code.delx.au - gnu-emacs/blob - lisp/gnus/nnrss.el

   1 ;;; nnrss.el --- interfacing with RSS
   2
   3 ;; Copyright (C) 2001, 2002, 2003, 2004, 2005 Free Software Foundation, Inc.
   4
   5 ;; Author: Shenghuo Zhu <zsh@cs.rochester.edu>
   6 ;; Keywords: RSS
   7
   8 ;; This file is part of GNU Emacs.
   9
  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.
  14
  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.
  19
  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.
  24
  25 ;;; Commentary:
  26
  27 ;;; Code:
  28
  29 (eval-when-compile (require 'cl))
  30
  31 (require 'gnus)
  32 (require 'nnoo)
  33 (require 'nnmail)
  34 (require 'message)
  35 (require 'mm-util)
  36 (require 'gnus-util)
  37 (require 'time-date)
  38 (require 'rfc2231)
  39 (require 'mm-url)
  40 (require 'rfc2047)
  41 (require 'mml)
  42 (eval-when-compile
  43   (ignore-errors
  44    (require 'xml)))
  45 (eval '(require 'xml))
  46
  47 (nnoo-declare nnrss)
  48
  49 (defvoo nnrss-directory (nnheader-concat gnus-directory "rss/")
  50   "Where nnrss will save its files.")
  51
  52 ;; (group max rss-url)
  53 (defvoo nnrss-server-data nil)
  54
  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 nil)
  61 (defvoo nnrss-status-string "")
  62
  63 (defconst nnrss-version "nnrss 1.0")
  64
  65 (defvar nnrss-group-alist '()
  66   "List of RSS addresses.")
  67
  68 (defvar nnrss-use-local nil)
  69
  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'.")
  73
  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'.")
  77
  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.")
  83
  84 (defvar nnrss-file-coding-system mm-universal-coding-system
  85   "Coding system used when reading and writing files.")
  86
  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.")
  92
  93 (nnoo-define-basics nnrss)
  94
  95 ;;; Interface functions
  96
  97 (defsubst nnrss-format-string (string)
  98   (gnus-replace-in-string string " *\n *" " "))
  99
 100 (defun nnrss-decode-group-name (group)
 101   (if (and group (mm-coding-system-p 'utf-8))
 102       (setq group (mm-decode-coding-string group 'utf-8))
 103     group))
 104
 105 (deffoo nnrss-retrieve-headers (articles &optional group server fetch-old)
 106   (setq group (nnrss-decode-group-name group))
 107   (nnrss-possibly-change-group group server)
 108   (let (e)
 109     (save-excursion
 110       (set-buffer nntp-server-buffer)
 111       (erase-buffer)
 112       (dolist (article articles)
 113         (if (setq e (assq article nnrss-group-data))
 114             (insert (number-to-string (car e)) "\t" ;; number
 115                     ;; subject
 116                     (or (nth 3 e) "")
 117                     "\t"
 118                     ;; from
 119                     (or (nth 4 e) "(nobody)")
 120                     "\t"
 121                     ;; date
 122                     (or (nth 5 e) "")
 123                     "\t"
 124                     ;; id
 125                     (format "<%d@%s.nnrss>" (car e) group)
 126                     "\t"
 127                     ;; refs
 128                     "\t"
 129                     ;; chars
 130                     "-1" "\t"
 131                     ;; lines
 132                     "-1" "\t"
 133                     ;; Xref
 134                     "" "\t"
 135                     (if (and (nth 6 e)
 136                              (memq nnrss-description-field
 137                                    nnmail-extra-headers))
 138                         (concat (symbol-name nnrss-description-field)
 139                                 ": "
 140                                 (nnrss-format-string (nth 6 e))
 141                                 "\t")
 142                       "")
 143                     (if (and (nth 2 e)
 144                              (memq nnrss-url-field
 145                                    nnmail-extra-headers))
 146                         (concat (symbol-name nnrss-url-field)
 147                                 ": "
 148                                 (nnrss-format-string (nth 2 e))
 149                                 "\t")
 150                       "")
 151                     "\n")))))
 152   'nov)
 153
 154 (deffoo nnrss-request-group (group &optional server dont-check)
 155   (setq group (nnrss-decode-group-name group))
 156   (nnheader-message 6 "nnrss: Requesting %s..." group)
 157   (nnrss-possibly-change-group group server)
 158   (prog1
 159       (if dont-check
 160           t
 161         (nnrss-check-group group server)
 162         (nnheader-report 'nnrss "Opened group %s" group)
 163         (nnheader-insert
 164          "211 %d %d %d %s\n" nnrss-group-max nnrss-group-min nnrss-group-max
 165          (prin1-to-string group)
 166          t))
 167     (nnheader-message 6 "nnrss: Requesting %s...done" group)))
 168
 169 (deffoo nnrss-close-group (group &optional server)
 170   t)
 171
 172 (deffoo nnrss-request-article (article &optional group server buffer)
 173   (setq group (nnrss-decode-group-name group))
 174   (when (stringp article)
 175     (setq article (if (string-match "\\`<\\([0-9]+\\)@" article)
 176                       (string-to-number (match-string 1 article))
 177                     0)))
 178   (nnrss-possibly-change-group group server)
 179   (let ((e (assq article nnrss-group-data))
 180         (nntp-server-buffer (or buffer nntp-server-buffer))
 181         post err)
 182     (when e
 183       (with-current-buffer nntp-server-buffer
 184         (erase-buffer)
 185         (if group
 186             (insert "Newsgroups: " group "\n"))
 187         (if (nth 3 e)
 188             (insert "Subject: " (nth 3 e) "\n"))
 189         (if (nth 4 e)
 190             (insert "From: " (nth 4 e) "\n"))
 191         (if (nth 5 e)
 192             (insert "Date: " (nnrss-format-string (nth 5 e)) "\n"))
 193         (let ((header (buffer-string))
 194               (text (if (nth 6 e)
 195                         (mapconcat 'identity
 196                                    (delete "" (split-string (nth 6 e) "\n+"))
 197                                    " ")))
 198               (link (nth 2 e))
 199               (enclosure (nth 7 e))
 200               ;; Enable encoding of Newsgroups header in XEmacs.
 201               (default-enable-multibyte-characters t)
 202               (rfc2047-header-encoding-alist
 203                (if (mm-coding-system-p 'utf-8)
 204                    (cons '("Newsgroups" . utf-8)
 205                          rfc2047-header-encoding-alist)
 206                  rfc2047-header-encoding-alist))
 207               rfc2047-encode-encoded-words body)
 208           (when (or text link enclosure)
 209             (insert "\n")
 210             (insert "<#multipart type=alternative>\n"
 211                     "<#part type=\"text/plain\">\n")
 212             (setq body (point))
 213             (when text
 214               (insert text "\n")
 215               (when (or link enclosure)
 216                 (insert "\n")))
 217             (when link
 218               (insert link "\n"))
 219             (when enclosure
 220               (insert (car enclosure) " "
 221                       (nth 2 enclosure) " "
 222                       (nth 3 enclosure) "\n"))
 223             (setq body (buffer-substring body (point)))
 224             (insert "<#/part>\n"
 225                     "<#part type=\"text/html\">\n"
 226                     "<html><head></head><body>\n")
 227             (when text
 228               (insert text "\n"))
 229             (when link
 230               (insert "<p><a href=\"" link "\">link</a></p>\n"))
 231             (when enclosure
 232               (insert "<p><a href=\"" (car enclosure) "\">"
 233                       (cadr enclosure) "</a> " (nth 2 enclosure)
 234                       " " (nth 3 enclosure) "</p>\n"))
 235             (insert "</body></html>\n"
 236                     "<#/part>\n"
 237                     "<#/multipart>\n"))
 238           (condition-case nil
 239               (mml-to-mime)
 240             (error
 241              (erase-buffer)
 242              (insert header
 243                      "Content-Type: text/plain; charset=gnus-decoded\n"
 244                      "Content-Transfer-Encoding: 8bit\n\n"
 245                      body)
 246              (nnheader-message
 247               3 "Warning - there might be invalid characters"))))
 248         (goto-char (point-min))
 249         (search-forward "\n\n")
 250         (forward-line -1)
 251         (insert (format "Message-ID: <%d@%s.nnrss>\n"
 252                         (car e)
 253                         (let ((rfc2047-encoding-type 'mime)
 254                               rfc2047-encode-max-chars)
 255                           (rfc2047-encode-string
 256                            (gnus-replace-in-string group "[\t\n ]+" "_")))))
 257         (when nnrss-content-function
 258           (funcall nnrss-content-function e group article))))
 259     (cond
 260      (err
 261       (nnheader-report 'nnrss err))
 262      ((not e)
 263       (nnheader-report 'nnrss "no such id: %d" article))
 264      (t
 265       (nnheader-report 'nnrss "article %s retrieved" (car e))
 266       ;; we return the article number.
 267       (cons nnrss-group (car e))))))
 268
 269 (deffoo nnrss-request-list (&optional server)
 270   (nnrss-possibly-change-group nil server)
 271   (nnrss-generate-active)
 272   t)
 273
 274 (deffoo nnrss-open-server (server &optional defs connectionless)
 275   (nnrss-read-server-data server)
 276   (nnoo-change-server 'nnrss server defs)
 277   t)
 278
 279 (deffoo nnrss-request-expire-articles
 280     (articles group &optional server force)
 281   (setq group (nnrss-decode-group-name group))
 282   (nnrss-possibly-change-group group server)
 283   (let (e days not-expirable changed)
 284     (dolist (art articles)
 285       (if (and (setq e (assq art nnrss-group-data))
 286                (nnmail-expired-article-p
 287                 group
 288                 (if (listp (setq days (nth 1 e))) days
 289                   (days-to-time (- days (time-to-days '(0 0)))))
 290                 force))
 291           (setq nnrss-group-data (delq e nnrss-group-data)
 292                 changed t)
 293         (push art not-expirable)))
 294     (if changed
 295         (nnrss-save-group-data group server))
 296     not-expirable))
 297
 298 (deffoo nnrss-request-delete-group (group &optional force server)
 299   (setq group (nnrss-decode-group-name group))
 300   (nnrss-possibly-change-group group server)
 301   (let (elem)
 302     ;; There may be two or more entries in `nnrss-group-alist' since
 303     ;; this function didn't delete them formerly.
 304     (while (setq elem (assoc group nnrss-group-alist))
 305       (setq nnrss-group-alist (delq elem nnrss-group-alist))))
 306   (setq nnrss-server-data
 307         (delq (assoc group nnrss-server-data) nnrss-server-data))
 308   (nnrss-save-server-data server)
 309   (ignore-errors
 310    (delete-file (nnrss-make-filename group server)))
 311   t)
 312
 313 (deffoo nnrss-request-list-newsgroups (&optional server)
 314   (nnrss-possibly-change-group nil server)
 315   (save-excursion
 316     (set-buffer nntp-server-buffer)
 317     (erase-buffer)
 318     (dolist (elem nnrss-group-alist)
 319       (if (third elem)
 320           (insert (car elem) "\t" (third elem) "\n"))))
 321   t)
 322
 323 (nnoo-define-skeleton nnrss)
 324
 325 ;;; Internal functions
 326 (eval-when-compile (defun xml-rpc-method-call (&rest args)))
 327
 328 (defun nnrss-get-encoding ()
 329   "Return an encoding attribute specified in the current xml contents.
 330 If `nnrss-compatible-encoding-alist' specifies the compatible encoding,
 331 it is used instead.  If the xml contents doesn't specify the encoding,
 332 return `utf-8' which is the default encoding for xml if it is available,
 333 otherwise return nil."
 334   (goto-char (point-min))
 335   (if (re-search-forward
 336        "<\\?[^>]*encoding=\\(\"\\([^\">]+\\)\"\\|'\\([^'>]+\\)'\\)"
 337        nil t)
 338       (let ((encoding (intern (downcase (or (match-string 2)
 339                                             (match-string 3))))))
 340         (or
 341          (mm-coding-system-p (cdr (assq encoding
 342                                         nnrss-compatible-encoding-alist)))
 343          (mm-coding-system-p encoding)
 344          (mm-coding-system-p (car (rassq encoding
 345                                          nnrss-compatible-encoding-alist)))))
 346     (mm-coding-system-p 'utf-8)))
 347
 348 (defun nnrss-fetch (url &optional local)
 349   "Fetch URL and put it in a the expected Lisp structure."
 350   (mm-with-unibyte-buffer
 351     ;;some CVS versions of url.el need this to close the connection quickly
 352     (let (cs xmlform htmlform)
 353       ;; bit o' work necessary for w3 pre-cvs and post-cvs
 354       (if local
 355           (let ((coding-system-for-read 'binary))
 356             (insert-file-contents url))
 357         ;; FIXME: shouldn't binding `coding-system-for-read' be moved
 358         ;; to `mm-url-insert'?
 359         (let ((coding-system-for-read 'binary))
 360           (mm-url-insert url)))
 361       (nnheader-remove-cr-followed-by-lf)
 362       ;; Decode text according to the encoding attribute.
 363       (when (setq cs (nnrss-get-encoding))
 364         (mm-decode-coding-region (point-min) (point-max) cs)
 365         (mm-enable-multibyte))
 366       (goto-char (point-min))
 367
 368       ;; Because xml-parse-region can't deal with anything that isn't
 369       ;; xml and w3-parse-buffer can't deal with some xml, we have to
 370       ;; parse with xml-parse-region first and, if that fails, parse
 371       ;; with w3-parse-buffer.  Yuck.  Eventually, someone should find out
 372       ;; why w3-parse-buffer fails to parse some well-formed xml and
 373       ;; fix it.
 374
 375       (condition-case err1
 376           (setq xmlform (xml-parse-region (point-min) (point-max)))
 377         (error
 378          (condition-case err2
 379              (setq htmlform (caddar (w3-parse-buffer
 380                                      (current-buffer))))
 381            (error
 382             (message "\
 383 nnrss: %s: Not valid XML %s and w3-parse doesn't work %s"
 384                      url err1 err2)))))
 385       (if htmlform
 386           htmlform
 387         xmlform))))
 388
 389 (defun nnrss-possibly-change-group (&optional group server)
 390   (when (and server
 391              (not (nnrss-server-opened server)))
 392     (nnrss-open-server server))
 393   (when (and group (not (equal group nnrss-group)))
 394     (nnrss-read-group-data group server)
 395     (setq nnrss-group group)))
 396
 397 (defvar nnrss-extra-categories '(nnrss-snarf-moreover-categories))
 398
 399 (defun nnrss-generate-active ()
 400   (when (y-or-n-p "Fetch extra categories? ")
 401     (dolist (func nnrss-extra-categories)
 402       (funcall func)))
 403   (save-excursion
 404     (set-buffer nntp-server-buffer)
 405     (erase-buffer)
 406     (dolist (elem nnrss-group-alist)
 407       (insert (prin1-to-string (car elem)) " 0 1 y\n"))
 408     (dolist (elem nnrss-server-data)
 409       (unless (assoc (car elem) nnrss-group-alist)
 410         (insert (prin1-to-string (car elem)) " 0 1 y\n")))))
 411
 412 ;;; data functions
 413
 414 (defun nnrss-read-server-data (server)
 415   (setq nnrss-server-data nil)
 416   (let ((file (nnrss-make-filename "nnrss" server)))
 417     (when (file-exists-p file)
 418       ;; In Emacs 21.3 and earlier, `load' doesn't support non-ASCII
 419       ;; file names.  So, we use `insert-file-contents' instead.
 420       (mm-with-multibyte-buffer
 421         (let ((coding-system-for-read nnrss-file-coding-system)
 422               (file-name-coding-system nnmail-pathname-coding-system))
 423           (insert-file-contents file)
 424           (eval-region (point-min) (point-max)))))))
 425
 426 (defun nnrss-save-server-data (server)
 427   (gnus-make-directory nnrss-directory)
 428   (let ((coding-system-for-write nnrss-file-coding-system)
 429         (file-name-coding-system nnmail-pathname-coding-system))
 430     (with-temp-file (nnrss-make-filename "nnrss" server)
 431       (insert (format ";; -*- coding: %s; -*-\n"
 432                       nnrss-file-coding-system))
 433       (gnus-prin1 `(setq nnrss-group-alist ',nnrss-group-alist))
 434       (insert "\n")
 435       (gnus-prin1 `(setq nnrss-server-data ',nnrss-server-data)))))
 436
 437 (defun nnrss-read-group-data (group server)
 438   (setq nnrss-group-data nil)
 439   (setq nnrss-group-hashtb (gnus-make-hashtable))
 440   (let ((pair (assoc group nnrss-server-data)))
 441     (setq nnrss-group-max (or (cadr pair) 0))
 442     (setq nnrss-group-min (+ nnrss-group-max 1)))
 443   (let ((file (nnrss-make-filename group server)))
 444     (when (file-exists-p file)
 445       ;; In Emacs 21.3 and earlier, `load' doesn't support non-ASCII
 446       ;; file names.  So, we use `insert-file-contents' instead.
 447       (mm-with-multibyte-buffer
 448         (let ((coding-system-for-read nnrss-file-coding-system)
 449               (file-name-coding-system nnmail-pathname-coding-system))
 450           (insert-file-contents file)
 451           (eval-region (point-min) (point-max))))
 452       (dolist (e nnrss-group-data)
 453         (gnus-sethash (or (nth 2 e) (nth 6 e)) t nnrss-group-hashtb)
 454         (when (and (car e) (> nnrss-group-min (car e)))
 455           (setq nnrss-group-min (car e)))
 456         (when (and (car e) (< nnrss-group-max (car e)))
 457           (setq nnrss-group-max (car e)))))))
 458
 459 (defun nnrss-save-group-data (group server)
 460   (gnus-make-directory nnrss-directory)
 461   (let ((coding-system-for-write nnrss-file-coding-system)
 462         (file-name-coding-system nnmail-pathname-coding-system))
 463     (with-temp-file (nnrss-make-filename group server)
 464       (insert (format ";; -*- coding: %s; -*-\n"
 465                       nnrss-file-coding-system))
 466       (gnus-prin1 `(setq nnrss-group-data ',nnrss-group-data)))))
 467
 468 (defun nnrss-make-filename (name server)
 469   (expand-file-name
 470    (nnrss-translate-file-chars
 471     (concat name
 472             (and server
 473                  (not (equal server ""))
 474                  "-")
 475             server
 476             ".el"))
 477    nnrss-directory))
 478
 479 (gnus-add-shutdown 'nnrss-close 'gnus)
 480
 481 (defun nnrss-close ()
 482   "Clear internal nnrss variables."
 483   (setq nnrss-group-data nil
 484         nnrss-server-data nil
 485         nnrss-group-hashtb nil
 486         nnrss-group-alist nil))
 487
 488 ;;; URL interface
 489
 490 (defun nnrss-no-cache (url)
 491   "")
 492
 493 (defun nnrss-insert-w3 (url)
 494   (mm-with-unibyte-current-buffer
 495     (mm-url-insert url)))
 496
 497 (defun nnrss-decode-entities-string (string)
 498   (if string
 499       (mm-with-multibyte-buffer
 500         (insert string)
 501         (mm-url-decode-entities-nbsp)
 502         (buffer-string))))
 503
 504 (defalias 'nnrss-insert 'nnrss-insert-w3)
 505
 506 (defun nnrss-mime-encode-string (string)
 507   (mm-with-multibyte-buffer
 508     (insert string)
 509     (mm-url-decode-entities-nbsp)
 510     (goto-char (point-min))
 511     (while (re-search-forward "[\t\n ]+" nil t)
 512       (replace-match " "))
 513     (goto-char (point-min))
 514     (skip-chars-forward " ")
 515     (delete-region (point-min) (point))
 516     (goto-char (point-max))
 517     (skip-chars-forward " ")
 518     (delete-region (point) (point-max))
 519     (let ((rfc2047-encoding-type 'mime)
 520           rfc2047-encode-max-chars)
 521       (rfc2047-encode-region (point-min) (point-max)))
 522     (goto-char (point-min))
 523     (while (search-forward "\n" nil t)
 524       (delete-backward-char 1))
 525     (buffer-string)))
 526
 527 ;;; Snarf functions
 528
 529 (defun nnrss-check-group (group server)
 530   (let (file xml subject url extra changed author date
 531              enclosure rss-ns rdf-ns content-ns dc-ns)
 532     (if (and nnrss-use-local
 533              (file-exists-p (setq file (expand-file-name
 534                                         (nnrss-translate-file-chars
 535                                          (concat group ".xml"))
 536                                         nnrss-directory))))
 537         (setq xml (nnrss-fetch file t))
 538       (setq url (or (nth 2 (assoc group nnrss-server-data))
 539                     (second (assoc group nnrss-group-alist))))
 540       (unless url
 541         (setq url
 542               (cdr
 543                (assoc 'href
 544                       (nnrss-discover-feed
 545                        (read-string
 546                         (format "URL to search for %s: " group) "http://")))))
 547         (let ((pair (assoc group nnrss-server-data)))
 548           (if pair
 549               (setcdr (cdr pair) (list url))
 550             (push (list group nnrss-group-max url) nnrss-server-data)))
 551         (setq changed t))
 552       (setq xml (nnrss-fetch url)))
 553     ;; See
 554     ;; http://feeds.archive.org/validator/docs/howto/declare_namespaces.html
 555     ;; for more RSS namespaces.
 556     (setq dc-ns (nnrss-get-namespace-prefix xml "http://purl.org/dc/elements/1.1/")
 557           rdf-ns (nnrss-get-namespace-prefix xml "http://www.w3.org/1999/02/22-rdf-syntax-ns#")
 558           rss-ns (nnrss-get-namespace-prefix xml "http://purl.org/rss/1.0/")
 559           content-ns (nnrss-get-namespace-prefix xml "http://purl.org/rss/1.0/modules/content/"))
 560     (dolist (item (nreverse (nnrss-find-el (intern (concat rss-ns "item")) xml)))
 561       (when (and (listp item)
 562                  (string= (concat rss-ns "item") (car item))
 563                  (if (setq url (nnrss-decode-entities-string
 564                                 (nnrss-node-text rss-ns 'link (cddr item))))
 565                      (not (gnus-gethash url nnrss-group-hashtb))
 566                    (setq extra (or (nnrss-node-text content-ns 'encoded item)
 567                                    (nnrss-node-text rss-ns 'description item)))
 568                    (not (gnus-gethash extra nnrss-group-hashtb))))
 569         (setq subject (nnrss-node-text rss-ns 'title item))
 570         (setq extra (or extra
 571                         (nnrss-node-text content-ns 'encoded item)
 572                         (nnrss-node-text rss-ns 'description item)))
 573         (setq author (or (nnrss-node-text rss-ns 'author item)
 574                          (nnrss-node-text dc-ns 'creator item)
 575                          (nnrss-node-text dc-ns 'contributor item)))
 576         (setq date (or (nnrss-node-text dc-ns 'date item)
 577                        (nnrss-node-text rss-ns 'pubDate item)
 578                        (message-make-date)))
 579         (when (setq enclosure (cadr (assq (intern (concat rss-ns "enclosure")) item)))
 580           (let ((url (cdr (assq 'url enclosure)))
 581                 (len (cdr (assq 'length enclosure)))
 582                 (type (cdr (assq 'type enclosure)))
 583                 (name))
 584             (setq len
 585                   (if (and len (integerp (setq len (string-to-number len))))
 586                       ;; actually already in `ls-lisp-format-file-size' but
 587                       ;; probably not worth to require it for one function
 588                       (do ((size (/ len 1.0) (/ size 1024.0))
 589                            (post-fixes (list "" "k" "M" "G" "T" "P" "E")
 590                                        (cdr post-fixes)))
 591                           ((< size 1024)
 592                            (format "%.1f%s" size (car post-fixes))))
 593                     "0"))
 594             (setq url (or url ""))
 595             (setq name (if (string-match "/\\([^/]*\\)$" url)
 596                            (match-string 1 url)
 597                          "file"))
 598             (setq type (or type ""))
 599             (setq enclosure (list url name len type))))
 600         (push
 601          (list
 602           (incf nnrss-group-max)
 603           (current-time)
 604           url
 605           (and subject (nnrss-mime-encode-string subject))
 606           (and author (nnrss-mime-encode-string author))
 607           date
 608           (and extra (nnrss-decode-entities-string extra))
 609           enclosure)
 610          nnrss-group-data)
 611         (gnus-sethash (or url extra) t nnrss-group-hashtb)
 612         (setq changed t))
 613       (setq extra nil))
 614     (when changed
 615       (nnrss-save-group-data group server)
 616       (let ((pair (assoc group nnrss-server-data)))
 617         (if pair
 618             (setcar (cdr pair) nnrss-group-max)
 619           (push (list group nnrss-group-max) nnrss-server-data)))
 620       (nnrss-save-server-data server))))
 621
 622 (defun nnrss-opml-import (opml-file)
 623   "OPML subscriptions import.
 624 Read the file and attempt to subscribe to each Feed in the file."
 625   (interactive "fImport file: ")
 626   (mapcar
 627    (lambda (node) (gnus-group-make-rss-group
 628                    (cdr (assq 'xmlUrl (cadr node)))))
 629    (nnrss-find-el 'outline
 630                   (progn
 631                     (find-file opml-file)
 632                     (xml-parse-region (point-min)
 633                                       (point-max))))))
 634
 635 (defun nnrss-opml-export ()
 636   "OPML subscription export.
 637 Export subscriptions to a buffer in OPML Format."
 638   (interactive)
 639   (with-current-buffer (get-buffer-create "*OPML Export*")
 640     (mm-set-buffer-file-coding-system 'utf-8)
 641     (insert "<?xml version=\"1.0\" encoding=\"utf-8\"?>\n"
 642             "<!-- OPML generated by Emacs Gnus' nnrss.el -->\n"
 643             "<opml version=\"1.1\">\n"
 644             "  <head>\n"
 645             "    <title>mySubscriptions</title>\n"
 646             "    <dateCreated>" (format-time-string "%a, %d %b %Y %T %z")
 647             "</dateCreated>\n"
 648             "    <ownerEmail>" user-mail-address "</ownerEmail>\n"
 649             "    <ownerName>" (user-full-name) "</ownerName>\n"
 650             "  </head>\n"
 651             "  <body>\n")
 652     (dolist (sub nnrss-group-alist)
 653       (insert "    <outline text=\"" (car sub)
 654               "\" xmlUrl=\"" (cadr sub) "\"/>\n"))
 655     (insert "  </body>\n"
 656             "</opml>\n"))
 657   (pop-to-buffer "*OPML Export*")
 658   (when (fboundp 'sgml-mode)
 659     (sgml-mode)))
 660
 661 (defun nnrss-generate-download-script ()
 662   "Generate a download script in the current buffer.
 663 It is useful when `(setq nnrss-use-local t)'."
 664   (interactive)
 665   (insert "#!/bin/sh\n")
 666   (insert "WGET=wget\n")
 667   (insert "RSSDIR='" (expand-file-name nnrss-directory) "'\n")
 668   (dolist (elem nnrss-server-data)
 669     (let ((url (or (nth 2 elem)
 670                    (second (assoc (car elem) nnrss-group-alist)))))
 671       (insert "$WGET -q -O \"$RSSDIR\"/'"
 672               (nnrss-translate-file-chars (concat (car elem) ".xml"))
 673               "' '" url "'\n"))))
 674
 675 (defun nnrss-translate-file-chars (name)
 676   (let ((nnheader-file-name-translation-alist
 677          (append nnheader-file-name-translation-alist '((?' . ?_)))))
 678     (nnheader-translate-file-chars name)))
 679
 680 (defvar nnrss-moreover-url
 681   "http://w.moreover.com/categories/category_list_rss.html"
 682   "The url of moreover.com categories.")
 683
 684 (defun nnrss-snarf-moreover-categories ()
 685   "Snarf RSS links from moreover.com."
 686   (interactive)
 687   (let (category name url changed)
 688     (with-temp-buffer
 689       (nnrss-insert nnrss-moreover-url)
 690       (goto-char (point-min))
 691       (while (re-search-forward
 692               "<a name=\"\\([^\"]+\\)\">\\|<a href=\"\\(http://[^\"]*moreover\\.com[^\"]+page\\?c=\\([^\"&]+\\)&o=rss\\)" nil t)
 693         (if (match-string 1)
 694             (setq category (match-string 1))
 695           (setq url (match-string 2)
 696                 name (mm-url-decode-entities-string
 697                       (rfc2231-decode-encoded-string
 698                        (match-string 3))))
 699           (if category
 700               (setq name (concat category "." name)))
 701           (unless (assoc name nnrss-server-data)
 702             (setq changed t)
 703             (push (list name 0 url) nnrss-server-data)))))
 704     (if changed
 705         (nnrss-save-server-data ""))))
 706
 707 (defun nnrss-node-text (namespace local-name element)
 708   (let* ((node (assq (intern (concat namespace (symbol-name local-name)))
 709                      element))
 710          (text (if (and node (listp node))
 711                    (nnrss-node-just-text node)
 712                  node))
 713          (cleaned-text (if text
 714                            (gnus-replace-in-string
 715                             (gnus-replace-in-string
 716                              text "^[\000-\037\177]+\\|^ +\\| +$" "")
 717                             "\r\n" "\n"))))
 718     (if (string-equal "" cleaned-text)
 719         nil
 720       cleaned-text)))
 721
 722 (defun nnrss-node-just-text (node)
 723   (if (and node (listp node))
 724       (mapconcat 'nnrss-node-just-text (cddr node) " ")
 725     node))
 726
 727 (defun nnrss-find-el (tag data &optional found-list)
 728   "Find the all matching elements in the data.
 729 Careful with this on large documents!"
 730   (when (consp data)
 731     (dolist (bit data)
 732       (when (car-safe bit)
 733         (when (equal tag (car bit))
 734           ;; Old xml.el may return a list of string.
 735           (when (and (consp (caddr bit))
 736                      (stringp (caaddr bit)))
 737             (setcar (cddr bit) (caaddr bit)))
 738           (setq found-list
 739                 (append found-list
 740                         (list bit))))
 741         (if (and (consp (car-safe (caddr bit)))
 742                  (not (stringp (caddr bit))))
 743             (setq found-list
 744                   (append found-list
 745                           (nnrss-find-el
 746                            tag (caddr bit))))
 747           (setq found-list
 748                 (append found-list
 749                         (nnrss-find-el
 750                          tag (cddr bit))))))))
 751   found-list)
 752
 753 (defun nnrss-rsslink-p (el)
 754   "Test if the element we are handed is an RSS autodiscovery link."
 755   (and (eq (car-safe el) 'link)
 756        (string-equal (cdr (assoc 'rel (cadr el))) "alternate")
 757        (or (string-equal (cdr (assoc 'type (cadr el)))
 758                          "application/rss+xml")
 759            (string-equal (cdr (assoc 'type (cadr el))) "text/xml"))))
 760
 761 (defun nnrss-get-rsslinks (data)
 762   "Extract the <link> elements that are links to RSS from the parsed data."
 763   (delq nil (mapcar
 764              (lambda (el)
 765                (if (nnrss-rsslink-p el) el))
 766              (nnrss-find-el 'link data))))
 767
 768 (defun nnrss-extract-hrefs (data)
 769   "Recursively extract hrefs from a page's source.
 770 DATA should be the output of `xml-parse-region' or
 771 `w3-parse-buffer'."
 772   (mapcar (lambda (ahref)
 773             (cdr (assoc 'href (cadr ahref))))
 774           (nnrss-find-el 'a data)))
 775
 776 (defmacro nnrss-match-macro (base-uri item onsite-list offsite-list)
 777   `(cond ((or (string-match (concat "^" ,base-uri) ,item)
 778               (not (string-match "://" ,item)))
 779           (setq ,onsite-list (append ,onsite-list (list ,item))))
 780          (t (setq ,offsite-list (append ,offsite-list (list ,item))))))
 781
 782 (defun nnrss-order-hrefs (base-uri hrefs)
 783   "Given a list of hrefs, sort them using the following priorities:
 784   1. links ending in .rss
 785   2. links ending in .rdf
 786   3. links ending in .xml
 787   4. links containing the above
 788   5. offsite links
 789
 790 BASE-URI is used to determine the location of the links and
 791 whether they are `offsite' or `onsite'."
 792   (let (rss-onsite-end  rdf-onsite-end  xml-onsite-end
 793         rss-onsite-in   rdf-onsite-in   xml-onsite-in
 794         rss-offsite-end rdf-offsite-end xml-offsite-end
 795         rss-offsite-in  rdf-offsite-in  xml-offsite-in)
 796     (dolist (href hrefs)
 797       (cond ((null href))
 798             ((string-match "\\.rss$" href)
 799              (nnrss-match-macro
 800               base-uri href rss-onsite-end rss-offsite-end))
 801             ((string-match "\\.rdf$" href)
 802              (nnrss-match-macro
 803               base-uri href rdf-onsite-end rdf-offsite-end))
 804             ((string-match "\\.xml$" href)
 805              (nnrss-match-macro
 806               base-uri href xml-onsite-end xml-offsite-end))
 807             ((string-match "rss" href)
 808              (nnrss-match-macro
 809               base-uri href rss-onsite-in rss-offsite-in))
 810             ((string-match "rdf" href)
 811              (nnrss-match-macro
 812               base-uri href rdf-onsite-in rdf-offsite-in))
 813             ((string-match "xml" href)
 814              (nnrss-match-macro
 815               base-uri href xml-onsite-in xml-offsite-in))))
 816     (append
 817      rss-onsite-end  rdf-onsite-end  xml-onsite-end
 818      rss-onsite-in   rdf-onsite-in   xml-onsite-in
 819      rss-offsite-end rdf-offsite-end xml-offsite-end
 820      rss-offsite-in rdf-offsite-in xml-offsite-in)))
 821
 822 (defun nnrss-discover-feed (url)
 823   "Given a page, find an RSS feed using Mark Pilgrim's
 824 `ultra-liberal rss locator' (http://diveintomark.org/2002/08/15.html)."
 825
 826   (let ((parsed-page (nnrss-fetch url)))
 827
 828 ;;    1. if this url is the rss, use it.
 829     (if (nnrss-rss-p parsed-page)
 830         (let ((rss-ns (nnrss-get-namespace-prefix parsed-page "http://purl.org/rss/1.0/")))
 831           (nnrss-rss-title-description rss-ns parsed-page url))
 832
 833 ;;    2. look for the <link rel="alternate"
 834 ;;    type="application/rss+xml" and use that if it is there.
 835       (let ((links (nnrss-get-rsslinks parsed-page)))
 836         (if links
 837             (let* ((xml (nnrss-fetch
 838                          (cdr (assoc 'href (cadar links)))))
 839                    (rss-ns (nnrss-get-namespace-prefix xml "http://purl.org/rss/1.0/")))
 840               (nnrss-rss-title-description rss-ns xml (cdr (assoc 'href (cadar links)))))
 841
 842 ;;    3. look for links on the site in the following order:
 843 ;;       - onsite links ending in .rss, .rdf, or .xml
 844 ;;       - onsite links containing any of the above
 845 ;;       - offsite links ending in .rss, .rdf, or .xml
 846 ;;       - offsite links containing any of the above
 847           (let* ((base-uri (progn (string-match ".*://[^/]+/?" url)
 848                                   (match-string 0 url)))
 849                  (hrefs (nnrss-order-hrefs
 850                          base-uri (nnrss-extract-hrefs parsed-page)))
 851                  (rss-link nil))
 852             (while (and (eq rss-link nil) (not (eq hrefs nil)))
 853               (let ((href-data (nnrss-fetch (car hrefs))))
 854                 (if (nnrss-rss-p href-data)
 855                     (let* ((rss-ns (nnrss-get-namespace-prefix href-data "http://purl.org/rss/1.0/")))
 856                       (setq rss-link (nnrss-rss-title-description
 857                                       rss-ns href-data (car hrefs))))
 858                   (setq hrefs (cdr hrefs)))))
 859             (if rss-link rss-link
 860
 861 ;;    4. check syndic8
 862               (nnrss-find-rss-via-syndic8 url))))))))
 863
 864 (defun nnrss-find-rss-via-syndic8 (url)
 865   "Query syndic8 for the rss feeds it has for URL."
 866   (if (not (locate-library "xml-rpc"))
 867       (progn
 868         (message "XML-RPC is not available... not checking Syndic8.")
 869         nil)
 870     (require 'xml-rpc)
 871     (let ((feedid (xml-rpc-method-call
 872                    "http://www.syndic8.com/xmlrpc.php"
 873                    'syndic8.FindSites
 874                    url)))
 875       (when feedid
 876         (let* ((feedinfo (xml-rpc-method-call
 877                           "http://www.syndic8.com/xmlrpc.php"
 878                           'syndic8.GetFeedInfo
 879                           feedid))
 880                (urllist
 881                 (delq nil
 882                       (mapcar
 883                        (lambda (listinfo)
 884                          (if (string-equal
 885                               (cdr (assoc "status" listinfo))
 886                               "Syndicated")
 887                              (cons
 888                               (cdr (assoc "sitename" listinfo))
 889                               (list
 890                                (cons 'title
 891                                      (cdr (assoc
 892                                            "sitename" listinfo)))
 893                                (cons 'href
 894                                      (cdr (assoc
 895                                            "dataurl" listinfo)))))))
 896                        feedinfo))))
 897           (if (not (> (length urllist) 1))
 898               (cdar urllist)
 899             (let ((completion-ignore-case t)
 900                   (selection
 901                    (mapcar (lambda (listinfo)
 902                              (cons (cdr (assoc "sitename" listinfo))
 903                                    (string-to-number
 904                                     (cdr (assoc "feedid" listinfo)))))
 905                            feedinfo)))
 906               (cdr (assoc
 907                     (completing-read
 908                      "Multiple feeds found.  Select one: "
 909                      selection nil t) urllist)))))))))
 910
 911 (defun nnrss-rss-p (data)
 912   "Test if DATA is an RSS feed.
 913 Simply ensures that the first element is rss or rdf."
 914   (or (eq (caar data) 'rss)
 915       (eq (caar data) 'rdf:RDF)))
 916
 917 (defun nnrss-rss-title-description (rss-namespace data url)
 918   "Return the title of an RSS feed."
 919   (if (nnrss-rss-p data)
 920       (let ((description (intern (concat rss-namespace "description")))
 921             (title (intern (concat rss-namespace "title")))
 922             (channel (nnrss-find-el (intern (concat rss-namespace "channel"))
 923                                     data)))
 924         (list
 925          (cons 'description (caddr (nth 0 (nnrss-find-el description channel))))
 926          (cons 'title (caddr (nth 0 (nnrss-find-el title channel))))
 927          (cons 'href url)))))
 928
 929 (defun nnrss-get-namespace-prefix (el uri)
 930   "Given EL (containing a parsed element) and URI (containing a string
 931 that gives the URI for which you want to retrieve the namespace
 932 prefix), return the prefix."
 933   (let* ((prefix (car (rassoc uri (cadar el))))
 934          (nslist (if prefix
 935                      (split-string (symbol-name prefix) ":")))
 936          (ns (cond ((eq (length nslist) 1) ; no prefix given
 937                     "")
 938                    ((eq (length nslist) 2) ; extract prefix
 939                     (cadr nslist)))))
 940     (if (and ns (not (string= ns "")))
 941         (concat ns ":")
 942       ns)))
 943
 944 (provide 'nnrss)
 945
 946
 947 ;;; nnrss.el ends here
 948
 949 ;;; arch-tag: 12910c07-0cdf-44fb-8d2c-416ded64c267