed-default-template'.
  134. :formatter formatter-function
  135. Instead of relying on a template, you may specify a function to format
  136. the outline node to be inserted as a child. This function gets passed
  137. a property list describing a single feed item, and it should return a
  138. string that is a properly formatted Org outline node of level 1.
  139. :new-handler function
  140. If adding new items as children to the outline is not what you want
  141. to do with new items, define a handler function that is called with
  142. a list of all new items in the feed, each one represented as a property
  143. list. The handler should do what needs to be done, and org-feed will
  144. mark all items given to this handler as \"handled\", i.e. they will not
  145. be passed to this handler again in future readings of the feed.
  146. When the handler is called, point will be at the feed headline.
  147. :changed-handler function
  148. This function gets passed a list of all entries that have been
  149. handled before, but are now still in the feed and have *changed*
  150. since last handled (as evidenced by a different sha1 hash).
  151. When the handler is called, point will be at the feed headline.
  152. :parse-feed function
  153. This function gets passed a buffer, and should return a list
  154. of entries, each being a property list containing the
  155. `:guid' and `:item-full-text' keys. The default is
  156. `org-feed-parse-rss-feed'; `org-feed-parse-atom-feed' is an
  157. alternative.
  158. :parse-entry function
  159. This function gets passed an entry as returned by the parse-feed
  160. function, and should return the entry with interesting properties added.
  161. The default is `org-feed-parse-rss-entry'; `org-feed-parse-atom-entry'
  162. is an alternative."
  163. :group 'org-feed
  164. :type '(repeat
  165. (list :value ("" "http://" "" "")
  166. (string :tag "Name")
  167. (string :tag "Feed URL")
  168. (file :tag "File for inbox")
  169. (string :tag "Headline for inbox")
  170. (repeat :inline t
  171. (choice
  172. (list :inline t :tag "Filter"
  173. (const :filter)
  174. (symbol :tag "Filter Function"))
  175. (list :inline t :tag "Template"
  176. (const :template)
  177. (string :tag "Template"))
  178. (list :inline t :tag "Formatter"
  179. (const :formatter)
  180. (symbol :tag "Formatter Function"))
  181. (list :inline t :tag "New items handler"
  182. (const :new-handler)
  183. (symbol :tag "Handler Function"))
  184. (list :inline t :tag "Changed items"
  185. (const :changed-handler)
  186. (symbol :tag "Handler Function"))
  187. (list :inline t :tag "Parse Feed"
  188. (const :parse-feed)
  189. (symbol :tag "Parse Feed Function"))
  190. (list :inline t :tag "Parse Entry"
  191. (const :parse-entry)
  192. (symbol :tag "Parse Entry Function"))
  193. )))))
  194. (defcustom org-feed-drawer "FEEDSTATUS"
  195. "The name of the drawer for feed status information.
  196. Each feed may also specify its own drawer name using the `:drawer'
  197. parameter in `org-feed-alist'."
  198. :group 'org-feed
  199. :type '(string :tag "Drawer Name"))
  200. (defcustom org-feed-default-template "\n* %h\n %U\n %description\n %a\n"
  201. "Template for the Org node created from RSS feed items.
  202. This is just the default, each feed can specify its own.
  203. Any fields from the feed item can be interpolated into the template with
  204. %name, for example %title, %description, %pubDate etc. In addition, the
  205. following special escapes are valid as well:
  206. %h The title, or the first line of the description
  207. %t The date as a stamp, either from <pubDate> (if present), or
  208. the current date
  209. %T Date and time
  210. %u,%U Like %t,%T, but inactive time stamps
  211. %a A link, from <guid> if that is a permalink, else from <link>
  212. %(sexp) Evaluate elisp `(sexp)' and replace with the result, the simple
  213. %-escapes above can be used as arguments, e.g. %(capitalize \\\"%h\\\")"
  214. :group 'org-feed
  215. :type '(string :tag "Template"))
  216. (defcustom org-feed-save-after-adding t
  217. "Non-nil means save buffer after adding new feed items."
  218. :group 'org-feed
  219. :type 'boolean)
  220. (defcustom org-feed-retrieve-method 'url-retrieve-synchronously
  221. "The method to be used to retrieve a feed URL.
  222. This can be `curl' or `wget' to call these external programs, or it can be
  223. an Emacs Lisp function that will return a buffer containing the content
  224. of the file pointed to by the URL."
  225. :group 'org-feed
  226. :type '(choice
  227. (const :tag "Internally with url.el" url-retrieve-synchronously)
  228. (const :tag "Externally with curl" curl)
  229. (const :tag "Externally with wget" wget)
  230. (function :tag "Function")))
  231. (defcustom org-feed-before-adding-hook nil
  232. "Hook that is run before adding new feed items to a file.
  233. You might want to commit the file in its current state to version control,
  234. for example."
  235. :group 'org-feed
  236. :type 'hook)
  237. (defcustom org-feed-after-adding-hook nil
  238. "Hook that is run after new items have been added to a file.
  239. Depending on `org-feed-save-after-adding', the buffer will already
  240. have been saved."
  241. :group 'org-feed
  242. :type 'hook)
  243. (defvar org-feed-buffer "*Org feed*"
  244. "The buffer used to retrieve a feed.")
  245. ;;;###autoload
  246. (defun org-feed-update-all ()
  247. "Get inbox items from all feeds in `org-feed-alist'."
  248. (interactive)
  249. (let ((nfeeds (length org-feed-alist))
  250. (nnew (apply '+ (mapcar 'org-feed-update org-feed-alist))))
  251. (message "%s from %d %s"
  252. (cond ((= nnew 0) "No new entries")
  253. ((= nnew 1) "1 new entry")
  254. (t (format "%d new entries" nnew)))
  255. nfeeds
  256. (if (= nfeeds 1) "feed" "feeds"))))
  257. ;;;###autoload
  258. (defun org-feed-update (feed &optional retrieve-only)
  259. "Get inbox items from FEED.
  260. FEED can be a string with an association in `org-feed-alist', or
  261. it can be a list structured like an entry in `org-feed-alist'."
  262. (interactive (list (org-completing-read "Feed name: " org-feed-alist)))
  263. (if (stringp feed) (setq feed (assoc feed org-feed-alist)))
  264. (unless feed
  265. (error "No such feed in `org-feed-alist"))
  266. (catch 'exit
  267. (let ((name (car feed))
  268. (url (nth 1 feed))
  269. (file (nth 2 feed))
  270. (headline (nth 3 feed))
  271. (filter (nth 1 (memq :filter feed)))
  272. (formatter (nth 1 (memq :formatter feed)))
  273. (new-handler (nth 1 (memq :new-handler feed)))
  274. (changed-handler (nth 1 (memq :changed-handler feed)))
  275. (template (or (nth 1 (memq :template feed))
  276. org-feed-default-template))
  277. (drawer (or (nth 1 (memq :drawer feed))
  278. org-feed-drawer))
  279. (parse-feed (or (nth 1 (memq :parse-feed feed))
  280. 'org-feed-parse-rss-feed))
  281. (parse-entry (or (nth 1 (memq :parse-entry feed))
  282. 'org-feed-parse-rss-entry))
  283. feed-buffer inbox-pos new-formatted
  284. entries old-status status new changed guid-alist e guid olds)
  285. (setq feed-buffer (org-feed-get-feed url))
  286. (unless (and feed-buffer (bufferp (get-buffer feed-buffer)))
  287. (error "Cannot get feed %s" name))
  288. (when retrieve-only
  289. (throw 'exit feed-buffer))
  290. (setq entries (funcall parse-feed feed-buffer))
  291. (ignore-errors (kill-buffer feed-buffer))
  292. (save-excursion
  293. (save-window-excursion
  294. (setq inbox-pos (org-feed-goto-inbox-internal file headline))
  295. (setq old-status (org-feed-read-previous-status inbox-pos drawer))
  296. ;; Add the "handled" status to the appropriate entries
  297. (setq entries (mapcar (lambda (e)
  298. (setq e
  299. (plist-put e :handled
  300. (nth 1 (assoc
  301. (plist-get e :guid)
  302. old-status)))))
  303. entries))
  304. ;; Find out which entries are new and which are changed
  305. (dolist (e entries)
  306. (if (not (plist-get e :handled))
  307. (push e new)
  308. (setq olds (nth 2 (assoc (plist-get e :guid) old-status)))
  309. (if (and olds
  310. (not (string= (sha1
  311. (plist-get e :item-full-text))
  312. olds)))
  313. (push e changed))))
  314. ;; Parse the relevant entries fully
  315. (setq new (mapcar parse-entry new)
  316. changed (mapcar parse-entry changed))
  317. ;; Run the filter
  318. (when filter
  319. (setq new (delq nil (mapcar filter new))
  320. changed (delq nil (mapcar filter new))))
  321. (when (not (or new changed))
  322. (message "No new items in feed %s" name)
  323. (throw 'exit 0))
  324. ;; Get alist based on guid, to look up entries
  325. (setq guid-alist
  326. (append
  327. (mapcar (lambda (e) (list (plist-get e :guid) e)) new)
  328. (mapcar (lambda (e) (list (plist-get e :guid) e)) changed)))
  329. ;; Construct the new status
  330. (setq status
  331. (mapcar
  332. (lambda (e)
  333. (setq guid (plist-get e :guid))
  334. (list guid
  335. ;; things count as handled if we handle them now,
  336. ;; or if they were handled previously
  337. (if (assoc guid guid-alist) t (plist-get e :handled))
  338. ;; A hash, to detect changes
  339. (sha1 (plist-get e :item-full-text))))
  340. entries))
  341. ;; Handle new items in the feed
  342. (when new
  343. (if new-handler
  344. (progn
  345. (goto-char inbox-pos)
  346. (funcall new-handler new))
  347. ;; No custom handler, do the default adding
  348. ;; Format the new entries into an alist with GUIDs in the car
  349. (setq new-formatted
  350. (mapcar
  351. (lambda (e) (org-feed-format-entry e template formatter))
  352. new)))
  353. ;; Insert the new items
  354. (org-feed-add-items inbox-pos new-formatted))
  355. ;; Handle changed items in the feed
  356. (when (and changed-handler changed)
  357. (goto-char inbox-pos)
  358. (funcall changed-handler changed))
  359. ;; Write the new status
  360. ;; We do this only now, in case something goes wrong above, so
  361. ;; that would would end up with a status that does not reflect
  362. ;; which items truely have been handled
  363. (org-feed-write-status inbox-pos drawer status)
  364. ;; Normalize the visibility of the inbox tree
  365. (goto-char inbox-pos)
  366. (hide-subtree)
  367. (show-children)
  368. (org-cycle-hide-drawers 'children)
  369. ;; Hooks and messages
  370. (when org-feed-save-after-adding (save-buffer))
  371. (message "Added %d new item%s from feed %s to file %s, heading %s"
  372. (length new) (if (> (length new) 1) "s" "")
  373. name
  374. (file-name-nondirectory file) headline)
  375. (run-hooks 'org-feed-after-adding-hook)
  376. (length new))))))
  377. ;;;###autoload
  378. (defun org-feed-goto-inbox (feed)
  379. "Go to the inbox that captures the feed named FEED."
  380. (interactive
  381. (list (if (= (length org-feed-alist) 1)
  382. (car org-feed-alist)
  383. (org-completing-read "Feed name: " org-feed-alist))))
  384. (if (stringp feed) (setq feed (assoc feed org-feed-alist)))
  385. (unless feed
  386. (error "No such feed in `org-feed-alist"))
  387. (org-feed-goto-inbox-internal (nth 2 feed) (nth 3 feed)))
  388. ;;;###autoload
  389. (defun org-feed-show-raw-feed (feed)
  390. "Show the raw feed buffer of a feed."
  391. (interactive
  392. (list (if (= (length org-feed-alist) 1)
  393. (car org-feed-alist)
  394. (org-completing-read "Feed name: " org-feed-alist))))
  395. (if (stringp feed) (setq feed (assoc feed org-feed-alist)))
  396. (unless feed
  397. (error "No such feed in `org-feed-alist"))
  398. (org-pop-to-buffer-same-window
  399. (org-feed-update feed 'retrieve-only))
  400. (goto-char (point-min)))
  401. (defun org-feed-goto-inbox-internal (file heading)
  402. "Find or create HEADING in FILE.
  403. Switch to that buffer, and return the position of that headline."
  404. (find-file file)
  405. (widen)
  406. (goto-char (point-min))
  407. (if (re-search-forward
  408. (concat "^\\*+[ \t]+" heading "[ \t]*\\(:.*?:[ \t]*\\)?$")
  409. nil t)
  410. (goto-char (match-beginning 0))
  411. (goto-char (point-max))
  412. (insert "\n\n* " heading "\n\n")
  413. (org-back-to-heading t))
  414. (point))
  415. (defun org-feed-read-previous-status (pos drawer)
  416. "Get the alist of old GUIDs from the entry at POS.
  417. This will find DRAWER and extract the alist."
  418. (save-excursion
  419. (goto-char pos)
  420. (let ((end (save-excursion (org-end-of-subtree t t))))
  421. (if (re-search-forward
  422. (concat "^[ \t]*:" drawer ":[ \t]*\n\\([^\000]*?\\)\n[ \t]*:END:")
  423. end t)
  424. (read (match-string 1))
  425. nil))))
  426. (defun org-feed-write-status (pos drawer status)
  427. "Write the feed STATUS to DRAWER in entry at POS."
  428. (save-excursion
  429. (goto-char pos)
  430. (let ((end (save-excursion (org-end-of-subtree t t)))
  431. guid)
  432. (if (re-search-forward (concat "^[ \t]*:" drawer ":[ \t]*\n")
  433. end t)
  434. (progn
  435. (goto-char (match-end 0))
  436. (delete-region (point)
  437. (save-excursion
  438. (and (re-search-forward "^[ \t]*:END:" nil t)
  439. (match-beginning 0)))))
  440. (outline-next-heading)
  441. (insert " :" drawer ":\n :END:\n")
  442. (beginning-of-line 0))
  443. (insert (pp-to-string status)))))
  444. (defun org-feed-add-items (pos entries)
  445. "Add the formatted items to the headline as POS."
  446. (let (entry level)
  447. (save-excursion
  448. (goto-char pos)
  449. (unless (looking-at org-complex-heading-regexp)
  450. (error "Wrong position"))
  451. (setq level (org-get-valid-level (length (match-string 1)) 1))
  452. (org-end-of-subtree t t)
  453. (skip-chars-backward " \t\n")
  454. (beginning-of-line 2)
  455. (setq pos (point))
  456. (while (setq entry (pop entries))
  457. (org-paste-subtree level entry 'yank))
  458. (org-mark-ring-push pos))))
  459. (defun org-feed-format-entry (entry template formatter)
  460. "Format ENTRY so that it can be inserted into an Org file.
  461. ENTRY is a property list. This function adds a `:formatted-for-org' property
  462. and returns the full property list.
  463. If that property is already present, nothing changes."
  464. (require 'org-capture)
  465. (if formatter
  466. (funcall formatter entry)
  467. (let (dlines time escape name tmp
  468. v-h v-t v-T v-u v-U v-a)
  469. (setq dlines (org-split-string (or (plist-get entry :description) "???")
  470. "\n")
  471. v-h (or (plist-get entry :title) (car dlines) "???")
  472. time (or (if (plist-get entry :pubDate)
  473. (org-read-date t t (plist-get entry :pubDate)))
  474. (current-time))
  475. v-t (format-time-string (org-time-stamp-format nil nil) time)
  476. v-T (format-time-string (org-time-stamp-format t nil) time)
  477. v-u (format-time-string (org-time-stamp-format nil t) time)
  478. v-U (format-time-string (org-time-stamp-format t t) time)
  479. v-a (if (setq tmp (or (and (plist-get entry :guid-permalink)
  480. (plist-get entry :guid))
  481. (plist-get entry :link)))
  482. (concat "[[" tmp "]]\n")
  483. ""))
  484. (with-temp-buffer
  485. (insert template)
  486. ;; Simple %-escapes
  487. ;; before embedded elisp to support simple %-escapes as
  488. ;; arguments for embedded elisp
  489. (goto-char (point-min))
  490. (while (re-search-forward "%\\([a-zA-Z]+\\)" nil t)
  491. (unless (org-capture-escaped-%)
  492. (setq name (match-string 1)
  493. escape (org-capture-inside-embedded-elisp-p))
  494. (cond
  495. ((member name '("h" "t" "T" "u" "U" "a"))
  496. (setq tmp (symbol-value (intern (concat "v-" name)))))
  497. ((setq tmp (plist-get entry (intern (concat ":" name))))
  498. (save-excursion
  499. (save-match-data
  500. (beginning-of-line 1)
  501. (when (looking-at
  502. (concat "^\\([ \t]*\\)%" name "[ \t]*$"))
  503. (setq tmp (org-feed-make-indented-block
  504. tmp (org-get-indentation))))))))
  505. (when tmp
  506. ;; escape string delimiters `"' when inside %() embedded lisp
  507. (when escape
  508. (setq tmp (replace-regexp-in-string "\"" "\\\\\"" tmp)))
  509. (replace-match tmp t t))))
  510. ;; %() embedded elisp
  511. (org-capture-expand-embedded-elisp)
  512. (decode-coding-string
  513. (buffer-string) (detect-coding-region (point-min) (point-max) t))))))
  514. (defun org-feed-make-indented-block (s n)
  515. "Add indentation of N spaces to a multiline string S."
  516. (if (not (string-match "\n" s))
  517. s
  518. (mapconcat 'identity
  519. (org-split-string s "\n")
  520. (concat "\n" (make-string n ?\ )))))
  521. (defun org-feed-skip-http-headers (buffer)
  522. "Remove HTTP headers from BUFFER, and return it.
  523. Assumes headers are indeed present!"
  524. (with-current-buffer buffer
  525. (widen)
  526. (goto-char (point-min))
  527. (search-forward "\n\n")
  528. (delete-region (point-min) (point))
  529. buffer))
  530. (defun org-feed-get-feed (url)
  531. "Get the RSS feed file at URL and return the buffer."
  532. (cond
  533. ((eq org-feed-retrieve-method 'url-retrieve-synchronously)
  534. (org-feed-skip-http-headers (url-retrieve-synchronously url)))
  535. ((eq org-feed-retrieve-method 'curl)
  536. (ignore-errors (kill-buffer org-feed-buffer))
  537. (call-process "curl" nil org-feed-buffer nil "--silent" url)
  538. org-feed-buffer)
  539. ((eq org-feed-retrieve-method 'wget)
  540. (ignore-errors (kill-buffer org-feed-buffer))
  541. (call-process "wget" nil org-feed-buffer nil "-q" "-O" "-" url)
  542. org-feed-buffer)
  543. ((functionp org-feed-retrieve-method)
  544. (funcall org-feed-retrieve-method url))))
  545. (defun org-feed-parse-rss-feed (buffer)
  546. "Parse BUFFER for RSS feed entries.
  547. Returns a list of entries, with each entry a property list,
  548. containing the properties `:guid' and `:item-full-text'."
  549. (let ((case-fold-search t)
  550. entries beg end item guid entry)
  551. (with-current-buffer buffer
  552. (widen)
  553. (goto-char (point-min))
  554. (while (re-search-forward "<item\\>.*?>" nil t)
  555. (setq beg (point)
  556. end (and (re-search-forward "</item>" nil t)
  557. (match-beginning 0)))
  558. (setq item (buffer-substring beg end)
  559. guid (if (string-match "<guid\\>.*?>\\(.*?\\)</guid>" item)
  560. (org-match-string-no-properties 1 item)))
  561. (setq entry (list :guid guid :item-full-text item))
  562. (push entry entries)
  563. (widen)
  564. (goto-char end))
  565. (nreverse entries))))
  566. (defun org-feed-parse-rss-entry (entry)
  567. "Parse the `:item-full-text' field for xml tags and create new properties."
  568. (require 'xml)
  569. (with-temp-buffer
  570. (insert (plist-get entry :item-full-text))
  571. (goto-char (point-min))
  572. (while (re-search-forward "<\\([a-zA-Z]+\\>\\).*?>\\([^\000]*?\\)</\\1>"
  573. nil t)
  574. (setq entry (plist-put entry
  575. (intern (concat ":" (match-string 1)))
  576. (xml-substitute-special (match-string 2)))))
  577. (goto-char (point-min))
  578. (unless (re-search-forward "isPermaLink[ \t]*=[ \t]*\"false\"" nil t)
  579. (setq entry (plist-put entry :guid-permalink t))))
  580. entry)
  581. (defun org-feed-parse-atom-feed (buffer)
  582. "Parse BUFFER for Atom feed entries.
  583. Returns a list of entries, with each entry a property list,
  584. containing the properties `:guid' and `:item-full-text'.
  585. The `:item-full-text' property actually contains the sexp
  586. formatted as a string, not the original XML data."
  587. (require 'xml)
  588. (with-current-buffer buffer
  589. (widen)
  590. (let ((feed (car (xml-parse-region (point-min) (point-max)))))
  591. (mapcar
  592. (lambda (entry)
  593. (list
  594. :guid (car (xml-node-children (car (xml-get-children entry 'id))))
  595. :item-full-text (prin1-to-string entry)))
  596. (xml-get-children feed 'entry)))))
  597. (defun org-feed-parse-atom-entry (entry)
  598. "Parse the `:item-full-text' as a sexp and create new properties."
  599. (let ((xml (car (read-from-string (plist-get entry :item-full-text)))))
  600. ;; Get first <link href='foo'/>.
  601. (setq entry (plist-put entry :link
  602. (xml-get-attribute
  603. (car (xml-get-children xml 'link))
  604. 'href)))
  605. ;; Add <title/> as :title.
  606. (setq entry (plist-put entry :title
  607. (xml-substitute-special
  608. (car (xml-node-children
  609. (car (xml-get-children xml 'title)))))))
  610. (let* ((content (car (xml-get-children xml 'content)))
  611. (type (xml-get-attribute-or-nil content 'type)))
  612. (when content
  613. (cond
  614. ((string= type "text")
  615. ;; We like plain text.
  616. (setq entry (plist-put entry :description
  617. (xml-substitute-special
  618. (car (xml-node-children content))))))
  619. ((string= type "html")
  620. ;; TODO: convert HTML to Org markup.
  621. (setq entry (plist-put entry :description
  622. (xml-substitute-special
  623. (car (xml-node-children content))))))
  624. ((string= type "xhtml")
  625. ;; TODO: convert XHTML to Org markup.
  626. (setq entry (plist-put entry :description
  627. (prin1-to-string
  628. (xml-node-children content)))))
  629. (t
  630. (setq entry (plist-put entry :description
  631. (format "Unknown '%s' content." type)))))))
  632. entry))
  633. (provide 'org-feed)
  634. ;; Local variables:
  635. ;; generated-autoload-file: "org-loaddefs.el"
  636. ;; End:
  637. ;;; org-feed.el ends here