lisp/gnus/nnrss.el

   1 ;;; nnrss.el --- interfacing with RSS
   2
   3 ;; Copyright (C) 2001-2017 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 by
  12 ;; the Free Software Foundation, either version 3 of the License, or
  13 ;; (at your option) any later version.
  14
  15 ;; GNU Emacs is distributed in the hope that it will be useful,
  16 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  17 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  18 ;; GNU General Public License for more details.
  19
  20 ;; You should have received a copy of the GNU General Public License
  21 ;; along with GNU Emacs.  If not, see <http://www.gnu.org/licenses/>.
  22
  23 ;;; Commentary:
  24
  25 ;;; Code:
  26
  27 (eval-when-compile (require 'cl))
  28
  29 (require 'gnus)
  30 (require 'nnoo)
  31 (require 'nnmail)
  32 (require 'message)
  33 (require 'mm-util)
  34 (require 'gnus-util)
  35 (require 'time-date)
  36 (require 'rfc2231)
  37 (require 'mm-url)
  38 (require 'rfc2047)
  39 (require 'mml)
  40 (require 'xml)
  41
  42 (defgroup nnrss nil
  43   "RSS access for Gnus."
  44   :group 'gnus)
  45
  46 (nnoo-declare nnrss)
  47
  48 (defvoo nnrss-directory (nnheader-concat gnus-directory "rss/")
  49   "Where nnrss will save its files.")
  50
  51 (defvoo nnrss-ignore-article-fields '(slash:comments)
  52   "*List of fields that should be ignored when comparing RSS articles.
  53 Some RSS feeds update article fields during their lives, e.g. to
  54 indicate the number of comments or the number of times the
  55 articles have been seen.  However, if there is a difference
  56 between the local article and the distant one, the latter is
  57 considered to be new.  To avoid this and discard some fields, set
  58 this variable to the list of fields to be ignored.")
  59
  60 ;; (group max rss-url)
  61 (defvoo nnrss-server-data nil)
  62
  63 ;; (num timestamp url subject author date extra)
  64 (defvoo nnrss-group-data nil)
  65 (defvoo nnrss-group-max 0)
  66 (defvoo nnrss-group-min 1)
  67 (defvoo nnrss-group nil)
  68 (defvoo nnrss-group-hashtb (make-hash-table :test 'equal))
  69 (defvoo nnrss-status-string "")
  70
  71 (defconst nnrss-version "nnrss 1.0")
  72
  73 (defvar nnrss-group-alist '()
  74   "List of RSS addresses.")
  75
  76 (defvar nnrss-use-local nil
  77   "If non-nil nnrss will read the feeds from local files in nnrss-directory.")
  78
  79 (defvar nnrss-description-field 'X-Gnus-Description
  80   "Field name used for DESCRIPTION.
  81 To use the description in headers, put this name into `nnmail-extra-headers'.")
  82
  83 (defvar nnrss-url-field 'X-Gnus-Url
  84   "Field name used for URL.
  85 To use the description in headers, put this name into `nnmail-extra-headers'.")
  86
  87 (defvar nnrss-content-function nil
  88   "A function which is called in `nnrss-request-article'.
  89 The arguments are (ENTRY GROUP ARTICLE).
  90 ENTRY is the record of the current headline.  GROUP is the group name.
  91 ARTICLE is the article number of the current headline.")
  92
  93 (defcustom nnrss-file-coding-system mm-universal-coding-system
  94   "Coding system used when reading and writing files.
  95 If you run Gnus with various versions of Emacsen, the value of this
  96 variable should be the coding system that all those Emacsen support.
  97 Note that you have to regenerate all the nnrss groups if you change
  98 the value.  Moreover, you should be patient even if you are made to
  99 read the same articles twice, that arises for the difference of the
 100 versions of xml.el."
 101   :group 'nnrss
 102   :type 'coding-system)
 103
 104 (defvar nnrss-compatible-encoding-alist
 105   (delq nil (mapcar (lambda (elem)
 106                       (if (and (mm-coding-system-p (car elem))
 107                                (mm-coding-system-p (cdr elem)))
 108                           elem))
 109                     mm-charset-override-alist))
 110   "Alist of encodings and those supersets.
 111 The cdr of each element is used to decode data if it is available when
 112 the car is what the data specify as the encoding.  Or, the car is used
 113 for decoding when the cdr that the data specify is not available.")
 114
 115 (nnoo-define-basics nnrss)
 116
 117 ;;; Interface functions
 118
 119 (defsubst nnrss-format-string (string)
 120   (replace-regexp-in-string " *\n *" " " string))
 121
 122 (defun nnrss-decode-group-name (group)
 123   (if (and group (mm-coding-system-p 'utf-8))
 124       (setq group (decode-coding-string group 'utf-8))
 125     group))
 126
 127 (deffoo nnrss-retrieve-headers (articles &optional group server fetch-old)
 128   (setq group (nnrss-decode-group-name group))
 129   (nnrss-possibly-change-group group server)
 130   (let (e)
 131     (with-current-buffer nntp-server-buffer
 132       (erase-buffer)
 133       (dolist (article articles)
 134         (if (setq e (assq article nnrss-group-data))
 135             (insert (number-to-string (car e)) "\t" ;; number
 136                     ;; subject
 137                     (or (nth 3 e) "")
 138                     "\t"
 139                     ;; from
 140                     (or (nth 4 e) "(nobody)")
 141                     "\t"
 142                     ;; date
 143                     (or (nth 5 e) "")
 144                     "\t"
 145                     ;; id
 146                     (format "<%d@%s.nnrss>" (car e) group)
 147                     "\t"
 148                     ;; refs
 149                     "\t"
 150                     ;; chars
 151                     "-1" "\t"
 152                     ;; lines
 153                     "-1" "\t"
 154                     ;; Xref
 155                     "" "\t"
 156                     (if (and (nth 6 e)
 157                              (memq nnrss-description-field
 158                                    nnmail-extra-headers))
 159                         (concat (symbol-name nnrss-description-field)
 160                                 ": "
 161                                 (nnrss-format-string (nth 6 e))
 162                                 "\t")
 163                       "")
 164                     (if (and (nth 2 e)
 165                              (memq nnrss-url-field
 166                                    nnmail-extra-headers))
 167                         (concat (symbol-name nnrss-url-field)
 168                                 ": "
 169                                 (nnrss-format-string (nth 2 e))
 170                                 "\t")
 171                       "")
 172                     "\n")))))
 173   'nov)
 174
 175 (deffoo nnrss-request-group (group &optional server dont-check info)
 176   (setq group (nnrss-decode-group-name group))
 177   (nnheader-message 6 "nnrss: Requesting %s..." group)
 178   (nnrss-possibly-change-group group server)
 179   (prog1
 180       (if dont-check
 181           t
 182         (nnrss-check-group group server)
 183         (nnheader-report 'nnrss "Opened group %s" group)
 184         (nnheader-insert
 185          "211 %d %d %d %s\n" nnrss-group-max nnrss-group-min nnrss-group-max
 186          (prin1-to-string group)
 187          t))
 188     (nnheader-message 6 "nnrss: Requesting %s...done" group)))
 189
 190 (deffoo nnrss-close-group (group &optional server)
 191   t)
 192
 193 (deffoo nnrss-request-article (article &optional group server buffer)
 194   (setq group (nnrss-decode-group-name group))
 195   (when (stringp article)
 196     (setq article (if (string-match "\\`<\\([0-9]+\\)@" article)
 197                       (string-to-number (match-string 1 article))
 198                     0)))
 199   (nnrss-possibly-change-group group server)
 200   (let ((e (assq article nnrss-group-data))
 201         (nntp-server-buffer (or buffer nntp-server-buffer))
 202         post err)
 203     (when e
 204       (with-current-buffer nntp-server-buffer
 205         (erase-buffer)
 206         (if group
 207             (insert "Newsgroups: " group "\n"))
 208         (if (nth 3 e)
 209             (insert "Subject: " (nth 3 e) "\n"))
 210         (if (nth 4 e)
 211             (insert "From: " (nth 4 e) "\n"))
 212         (if (nth 5 e)
 213             (insert "Date: " (nnrss-format-string (nth 5 e)) "\n"))
 214         (let ((header (buffer-string))
 215               (text (nth 6 e))
 216               (link (nth 2 e))
 217               (enclosure (nth 7 e))
 218               (comments (nth 8 e))
 219               (rfc2047-header-encoding-alist
 220                (if (mm-coding-system-p 'utf-8)
 221                    (cons '("Newsgroups" . utf-8)
 222                          rfc2047-header-encoding-alist)
 223                  rfc2047-header-encoding-alist))
 224               rfc2047-encode-encoded-words body fn)
 225           (when (or text link enclosure comments)
 226             (insert "\n")
 227             (insert "<#multipart type=alternative>\n"
 228                     "<#part type=\"text/plain\">\n")
 229             (setq body (point))
 230             (when text
 231               (insert text)
 232               (goto-char body)
 233               (while (re-search-forward "\n+" nil t)
 234                 (replace-match " "))
 235               (goto-char body)
 236               ;; See `nnrss-check-group', which inserts "<br /><br />".
 237               (when (search-forward "<br /><br />" nil t)
 238                 (if (eobp)
 239                     (replace-match "\n")
 240                   (replace-match "\n\n")))
 241               (unless (eobp)
 242                 (let ((fill-column (default-value 'fill-column))
 243                       (window (get-buffer-window nntp-server-buffer)))
 244                   (when window
 245                     (setq fill-column
 246                           (max 1 (/ (* (window-width window) 7) 8))))
 247                   (fill-region (point) (point-max))
 248                   (goto-char (point-max))
 249                   (unless (bolp)
 250                     (insert "\n"))))
 251               (when (or link enclosure)
 252                 (insert "\n")))
 253             (when link
 254               (insert link "\n"))
 255             (when enclosure
 256               (insert (car enclosure) " "
 257                       (nth 2 enclosure) " "
 258                       (nth 3 enclosure) "\n"))
 259             (when comments
 260               (insert comments "\n"))
 261             (setq body (buffer-substring body (point)))
 262             (insert "<#/part>\n"
 263                     "<#part type=\"text/html\">\n"
 264                     "<html><head></head><body>\n")
 265             (when text
 266               (insert text "\n"))
 267             (when link
 268               (insert "<p><a href=\"" link "\">link</a></p>\n"))
 269             (when enclosure
 270               (insert "<p><a href=\"" (car enclosure) "\">"
 271                       (cadr enclosure) "</a> " (nth 2 enclosure)
 272                       " " (nth 3 enclosure) "</p>\n"))
 273             (when comments
 274               (insert "<p><a href=\"" comments "\">comments</a></p>\n"))
 275             (insert "</body></html>\n"
 276                     "<#/part>\n"
 277                     "<#/multipart>\n"))
 278           (condition-case nil
 279               ;; Allow `mml-to-mime' to generate MIME article without
 280               ;; making inquiry to a user for unknown encoding.
 281               (let ((mml-confirmation-set
 282                      (cons 'unknown-encoding mml-confirmation-set)))
 283                 (mml-to-mime))
 284             (error
 285              (erase-buffer)
 286              (insert header
 287                      "Content-Type: text/plain; charset=gnus-decoded\n"
 288                      "Content-Transfer-Encoding: 8bit\n\n"
 289                      body)
 290              (nnheader-message
 291               3 "Warning - there might be invalid characters"))))
 292         (goto-char (point-min))
 293         (search-forward "\n\n")
 294         (forward-line -1)
 295         (insert (format "Message-ID: <%d@%s.nnrss>\n"
 296                         (car e)
 297                         (let ((rfc2047-encoding-type 'mime)
 298                               rfc2047-encode-max-chars)
 299                           (rfc2047-encode-string
 300                            (replace-regexp-in-string "[\t\n ]+" "_" group)))))
 301         (when nnrss-content-function
 302           (funcall nnrss-content-function e group article))))
 303     (cond
 304      (err
 305       (nnheader-report 'nnrss err))
 306      ((not e)
 307       (nnheader-report 'nnrss "no such id: %d" article))
 308      (t
 309       (nnheader-report 'nnrss "article %s retrieved" (car e))
 310       ;; we return the article number.
 311       (cons nnrss-group (car e))))))
 312
 313 (deffoo nnrss-open-server (server &optional defs connectionless)
 314   (nnrss-read-server-data server)
 315   (nnoo-change-server 'nnrss server defs)
 316   t)
 317
 318 (deffoo nnrss-request-expire-articles
 319     (articles group &optional server force)
 320   (setq group (nnrss-decode-group-name group))
 321   (nnrss-possibly-change-group group server)
 322   (let (e days not-expirable changed)
 323     (dolist (art articles)
 324       (if (and (setq e (assq art nnrss-group-data))
 325                (nnmail-expired-article-p
 326                 group
 327                 (if (listp (setq days (nth 1 e))) days
 328                   (days-to-time (- days (time-to-days '(0 0)))))
 329                 force))
 330           (setq nnrss-group-data (delq e nnrss-group-data)
 331                 changed t)
 332         (push art not-expirable)))
 333     (if changed
 334         (nnrss-save-group-data group server))
 335     not-expirable))
 336
 337 (deffoo nnrss-request-delete-group (group &optional force server)
 338   (setq group (nnrss-decode-group-name group))
 339   (nnrss-possibly-change-group group server)
 340   (let (elem)
 341     ;; There may be two or more entries in `nnrss-group-alist' since
 342     ;; this function didn't delete them formerly.
 343     (while (setq elem (assoc group nnrss-group-alist))
 344       (setq nnrss-group-alist (delq elem nnrss-group-alist))))
 345   (setq nnrss-server-data
 346         (delq (assoc group nnrss-server-data) nnrss-server-data))
 347   (nnrss-save-server-data server)
 348   (ignore-errors
 349     (let ((file-name-coding-system nnmail-pathname-coding-system))
 350       (delete-file (nnrss-make-filename group server))))
 351   t)
 352
 353 (deffoo nnrss-request-list-newsgroups (&optional server)
 354   (nnrss-possibly-change-group nil server)
 355   (with-current-buffer nntp-server-buffer
 356     (erase-buffer)
 357     (dolist (elem nnrss-group-alist)
 358       (if (third elem)
 359           (insert (car elem) "\t" (third elem) "\n"))))
 360   t)
 361
 362 (deffoo nnrss-retrieve-groups (groups &optional server)
 363   (dolist (group groups)
 364     (setq group (nnrss-decode-group-name group))
 365     (nnrss-possibly-change-group group server)
 366     (nnrss-check-group group server))
 367   (with-current-buffer nntp-server-buffer
 368     (erase-buffer)
 369     (dolist (group groups)
 370       (let ((elem (assoc (gnus-group-decoded-name group) nnrss-server-data)))
 371         (insert (format "%S %s 1 y\n" group (or (cadr elem) 0)))))
 372     'active))
 373
 374 (nnoo-define-skeleton nnrss)
 375
 376 ;;; Internal functions
 377 (defun nnrss-get-encoding ()
 378   "Return an encoding attribute specified in the current xml contents.
 379 If `nnrss-compatible-encoding-alist' specifies the compatible encoding,
 380 it is used instead.  If the xml contents doesn't specify the encoding,
 381 return `utf-8' which is the default encoding for xml if it is available,
 382 otherwise return nil."
 383   (goto-char (point-min))
 384   (if (re-search-forward
 385        "<\\?[^>]*encoding=\\(?:\"\\([^\">]+\\)\"\\|'\\([^'>]+\\)'\\)"
 386        nil t)
 387       (let ((encoding (intern (downcase (or (match-string 1)
 388                                             (match-string 2))))))
 389         (or
 390          (mm-coding-system-p (cdr (assq encoding
 391                                         nnrss-compatible-encoding-alist)))
 392          (mm-coding-system-p encoding)
 393          (mm-coding-system-p (car (rassq encoding
 394                                          nnrss-compatible-encoding-alist)))))
 395     (mm-coding-system-p 'utf-8)))
 396
 397 (declare-function libxml-parse-html-region "xml.c"
 398                   (start end &optional base-url discard-comments))
 399 (defun nnrss-fetch (url &optional local)
 400   "Fetch URL and put it in a the expected Lisp structure."
 401   (mm-with-unibyte-buffer
 402     ;;some versions of url.el need this to close the connection quickly
 403     (let (cs xmlform htmlform)
 404       ;; bit o' work necessary for w3 pre-cvs and post-cvs
 405       (if local
 406           (let ((coding-system-for-read 'binary))
 407             (insert-file-contents url))
 408         ;; FIXME: shouldn't binding `coding-system-for-read' be moved
 409         ;; to `mm-url-insert'?
 410         (let ((coding-system-for-read 'binary))
 411           (condition-case err
 412               (mm-url-insert url)
 413             (error (if (or debug-on-quit debug-on-error)
 414                        (signal (car err) (cdr err))
 415                      (message "nnrss: Failed to fetch %s" url))))))
 416       (nnheader-remove-cr-followed-by-lf)
 417       ;; Decode text according to the encoding attribute.
 418       (when (setq cs (nnrss-get-encoding))
 419         (insert (prog1
 420                     (decode-coding-string (buffer-string) cs)
 421                   (erase-buffer)
 422                   (mm-enable-multibyte))))
 423       (goto-char (point-min))
 424
 425       (condition-case err1
 426           (setq xmlform (xml-parse-region (point-min) (point-max)))
 427         (error
 428          (condition-case err2
 429              (setq htmlform (libxml-parse-html-region (point-min) (point-max)))
 430            (error
 431             (message "\
 432 nnrss: %s: Not valid XML %s and libxml-parse-html-region doesn't work %s"
 433                      url err1 err2)))))
 434       (if htmlform
 435           htmlform
 436         xmlform))))
 437
 438 (defun nnrss-possibly-change-group (&optional group server)
 439   (when (and server
 440              (not (nnrss-server-opened server)))
 441     (nnrss-open-server server))
 442   (when (and group (not (equal group nnrss-group)))
 443     (nnrss-read-group-data group server)
 444     (setq nnrss-group group)))
 445
 446 (autoload 'timezone-parse-date "timezone")
 447
 448 (defun nnrss-normalize-date (date)
 449   "Return a date string of DATE in the RFC822 style.
 450 This function handles the ISO 8601 date format described in
 451 URL `http://www.w3.org/TR/NOTE-datetime', and also the RFC822 style
 452 which RSS 2.0 allows."
 453   (let (case-fold-search vector year month day time zone cts given)
 454     (cond ((null date))                 ; do nothing for this case
 455           ;; if the date is just digits (unix time stamp):
 456           ((string-match "^[0-9]+$" date)
 457            (setq given (seconds-to-time (string-to-number date))))
 458           ;; RFC822
 459           ((string-match " [0-9]+ " date)
 460            (setq vector (timezone-parse-date date)
 461                  year (string-to-number (aref vector 0)))
 462            (when (>= year 1969)
 463              (setq month (string-to-number (aref vector 1))
 464                    day (string-to-number (aref vector 2)))
 465              (unless (>= (length (setq time (aref vector 3))) 3)
 466                (setq time "00:00:00"))
 467              (when (and (setq zone (aref vector 4))
 468                         (not (string-match "\\`[A-Z+-]" zone)))
 469                (setq zone nil))))
 470           ;; ISO 8601
 471           ((string-match
 472             (eval-when-compile
 473               (concat
 474                ;; 1. year
 475                "\\(199[0-9]\\|20[0-9][0-9]\\)"
 476                "\\(?:-"
 477                ;; 2. month
 478                "\\([01][0-9]\\)"
 479                "\\(?:-"
 480                ;; 3. day
 481                "\\([0-3][0-9]\\)"
 482                "\\)?\\)?\\(?:T"
 483                ;; 4. hh:mm
 484                "\\([012][0-9]:[0-5][0-9]\\)"
 485                "\\(?:"
 486                ;; 5. :ss
 487                "\\(:[0-5][0-9]\\)"
 488                "\\(?:\\.[0-9]+\\)?\\)?\\)?"
 489                ;; 6+7,8,9. zone
 490                "\\(?:\\(?:\\([+-][012][0-9]\\):\\([0-5][0-9]\\)\\)"
 491                "\\|\\([+-][012][0-9][0-5][0-9]\\)"
 492                "\\|\\(Z\\)\\)?"))
 493             date)
 494            (setq year (string-to-number (match-string 1 date))
 495                  month (string-to-number (or (match-string 2 date) "1"))
 496                  day (string-to-number (or (match-string 3 date) "1"))
 497                  time (if (match-beginning 5)
 498                           (substring date (match-beginning 4) (match-end 5))
 499                         (concat (or (match-string 4 date) "00:00") ":00"))
 500                  zone (cond ((match-beginning 6)
 501                              (concat (match-string 6 date)
 502                                      (match-string 7 date)))
 503                             ((match-beginning 9) ;; Z
 504                              "+0000")
 505                             (t ;; nil if zone is not provided.
 506                              (match-string 8 date))))))
 507     (if month
 508         (progn
 509           (setq cts (current-time-string (encode-time 0 0 0 day month year)))
 510           (format "%s, %02d %s %04d %s%s"
 511                   (substring cts 0 3) day (substring cts 4 7) year time
 512                   (if zone
 513                       (concat " " zone)
 514                     "")))
 515       (message-make-date given))))
 516
 517 ;;; data functions
 518
 519 (defun nnrss-read-server-data (server)
 520   (setq nnrss-server-data nil)
 521   (let ((file (nnrss-make-filename "nnrss" server))
 522         (file-name-coding-system nnmail-pathname-coding-system))
 523     (when (file-exists-p file)
 524       (load file nil t t))))
 525
 526 (defun nnrss-save-server-data (server)
 527   (gnus-make-directory nnrss-directory)
 528   (let ((coding-system-for-write nnrss-file-coding-system)
 529         (file-name-coding-system nnmail-pathname-coding-system))
 530     (with-temp-file (nnrss-make-filename "nnrss" server)
 531       (insert (format ";; -*- coding: %s; -*-\n"
 532                       nnrss-file-coding-system))
 533       (gnus-prin1 `(setq nnrss-group-alist ',nnrss-group-alist))
 534       (insert "\n")
 535       (gnus-prin1 `(setq nnrss-server-data ',nnrss-server-data)))))
 536
 537 (defun nnrss-read-group-data (group server)
 538   (setq nnrss-group-data nil)
 539   (if (hash-table-p nnrss-group-hashtb)
 540       (clrhash nnrss-group-hashtb)
 541     (setq nnrss-group-hashtb (make-hash-table :test 'equal)))
 542   (let ((pair (assoc group nnrss-server-data)))
 543     (setq nnrss-group-max (or (cadr pair) 0))
 544     (setq nnrss-group-min (+ nnrss-group-max 1)))
 545   (let ((file (nnrss-make-filename group server))
 546         (file-name-coding-system nnmail-pathname-coding-system))
 547     (when (file-exists-p file)
 548       (load file nil t t)
 549       (dolist (e nnrss-group-data)
 550         (puthash (nth 9 e) t nnrss-group-hashtb)
 551         (when (and (car e) (> nnrss-group-min (car e)))
 552           (setq nnrss-group-min (car e)))
 553         (when (and (car e) (< nnrss-group-max (car e)))
 554           (setq nnrss-group-max (car e)))))))
 555
 556 (defun nnrss-save-group-data (group server)
 557   (gnus-make-directory nnrss-directory)
 558   (let ((coding-system-for-write nnrss-file-coding-system)
 559         (file-name-coding-system nnmail-pathname-coding-system))
 560     (with-temp-file (nnrss-make-filename group server)
 561       (insert (format ";; -*- coding: %s; -*-\n"
 562                       nnrss-file-coding-system))
 563       (gnus-prin1 `(setq nnrss-group-data ',nnrss-group-data)))))
 564
 565 (defun nnrss-make-filename (name server)
 566   (expand-file-name
 567    (nnrss-translate-file-chars
 568     (concat name
 569             (and server
 570                  (not (equal server ""))
 571                  "-")
 572             server
 573             ".el"))
 574    nnrss-directory))
 575
 576 (gnus-add-shutdown 'nnrss-close 'gnus)
 577
 578 (defun nnrss-close ()
 579   "Clear internal nnrss variables."
 580   (setq nnrss-group-data nil
 581         nnrss-server-data nil
 582         nnrss-group-hashtb nil
 583         nnrss-group-alist nil))
 584
 585 ;;; URL interface
 586
 587 (defun nnrss-no-cache (url)
 588   "")
 589
 590 (defun nnrss-insert (url)
 591   (condition-case err
 592       (mm-url-insert url)
 593     (error (if (or debug-on-quit debug-on-error)
 594                (signal (car err) (cdr err))
 595              (message "nnrss: Failed to fetch %s" url)))))
 596
 597 (defun nnrss-decode-entities-string (string)
 598   (if string
 599       (mm-with-multibyte-buffer
 600         (insert string)
 601         (mm-url-decode-entities-nbsp)
 602         (buffer-string))))
 603
 604 (defun nnrss-mime-encode-string (string)
 605   (mm-with-multibyte-buffer
 606     (insert string)
 607     (mm-url-decode-entities-nbsp)
 608     (goto-char (point-min))
 609     (while (re-search-forward "[\t\n ]+" nil t)
 610       (replace-match " "))
 611     (goto-char (point-min))
 612     (skip-chars-forward " ")
 613     (delete-region (point-min) (point))
 614     (goto-char (point-max))
 615     (skip-chars-forward " ")
 616     (delete-region (point) (point-max))
 617     (let ((rfc2047-encoding-type 'mime)
 618           rfc2047-encode-max-chars)
 619       (rfc2047-encode-region (point-min) (point-max)))
 620     (goto-char (point-min))
 621     (while (search-forward "\n" nil t)
 622       (delete-char -1))
 623     (buffer-string)))
 624
 625 ;;; Snarf functions
 626 (defun nnrss-make-hash-index (item)
 627   (gnus-message 9 "nnrss: Making hash index of %s" (gnus-prin1-to-string item))
 628   (setq item (gnus-remove-if
 629               (lambda (field)
 630                 (when (listp field)
 631                   (memq (car field) nnrss-ignore-article-fields)))
 632               item))
 633   (md5 (gnus-prin1-to-string item)
 634        nil nil
 635        nnrss-file-coding-system))
 636
 637 (defun nnrss-check-group (group server)
 638   (let (file xml subject url extra changed author date feed-subject
 639              enclosure comments rss-ns rdf-ns content-ns dc-ns
 640              hash-index)
 641     (if (and nnrss-use-local
 642              (file-exists-p (setq file (expand-file-name
 643                                         (nnrss-translate-file-chars
 644                                          (concat group ".xml"))
 645                                         nnrss-directory))))
 646         (setq xml (nnrss-fetch file t))
 647       (setq url (or (nth 2 (assoc group nnrss-server-data))
 648                     (second (assoc group nnrss-group-alist))))
 649       (unless url
 650         (setq url
 651               (cdr
 652                (assoc 'href
 653                       (nnrss-discover-feed
 654                        (read-string
 655                         (format "URL to search for %s: " group) "http://")))))
 656         (let ((pair (assoc group nnrss-server-data)))
 657           (if pair
 658               (setcdr (cdr pair) (list url))
 659             (push (list group nnrss-group-max url) nnrss-server-data)))
 660         (setq changed t))
 661       (setq xml (nnrss-fetch url)))
 662     (setq dc-ns (nnrss-get-namespace-prefix xml "http://purl.org/dc/elements/1.1/")
 663           rdf-ns (nnrss-get-namespace-prefix xml "http://www.w3.org/1999/02/22-rdf-syntax-ns#")
 664           rss-ns (nnrss-get-namespace-prefix xml "http://purl.org/rss/1.0/")
 665           content-ns (nnrss-get-namespace-prefix xml "http://purl.org/rss/1.0/modules/content/"))
 666     (dolist (item (nreverse (nnrss-find-el (intern (concat rss-ns "item")) xml)))
 667       (when (and (listp item)
 668                  (string= (concat rss-ns "item") (car item))
 669                  (progn (setq hash-index (nnrss-make-hash-index item))
 670                         (not (gethash hash-index nnrss-group-hashtb))))
 671         (setq subject (nnrss-node-text rss-ns 'title item))
 672         (setq url (nnrss-decode-entities-string
 673                    (nnrss-node-text rss-ns 'link (cddr item))))
 674         (setq extra (or (nnrss-node-text content-ns 'encoded item)
 675                         (nnrss-node-text rss-ns 'description item)))
 676         (if (setq feed-subject (nnrss-node-text dc-ns 'subject item))
 677             (setq extra (concat feed-subject "<br /><br />" extra)))
 678         (setq author (or (nnrss-node-text rss-ns 'author item)
 679                          (nnrss-node-text dc-ns 'creator item)
 680                          (nnrss-node-text dc-ns 'contributor item)))
 681         (setq date (nnrss-normalize-date
 682                     (or (nnrss-node-text dc-ns 'date item)
 683                         (nnrss-node-text rss-ns 'pubDate item))))
 684         (setq comments (nnrss-node-text rss-ns 'comments item))
 685         (when (setq enclosure (cadr (assq (intern (concat rss-ns "enclosure")) item)))
 686           (let ((url (cdr (assq 'url enclosure)))
 687                 (len (cdr (assq 'length enclosure)))
 688                 (type (cdr (assq 'type enclosure)))
 689                 (name))
 690             (setq len
 691                   (if (and len (integerp (setq len (string-to-number len))))
 692                       ;; actually already in `ls-lisp-format-file-size' but
 693                       ;; probably not worth to require it for one function
 694                       (do ((size (/ len 1.0) (/ size 1024.0))
 695                            (post-fixes (list "" "k" "M" "G" "T" "P" "E")
 696                                        (cdr post-fixes)))
 697                           ((< size 1024)
 698                            (format "%.1f%s" size (car post-fixes))))
 699                     "0"))
 700             (setq url (or url ""))
 701             (setq name (if (string-match "/\\([^/]*\\)$" url)
 702                            (match-string 1 url)
 703                          "file"))
 704             (setq type (or type ""))
 705             (setq enclosure (list url name len type))))
 706         (push
 707          (list
 708           (incf nnrss-group-max)
 709           (current-time)
 710           url
 711           (and subject (nnrss-mime-encode-string subject))
 712           (and author (nnrss-mime-encode-string author))
 713           date
 714           (and extra (nnrss-decode-entities-string extra))
 715           enclosure
 716           comments
 717           hash-index)
 718          nnrss-group-data)
 719         (puthash hash-index t nnrss-group-hashtb)
 720         (setq changed t))
 721       (setq extra nil))
 722     (when changed
 723       (nnrss-save-group-data group server)
 724       (let ((pair (assoc group nnrss-server-data)))
 725         (if pair
 726             (setcar (cdr pair) nnrss-group-max)
 727           (push (list group nnrss-group-max) nnrss-server-data)))
 728       (nnrss-save-server-data server))))
 729
 730 (declare-function gnus-group-make-rss-group "gnus-group" (&optional url))
 731
 732 (defun nnrss-opml-import (opml-file)
 733   "OPML subscriptions import.
 734 Read the file and attempt to subscribe to each Feed in the file."
 735   (interactive "fImport file: ")
 736   (mapc
 737    (lambda (node)
 738      (let ((xmlurl (cdr (assq 'xmlUrl (cadr node)))))
 739        (when (and xmlurl
 740                   (not (string-match "\\`[\t ]*\\'" xmlurl))
 741                   (prog1
 742                       (y-or-n-p (format "Subscribe to %s " xmlurl))
 743                     (message "")))
 744          (condition-case err
 745              (progn
 746                (gnus-group-make-rss-group xmlurl)
 747                (forward-line 1))
 748            (error
 749             (message
 750              "Failed to subscribe to %s (%s); type any key to continue: "
 751              xmlurl
 752              (error-message-string err))
 753             (let ((echo-keystrokes 0))
 754               (read-char)))))))
 755    (nnrss-find-el 'outline
 756                   (mm-with-multibyte-buffer
 757                     (insert-file-contents opml-file)
 758                     (xml-parse-region (point-min) (point-max))))))
 759
 760 (defun nnrss-opml-export ()
 761   "OPML subscription export.
 762 Export subscriptions to a buffer in OPML Format."
 763   (interactive)
 764   (with-current-buffer (get-buffer-create "*OPML Export*")
 765     (set-buffer-file-coding-system 'utf-8)
 766     (insert "<?xml version=\"1.0\" encoding=\"utf-8\"?>\n"
 767             "<!-- OPML generated by Emacs Gnus' nnrss.el -->\n"
 768             "<opml version=\"1.1\">\n"
 769             "  <head>\n"
 770             "    <title>mySubscriptions</title>\n"
 771             "    <dateCreated>" (format-time-string "%a, %d %b %Y %T %z")
 772             "</dateCreated>\n"
 773             "    <ownerEmail>" user-mail-address "</ownerEmail>\n"
 774             "    <ownerName>" (user-full-name) "</ownerName>\n"
 775             "  </head>\n"
 776             "  <body>\n")
 777     (dolist (sub nnrss-group-alist)
 778       (insert "    <outline text=\"" (car sub)
 779               "\" xmlUrl=\"" (cadr sub) "\"/>\n"))
 780     (insert "  </body>\n"
 781             "</opml>\n"))
 782   (pop-to-buffer "*OPML Export*")
 783   (when (fboundp 'sgml-mode)
 784     (sgml-mode)))
 785
 786 (defun nnrss-generate-download-script ()
 787   "Generate a download script in the current buffer.
 788 It is useful when `(setq nnrss-use-local t)'."
 789   (interactive)
 790   (insert "#!/bin/sh\n")
 791   (insert "WGET=wget\n")
 792   (insert "RSSDIR='" (expand-file-name nnrss-directory) "'\n")
 793   (dolist (elem nnrss-server-data)
 794     (let ((url (or (nth 2 elem)
 795                    (second (assoc (car elem) nnrss-group-alist)))))
 796       (insert "$WGET -q -O \"$RSSDIR\"/'"
 797               (nnrss-translate-file-chars (concat (car elem) ".xml"))
 798               "' '" url "'\n"))))
 799
 800 (defun nnrss-translate-file-chars (name)
 801   (let ((nnheader-file-name-translation-alist
 802          (append nnheader-file-name-translation-alist '((?' . ?_)))))
 803     (nnheader-translate-file-chars name)))
 804
 805 (defun nnrss-node-text (namespace local-name element)
 806   (let* ((node (assq (intern (concat namespace (symbol-name local-name)))
 807                      element))
 808          (text (if (and node (listp node))
 809                    (nnrss-node-just-text node)
 810                  node))
 811          (cleaned-text (if text
 812                            (replace-regexp-in-string
 813                             "\r\n" "\n"
 814                             (replace-regexp-in-string
 815                              "^[\000-\037\177]+\\|^ +\\| +$" ""
 816                              text)))))
 817     (if (string-equal "" cleaned-text)
 818         nil
 819       cleaned-text)))
 820
 821 (defun nnrss-node-just-text (node)
 822   (if (and node (listp node))
 823       (mapconcat 'nnrss-node-just-text (cddr node) " ")
 824     node))
 825
 826 (defun nnrss-find-el (tag data &optional found-list)
 827   "Find the all matching elements in the data.
 828 Careful with this on large documents!"
 829   (when (consp data)
 830     (dolist (bit data)
 831       (when (car-safe bit)
 832         (when (equal tag (car bit))
 833           ;; Old xml.el may return a list of string.
 834           (when (and (consp (caddr bit))
 835                      (stringp (caaddr bit)))
 836             (setcar (cddr bit) (caaddr bit)))
 837           (setq found-list
 838                 (append found-list
 839                         (list bit))))
 840         (if (and (consp (car-safe (caddr bit)))
 841                  (not (stringp (caddr bit))))
 842             (setq found-list
 843                   (append found-list
 844                           (nnrss-find-el
 845                            tag (caddr bit))))
 846           (setq found-list
 847                 (append found-list
 848                         (nnrss-find-el
 849                          tag (cddr bit))))))))
 850   found-list)
 851
 852 (defun nnrss-rsslink-p (el)
 853   "Test if the element we are handed is an RSS autodiscovery link."
 854   (and (eq (car-safe el) 'link)
 855        (string-equal (cdr (assoc 'rel (cadr el))) "alternate")
 856        (or (string-equal (cdr (assoc 'type (cadr el)))
 857                          "application/rss+xml")
 858            (string-equal (cdr (assoc 'type (cadr el))) "text/xml"))))
 859
 860 (defun nnrss-get-rsslinks (data)
 861   "Extract the <link> elements that are links to RSS from the parsed data."
 862   (delq nil (mapcar
 863              (lambda (el)
 864                (if (nnrss-rsslink-p el) el))
 865              (nnrss-find-el 'link data))))
 866
 867 (defun nnrss-extract-hrefs (data)
 868   "Recursively extract hrefs from a page's source.
 869 DATA should be the output of `xml-parse-region'."
 870   (mapcar (lambda (ahref)
 871             (cdr (assoc 'href (cadr ahref))))
 872           (nnrss-find-el 'a data)))
 873
 874 (defmacro nnrss-match-macro (base-uri item onsite-list offsite-list)
 875   `(cond ((or (string-match (concat "^" ,base-uri) ,item)
 876               (not (string-match "://" ,item)))
 877           (setq ,onsite-list (append ,onsite-list (list ,item))))
 878          (t (setq ,offsite-list (append ,offsite-list (list ,item))))))
 879
 880 (defun nnrss-order-hrefs (base-uri hrefs)
 881   "Given a list of hrefs, sort them using the following priorities:
 882   1. links ending in .rss
 883   2. links ending in .rdf
 884   3. links ending in .xml
 885   4. links containing the above
 886   5. offsite links
 887
 888 BASE-URI is used to determine the location of the links and
 889 whether they are `offsite' or `onsite'."
 890   (let (rss-onsite-end  rdf-onsite-end  xml-onsite-end
 891         rss-onsite-in   rdf-onsite-in   xml-onsite-in
 892         rss-offsite-end rdf-offsite-end xml-offsite-end
 893         rss-offsite-in rdf-offsite-in xml-offsite-in)
 894     (dolist (href hrefs)
 895       (cond ((null href))
 896             ((string-match "\\.rss$" href)
 897              (nnrss-match-macro
 898               base-uri href rss-onsite-end rss-offsite-end))
 899             ((string-match "\\.rdf$" href)
 900              (nnrss-match-macro
 901               base-uri href rdf-onsite-end rdf-offsite-end))
 902             ((string-match "\\.xml$" href)
 903              (nnrss-match-macro
 904               base-uri href xml-onsite-end xml-offsite-end))
 905             ((string-match "rss" href)
 906              (nnrss-match-macro
 907               base-uri href rss-onsite-in rss-offsite-in))
 908             ((string-match "rdf" href)
 909              (nnrss-match-macro
 910               base-uri href rdf-onsite-in rdf-offsite-in))
 911             ((string-match "xml" href)
 912              (nnrss-match-macro
 913               base-uri href xml-onsite-in xml-offsite-in))))
 914     (append
 915      rss-onsite-end  rdf-onsite-end  xml-onsite-end
 916      rss-onsite-in   rdf-onsite-in   xml-onsite-in
 917      rss-offsite-end rdf-offsite-end xml-offsite-end
 918      rss-offsite-in rdf-offsite-in xml-offsite-in)))
 919
 920 (defun nnrss-discover-feed (url)
 921   "Given a page, find an RSS feed.
 922 Use Mark Pilgrim's `ultra-liberal rss locator'."
 923   (let ((parsed-page (nnrss-fetch url)))
 924     ;;    1. if this url is the rss, use it.
 925     (if (nnrss-rss-p parsed-page)
 926         (let ((rss-ns (nnrss-get-namespace-prefix parsed-page "http://purl.org/rss/1.0/")))
 927           (nnrss-rss-title-description rss-ns parsed-page url))
 928
 929       ;;    2. look for the <link rel="alternate"
 930       ;;    type="application/rss+xml" and use that if it is there.
 931       (let ((links (nnrss-get-rsslinks parsed-page)))
 932         (if links
 933             (let* ((xml (nnrss-fetch
 934                          (cdr (assoc 'href (cadar links)))))
 935                    (rss-ns (nnrss-get-namespace-prefix
 936                             xml "http://purl.org/rss/1.0/")))
 937               (nnrss-rss-title-description
 938                rss-ns xml (cdr (assoc 'href (cadar links)))))
 939
 940           ;;    3. look for links on the site in the following order:
 941           ;;       - onsite links ending in .rss, .rdf, or .xml
 942           ;;       - onsite links containing any of the above
 943           ;;       - offsite links ending in .rss, .rdf, or .xml
 944           ;;       - offsite links containing any of the above
 945           (let* ((base-uri (progn (string-match ".*://[^/]+/?" url)
 946                                   (match-string 0 url)))
 947                  (hrefs (nnrss-order-hrefs
 948                          base-uri (nnrss-extract-hrefs parsed-page)))
 949                  (rss-link nil))
 950             (while (and (eq rss-link nil) (not (eq hrefs nil)))
 951               (let ((href-data (nnrss-fetch (car hrefs))))
 952                 (if (nnrss-rss-p href-data)
 953                     (let* ((rss-ns (nnrss-get-namespace-prefix href-data "http://purl.org/rss/1.0/")))
 954                       (setq rss-link (nnrss-rss-title-description
 955                                       rss-ns href-data (car hrefs))))
 956                   (setq hrefs (cdr hrefs)))))
 957             (if rss-link
 958                 rss-link
 959               ;;    4. check syndic8
 960               (nnrss-find-rss-via-syndic8 url))))))))
 961
 962 (declare-function xml-rpc-method-call "ext:xml-rpc"
 963                   (server-url method &rest params))
 964
 965 (defun nnrss-find-rss-via-syndic8 (url)
 966   "Query syndic8 for the rss feeds it has for URL."
 967   (if (not (locate-library "xml-rpc"))
 968       (progn
 969         (message "XML-RPC is not available... not checking Syndic8.")
 970         nil)
 971     (require 'xml-rpc)
 972     (let ((feedid (xml-rpc-method-call
 973                    "http://www.syndic8.com/xmlrpc.php"
 974                    'syndic8.FindSites
 975                    url)))
 976       (when feedid
 977         (let* ((feedinfo (xml-rpc-method-call
 978                           "http://www.syndic8.com/xmlrpc.php"
 979                           'syndic8.GetFeedInfo
 980                           feedid))
 981                (urllist
 982                 (delq nil
 983                       (mapcar
 984                        (lambda (listinfo)
 985                          (if (string-equal
 986                               (cdr (assoc "status" listinfo))
 987                               "Syndicated")
 988                              (cons
 989                               (cdr (assoc "sitename" listinfo))
 990                               (list
 991                                (cons 'title
 992                                      (cdr (assoc
 993                                            "sitename" listinfo)))
 994                                (cons 'href
 995                                      (cdr (assoc
 996                                            "dataurl" listinfo)))))))
 997                        feedinfo))))
 998           (if (not (> (length urllist) 1))
 999               (cdar urllist)
1000             (let ((completion-ignore-case t)
1001                   (selection
1002                    (mapcar (lambda (listinfo)
1003                              (cons (cdr (assoc "sitename" listinfo))
1004                                    (string-to-number
1005                                     (cdr (assoc "feedid" listinfo)))))
1006                            feedinfo)))
1007               (cdr (assoc
1008                     (gnus-completing-read
1009                      "Multiple feeds found. Select one"
1010                      selection t) urllist)))))))))
1011
1012 (defun nnrss-rss-p (data)
1013   "Test if DATA is an RSS feed.
1014 Simply ensures that the first element is rss or rdf."
1015   (or (eq (caar data) 'rss)
1016       (eq (caar data) 'rdf:RDF)))
1017
1018 (defun nnrss-rss-title-description (rss-namespace data url)
1019   "Return the title of an RSS feed."
1020   (if (nnrss-rss-p data)
1021       (let ((description (intern (concat rss-namespace "description")))
1022             (title (intern (concat rss-namespace "title")))
1023             (channel (nnrss-find-el (intern (concat rss-namespace "channel"))
1024                                     data)))
1025         (list
1026          (cons 'description (caddr (nth 0 (nnrss-find-el description channel))))
1027          (cons 'title (caddr (nth 0 (nnrss-find-el title channel))))
1028          (cons 'href url)))))
1029
1030 (defun nnrss-get-namespace-prefix (el uri)
1031   "Given EL (containing a parsed element) and URI (containing a string
1032 that gives the URI for which you want to retrieve the namespace
1033 prefix), return the prefix."
1034   (let* ((prefix (car (rassoc uri (cadar el))))
1035          (nslist (if prefix
1036                      (split-string (symbol-name prefix) ":")))
1037          (ns (cond ((eq (length nslist) 1) ; no prefix given
1038                     "")
1039                    ((eq (length nslist) 2) ; extract prefix
1040                     (cadr nslist)))))
1041     (if (and ns (not (string= ns "")))
1042         (concat ns ":")
1043       ns)))
1044
1045 (provide 'nnrss)
1046
1047 ;;; nnrss.el ends here