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,
   4 ;;   2006 Free Software Foundation, Inc.
   5
   6 ;; Author: Shenghuo Zhu <zsh@cs.rochester.edu>
   7 ;; Keywords: RSS
   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
  13 ;; by the Free Software Foundation; either version 2, or (at your
  14 ;; option) any later version.
  15
  16 ;; GNU Emacs is distributed in the hope that it will be useful, but
  17 ;; WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  19 ;; General Public License for more details.
  20
  21 ;; You should have received a copy of the GNU General Public License
  22 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  23 ;; Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
  24 ;; Boston, MA 02110-1301, USA.
  25
  26 ;;; Commentary:
  27
  28 ;;; Code:
  29
  30 (eval-when-compile (require 'cl))
  31
  32 (require 'gnus)
  33 (require 'nnoo)
  34 (require 'nnmail)
  35 (require 'message)
  36 (require 'mm-util)
  37 (require 'gnus-util)
  38 (require 'time-date)
  39 (require 'rfc2231)
  40 (require 'mm-url)
  41 (require 'rfc2047)
  42 (require 'mml)
  43 (eval-when-compile
  44   (ignore-errors
  45    (require 'xml)))
  46 (eval '(require 'xml))
  47
  48 (nnoo-declare nnrss)
  49
  50 (defvoo nnrss-directory (nnheader-concat gnus-directory "rss/")
  51   "Where nnrss will save its files.")
  52
  53 ;; (group max rss-url)
  54 (defvoo nnrss-server-data nil)
  55
  56 ;; (num timestamp url subject author date extra)
  57 (defvoo nnrss-group-data nil)
  58 (defvoo nnrss-group-max 0)
  59 (defvoo nnrss-group-min 1)
  60 (defvoo nnrss-group nil)
  61 (defvoo nnrss-group-hashtb nil)
  62 (defvoo nnrss-status-string "")
  63
  64 (defconst nnrss-version "nnrss 1.0")
  65
  66 (defvar nnrss-group-alist '()
  67   "List of RSS addresses.")
  68
  69 (defvar nnrss-use-local nil)
  70
  71 (defvar nnrss-description-field 'X-Gnus-Description
  72   "Field name used for DESCRIPTION.
  73 To use the description in headers, put this name into `nnmail-extra-headers'.")
  74
  75 (defvar nnrss-url-field 'X-Gnus-Url
  76   "Field name used for URL.
  77 To use the description in headers, put this name into `nnmail-extra-headers'.")
  78
  79 (defvar nnrss-content-function nil
  80   "A function which is called in `nnrss-request-article'.
  81 The arguments are (ENTRY GROUP ARTICLE).
  82 ENTRY is the record of the current headline.  GROUP is the group name.
  83 ARTICLE is the article number of the current headline.")
  84
  85 (defvar nnrss-file-coding-system mm-universal-coding-system
  86   "Coding system used when reading and writing files.")
  87
  88 (defvar nnrss-compatible-encoding-alist '((iso-8859-1 . windows-1252))
  89   "Alist of encodings and those supersets.
  90 The cdr of each element is used to decode data if it is available when
  91 the car is what the data specify as the encoding.  Or, the car is used
  92 for decoding when the cdr that the data specify is not available.")
  93
  94 (defvar nnrss-wash-html-in-text-plain-parts nil
  95   "*Non-nil means render text in text/plain parts as HTML.
  96 The function specified by the `mm-text-html-renderer' variable will be
  97 used to render text.  If it is nil, text will simply be folded.")
  98
  99 (nnoo-define-basics nnrss)
 100
 101 ;;; Interface functions
 102
 103 (defsubst nnrss-format-string (string)
 104   (gnus-replace-in-string string " *\n *" " "))
 105
 106 (defun nnrss-decode-group-name (group)
 107   (if (and group (mm-coding-system-p 'utf-8))
 108       (setq group (mm-decode-coding-string group 'utf-8))
 109     group))
 110
 111 (deffoo nnrss-retrieve-headers (articles &optional group server fetch-old)
 112   (setq group (nnrss-decode-group-name group))
 113   (nnrss-possibly-change-group group server)
 114   (let (e)
 115     (save-excursion
 116       (set-buffer nntp-server-buffer)
 117       (erase-buffer)
 118       (dolist (article articles)
 119         (if (setq e (assq article nnrss-group-data))
 120             (insert (number-to-string (car e)) "\t" ;; number
 121                     ;; subject
 122                     (or (nth 3 e) "")
 123                     "\t"
 124                     ;; from
 125                     (or (nth 4 e) "(nobody)")
 126                     "\t"
 127                     ;; date
 128                     (or (nth 5 e) "")
 129                     "\t"
 130                     ;; id
 131                     (format "<%d@%s.nnrss>" (car e) group)
 132                     "\t"
 133                     ;; refs
 134                     "\t"
 135                     ;; chars
 136                     "-1" "\t"
 137                     ;; lines
 138                     "-1" "\t"
 139                     ;; Xref
 140                     "" "\t"
 141                     (if (and (nth 6 e)
 142                              (memq nnrss-description-field
 143                                    nnmail-extra-headers))
 144                         (concat (symbol-name nnrss-description-field)
 145                                 ": "
 146                                 (nnrss-format-string (nth 6 e))
 147                                 "\t")
 148                       "")
 149                     (if (and (nth 2 e)
 150                              (memq nnrss-url-field
 151                                    nnmail-extra-headers))
 152                         (concat (symbol-name nnrss-url-field)
 153                                 ": "
 154                                 (nnrss-format-string (nth 2 e))
 155                                 "\t")
 156                       "")
 157                     "\n")))))
 158   'nov)
 159
 160 (deffoo nnrss-request-group (group &optional server dont-check)
 161   (setq group (nnrss-decode-group-name group))
 162   (nnheader-message 6 "nnrss: Requesting %s..." group)
 163   (nnrss-possibly-change-group group server)
 164   (prog1
 165       (if dont-check
 166           t
 167         (nnrss-check-group group server)
 168         (nnheader-report 'nnrss "Opened group %s" group)
 169         (nnheader-insert
 170          "211 %d %d %d %s\n" nnrss-group-max nnrss-group-min nnrss-group-max
 171          (prin1-to-string group)
 172          t))
 173     (nnheader-message 6 "nnrss: Requesting %s...done" group)))
 174
 175 (deffoo nnrss-close-group (group &optional server)
 176   t)
 177
 178 (eval-when-compile
 179   (defvar mm-text-html-renderer)
 180   (defvar mm-text-html-washer-alist))
 181
 182 (deffoo nnrss-request-article (article &optional group server buffer)
 183   (setq group (nnrss-decode-group-name group))
 184   (when (stringp article)
 185     (setq article (if (string-match "\\`<\\([0-9]+\\)@" article)
 186                       (string-to-number (match-string 1 article))
 187                     0)))
 188   (nnrss-possibly-change-group group server)
 189   (let ((e (assq article nnrss-group-data))
 190         (nntp-server-buffer (or buffer nntp-server-buffer))
 191         post err)
 192     (when e
 193       (with-current-buffer nntp-server-buffer
 194         (erase-buffer)
 195         (if group
 196             (insert "Newsgroups: " group "\n"))
 197         (if (nth 3 e)
 198             (insert "Subject: " (nth 3 e) "\n"))
 199         (if (nth 4 e)
 200             (insert "From: " (nth 4 e) "\n"))
 201         (if (nth 5 e)
 202             (insert "Date: " (nnrss-format-string (nth 5 e)) "\n"))
 203         (let ((header (buffer-string))
 204               (text (nth 6 e))
 205               (link (nth 2 e))
 206               (enclosure (nth 7 e))
 207               (comments (nth 8 e))
 208               ;; Enable encoding of Newsgroups header in XEmacs.
 209               (default-enable-multibyte-characters t)
 210               (rfc2047-header-encoding-alist
 211                (if (mm-coding-system-p 'utf-8)
 212                    (cons '("Newsgroups" . utf-8)
 213                          rfc2047-header-encoding-alist)
 214                  rfc2047-header-encoding-alist))
 215               rfc2047-encode-encoded-words body fn)
 216           (when (or text link enclosure comments)
 217             (insert "\n")
 218             (insert "<#multipart type=alternative>\n"
 219                     "<#part type=\"text/plain\">\n")
 220             (setq body (point))
 221             (when text
 222               (insert text)
 223               (goto-char body)
 224               (if (and nnrss-wash-html-in-text-plain-parts
 225                        (progn
 226                          (require 'mm-view)
 227                          (setq fn (or (cdr (assq mm-text-html-renderer
 228                                                  mm-text-html-washer-alist))
 229                                       mm-text-html-renderer))))
 230                   (progn
 231                     (narrow-to-region body (point-max))
 232                     (if (functionp fn)
 233                         (funcall fn)
 234                       (apply (car fn) (cdr fn)))
 235                     (widen)
 236                     (goto-char body)
 237                     (re-search-forward "[^\t\n ]" nil t)
 238                     (beginning-of-line)
 239                     (delete-region body (point))
 240                     (goto-char (point-max))
 241                     (skip-chars-backward "\t\n ")
 242                     (end-of-line)
 243                     (delete-region (point) (point-max))
 244                     (insert "\n"))
 245                 (while (re-search-forward "\n+" nil t)
 246                   (replace-match " "))
 247                 (goto-char body)
 248                 ;; See `nnrss-check-group', which inserts "<br /><br />".
 249                 (when (search-forward "<br /><br />" nil t)
 250                   (if (eobp)
 251                       (replace-match "\n")
 252                     (replace-match "\n\n")))
 253                 (unless (eobp)
 254                   (let ((fill-column default-fill-column)
 255                         (window (get-buffer-window nntp-server-buffer)))
 256                     (when window
 257                       (setq fill-column
 258                             (max 1 (/ (* (window-width window) 7) 8))))
 259                     (fill-region (point) (point-max))
 260                     (goto-char (point-max))
 261                     ;; XEmacs version of `fill-region' inserts newline.
 262                     (unless (bolp)
 263                       (insert "\n")))))
 264               (when (or link enclosure)
 265                 (insert "\n")))
 266             (when link
 267               (insert link "\n"))
 268             (when enclosure
 269               (insert (car enclosure) " "
 270                       (nth 2 enclosure) " "
 271                       (nth 3 enclosure) "\n"))
 272             (when comments
 273               (insert comments "\n"))
 274             (setq body (buffer-substring body (point)))
 275             (insert "<#/part>\n"
 276                     "<#part type=\"text/html\">\n"
 277                     "<html><head></head><body>\n")
 278             (when text
 279               (insert text "\n"))
 280             (when link
 281               (insert "<p><a href=\"" link "\">link</a></p>\n"))
 282             (when enclosure
 283               (insert "<p><a href=\"" (car enclosure) "\">"
 284                       (cadr enclosure) "</a> " (nth 2 enclosure)
 285                       " " (nth 3 enclosure) "</p>\n"))
 286             (when comments
 287               (insert "<p><a href=\"" comments "\">comments</a></p>\n"))
 288             (insert "</body></html>\n"
 289                     "<#/part>\n"
 290                     "<#/multipart>\n"))
 291           (condition-case nil
 292               (mml-to-mime)
 293             (error
 294              (erase-buffer)
 295              (insert header
 296                      "Content-Type: text/plain; charset=gnus-decoded\n"
 297                      "Content-Transfer-Encoding: 8bit\n\n"
 298                      body)
 299              (nnheader-message
 300               3 "Warning - there might be invalid characters"))))
 301         (goto-char (point-min))
 302         (search-forward "\n\n")
 303         (forward-line -1)
 304         (insert (format "Message-ID: <%d@%s.nnrss>\n"
 305                         (car e)
 306                         (let ((rfc2047-encoding-type 'mime)
 307                               rfc2047-encode-max-chars)
 308                           (rfc2047-encode-string
 309                            (gnus-replace-in-string group "[\t\n ]+" "_")))))
 310         (when nnrss-content-function
 311           (funcall nnrss-content-function e group article))))
 312     (cond
 313      (err
 314       (nnheader-report 'nnrss err))
 315      ((not e)
 316       (nnheader-report 'nnrss "no such id: %d" article))
 317      (t
 318       (nnheader-report 'nnrss "article %s retrieved" (car e))
 319       ;; we return the article number.
 320       (cons nnrss-group (car e))))))
 321
 322 (deffoo nnrss-request-list (&optional server)
 323   (nnrss-possibly-change-group nil server)
 324   (nnrss-generate-active)
 325   t)
 326
 327 (deffoo nnrss-open-server (server &optional defs connectionless)
 328   (nnrss-read-server-data server)
 329   (nnoo-change-server 'nnrss server defs)
 330   t)
 331
 332 (deffoo nnrss-request-expire-articles
 333     (articles group &optional server force)
 334   (setq group (nnrss-decode-group-name group))
 335   (nnrss-possibly-change-group group server)
 336   (let (e days not-expirable changed)
 337     (dolist (art articles)
 338       (if (and (setq e (assq art nnrss-group-data))
 339                (nnmail-expired-article-p
 340                 group
 341                 (if (listp (setq days (nth 1 e))) days
 342                   (days-to-time (- days (time-to-days '(0 0)))))
 343                 force))
 344           (setq nnrss-group-data (delq e nnrss-group-data)
 345                 changed t)
 346         (push art not-expirable)))
 347     (if changed
 348         (nnrss-save-group-data group server))
 349     not-expirable))
 350
 351 (deffoo nnrss-request-delete-group (group &optional force server)
 352   (setq group (nnrss-decode-group-name group))
 353   (nnrss-possibly-change-group group server)
 354   (let (elem)
 355     ;; There may be two or more entries in `nnrss-group-alist' since
 356     ;; this function didn't delete them formerly.
 357     (while (setq elem (assoc group nnrss-group-alist))
 358       (setq nnrss-group-alist (delq elem nnrss-group-alist))))
 359   (setq nnrss-server-data
 360         (delq (assoc group nnrss-server-data) nnrss-server-data))
 361   (nnrss-save-server-data server)
 362   (ignore-errors
 363    (delete-file (nnrss-make-filename group server)))
 364   t)
 365
 366 (deffoo nnrss-request-list-newsgroups (&optional server)
 367   (nnrss-possibly-change-group nil server)
 368   (save-excursion
 369     (set-buffer nntp-server-buffer)
 370     (erase-buffer)
 371     (dolist (elem nnrss-group-alist)
 372       (if (third elem)
 373           (insert (car elem) "\t" (third elem) "\n"))))
 374   t)
 375
 376 (nnoo-define-skeleton nnrss)
 377
 378 ;;; Internal functions
 379 (eval-when-compile (defun xml-rpc-method-call (&rest args)))
 380
 381 (defun nnrss-get-encoding ()
 382   "Return an encoding attribute specified in the current xml contents.
 383 If `nnrss-compatible-encoding-alist' specifies the compatible encoding,
 384 it is used instead.  If the xml contents doesn't specify the encoding,
 385 return `utf-8' which is the default encoding for xml if it is available,
 386 otherwise return nil."
 387   (goto-char (point-min))
 388   (if (re-search-forward
 389        "<\\?[^>]*encoding=\\(\"\\([^\">]+\\)\"\\|'\\([^'>]+\\)'\\)"
 390        nil t)
 391       (let ((encoding (intern (downcase (or (match-string 2)
 392                                             (match-string 3))))))
 393         (or
 394          (mm-coding-system-p (cdr (assq encoding
 395                                         nnrss-compatible-encoding-alist)))
 396          (mm-coding-system-p encoding)
 397          (mm-coding-system-p (car (rassq encoding
 398                                          nnrss-compatible-encoding-alist)))))
 399     (mm-coding-system-p 'utf-8)))
 400
 401 (defun nnrss-fetch (url &optional local)
 402   "Fetch URL and put it in a the expected Lisp structure."
 403   (mm-with-unibyte-buffer
 404     ;;some CVS versions of url.el need this to close the connection quickly
 405     (let (cs xmlform htmlform)
 406       ;; bit o' work necessary for w3 pre-cvs and post-cvs
 407       (if local
 408           (let ((coding-system-for-read 'binary))
 409             (insert-file-contents url))
 410         ;; FIXME: shouldn't binding `coding-system-for-read' be moved
 411         ;; to `mm-url-insert'?
 412         (let ((coding-system-for-read 'binary))
 413           (condition-case err
 414               (mm-url-insert url)
 415             (error (if (or debug-on-quit debug-on-error)
 416                        (signal (car err) (cdr err))
 417                      (message "nnrss: Failed to fetch %s" url))))))
 418       (nnheader-remove-cr-followed-by-lf)
 419       ;; Decode text according to the encoding attribute.
 420       (when (setq cs (nnrss-get-encoding))
 421         (mm-decode-coding-region (point-min) (point-max) cs)
 422         (mm-enable-multibyte))
 423       (goto-char (point-min))
 424
 425       ;; Because xml-parse-region can't deal with anything that isn't
 426       ;; xml and w3-parse-buffer can't deal with some xml, we have to
 427       ;; parse with xml-parse-region first and, if that fails, parse
 428       ;; with w3-parse-buffer.  Yuck.  Eventually, someone should find out
 429       ;; why w3-parse-buffer fails to parse some well-formed xml and
 430       ;; fix it.
 431
 432       (condition-case err1
 433           (setq xmlform (xml-parse-region (point-min) (point-max)))
 434         (error
 435          (condition-case err2
 436              (setq htmlform (caddar (w3-parse-buffer
 437                                      (current-buffer))))
 438            (error
 439             (message "\
 440 nnrss: %s: Not valid XML %s and w3-parse doesn't work %s"
 441                      url err1 err2)))))
 442       (if htmlform
 443           htmlform
 444         xmlform))))
 445
 446 (defun nnrss-possibly-change-group (&optional group server)
 447   (when (and server
 448              (not (nnrss-server-opened server)))
 449     (nnrss-open-server server))
 450   (when (and group (not (equal group nnrss-group)))
 451     (nnrss-read-group-data group server)
 452     (setq nnrss-group group)))
 453
 454 (defvar nnrss-extra-categories '(nnrss-snarf-moreover-categories))
 455
 456 (defun nnrss-generate-active ()
 457   (when (y-or-n-p "Fetch extra categories? ")
 458     (dolist (func nnrss-extra-categories)
 459       (funcall func)))
 460   (save-excursion
 461     (set-buffer nntp-server-buffer)
 462     (erase-buffer)
 463     (dolist (elem nnrss-group-alist)
 464       (insert (prin1-to-string (car elem)) " 0 1 y\n"))
 465     (dolist (elem nnrss-server-data)
 466       (unless (assoc (car elem) nnrss-group-alist)
 467         (insert (prin1-to-string (car elem)) " 0 1 y\n")))))
 468
 469 (eval-and-compile (autoload 'timezone-parse-date "timezone"))
 470
 471 (defun nnrss-normalize-date (date)
 472   "Return a date string of DATE in the RFC822 style.
 473 This function handles the ISO 8601 date format described in
 474 <URL:http://www.w3.org/TR/NOTE-datetime>, and also the RFC822 style
 475 which RSS 2.0 allows."
 476   (let (case-fold-search vector year month day time zone cts)
 477     (cond ((null date))
 478           ;; RFC822
 479           ((string-match " [0-9]+ " date)
 480            (setq vector (timezone-parse-date date)
 481                  year (string-to-number (aref vector 0)))
 482            (when (>= year 1969)
 483              (setq month (string-to-number (aref vector 1))
 484                    day (string-to-number (aref vector 2)))
 485              (unless (>= (length (setq time (aref vector 3))) 3)
 486                (setq time "00:00:00"))
 487              (when (and (setq zone (aref vector 4))
 488                         (not (string-match "\\`[A-Z+-]" zone)))
 489                (setq zone nil))))
 490           ;; ISO 8601
 491           ((string-match
 492             (eval-when-compile
 493               (concat
 494                ;; 1. year
 495                "\\(199[0-9]\\|20[0-9][0-9]\\)"
 496                "\\(-"
 497                ;; 3. month
 498                "\\([01][0-9]\\)"
 499                "\\(-"
 500                ;; 5. day
 501                "\\([0-3][0-9]\\)"
 502                "\\)?\\)?\\(T"
 503                ;; 7. hh:mm
 504                "\\([012][0-9]:[0-5][0-9]\\)"
 505                "\\("
 506                ;; 9. :ss
 507                "\\(:[0-5][0-9]\\)"
 508                "\\(\\.[0-9]+\\)?\\)?\\)?"
 509                ;; 13+14,15,16. zone
 510                "\\(\\(\\([+-][012][0-9]\\):\\([0-5][0-9]\\)\\)"
 511                "\\|\\([+-][012][0-9][0-5][0-9]\\)"
 512                "\\|\\(Z\\)\\)?"))
 513             date)
 514            (setq year (string-to-number (match-string 1 date))
 515                  month (string-to-number (or (match-string 3 date) "1"))
 516                  day (string-to-number (or (match-string 5 date) "1"))
 517                  time (if (match-beginning 9)
 518                           (substring date (match-beginning 7) (match-end 9))
 519                         (concat (or (match-string 7 date) "00:00") ":00"))
 520                  zone (cond ((match-beginning 13)
 521                              (concat (match-string 13 date)
 522                                      (match-string 14 date)))
 523                             ((match-beginning 16) ;; Z
 524                              "+0000")
 525                             (t ;; nil if zone is not provided.
 526                              (match-string 15 date))))))
 527     (if month
 528         (progn
 529           (setq cts (current-time-string (encode-time 0 0 0 day month year)))
 530           (format "%s, %02d %s %04d %s%s"
 531                   (substring cts 0 3) day (substring cts 4 7) year time
 532                   (if zone
 533                       (concat " " zone)
 534                     "")))
 535       (message-make-date))))
 536
 537 ;;; data functions
 538
 539 (defun nnrss-read-server-data (server)
 540   (setq nnrss-server-data nil)
 541   (let ((file (nnrss-make-filename "nnrss" server)))
 542     (when (file-exists-p file)
 543       ;; In Emacs 21.3 and earlier, `load' doesn't support non-ASCII
 544       ;; file names.  So, we use `insert-file-contents' instead.
 545       (mm-with-multibyte-buffer
 546         (let ((coding-system-for-read nnrss-file-coding-system)
 547               (file-name-coding-system nnmail-pathname-coding-system))
 548           (insert-file-contents file)
 549           (eval-region (point-min) (point-max)))))))
 550
 551 (defun nnrss-save-server-data (server)
 552   (gnus-make-directory nnrss-directory)
 553   (let ((coding-system-for-write nnrss-file-coding-system)
 554         (file-name-coding-system nnmail-pathname-coding-system))
 555     (with-temp-file (nnrss-make-filename "nnrss" server)
 556       (insert (format ";; -*- coding: %s; -*-\n"
 557                       nnrss-file-coding-system))
 558       (gnus-prin1 `(setq nnrss-group-alist ',nnrss-group-alist))
 559       (insert "\n")
 560       (gnus-prin1 `(setq nnrss-server-data ',nnrss-server-data)))))
 561
 562 (defun nnrss-read-group-data (group server)
 563   (setq nnrss-group-data nil)
 564   (setq nnrss-group-hashtb (gnus-make-hashtable))
 565   (let ((pair (assoc group nnrss-server-data)))
 566     (setq nnrss-group-max (or (cadr pair) 0))
 567     (setq nnrss-group-min (+ nnrss-group-max 1)))
 568   (let ((file (nnrss-make-filename group server)))
 569     (when (file-exists-p file)
 570       ;; In Emacs 21.3 and earlier, `load' doesn't support non-ASCII
 571       ;; file names.  So, we use `insert-file-contents' instead.
 572       (mm-with-multibyte-buffer
 573         (let ((coding-system-for-read nnrss-file-coding-system)
 574               (file-name-coding-system nnmail-pathname-coding-system))
 575           (insert-file-contents file)
 576           (eval-region (point-min) (point-max))))
 577       (dolist (e nnrss-group-data)
 578         (gnus-sethash (or (nth 2 e) (nth 6 e)) t nnrss-group-hashtb)
 579         (when (and (car e) (> nnrss-group-min (car e)))
 580           (setq nnrss-group-min (car e)))
 581         (when (and (car e) (< nnrss-group-max (car e)))
 582           (setq nnrss-group-max (car e)))))))
 583
 584 (defun nnrss-save-group-data (group server)
 585   (gnus-make-directory nnrss-directory)
 586   (let ((coding-system-for-write nnrss-file-coding-system)
 587         (file-name-coding-system nnmail-pathname-coding-system))
 588     (with-temp-file (nnrss-make-filename group server)
 589       (insert (format ";; -*- coding: %s; -*-\n"
 590                       nnrss-file-coding-system))
 591       (gnus-prin1 `(setq nnrss-group-data ',nnrss-group-data)))))
 592
 593 (defun nnrss-make-filename (name server)
 594   (expand-file-name
 595    (nnrss-translate-file-chars
 596     (concat name
 597             (and server
 598                  (not (equal server ""))
 599                  "-")
 600             server
 601             ".el"))
 602    nnrss-directory))
 603
 604 (gnus-add-shutdown 'nnrss-close 'gnus)
 605
 606 (defun nnrss-close ()
 607   "Clear internal nnrss variables."
 608   (setq nnrss-group-data nil
 609         nnrss-server-data nil
 610         nnrss-group-hashtb nil
 611         nnrss-group-alist nil))
 612
 613 ;;; URL interface
 614
 615 (defun nnrss-no-cache (url)
 616   "")
 617
 618 (defun nnrss-insert-w3 (url)
 619   (mm-with-unibyte-current-buffer
 620     (condition-case err
 621         (mm-url-insert url)
 622       (error (if (or debug-on-quit debug-on-error)
 623                  (signal (car err) (cdr err))
 624                (message "nnrss: Failed to fetch %s" url))))))
 625
 626 (defun nnrss-decode-entities-string (string)
 627   (if string
 628       (mm-with-multibyte-buffer
 629         (insert string)
 630         (mm-url-decode-entities-nbsp)
 631         (buffer-string))))
 632
 633 (defalias 'nnrss-insert 'nnrss-insert-w3)
 634
 635 (defun nnrss-mime-encode-string (string)
 636   (mm-with-multibyte-buffer
 637     (insert string)
 638     (mm-url-decode-entities-nbsp)
 639     (goto-char (point-min))
 640     (while (re-search-forward "[\t\n ]+" nil t)
 641       (replace-match " "))
 642     (goto-char (point-min))
 643     (skip-chars-forward " ")
 644     (delete-region (point-min) (point))
 645     (goto-char (point-max))
 646     (skip-chars-forward " ")
 647     (delete-region (point) (point-max))
 648     (let ((rfc2047-encoding-type 'mime)
 649           rfc2047-encode-max-chars)
 650       (rfc2047-encode-region (point-min) (point-max)))
 651     (goto-char (point-min))
 652     (while (search-forward "\n" nil t)
 653       (delete-backward-char 1))
 654     (buffer-string)))
 655
 656 ;;; Snarf functions
 657
 658 (defun nnrss-check-group (group server)
 659   (let (file xml subject url extra changed author date feed-subject
 660              enclosure comments rss-ns rdf-ns content-ns dc-ns)
 661     (if (and nnrss-use-local
 662              (file-exists-p (setq file (expand-file-name
 663                                         (nnrss-translate-file-chars
 664                                          (concat group ".xml"))
 665                                         nnrss-directory))))
 666         (setq xml (nnrss-fetch file t))
 667       (setq url (or (nth 2 (assoc group nnrss-server-data))
 668                     (second (assoc group nnrss-group-alist))))
 669       (unless url
 670         (setq url
 671               (cdr
 672                (assoc 'href
 673                       (nnrss-discover-feed
 674                        (read-string
 675                         (format "URL to search for %s: " group) "http://")))))
 676         (let ((pair (assoc group nnrss-server-data)))
 677           (if pair
 678               (setcdr (cdr pair) (list url))
 679             (push (list group nnrss-group-max url) nnrss-server-data)))
 680         (setq changed t))
 681       (setq xml (nnrss-fetch url)))
 682     ;; See
 683     ;; http://feeds.archive.org/validator/docs/howto/declare_namespaces.html
 684     ;; for more RSS namespaces.
 685     (setq dc-ns (nnrss-get-namespace-prefix xml "http://purl.org/dc/elements/1.1/")
 686           rdf-ns (nnrss-get-namespace-prefix xml "http://www.w3.org/1999/02/22-rdf-syntax-ns#")
 687           rss-ns (nnrss-get-namespace-prefix xml "http://purl.org/rss/1.0/")
 688           content-ns (nnrss-get-namespace-prefix xml "http://purl.org/rss/1.0/modules/content/"))
 689     (dolist (item (nreverse (nnrss-find-el (intern (concat rss-ns "item")) xml)))
 690       (when (and (listp item)
 691                  (string= (concat rss-ns "item") (car item))
 692                  (if (setq url (nnrss-decode-entities-string
 693                                 (nnrss-node-text rss-ns 'link (cddr item))))
 694                      (not (gnus-gethash url nnrss-group-hashtb))
 695                    (setq extra (or (nnrss-node-text content-ns 'encoded item)
 696                                    (nnrss-node-text rss-ns 'description item)))
 697                    (not (gnus-gethash extra nnrss-group-hashtb))))
 698         (setq subject (nnrss-node-text rss-ns 'title item))
 699         (setq extra (or extra
 700                         (nnrss-node-text content-ns 'encoded item)
 701                         (nnrss-node-text rss-ns 'description item)))
 702         (if (setq feed-subject (nnrss-node-text dc-ns 'subject item))
 703             (setq extra (concat feed-subject "<br /><br />" extra)))
 704         (setq author (or (nnrss-node-text rss-ns 'author item)
 705                          (nnrss-node-text dc-ns 'creator item)
 706                          (nnrss-node-text dc-ns 'contributor item)))
 707         (setq date (nnrss-normalize-date
 708                     (or (nnrss-node-text dc-ns 'date item)
 709                         (nnrss-node-text rss-ns 'pubDate item))))
 710         (setq comments (nnrss-node-text rss-ns 'comments item))
 711         (when (setq enclosure (cadr (assq (intern (concat rss-ns "enclosure")) item)))
 712           (let ((url (cdr (assq 'url enclosure)))
 713                 (len (cdr (assq 'length enclosure)))
 714                 (type (cdr (assq 'type enclosure)))
 715                 (name))
 716             (setq len
 717                   (if (and len (integerp (setq len (string-to-number len))))
 718                       ;; actually already in `ls-lisp-format-file-size' but
 719                       ;; probably not worth to require it for one function
 720                       (do ((size (/ len 1.0) (/ size 1024.0))
 721                            (post-fixes (list "" "k" "M" "G" "T" "P" "E")
 722                                        (cdr post-fixes)))
 723                           ((< size 1024)
 724                            (format "%.1f%s" size (car post-fixes))))
 725                     "0"))
 726             (setq url (or url ""))
 727             (setq name (if (string-match "/\\([^/]*\\)$" url)
 728                            (match-string 1 url)
 729                          "file"))
 730             (setq type (or type ""))
 731             (setq enclosure (list url name len type))))
 732         (push
 733          (list
 734           (incf nnrss-group-max)
 735           (current-time)
 736           url
 737           (and subject (nnrss-mime-encode-string subject))
 738           (and author (nnrss-mime-encode-string author))
 739           date
 740           (and extra (nnrss-decode-entities-string extra))
 741           enclosure
 742           comments)
 743          nnrss-group-data)
 744         (gnus-sethash (or url extra) t nnrss-group-hashtb)
 745         (setq changed t))
 746       (setq extra nil))
 747     (when changed
 748       (nnrss-save-group-data group server)
 749       (let ((pair (assoc group nnrss-server-data)))
 750         (if pair
 751             (setcar (cdr pair) nnrss-group-max)
 752           (push (list group nnrss-group-max) nnrss-server-data)))
 753       (nnrss-save-server-data server))))
 754
 755 (defun nnrss-opml-import (opml-file)
 756   "OPML subscriptions import.
 757 Read the file and attempt to subscribe to each Feed in the file."
 758   (interactive "fImport file: ")
 759   (mapc
 760    (lambda (node)
 761      (let ((xmlurl (cdr (assq 'xmlUrl (cadr node)))))
 762        (when (and xmlurl
 763                   (not (string-match "\\`[\t ]*\\'" xmlurl))
 764                   (prog1
 765                       (y-or-n-p (format "Subscribe to %s " xmlurl))
 766                     (message "")))
 767          (condition-case err
 768              (progn
 769                (gnus-group-make-rss-group xmlurl)
 770                (forward-line 1))
 771            (error
 772             (message
 773              "Failed to subscribe to %s (%s); type any key to continue: "
 774              xmlurl
 775              (error-message-string err))
 776             (let ((echo-keystrokes 0))
 777               (read-char)))))))
 778    (nnrss-find-el 'outline
 779                   (mm-with-multibyte-buffer
 780                     (insert-file-contents opml-file)
 781                     (xml-parse-region (point-min) (point-max))))))
 782
 783 (defun nnrss-opml-export ()
 784   "OPML subscription export.
 785 Export subscriptions to a buffer in OPML Format."
 786   (interactive)
 787   (with-current-buffer (get-buffer-create "*OPML Export*")
 788     (mm-set-buffer-file-coding-system 'utf-8)
 789     (insert "<?xml version=\"1.0\" encoding=\"utf-8\"?>\n"
 790             "<!-- OPML generated by Emacs Gnus' nnrss.el -->\n"
 791             "<opml version=\"1.1\">\n"
 792             "  <head>\n"
 793             "    <title>mySubscriptions</title>\n"
 794             "    <dateCreated>" (format-time-string "%a, %d %b %Y %T %z")
 795             "</dateCreated>\n"
 796             "    <ownerEmail>" user-mail-address "</ownerEmail>\n"
 797             "    <ownerName>" (user-full-name) "</ownerName>\n"
 798             "  </head>\n"
 799             "  <body>\n")
 800     (dolist (sub nnrss-group-alist)
 801       (insert "    <outline text=\"" (car sub)
 802               "\" xmlUrl=\"" (cadr sub) "\"/>\n"))
 803     (insert "  </body>\n"
 804             "</opml>\n"))
 805   (pop-to-buffer "*OPML Export*")
 806   (when (fboundp 'sgml-mode)
 807     (sgml-mode)))
 808
 809 (defun nnrss-generate-download-script ()
 810   "Generate a download script in the current buffer.
 811 It is useful when `(setq nnrss-use-local t)'."
 812   (interactive)
 813   (insert "#!/bin/sh\n")
 814   (insert "WGET=wget\n")
 815   (insert "RSSDIR='" (expand-file-name nnrss-directory) "'\n")
 816   (dolist (elem nnrss-server-data)
 817     (let ((url (or (nth 2 elem)
 818                    (second (assoc (car elem) nnrss-group-alist)))))
 819       (insert "$WGET -q -O \"$RSSDIR\"/'"
 820               (nnrss-translate-file-chars (concat (car elem) ".xml"))
 821               "' '" url "'\n"))))
 822
 823 (defun nnrss-translate-file-chars (name)
 824   (let ((nnheader-file-name-translation-alist
 825          (append nnheader-file-name-translation-alist '((?' . ?_)))))
 826     (nnheader-translate-file-chars name)))
 827
 828 (defvar nnrss-moreover-url
 829   "http://w.moreover.com/categories/category_list_rss.html"
 830   "The url of moreover.com categories.")
 831
 832 (defun nnrss-snarf-moreover-categories ()
 833   "Snarf RSS links from moreover.com."
 834   (interactive)
 835   (let (category name url changed)
 836     (with-temp-buffer
 837       (nnrss-insert nnrss-moreover-url)
 838       (goto-char (point-min))
 839       (while (re-search-forward
 840               "<a name=\"\\([^\"]+\\)\">\\|<a href=\"\\(http://[^\"]*moreover\\.com[^\"]+page\\?c=\\([^\"&]+\\)&o=rss\\)" nil t)
 841         (if (match-string 1)
 842             (setq category (match-string 1))
 843           (setq url (match-string 2)
 844                 name (mm-url-decode-entities-string
 845                       (rfc2231-decode-encoded-string
 846                        (match-string 3))))
 847           (if category
 848               (setq name (concat category "." name)))
 849           (unless (assoc name nnrss-server-data)
 850             (setq changed t)
 851             (push (list name 0 url) nnrss-server-data)))))
 852     (if changed
 853         (nnrss-save-server-data ""))))
 854
 855 (defun nnrss-node-text (namespace local-name element)
 856   (let* ((node (assq (intern (concat namespace (symbol-name local-name)))
 857                      element))
 858          (text (if (and node (listp node))
 859                    (nnrss-node-just-text node)
 860                  node))
 861          (cleaned-text (if text
 862                            (gnus-replace-in-string
 863                             (gnus-replace-in-string
 864                              text "^[\000-\037\177]+\\|^ +\\| +$" "")
 865                             "\r\n" "\n"))))
 866     (if (string-equal "" cleaned-text)
 867         nil
 868       cleaned-text)))
 869
 870 (defun nnrss-node-just-text (node)
 871   (if (and node (listp node))
 872       (mapconcat 'nnrss-node-just-text (cddr node) " ")
 873     node))
 874
 875 (defun nnrss-find-el (tag data &optional found-list)
 876   "Find the all matching elements in the data.
 877 Careful with this on large documents!"
 878   (when (consp data)
 879     (dolist (bit data)
 880       (when (car-safe bit)
 881         (when (equal tag (car bit))
 882           ;; Old xml.el may return a list of string.
 883           (when (and (consp (caddr bit))
 884                      (stringp (caaddr bit)))
 885             (setcar (cddr bit) (caaddr bit)))
 886           (setq found-list
 887                 (append found-list
 888                         (list bit))))
 889         (if (and (consp (car-safe (caddr bit)))
 890                  (not (stringp (caddr bit))))
 891             (setq found-list
 892                   (append found-list
 893                           (nnrss-find-el
 894                            tag (caddr bit))))
 895           (setq found-list
 896                 (append found-list
 897                         (nnrss-find-el
 898                          tag (cddr bit))))))))
 899   found-list)
 900
 901 (defun nnrss-rsslink-p (el)
 902   "Test if the element we are handed is an RSS autodiscovery link."
 903   (and (eq (car-safe el) 'link)
 904        (string-equal (cdr (assoc 'rel (cadr el))) "alternate")
 905        (or (string-equal (cdr (assoc 'type (cadr el)))
 906                          "application/rss+xml")
 907            (string-equal (cdr (assoc 'type (cadr el))) "text/xml"))))
 908
 909 (defun nnrss-get-rsslinks (data)
 910   "Extract the <link> elements that are links to RSS from the parsed data."
 911   (delq nil (mapcar
 912              (lambda (el)
 913                (if (nnrss-rsslink-p el) el))
 914              (nnrss-find-el 'link data))))
 915
 916 (defun nnrss-extract-hrefs (data)
 917   "Recursively extract hrefs from a page's source.
 918 DATA should be the output of `xml-parse-region' or
 919 `w3-parse-buffer'."
 920   (mapcar (lambda (ahref)
 921             (cdr (assoc 'href (cadr ahref))))
 922           (nnrss-find-el 'a data)))
 923
 924 (defmacro nnrss-match-macro (base-uri item onsite-list offsite-list)
 925   `(cond ((or (string-match (concat "^" ,base-uri) ,item)
 926               (not (string-match "://" ,item)))
 927           (setq ,onsite-list (append ,onsite-list (list ,item))))
 928          (t (setq ,offsite-list (append ,offsite-list (list ,item))))))
 929
 930 (defun nnrss-order-hrefs (base-uri hrefs)
 931   "Given a list of hrefs, sort them using the following priorities:
 932   1. links ending in .rss
 933   2. links ending in .rdf
 934   3. links ending in .xml
 935   4. links containing the above
 936   5. offsite links
 937
 938 BASE-URI is used to determine the location of the links and
 939 whether they are `offsite' or `onsite'."
 940   (let (rss-onsite-end  rdf-onsite-end  xml-onsite-end
 941         rss-onsite-in   rdf-onsite-in   xml-onsite-in
 942         rss-offsite-end rdf-offsite-end xml-offsite-end
 943         rss-offsite-in  rdf-offsite-in  xml-offsite-in)
 944     (dolist (href hrefs)
 945       (cond ((null href))
 946             ((string-match "\\.rss$" href)
 947              (nnrss-match-macro
 948               base-uri href rss-onsite-end rss-offsite-end))
 949             ((string-match "\\.rdf$" href)
 950              (nnrss-match-macro
 951               base-uri href rdf-onsite-end rdf-offsite-end))
 952             ((string-match "\\.xml$" href)
 953              (nnrss-match-macro
 954               base-uri href xml-onsite-end xml-offsite-end))
 955             ((string-match "rss" href)
 956              (nnrss-match-macro
 957               base-uri href rss-onsite-in rss-offsite-in))
 958             ((string-match "rdf" href)
 959              (nnrss-match-macro
 960               base-uri href rdf-onsite-in rdf-offsite-in))
 961             ((string-match "xml" href)
 962              (nnrss-match-macro
 963               base-uri href xml-onsite-in xml-offsite-in))))
 964     (append
 965      rss-onsite-end  rdf-onsite-end  xml-onsite-end
 966      rss-onsite-in   rdf-onsite-in   xml-onsite-in
 967      rss-offsite-end rdf-offsite-end xml-offsite-end
 968      rss-offsite-in rdf-offsite-in xml-offsite-in)))
 969
 970 (defun nnrss-discover-feed (url)
 971   "Given a page, find an RSS feed using Mark Pilgrim's
 972 `ultra-liberal rss locator' (http://diveintomark.org/2002/08/15.html)."
 973
 974   (let ((parsed-page (nnrss-fetch url)))
 975
 976 ;;    1. if this url is the rss, use it.
 977     (if (nnrss-rss-p parsed-page)
 978         (let ((rss-ns (nnrss-get-namespace-prefix parsed-page "http://purl.org/rss/1.0/")))
 979           (nnrss-rss-title-description rss-ns parsed-page url))
 980
 981 ;;    2. look for the <link rel="alternate"
 982 ;;    type="application/rss+xml" and use that if it is there.
 983       (let ((links (nnrss-get-rsslinks parsed-page)))
 984         (if links
 985             (let* ((xml (nnrss-fetch
 986                          (cdr (assoc 'href (cadar links)))))
 987                    (rss-ns (nnrss-get-namespace-prefix xml "http://purl.org/rss/1.0/")))
 988               (nnrss-rss-title-description rss-ns xml (cdr (assoc 'href (cadar links)))))
 989
 990 ;;    3. look for links on the site in the following order:
 991 ;;       - onsite links ending in .rss, .rdf, or .xml
 992 ;;       - onsite links containing any of the above
 993 ;;       - offsite links ending in .rss, .rdf, or .xml
 994 ;;       - offsite links containing any of the above
 995           (let* ((base-uri (progn (string-match ".*://[^/]+/?" url)
 996                                   (match-string 0 url)))
 997                  (hrefs (nnrss-order-hrefs
 998                          base-uri (nnrss-extract-hrefs parsed-page)))
 999                  (rss-link nil))
1000             (while (and (eq rss-link nil) (not (eq hrefs nil)))
1001               (let ((href-data (nnrss-fetch (car hrefs))))
1002                 (if (nnrss-rss-p href-data)
1003                     (let* ((rss-ns (nnrss-get-namespace-prefix href-data "http://purl.org/rss/1.0/")))
1004                       (setq rss-link (nnrss-rss-title-description
1005                                       rss-ns href-data (car hrefs))))
1006                   (setq hrefs (cdr hrefs)))))
1007             (if rss-link rss-link
1008
1009 ;;    4. check syndic8
1010               (nnrss-find-rss-via-syndic8 url))))))))
1011
1012 (defun nnrss-find-rss-via-syndic8 (url)
1013   "Query syndic8 for the rss feeds it has for URL."
1014   (if (not (locate-library "xml-rpc"))
1015       (progn
1016         (message "XML-RPC is not available... not checking Syndic8.")
1017         nil)
1018     (require 'xml-rpc)
1019     (let ((feedid (xml-rpc-method-call
1020                    "http://www.syndic8.com/xmlrpc.php"
1021                    'syndic8.FindSites
1022                    url)))
1023       (when feedid
1024         (let* ((feedinfo (xml-rpc-method-call
1025                           "http://www.syndic8.com/xmlrpc.php"
1026                           'syndic8.GetFeedInfo
1027                           feedid))
1028                (urllist
1029                 (delq nil
1030                       (mapcar
1031                        (lambda (listinfo)
1032                          (if (string-equal
1033                               (cdr (assoc "status" listinfo))
1034                               "Syndicated")
1035                              (cons
1036                               (cdr (assoc "sitename" listinfo))
1037                               (list
1038                                (cons 'title
1039                                      (cdr (assoc
1040                                            "sitename" listinfo)))
1041                                (cons 'href
1042                                      (cdr (assoc
1043                                            "dataurl" listinfo)))))))
1044                        feedinfo))))
1045           (if (not (> (length urllist) 1))
1046               (cdar urllist)
1047             (let ((completion-ignore-case t)
1048                   (selection
1049                    (mapcar (lambda (listinfo)
1050                              (cons (cdr (assoc "sitename" listinfo))
1051                                    (string-to-number
1052                                     (cdr (assoc "feedid" listinfo)))))
1053                            feedinfo)))
1054               (cdr (assoc
1055                     (completing-read
1056                      "Multiple feeds found.  Select one: "
1057                      selection nil t) urllist)))))))))
1058
1059 (defun nnrss-rss-p (data)
1060   "Test if DATA is an RSS feed.
1061 Simply ensures that the first element is rss or rdf."
1062   (or (eq (caar data) 'rss)
1063       (eq (caar data) 'rdf:RDF)))
1064
1065 (defun nnrss-rss-title-description (rss-namespace data url)
1066   "Return the title of an RSS feed."
1067   (if (nnrss-rss-p data)
1068       (let ((description (intern (concat rss-namespace "description")))
1069             (title (intern (concat rss-namespace "title")))
1070             (channel (nnrss-find-el (intern (concat rss-namespace "channel"))
1071                                     data)))
1072         (list
1073          (cons 'description (caddr (nth 0 (nnrss-find-el description channel))))
1074          (cons 'title (caddr (nth 0 (nnrss-find-el title channel))))
1075          (cons 'href url)))))
1076
1077 (defun nnrss-get-namespace-prefix (el uri)
1078   "Given EL (containing a parsed element) and URI (containing a string
1079 that gives the URI for which you want to retrieve the namespace
1080 prefix), return the prefix."
1081   (let* ((prefix (car (rassoc uri (cadar el))))
1082          (nslist (if prefix
1083                      (split-string (symbol-name prefix) ":")))
1084          (ns (cond ((eq (length nslist) 1) ; no prefix given
1085                     "")
1086                    ((eq (length nslist) 2) ; extract prefix
1087                     (cadr nslist)))))
1088     (if (and ns (not (string= ns "")))
1089         (concat ns ":")
1090       ns)))
1091
1092 (provide 'nnrss)
1093
1094
1095 ;;; nnrss.el ends here
1096
1097 ;;; arch-tag: 12910c07-0cdf-44fb-8d2c-416ded64c267