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

   1 ;;; nnrss.el --- interfacing with RSS
   2 ;; Copyright (C) 2001, 2002, 2003, 2004, 2005  Free Software Foundation, Inc.
   3
   4 ;; Author: Shenghuo Zhu <zsh@cs.rochester.edu>
   5 ;; Keywords: RSS
   6
   7 ;; This file is part of GNU Emacs.
   8
   9 ;; GNU Emacs is free software; you can redistribute it and/or modify
  10 ;; it under the terms of the GNU General Public License as published
  11 ;; by the Free Software Foundation; either version 2, or (at your
  12 ;; option) any later version.
  13
  14 ;; GNU Emacs is distributed in the hope that it will be useful, but
  15 ;; WITHOUT ANY WARRANTY; without even the implied warranty of
  16 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  17 ;; General Public License for more details.
  18
  19 ;; You should have received a copy of the GNU General Public License
  20 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  21 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  22 ;; Boston, MA 02111-1307, USA.
  23
  24 ;;; Commentary:
  25
  26 ;;; Code:
  27
  28 (eval-when-compile (require 'cl))
  29
  30 (require 'gnus)
  31 (require 'nnoo)
  32 (require 'nnmail)
  33 (require 'message)
  34 (require 'mm-util)
  35 (require 'gnus-util)
  36 (require 'time-date)
  37 (require 'rfc2231)
  38 (require 'mm-url)
  39 (require 'rfc2047)
  40 (require 'mml)
  41 (eval-when-compile
  42   (ignore-errors
  43    (require 'xml)))
  44 (eval '(require 'xml))
  45
  46 (nnoo-declare nnrss)
  47
  48 (defvoo nnrss-directory (nnheader-concat gnus-directory "rss/")
  49   "Where nnrss will save its files.")
  50
  51 ;; (group max rss-url)
  52 (defvoo nnrss-server-data nil)
  53
  54 ;; (num timestamp url subject author date extra)
  55 (defvoo nnrss-group-data nil)
  56 (defvoo nnrss-group-max 0)
  57 (defvoo nnrss-group-min 1)
  58 (defvoo nnrss-group nil)
  59 (defvoo nnrss-group-hashtb nil)
  60 (defvoo nnrss-status-string "")
  61
  62 (defconst nnrss-version "nnrss 1.0")
  63
  64 (defvar nnrss-group-alist '()
  65   "List of RSS addresses.")
  66
  67 (defvar nnrss-use-local nil)
  68
  69 (defvar nnrss-description-field 'X-Gnus-Description
  70   "Field name used for DESCRIPTION.
  71 To use the description in headers, put this name into `nnmail-extra-headers'.")
  72
  73 (defvar nnrss-url-field 'X-Gnus-Url
  74   "Field name used for URL.
  75 To use the description in headers, put this name into `nnmail-extra-headers'.")
  76
  77 (defvar nnrss-content-function nil
  78   "A function which is called in `nnrss-request-article'.
  79 The arguments are (ENTRY GROUP ARTICLE).
  80 ENTRY is the record of the current headline.  GROUP is the group name.
  81 ARTICLE is the article number of the current headline.")
  82
  83 (defvar nnrss-file-coding-system mm-universal-coding-system
  84   "Coding system used when reading and writing files.")
  85
  86 (defvar nnrss-compatible-encoding-alist '((iso-8859-1 . windows-1252))
  87   "Alist of encodings and those supersets.
  88 The cdr of each element is used to decode data if it is available when
  89 the car is what the data specify as the encoding. Or, the car is used
  90 for decoding when the cdr that the data specify is not available.")
  91
  92 (nnoo-define-basics nnrss)
  93
  94 ;;; Interface functions
  95
  96 (defsubst nnrss-format-string (string)
  97   (gnus-replace-in-string string " *\n *" " "))
  98
  99 (defun nnrss-decode-group-name (group)
 100   (if (and group (mm-coding-system-p 'utf-8))
 101       (setq group (mm-decode-coding-string group 'utf-8))
 102     group))
 103
 104 (deffoo nnrss-retrieve-headers (articles &optional group server fetch-old)
 105   (setq group (nnrss-decode-group-name group))
 106   (nnrss-possibly-change-group group server)
 107   (let (e)
 108     (save-excursion
 109       (set-buffer nntp-server-buffer)
 110       (erase-buffer)
 111       (dolist (article articles)
 112         (if (setq e (assq article nnrss-group-data))
 113             (insert (number-to-string (car e)) "\t" ;; number
 114                     ;; subject
 115                     (or (nth 3 e) "")
 116                     "\t"
 117                     ;; from
 118                     (or (nth 4 e) "(nobody)")
 119                     "\t"
 120                     ;; date
 121                     (or (nth 5 e) "")
 122                     "\t"
 123                     ;; id
 124                     (format "<%d@%s.nnrss>" (car e) group)
 125                     "\t"
 126                     ;; refs
 127                     "\t"
 128                     ;; chars
 129                     "-1" "\t"
 130                     ;; lines
 131                     "-1" "\t"
 132                     ;; Xref
 133                     "" "\t"
 134                     (if (and (nth 6 e)
 135                              (memq nnrss-description-field
 136                                    nnmail-extra-headers))
 137                         (concat (symbol-name nnrss-description-field)
 138                                 ": "
 139                                 (nnrss-format-string (nth 6 e))
 140                                 "\t")
 141                       "")
 142                     (if (and (nth 2 e)
 143                              (memq nnrss-url-field
 144                                    nnmail-extra-headers))
 145                         (concat (symbol-name nnrss-url-field)
 146                                 ": "
 147                                 (nnrss-format-string (nth 2 e))
 148                                 "\t")
 149                       "")
 150                     "\n")))))
 151   'nov)
 152
 153 (deffoo nnrss-request-group (group &optional server dont-check)
 154   (setq group (nnrss-decode-group-name group))
 155   (nnheader-message 6 "nnrss: Requesting %s..." group)
 156   (nnrss-possibly-change-group group server)
 157   (prog1
 158       (if dont-check
 159           t
 160         (nnrss-check-group group server)
 161         (nnheader-report 'nnrss "Opened group %s" group)
 162         (nnheader-insert
 163          "211 %d %d %d %s\n" nnrss-group-max nnrss-group-min nnrss-group-max
 164          (prin1-to-string group)
 165          t))
 166     (nnheader-message 6 "nnrss: Requesting %s...done" group)))
 167
 168 (deffoo nnrss-close-group (group &optional server)
 169   t)
 170
 171 (deffoo nnrss-request-article (article &optional group server buffer)
 172   (setq group (nnrss-decode-group-name group))
 173   (when (stringp article)
 174     (setq article (if (string-match "\\`<\\([0-9]+\\)@" article)
 175                       (string-to-number (match-string 1 article))
 176                     0)))
 177   (nnrss-possibly-change-group group server)
 178   (let ((e (assq article nnrss-group-data))
 179         (nntp-server-buffer (or buffer nntp-server-buffer))
 180         post err)
 181     (when e
 182       (with-current-buffer nntp-server-buffer
 183         (erase-buffer)
 184         (if group
 185             (insert "Newsgroups: " group "\n"))
 186         (if (nth 3 e)
 187             (insert "Subject: " (nth 3 e) "\n"))
 188         (if (nth 4 e)
 189             (insert "From: " (nth 4 e) "\n"))
 190         (if (nth 5 e)
 191             (insert "Date: " (nnrss-format-string (nth 5 e)) "\n"))
 192         (let ((header (buffer-string))
 193               (text (if (nth 6 e)
 194                         (mapconcat 'identity
 195                                    (delete "" (split-string (nth 6 e) "\n+"))
 196                                    " ")))
 197               (link (nth 2 e))
 198               ;; Enable encoding of Newsgroups header in XEmacs.
 199               (default-enable-multibyte-characters t)
 200               (rfc2047-header-encoding-alist
 201                (if (mm-coding-system-p 'utf-8)
 202                    (cons '("Newsgroups" . utf-8)
 203                          rfc2047-header-encoding-alist)
 204                  rfc2047-header-encoding-alist))
 205               rfc2047-encode-encoded-words body)
 206           (when (or text link)
 207             (insert "\n")
 208             (insert "<#multipart type=alternative>\n"
 209                     "<#part type=\"text/plain\">\n")
 210             (setq body (point))
 211             (if text
 212                 (progn
 213                   (insert text "\n")
 214                   (when link
 215                     (insert "\n" link "\n")))
 216               (when link
 217                 (insert link "\n")))
 218             (setq body (buffer-substring body (point)))
 219             (insert "<#/part>\n"
 220                     "<#part type=\"text/html\">\n"
 221                     "<html><head></head><body>\n")
 222             (when text
 223               (insert text "\n"))
 224             (when link
 225               (insert "<p><a href=\"" link "\">link</a></p>\n"))
 226             (insert "</body></html>\n"
 227                     "<#/part>\n"
 228                     "<#/multipart>\n"))
 229           (condition-case nil
 230               (mml-to-mime)
 231             (error
 232              (erase-buffer)
 233              (insert header
 234                      "Content-Type: text/plain; charset=gnus-decoded\n"
 235                      "Content-Transfer-Encoding: 8bit\n\n"
 236                      body)
 237              (nnheader-message
 238               3 "Warning - there might be invalid characters"))))
 239         (goto-char (point-min))
 240         (search-forward "\n\n")
 241         (forward-line -1)
 242         (insert (format "Message-ID: <%d@%s.nnrss>\n"
 243                         (car e)
 244                         (let ((rfc2047-encoding-type 'mime)
 245                               rfc2047-encode-max-chars)
 246                           (rfc2047-encode-string
 247                            (gnus-replace-in-string group "[\t\n ]+" "_")))))
 248         (when nnrss-content-function
 249           (funcall nnrss-content-function e group article))))
 250     (cond
 251      (err
 252       (nnheader-report 'nnrss err))
 253      ((not e)
 254       (nnheader-report 'nnrss "no such id: %d" article))
 255      (t
 256       (nnheader-report 'nnrss "article %s retrieved" (car e))
 257       ;; we return the article number.
 258       (cons nnrss-group (car e))))))
 259
 260 (deffoo nnrss-request-list (&optional server)
 261   (nnrss-possibly-change-group nil server)
 262   (nnrss-generate-active)
 263   t)
 264
 265 (deffoo nnrss-open-server (server &optional defs connectionless)
 266   (nnrss-read-server-data server)
 267   (nnoo-change-server 'nnrss server defs)
 268   t)
 269
 270 (deffoo nnrss-request-expire-articles
 271     (articles group &optional server force)
 272   (setq group (nnrss-decode-group-name group))
 273   (nnrss-possibly-change-group group server)
 274   (let (e days not-expirable changed)
 275     (dolist (art articles)
 276       (if (and (setq e (assq art nnrss-group-data))
 277                (nnmail-expired-article-p
 278                 group
 279                 (if (listp (setq days (nth 1 e))) days
 280                   (days-to-time (- days (time-to-days '(0 0)))))
 281                 force))
 282           (setq nnrss-group-data (delq e nnrss-group-data)
 283                 changed t)
 284         (push art not-expirable)))
 285     (if changed
 286         (nnrss-save-group-data group server))
 287     not-expirable))
 288
 289 (deffoo nnrss-request-delete-group (group &optional force server)
 290   (setq group (nnrss-decode-group-name group))
 291   (nnrss-possibly-change-group group server)
 292   (let (elem)
 293     ;; There may be two or more entries in `nnrss-group-alist' since
 294     ;; this function didn't delete them formerly.
 295     (while (setq elem (assoc group nnrss-group-alist))
 296       (setq nnrss-group-alist (delq elem nnrss-group-alist))))
 297   (setq nnrss-server-data
 298         (delq (assoc group nnrss-server-data) nnrss-server-data))
 299   (nnrss-save-server-data server)
 300   (ignore-errors
 301    (delete-file (nnrss-make-filename group server)))
 302   t)
 303
 304 (deffoo nnrss-request-list-newsgroups (&optional server)
 305   (nnrss-possibly-change-group nil server)
 306   (save-excursion
 307     (set-buffer nntp-server-buffer)
 308     (erase-buffer)
 309     (dolist (elem nnrss-group-alist)
 310       (if (third elem)
 311           (insert (car elem) "\t" (third elem) "\n"))))
 312   t)
 313
 314 (nnoo-define-skeleton nnrss)
 315
 316 ;;; Internal functions
 317 (eval-when-compile (defun xml-rpc-method-call (&rest args)))
 318
 319 (defun nnrss-get-encoding ()
 320   "Return an encoding attribute specified in the current xml contents.
 321 If `nnrss-compatible-encoding-alist' specifies the compatible encoding,
 322 it is used instead.  If the xml contents doesn't specify the encoding,
 323 return `utf-8' which is the default encoding for xml if it is available,
 324 otherwise return nil."
 325   (goto-char (point-min))
 326   (if (re-search-forward
 327        "<\\?[^>]*encoding=\\(\"\\([^\">]+\\)\"\\|'\\([^'>]+\\)'\\)"
 328        nil t)
 329       (let ((encoding (intern (downcase (or (match-string 2)
 330                                             (match-string 3))))))
 331         (or
 332          (mm-coding-system-p (cdr (assq encoding
 333                                         nnrss-compatible-encoding-alist)))
 334          (mm-coding-system-p encoding)
 335          (mm-coding-system-p (car (rassq encoding
 336                                          nnrss-compatible-encoding-alist)))))
 337     (mm-coding-system-p 'utf-8)))
 338
 339 (defun nnrss-fetch (url &optional local)
 340   "Fetch URL and put it in a the expected Lisp structure."
 341   (mm-with-unibyte-buffer
 342     ;;some CVS versions of url.el need this to close the connection quickly
 343     (let (cs xmlform htmlform)
 344       ;; bit o' work necessary for w3 pre-cvs and post-cvs
 345       (if local
 346           (let ((coding-system-for-read 'binary))
 347             (insert-file-contents url))
 348         ;; FIXME: shouldn't binding `coding-system-for-read' be moved
 349         ;; to `mm-url-insert'?
 350         (let ((coding-system-for-read 'binary))
 351           (mm-url-insert url)))
 352       (nnheader-remove-cr-followed-by-lf)
 353       ;; Decode text according to the encoding attribute.
 354       (when (setq cs (nnrss-get-encoding))
 355         (mm-decode-coding-region (point-min) (point-max) cs)
 356         (mm-enable-multibyte))
 357       (goto-char (point-min))
 358
 359       ;; Because xml-parse-region can't deal with anything that isn't
 360       ;; xml and w3-parse-buffer can't deal with some xml, we have to
 361       ;; parse with xml-parse-region first and, if that fails, parse
 362       ;; with w3-parse-buffer.  Yuck.  Eventually, someone should find out
 363       ;; why w3-parse-buffer fails to parse some well-formed xml and
 364       ;; fix it.
 365
 366       (condition-case err1
 367           (setq xmlform (xml-parse-region (point-min) (point-max)))
 368         (error
 369          (condition-case err2
 370              (setq htmlform (caddar (w3-parse-buffer
 371                                      (current-buffer))))
 372            (error
 373             (message "\
 374 nnrss: %s: Not valid XML %s and w3-parse doesn't work %s"
 375                      url err1 err2)))))
 376       (if htmlform
 377           htmlform
 378         xmlform))))
 379
 380 (defun nnrss-possibly-change-group (&optional group server)
 381   (when (and server
 382              (not (nnrss-server-opened server)))
 383     (nnrss-open-server server))
 384   (when (and group (not (equal group nnrss-group)))
 385     (nnrss-read-group-data group server)
 386     (setq nnrss-group group)))
 387
 388 (defvar nnrss-extra-categories '(nnrss-snarf-moreover-categories))
 389
 390 (defun nnrss-generate-active ()
 391   (when (y-or-n-p "Fetch extra categories? ")
 392     (dolist (func nnrss-extra-categories)
 393       (funcall func)))
 394   (save-excursion
 395     (set-buffer nntp-server-buffer)
 396     (erase-buffer)
 397     (dolist (elem nnrss-group-alist)
 398       (insert (prin1-to-string (car elem)) " 0 1 y\n"))
 399     (dolist (elem nnrss-server-data)
 400       (unless (assoc (car elem) nnrss-group-alist)
 401         (insert (prin1-to-string (car elem)) " 0 1 y\n")))))
 402
 403 ;;; data functions
 404
 405 (defun nnrss-read-server-data (server)
 406   (setq nnrss-server-data nil)
 407   (let ((file (nnrss-make-filename "nnrss" server)))
 408     (when (file-exists-p file)
 409       ;; In Emacs 21.3 and earlier, `load' doesn't support non-ASCII
 410       ;; file names.  So, we use `insert-file-contents' instead.
 411       (mm-with-multibyte-buffer
 412         (let ((coding-system-for-read nnrss-file-coding-system)
 413               (file-name-coding-system nnmail-pathname-coding-system))
 414           (insert-file-contents file)
 415           (eval-region (point-min) (point-max)))))))
 416
 417 (defun nnrss-save-server-data (server)
 418   (gnus-make-directory nnrss-directory)
 419   (let ((coding-system-for-write nnrss-file-coding-system)
 420         (file-name-coding-system nnmail-pathname-coding-system))
 421     (with-temp-file (nnrss-make-filename "nnrss" server)
 422       (insert (format ";; -*- coding: %s; -*-\n"
 423                       nnrss-file-coding-system))
 424       (gnus-prin1 `(setq nnrss-group-alist ',nnrss-group-alist))
 425       (insert "\n")
 426       (gnus-prin1 `(setq nnrss-server-data ',nnrss-server-data)))))
 427
 428 (defun nnrss-read-group-data (group server)
 429   (setq nnrss-group-data nil)
 430   (setq nnrss-group-hashtb (gnus-make-hashtable))
 431   (let ((pair (assoc group nnrss-server-data)))
 432     (setq nnrss-group-max (or (cadr pair) 0))
 433     (setq nnrss-group-min (+ nnrss-group-max 1)))
 434   (let ((file (nnrss-make-filename group server)))
 435     (when (file-exists-p file)
 436       ;; In Emacs 21.3 and earlier, `load' doesn't support non-ASCII
 437       ;; file names.  So, we use `insert-file-contents' instead.
 438       (mm-with-multibyte-buffer
 439         (let ((coding-system-for-read nnrss-file-coding-system)
 440               (file-name-coding-system nnmail-pathname-coding-system))
 441           (insert-file-contents file)
 442           (eval-region (point-min) (point-max))))
 443       (dolist (e nnrss-group-data)
 444         (gnus-sethash (or (nth 2 e) (nth 6 e)) t nnrss-group-hashtb)
 445         (when (and (car e) (> nnrss-group-min (car e)))
 446           (setq nnrss-group-min (car e)))
 447         (when (and (car e) (< nnrss-group-max (car e)))
 448           (setq nnrss-group-max (car e)))))))
 449
 450 (defun nnrss-save-group-data (group server)
 451   (gnus-make-directory nnrss-directory)
 452   (let ((coding-system-for-write nnrss-file-coding-system)
 453         (file-name-coding-system nnmail-pathname-coding-system))
 454     (with-temp-file (nnrss-make-filename group server)
 455       (insert (format ";; -*- coding: %s; -*-\n"
 456                       nnrss-file-coding-system))
 457       (gnus-prin1 `(setq nnrss-group-data ',nnrss-group-data)))))
 458
 459 (defun nnrss-make-filename (name server)
 460   (expand-file-name
 461    (nnrss-translate-file-chars
 462     (concat name
 463             (and server
 464                  (not (equal server ""))
 465                  "-")
 466             server
 467             ".el"))
 468    nnrss-directory))
 469
 470 (gnus-add-shutdown 'nnrss-close 'gnus)
 471
 472 (defun nnrss-close ()
 473   "Clear internal nnrss variables."
 474   (setq nnrss-group-data nil
 475         nnrss-server-data nil
 476         nnrss-group-hashtb nil
 477         nnrss-group-alist nil))
 478
 479 ;;; URL interface
 480
 481 (defun nnrss-no-cache (url)
 482   "")
 483
 484 (defun nnrss-insert-w3 (url)
 485   (mm-with-unibyte-current-buffer
 486     (mm-url-insert url)))
 487
 488 (defun nnrss-decode-entities-string (string)
 489   (if string
 490       (mm-with-multibyte-buffer
 491         (insert string)
 492         (mm-url-decode-entities-nbsp)
 493         (buffer-string))))
 494
 495 (defalias 'nnrss-insert 'nnrss-insert-w3)
 496
 497 (defun nnrss-mime-encode-string (string)
 498   (mm-with-multibyte-buffer
 499     (insert string)
 500     (mm-url-decode-entities-nbsp)
 501     (goto-char (point-min))
 502     (while (re-search-forward "[\t\n ]+" nil t)
 503       (replace-match " "))
 504     (goto-char (point-min))
 505     (skip-chars-forward " ")
 506     (delete-region (point-min) (point))
 507     (goto-char (point-max))
 508     (skip-chars-forward " ")
 509     (delete-region (point) (point-max))
 510     (let ((rfc2047-encoding-type 'mime)
 511           rfc2047-encode-max-chars)
 512       (rfc2047-encode-region (point-min) (point-max)))
 513     (goto-char (point-min))
 514     (while (search-forward "\n" nil t)
 515       (delete-backward-char 1))
 516     (buffer-string)))
 517
 518 ;;; Snarf functions
 519
 520 (defun nnrss-check-group (group server)
 521   (let (file xml subject url extra changed author
 522              date rss-ns rdf-ns content-ns dc-ns)
 523     (if (and nnrss-use-local
 524              (file-exists-p (setq file (expand-file-name
 525                                         (nnrss-translate-file-chars
 526                                          (concat group ".xml"))
 527                                         nnrss-directory))))
 528         (setq xml (nnrss-fetch file t))
 529       (setq url (or (nth 2 (assoc group nnrss-server-data))
 530                     (second (assoc group nnrss-group-alist))))
 531       (unless url
 532         (setq url
 533               (cdr
 534                (assoc 'href
 535                       (nnrss-discover-feed
 536                        (read-string
 537                         (format "URL to search for %s: " group) "http://")))))
 538         (let ((pair (assoc group nnrss-server-data)))
 539           (if pair
 540               (setcdr (cdr pair) (list url))
 541             (push (list group nnrss-group-max url) nnrss-server-data)))
 542         (setq changed t))
 543       (setq xml (nnrss-fetch url)))
 544     ;; See
 545     ;; http://feeds.archive.org/validator/docs/howto/declare_namespaces.html
 546     ;; for more RSS namespaces.
 547     (setq dc-ns (nnrss-get-namespace-prefix xml "http://purl.org/dc/elements/1.1/")
 548           rdf-ns (nnrss-get-namespace-prefix xml "http://www.w3.org/1999/02/22-rdf-syntax-ns#")
 549           rss-ns (nnrss-get-namespace-prefix xml "http://purl.org/rss/1.0/")
 550           content-ns (nnrss-get-namespace-prefix xml "http://purl.org/rss/1.0/modules/content/"))
 551     (dolist (item (nreverse (nnrss-find-el (intern (concat rss-ns "item")) xml)))
 552       (when (and (listp item)
 553                  (string= (concat rss-ns "item") (car item))
 554                  (if (setq url (nnrss-decode-entities-string
 555                                 (nnrss-node-text rss-ns 'link (cddr item))))
 556                      (not (gnus-gethash url nnrss-group-hashtb))
 557                    (setq extra (or (nnrss-node-text content-ns 'encoded item)
 558                                    (nnrss-node-text rss-ns 'description item)))
 559                    (not (gnus-gethash extra nnrss-group-hashtb))))
 560         (setq subject (nnrss-node-text rss-ns 'title item))
 561         (setq extra (or extra
 562                         (nnrss-node-text content-ns 'encoded item)
 563                         (nnrss-node-text rss-ns 'description item)))
 564         (setq author (or (nnrss-node-text rss-ns 'author item)
 565                          (nnrss-node-text dc-ns 'creator item)
 566                          (nnrss-node-text dc-ns 'contributor item)))
 567         (setq date (or (nnrss-node-text dc-ns 'date item)
 568                        (nnrss-node-text rss-ns 'pubDate item)
 569                        (message-make-date)))
 570         (push
 571          (list
 572           (incf nnrss-group-max)
 573           (current-time)
 574           url
 575           (and subject (nnrss-mime-encode-string subject))
 576           (and author (nnrss-mime-encode-string author))
 577           date
 578           (and extra (nnrss-decode-entities-string extra)))
 579          nnrss-group-data)
 580         (gnus-sethash (or url extra) t nnrss-group-hashtb)
 581         (setq changed t))
 582       (setq extra nil))
 583     (when changed
 584       (nnrss-save-group-data group server)
 585       (let ((pair (assoc group nnrss-server-data)))
 586         (if pair
 587             (setcar (cdr pair) nnrss-group-max)
 588           (push (list group nnrss-group-max) nnrss-server-data)))
 589       (nnrss-save-server-data server))))
 590
 591 (defun nnrss-opml-import (opml-file)
 592   "OPML subscriptions import.
 593 Read the file and attempt to subscribe to each Feed in the file."
 594   (interactive "fImport file: ")
 595   (mapcar
 596    (lambda (node) (gnus-group-make-rss-group
 597                    (cdr (assq 'xmlUrl (cadr node)))))
 598    (nnrss-find-el 'outline
 599                   (progn
 600                     (find-file opml-file)
 601                     (xml-parse-region (point-min)
 602                                       (point-max))))))
 603
 604 (defun nnrss-opml-export ()
 605   "OPML subscription export.
 606 Export subscriptions to a buffer in OPML Format."
 607   (interactive)
 608   (with-current-buffer (get-buffer-create "*OPML Export*")
 609     (mm-set-buffer-file-coding-system 'utf-8)
 610     (insert "<?xml version=\"1.0\" encoding=\"utf-8\"?>\n"
 611             "<!-- OPML generated by Emacs Gnus' nnrss.el -->\n"
 612             "<opml version=\"1.1\">\n"
 613             "  <head>\n"
 614             "    <title>mySubscriptions</title>\n"
 615             "    <dateCreated>" (format-time-string "%a, %d %b %Y %T %z")
 616             "</dateCreated>\n"
 617             "    <ownerEmail>" user-mail-address "</ownerEmail>\n"
 618             "    <ownerName>" (user-full-name) "</ownerName>\n"
 619             "  </head>\n"
 620             "  <body>\n")
 621     (dolist (sub nnrss-group-alist)
 622       (insert "    <outline text=\"" (car sub)
 623               "\" xmlUrl=\"" (cadr sub) "\"/>\n"))
 624     (insert "  </body>\n"
 625             "</opml>\n"))
 626   (pop-to-buffer "*OPML Export*")
 627   (when (fboundp 'sgml-mode)
 628     (sgml-mode)))
 629
 630 (defun nnrss-generate-download-script ()
 631   "Generate a download script in the current buffer.
 632 It is useful when `(setq nnrss-use-local t)'."
 633   (interactive)
 634   (insert "#!/bin/sh\n")
 635   (insert "WGET=wget\n")
 636   (insert "RSSDIR='" (expand-file-name nnrss-directory) "'\n")
 637   (dolist (elem nnrss-server-data)
 638     (let ((url (or (nth 2 elem)
 639                    (second (assoc (car elem) nnrss-group-alist)))))
 640       (insert "$WGET -q -O \"$RSSDIR\"/'"
 641               (nnrss-translate-file-chars (concat (car elem) ".xml"))
 642               "' '" url "'\n"))))
 643
 644 (defun nnrss-translate-file-chars (name)
 645   (let ((nnheader-file-name-translation-alist
 646          (append nnheader-file-name-translation-alist '((?' . ?_)))))
 647     (nnheader-translate-file-chars name)))
 648
 649 (defvar nnrss-moreover-url
 650   "http://w.moreover.com/categories/category_list_rss.html"
 651   "The url of moreover.com categories.")
 652
 653 (defun nnrss-snarf-moreover-categories ()
 654   "Snarf RSS links from moreover.com."
 655   (interactive)
 656   (let (category name url changed)
 657     (with-temp-buffer
 658       (nnrss-insert nnrss-moreover-url)
 659       (goto-char (point-min))
 660       (while (re-search-forward
 661               "<a name=\"\\([^\"]+\\)\">\\|<a href=\"\\(http://[^\"]*moreover\\.com[^\"]+page\\?c=\\([^\"&]+\\)&o=rss\\)" nil t)
 662         (if (match-string 1)
 663             (setq category (match-string 1))
 664           (setq url (match-string 2)
 665                 name (mm-url-decode-entities-string
 666                       (rfc2231-decode-encoded-string
 667                        (match-string 3))))
 668           (if category
 669               (setq name (concat category "." name)))
 670           (unless (assoc name nnrss-server-data)
 671             (setq changed t)
 672             (push (list name 0 url) nnrss-server-data)))))
 673     (if changed
 674         (nnrss-save-server-data ""))))
 675
 676 (defun nnrss-node-text (namespace local-name element)
 677   (let* ((node (assq (intern (concat namespace (symbol-name local-name)))
 678                      element))
 679          (text (if (and node (listp node))
 680                    (nnrss-node-just-text node)
 681                  node))
 682          (cleaned-text (if text (gnus-replace-in-string
 683                                  text "^[\000-\037\177]+\\|^ +\\| +$" ""))))
 684     (if (string-equal "" cleaned-text)
 685         nil
 686       cleaned-text)))
 687
 688 (defun nnrss-node-just-text (node)
 689   (if (and node (listp node))
 690       (mapconcat 'nnrss-node-just-text (cddr node) " ")
 691     node))
 692
 693 (defun nnrss-find-el (tag data &optional found-list)
 694   "Find the all matching elements in the data.
 695 Careful with this on large documents!"
 696   (when (consp data)
 697     (dolist (bit data)
 698       (when (car-safe bit)
 699         (when (equal tag (car bit))
 700           ;; Old xml.el may return a list of string.
 701           (when (and (consp (caddr bit))
 702                      (stringp (caaddr bit)))
 703             (setcar (cddr bit) (caaddr bit)))
 704           (setq found-list
 705                 (append found-list
 706                         (list bit))))
 707         (if (and (consp (car-safe (caddr bit)))
 708                  (not (stringp (caddr bit))))
 709             (setq found-list
 710                   (append found-list
 711                           (nnrss-find-el
 712                            tag (caddr bit))))
 713           (setq found-list
 714                 (append found-list
 715                         (nnrss-find-el
 716                          tag (cddr bit))))))))
 717   found-list)
 718
 719 (defun nnrss-rsslink-p (el)
 720   "Test if the element we are handed is an RSS autodiscovery link."
 721   (and (eq (car-safe el) 'link)
 722        (string-equal (cdr (assoc 'rel (cadr el))) "alternate")
 723        (or (string-equal (cdr (assoc 'type (cadr el)))
 724                          "application/rss+xml")
 725            (string-equal (cdr (assoc 'type (cadr el))) "text/xml"))))
 726
 727 (defun nnrss-get-rsslinks (data)
 728   "Extract the <link> elements that are links to RSS from the parsed data."
 729   (delq nil (mapcar
 730              (lambda (el)
 731                (if (nnrss-rsslink-p el) el))
 732              (nnrss-find-el 'link data))))
 733
 734 (defun nnrss-extract-hrefs (data)
 735   "Recursively extract hrefs from a page's source.
 736 DATA should be the output of `xml-parse-region' or
 737 `w3-parse-buffer'."
 738   (mapcar (lambda (ahref)
 739             (cdr (assoc 'href (cadr ahref))))
 740           (nnrss-find-el 'a data)))
 741
 742 (defmacro nnrss-match-macro (base-uri item onsite-list offsite-list)
 743   `(cond ((or (string-match (concat "^" ,base-uri) ,item)
 744               (not (string-match "://" ,item)))
 745           (setq ,onsite-list (append ,onsite-list (list ,item))))
 746          (t (setq ,offsite-list (append ,offsite-list (list ,item))))))
 747
 748 (defun nnrss-order-hrefs (base-uri hrefs)
 749   "Given a list of hrefs, sort them using the following priorities:
 750   1. links ending in .rss
 751   2. links ending in .rdf
 752   3. links ending in .xml
 753   4. links containing the above
 754   5. offsite links
 755
 756 BASE-URI is used to determine the location of the links and
 757 whether they are `offsite' or `onsite'."
 758   (let (rss-onsite-end  rdf-onsite-end  xml-onsite-end
 759         rss-onsite-in   rdf-onsite-in   xml-onsite-in
 760         rss-offsite-end rdf-offsite-end xml-offsite-end
 761         rss-offsite-in  rdf-offsite-in  xml-offsite-in)
 762     (dolist (href hrefs)
 763       (cond ((null href))
 764             ((string-match "\\.rss$" href)
 765              (nnrss-match-macro
 766               base-uri href rss-onsite-end rss-offsite-end))
 767             ((string-match "\\.rdf$" href)
 768              (nnrss-match-macro
 769               base-uri href rdf-onsite-end rdf-offsite-end))
 770             ((string-match "\\.xml$" href)
 771              (nnrss-match-macro
 772               base-uri href xml-onsite-end xml-offsite-end))
 773             ((string-match "rss" href)
 774              (nnrss-match-macro
 775               base-uri href rss-onsite-in rss-offsite-in))
 776             ((string-match "rdf" href)
 777              (nnrss-match-macro
 778               base-uri href rdf-onsite-in rdf-offsite-in))
 779             ((string-match "xml" href)
 780              (nnrss-match-macro
 781               base-uri href xml-onsite-in xml-offsite-in))))
 782     (append
 783      rss-onsite-end  rdf-onsite-end  xml-onsite-end
 784      rss-onsite-in   rdf-onsite-in   xml-onsite-in
 785      rss-offsite-end rdf-offsite-end xml-offsite-end
 786      rss-offsite-in rdf-offsite-in xml-offsite-in)))
 787
 788 (defun nnrss-discover-feed (url)
 789   "Given a page, find an RSS feed using Mark Pilgrim's
 790 `ultra-liberal rss locator' (http://diveintomark.org/2002/08/15.html)."
 791
 792   (let ((parsed-page (nnrss-fetch url)))
 793
 794 ;;    1. if this url is the rss, use it.
 795     (if (nnrss-rss-p parsed-page)
 796         (let ((rss-ns (nnrss-get-namespace-prefix parsed-page "http://purl.org/rss/1.0/")))
 797           (nnrss-rss-title-description rss-ns parsed-page url))
 798
 799 ;;    2. look for the <link rel="alternate"
 800 ;;    type="application/rss+xml" and use that if it is there.
 801       (let ((links (nnrss-get-rsslinks parsed-page)))
 802         (if links
 803             (let* ((xml (nnrss-fetch
 804                          (cdr (assoc 'href (cadar links)))))
 805                    (rss-ns (nnrss-get-namespace-prefix xml "http://purl.org/rss/1.0/")))
 806               (nnrss-rss-title-description rss-ns xml (cdr (assoc 'href (cadar links)))))
 807
 808 ;;    3. look for links on the site in the following order:
 809 ;;       - onsite links ending in .rss, .rdf, or .xml
 810 ;;       - onsite links containing any of the above
 811 ;;       - offsite links ending in .rss, .rdf, or .xml
 812 ;;       - offsite links containing any of the above
 813           (let* ((base-uri (progn (string-match ".*://[^/]+/?" url)
 814                                   (match-string 0 url)))
 815                  (hrefs (nnrss-order-hrefs
 816                          base-uri (nnrss-extract-hrefs parsed-page)))
 817                  (rss-link nil))
 818             (while (and (eq rss-link nil) (not (eq hrefs nil)))
 819               (let ((href-data (nnrss-fetch (car hrefs))))
 820                 (if (nnrss-rss-p href-data)
 821                     (let* ((rss-ns (nnrss-get-namespace-prefix href-data "http://purl.org/rss/1.0/")))
 822                       (setq rss-link (nnrss-rss-title-description
 823                                       rss-ns href-data (car hrefs))))
 824                   (setq hrefs (cdr hrefs)))))
 825             (if rss-link rss-link
 826
 827 ;;    4. check syndic8
 828               (nnrss-find-rss-via-syndic8 url))))))))
 829
 830 (defun nnrss-find-rss-via-syndic8 (url)
 831   "Query syndic8 for the rss feeds it has for URL."
 832   (if (not (locate-library "xml-rpc"))
 833       (progn
 834         (message "XML-RPC is not available... not checking Syndic8.")
 835         nil)
 836     (require 'xml-rpc)
 837     (let ((feedid (xml-rpc-method-call
 838                    "http://www.syndic8.com/xmlrpc.php"
 839                    'syndic8.FindSites
 840                    url)))
 841       (when feedid
 842         (let* ((feedinfo (xml-rpc-method-call
 843                           "http://www.syndic8.com/xmlrpc.php"
 844                           'syndic8.GetFeedInfo
 845                           feedid))
 846                (urllist
 847                 (delq nil
 848                       (mapcar
 849                        (lambda (listinfo)
 850                          (if (string-equal
 851                               (cdr (assoc "status" listinfo))
 852                               "Syndicated")
 853                              (cons
 854                               (cdr (assoc "sitename" listinfo))
 855                               (list
 856                                (cons 'title
 857                                      (cdr (assoc
 858                                            "sitename" listinfo)))
 859                                (cons 'href
 860                                      (cdr (assoc
 861                                            "dataurl" listinfo)))))))
 862                        feedinfo))))
 863           (if (not (> (length urllist) 1))
 864               (cdar urllist)
 865             (let ((completion-ignore-case t)
 866                   (selection
 867                    (mapcar (lambda (listinfo)
 868                              (cons (cdr (assoc "sitename" listinfo))
 869                                    (string-to-int
 870                                     (cdr (assoc "feedid" listinfo)))))
 871                            feedinfo)))
 872               (cdr (assoc
 873                     (completing-read
 874                      "Multiple feeds found.  Select one: "
 875                      selection nil t) urllist)))))))))
 876
 877 (defun nnrss-rss-p (data)
 878   "Test if DATA is an RSS feed.
 879 Simply ensures that the first element is rss or rdf."
 880   (or (eq (caar data) 'rss)
 881       (eq (caar data) 'rdf:RDF)))
 882
 883 (defun nnrss-rss-title-description (rss-namespace data url)
 884   "Return the title of an RSS feed."
 885   (if (nnrss-rss-p data)
 886       (let ((description (intern (concat rss-namespace "description")))
 887             (title (intern (concat rss-namespace "title")))
 888             (channel (nnrss-find-el (intern (concat rss-namespace "channel"))
 889                                     data)))
 890         (list
 891          (cons 'description (caddr (nth 0 (nnrss-find-el description channel))))
 892          (cons 'title (caddr (nth 0 (nnrss-find-el title channel))))
 893          (cons 'href url)))))
 894
 895 (defun nnrss-get-namespace-prefix (el uri)
 896   "Given EL (containing a parsed element) and URI (containing a string
 897 that gives the URI for which you want to retrieve the namespace
 898 prefix), return the prefix."
 899   (let* ((prefix (car (rassoc uri (cadar el))))
 900          (nslist (if prefix
 901                      (split-string (symbol-name prefix) ":")))
 902          (ns (cond ((eq (length nslist) 1) ; no prefix given
 903                     "")
 904                    ((eq (length nslist) 2) ; extract prefix
 905                     (cadr nslist)))))
 906     (if (and ns (not (string= ns "")))
 907         (concat ns ":")
 908       ns)))
 909
 910 (provide 'nnrss)
 911
 912
 913 ;;; nnrss.el ends here
 914
 915 ;;; arch-tag: 12910c07-0cdf-44fb-8d2c-416ded64c267