lisp/gnus/nnweb.el

   1 ;;; nnweb.el --- retrieving articles via web search engines
   2
   3 ;; Copyright (C) 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
   4 ;;   2004, 2005 Free Software Foundation, Inc.
   5
   6 ;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
   7 ;; Keywords: news
   8
   9 ;; This file is part of GNU Emacs.
  10
  11 ;; GNU Emacs is free software; you can redistribute it and/or modify
  12 ;; it under the terms of the GNU General Public License as published by
  13 ;; the Free Software Foundation; either version 2, or (at your option)
  14 ;; any later version.
  15
  16 ;; GNU Emacs is distributed in the hope that it will be useful,
  17 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  18 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  19 ;; GNU 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 ;; Note: You need to have `w3' installed for some functions to work.
  29
  30 ;; FIXME: Due to changes in the HTML output of Google Groups and Gmane, stuff
  31 ;; related to web groups (gnus-group-make-web-group) doesn't work anymore.
  32
  33 ;; Fetching an article by MID (cf. gnus-refer-article-method) over Google
  34 ;; Groups should work.
  35
  36 ;;; Code:
  37
  38 (eval-when-compile (require 'cl))
  39
  40 (require 'nnoo)
  41 (require 'message)
  42 (require 'gnus-util)
  43 (require 'gnus)
  44 (require 'nnmail)
  45 (require 'mm-util)
  46 (require 'mm-url)
  47 (eval-and-compile
  48   (ignore-errors
  49     (require 'url)))
  50 (autoload 'w3-parse-buffer "w3-parse")
  51
  52 (nnoo-declare nnweb)
  53
  54 (defvoo nnweb-directory (nnheader-concat gnus-directory "nnweb/")
  55   "Where nnweb will save its files.")
  56
  57 (defvoo nnweb-type 'google
  58   "What search engine type is being used.
  59 Valid types include `google', `dejanews', and `gmane'.")
  60
  61 (defvar nnweb-type-definition
  62   '((google
  63      (id . "http://www.google.com/groups?as_umsgid=%s&hl=en&dmode=source")
  64      (article . nnweb-google-wash-article)
  65      (reference . identity)
  66      (map . nnweb-google-create-mapping)
  67      (search . nnweb-google-search)
  68      (address . "http://groups.google.com/groups")
  69      (base    . "http://groups.google.com")
  70      (identifier . nnweb-google-identity))
  71     (dejanews ;; alias of google
  72      (article . ignore)
  73      (id . "http://groups.google.com/groups?selm=%s&output=gplain")
  74      (reference . identity)
  75      (map . nnweb-google-create-mapping)
  76      (search . nnweb-google-search)
  77      (address . "http://groups.google.com/groups")
  78      (base    . "http://groups.google.com")
  79      (identifier . nnweb-google-identity))
  80     (gmane
  81      (article . nnweb-gmane-wash-article)
  82      (id . "http://gmane.org/view.php?group=%s")
  83      (reference . identity)
  84      (map . nnweb-gmane-create-mapping)
  85      (search . nnweb-gmane-search)
  86      (address . "http://gmane.org/")
  87      (identifier . nnweb-gmane-identity)))
  88   "Type-definition alist.")
  89
  90 (defvoo nnweb-search nil
  91   "Search string to feed to Google.")
  92
  93 (defvoo nnweb-max-hits 999
  94   "Maximum number of hits to display.")
  95
  96 (defvoo nnweb-ephemeral-p nil
  97   "Whether this nnweb server is ephemeral.")
  98
  99 ;;; Internal variables
 100
 101 (defvoo nnweb-articles nil)
 102 (defvoo nnweb-buffer nil)
 103 (defvoo nnweb-group-alist nil)
 104 (defvoo nnweb-group nil)
 105 (defvoo nnweb-hashtb nil)
 106
 107 ;;; Interface functions
 108
 109 (nnoo-define-basics nnweb)
 110
 111 (deffoo nnweb-retrieve-headers (articles &optional group server fetch-old)
 112   (nnweb-possibly-change-server group server)
 113   (save-excursion
 114     (set-buffer nntp-server-buffer)
 115     (erase-buffer)
 116     (let (article header)
 117       (mm-with-unibyte-current-buffer
 118         (while (setq article (pop articles))
 119           (when (setq header (cadr (assq article nnweb-articles)))
 120             (nnheader-insert-nov header))))
 121       'nov)))
 122
 123 (deffoo nnweb-request-scan (&optional group server)
 124   (nnweb-possibly-change-server group server)
 125   (if nnweb-ephemeral-p
 126       (setq nnweb-hashtb (gnus-make-hashtable 4095)))
 127   (funcall (nnweb-definition 'map))
 128   (unless nnweb-ephemeral-p
 129     (nnweb-write-active)
 130     (nnweb-write-overview group)))
 131
 132 (deffoo nnweb-request-group (group &optional server dont-check)
 133   (nnweb-possibly-change-server nil server)
 134   (when (and group
 135              (not (equal group nnweb-group))
 136              (not nnweb-ephemeral-p))
 137     (setq nnweb-group group
 138           nnweb-articles nil)
 139     (let ((info (assoc group nnweb-group-alist)))
 140       (when info
 141         (setq nnweb-type (nth 2 info))
 142         (setq nnweb-search (nth 3 info))
 143         (unless dont-check
 144           (nnweb-read-overview group)))))
 145   (cond
 146    ((not nnweb-articles)
 147     (nnheader-report 'nnweb "No matching articles"))
 148    (t
 149     (let ((active (if nnweb-ephemeral-p
 150                       (cons (caar nnweb-articles)
 151                             (caar (last nnweb-articles)))
 152                     (cadr (assoc group nnweb-group-alist)))))
 153       (nnheader-report 'nnweb "Opened group %s" group)
 154       (nnheader-insert
 155        "211 %d %d %d %s\n" (length nnweb-articles)
 156        (car active) (cdr active) group)))))
 157
 158 (deffoo nnweb-close-group (group &optional server)
 159   (nnweb-possibly-change-server group server)
 160   (when (gnus-buffer-live-p nnweb-buffer)
 161     (save-excursion
 162       (set-buffer nnweb-buffer)
 163       (set-buffer-modified-p nil)
 164       (kill-buffer nnweb-buffer)))
 165   t)
 166
 167 (deffoo nnweb-request-article (article &optional group server buffer)
 168   (nnweb-possibly-change-server group server)
 169   (save-excursion
 170     (set-buffer (or buffer nntp-server-buffer))
 171     (let* ((header (cadr (assq article nnweb-articles)))
 172            (url (and header (mail-header-xref header))))
 173       (when (or (and url
 174                      (mm-with-unibyte-current-buffer
 175                        (mm-url-insert url)))
 176                 (and (stringp article)
 177                      (nnweb-definition 'id t)
 178                      (let ((fetch (nnweb-definition 'id))
 179                            art active)
 180                        (when (string-match "^<\\(.*\\)>$" article)
 181                          (setq art (match-string 1 article)))
 182                        (when (and fetch art)
 183                          (setq url (format fetch art))
 184                          (mm-with-unibyte-current-buffer
 185                            (mm-url-insert url))
 186                          (if (nnweb-definition 'reference t)
 187                              (setq article
 188                                    (funcall (nnweb-definition
 189                                              'reference) article)))))))
 190         (unless nnheader-callback-function
 191           (funcall (nnweb-definition 'article)))
 192         (nnheader-report 'nnweb "Fetched article %s" article)
 193         (cons group (and (numberp article) article))))))
 194
 195 (deffoo nnweb-close-server (&optional server)
 196   (when (and (nnweb-server-opened server)
 197              (gnus-buffer-live-p nnweb-buffer))
 198     (save-excursion
 199       (set-buffer nnweb-buffer)
 200       (set-buffer-modified-p nil)
 201       (kill-buffer nnweb-buffer)))
 202   (nnoo-close-server 'nnweb server))
 203
 204 (deffoo nnweb-request-list (&optional server)
 205   (nnweb-possibly-change-server nil server)
 206   (save-excursion
 207     (set-buffer nntp-server-buffer)
 208     (nnmail-generate-active nnweb-group-alist)
 209     t))
 210
 211 (deffoo nnweb-request-update-info (group info &optional server)
 212   (nnweb-possibly-change-server group server))
 213
 214 (deffoo nnweb-asynchronous-p ()
 215   nil)
 216
 217 (deffoo nnweb-request-create-group (group &optional server args)
 218   (nnweb-possibly-change-server nil server)
 219   (nnweb-request-delete-group group)
 220   (push `(,group ,(cons 1 0) ,@args) nnweb-group-alist)
 221   (nnweb-write-active)
 222   t)
 223
 224 (deffoo nnweb-request-delete-group (group &optional force server)
 225   (nnweb-possibly-change-server group server)
 226   (gnus-pull group nnweb-group-alist t)
 227   (nnweb-write-active)
 228   (gnus-delete-file (nnweb-overview-file group))
 229   t)
 230
 231 (nnoo-define-skeleton nnweb)
 232
 233 ;;; Internal functions
 234
 235 (defun nnweb-read-overview (group)
 236   "Read the overview of GROUP and build the map."
 237   (when (file-exists-p (nnweb-overview-file group))
 238     (mm-with-unibyte-buffer
 239       (nnheader-insert-file-contents (nnweb-overview-file group))
 240       (goto-char (point-min))
 241       (let (header)
 242         (while (not (eobp))
 243           (setq header (nnheader-parse-nov))
 244           (forward-line 1)
 245           (push (list (mail-header-number header)
 246                       header (mail-header-xref header))
 247                 nnweb-articles)
 248           (nnweb-set-hashtb header (car nnweb-articles)))))))
 249
 250 (defun nnweb-write-overview (group)
 251   "Write the overview file for GROUP."
 252   (with-temp-file (nnweb-overview-file group)
 253     (let ((articles nnweb-articles))
 254       (while articles
 255         (nnheader-insert-nov (cadr (pop articles)))))))
 256
 257 (defun nnweb-set-hashtb (header data)
 258   (gnus-sethash (nnweb-identifier (mail-header-xref header))
 259                 data nnweb-hashtb))
 260
 261 (defun nnweb-get-hashtb (url)
 262   (gnus-gethash (nnweb-identifier url) nnweb-hashtb))
 263
 264 (defun nnweb-identifier (ident)
 265   (funcall (nnweb-definition 'identifier) ident))
 266
 267 (defun nnweb-overview-file (group)
 268   "Return the name of the overview file of GROUP."
 269   (nnheader-concat nnweb-directory group ".overview"))
 270
 271 (defun nnweb-write-active ()
 272   "Save the active file."
 273   (gnus-make-directory nnweb-directory)
 274   (with-temp-file (nnheader-concat nnweb-directory "active")
 275     (prin1 `(setq nnweb-group-alist ',nnweb-group-alist) (current-buffer))))
 276
 277 (defun nnweb-read-active ()
 278   "Read the active file."
 279   (load (nnheader-concat nnweb-directory "active") t t t))
 280
 281 (defun nnweb-definition (type &optional noerror)
 282   "Return the definition of TYPE."
 283   (let ((def (cdr (assq type (assq nnweb-type nnweb-type-definition)))))
 284     (when (and (not def)
 285                (not noerror))
 286       (error "Undefined definition %s" type))
 287     def))
 288
 289 (defun nnweb-possibly-change-server (&optional group server)
 290   (nnweb-init server)
 291   (when server
 292     (unless (nnweb-server-opened server)
 293       (nnweb-open-server server)))
 294   (unless nnweb-group-alist
 295     (nnweb-read-active))
 296   (unless nnweb-hashtb
 297     (setq nnweb-hashtb (gnus-make-hashtable 4095)))
 298   (when group
 299     (when (and (not nnweb-ephemeral-p)
 300                (equal group nnweb-group))
 301       (nnweb-request-group group nil t))))
 302
 303 (defun nnweb-init (server)
 304   "Initialize buffers and such."
 305   (unless (gnus-buffer-live-p nnweb-buffer)
 306     (setq nnweb-buffer
 307           (save-excursion
 308             (mm-with-unibyte
 309               (nnheader-set-temp-buffer
 310                (format " *nnweb %s %s %s*"
 311                        nnweb-type nnweb-search server))
 312               (current-buffer))))))
 313
 314 ;;;
 315 ;;; groups.google.com
 316 ;;;
 317
 318 (defun nnweb-google-wash-article ()
 319   ;; We have Google's masked e-mail addresses here.  :-/
 320   (let ((case-fold-search t))
 321     (goto-char (point-min))
 322     (if (save-excursion
 323           (or (re-search-forward "The requested message.*could not be found."
 324                                  nil t)
 325               (not (and (re-search-forward "^<pre>" nil t)
 326                         (re-search-forward "^</pre>" nil t)))))
 327         ;; FIXME: Don't know how to indicate "not found".
 328         ;; Should this function throw an error?  --rsteib
 329         (progn
 330           (gnus-message 3 "Requested article not found")
 331           (erase-buffer))
 332       (delete-region (point-min)
 333                      (1+ (re-search-forward "^<pre>" nil t)))
 334       (goto-char (point-min))
 335       (delete-region (- (re-search-forward "^</pre>" nil t) (length "</pre>"))
 336                      (point-max))
 337       (mm-url-decode-entities))))
 338
 339 (defun nnweb-google-parse-1 (&optional Message-ID)
 340   (let ((i 0)
 341         (case-fold-search t)
 342         (active (cadr (assoc nnweb-group nnweb-group-alist)))
 343         Subject Score Date Newsgroups From
 344         map url mid)
 345     (unless active
 346       (push (list nnweb-group (setq active (cons 1 0))
 347                   nnweb-type nnweb-search)
 348             nnweb-group-alist))
 349     ;; Go through all the article hits on this page.
 350     (goto-char (point-min))
 351     (while (re-search-forward
 352             "a href=/groups\\(\\?[^ \">]*selm=\\([^ &\">]+\\)\\)" nil t)
 353       (setq mid (match-string 2)
 354             url (format
 355                  (nnweb-definition 'id) mid))
 356       (narrow-to-region (search-forward ">" nil t)
 357                         (search-forward "</a>" nil t))
 358       (mm-url-remove-markup)
 359       (mm-url-decode-entities)
 360       (setq Subject (buffer-string))
 361       (goto-char (point-max))
 362       (widen)
 363       (forward-line 2)
 364       (when (looking-at "<br><font[^>]+>")
 365         (goto-char (match-end 0)))
 366       (if (not (looking-at "<a[^>]+>"))
 367           (skip-chars-forward " \t")
 368         (narrow-to-region (point)
 369                           (search-forward "</a>" nil t))
 370         (mm-url-remove-markup)
 371         (mm-url-decode-entities)
 372         (setq Newsgroups (buffer-string))
 373         (goto-char (point-max))
 374         (widen)
 375         (skip-chars-forward "- \t"))
 376       (when (looking-at
 377              "\\([0-9]+\\)[/ ]\\([A-Za-z]+\\)[/ ]\\([0-9]+\\)[ \t]*by[ \t]*\\([^<]*\\) - <a")
 378         (setq From (match-string 4)
 379               Date (format "%s %s 00:00:00 %s"
 380                            (match-string 2) (match-string 1)
 381                            (match-string 3))))
 382       (forward-line 1)
 383       (incf i)
 384       (unless (nnweb-get-hashtb url)
 385         (push
 386          (list
 387           (incf (cdr active))
 388           (make-full-mail-header
 389            (cdr active) (if Newsgroups
 390                             (concat  "(" Newsgroups ") " Subject)
 391                           Subject)
 392            From Date (or Message-ID mid)
 393            nil 0 0 url))
 394          map)
 395         (nnweb-set-hashtb (cadar map) (car map))))
 396     map))
 397
 398 (defun nnweb-google-reference (id)
 399   (let ((map (nnweb-google-parse-1 id)) header)
 400     (setq nnweb-articles
 401           (nconc nnweb-articles map))
 402     (when (setq header (cadar map))
 403       (mm-with-unibyte-current-buffer
 404         (mm-url-insert (mail-header-xref header)))
 405       (caar map))))
 406
 407 (defun nnweb-google-create-mapping ()
 408   "Perform the search and create a number-to-url alist."
 409   (save-excursion
 410     (set-buffer nnweb-buffer)
 411     (erase-buffer)
 412     (when (funcall (nnweb-definition 'search) nnweb-search)
 413         (let ((more t)
 414               (i 0))
 415           (while more
 416             (setq nnweb-articles
 417                   (nconc nnweb-articles (nnweb-google-parse-1)))
 418             ;; Check if there are more articles to fetch
 419             (goto-char (point-min))
 420             (incf i 100)
 421             (if (or (not (re-search-forward
 422                           "<td nowrap><a href=\\([^>]+\\).*<span class=b>Next</span>" nil t))
 423                     (>= i nnweb-max-hits))
 424                 (setq more nil)
 425               ;; Yup, there are more articles
 426               (setq more (concat (nnweb-definition 'base) (match-string 1)))
 427             (when more
 428               (erase-buffer)
 429               (mm-url-insert more))))
 430           ;; Return the articles in the right order.
 431           (setq nnweb-articles
 432                 (sort nnweb-articles 'car-less-than-car))))))
 433
 434 (defun nnweb-google-search (search)
 435   (mm-url-insert
 436    (concat
 437     (nnweb-definition 'address)
 438     "?"
 439     (mm-url-encode-www-form-urlencoded
 440      `(("q" . ,search)
 441        ("num" . "100")
 442        ("hq" . "")
 443        ("hl" . "en")
 444        ("lr" . "")
 445        ("safe" . "off")
 446        ("sites" . "groups")))))
 447   t)
 448
 449 (defun nnweb-google-identity (url)
 450   "Return an unique identifier based on URL."
 451   (if (string-match "selm=\\([^ &>]+\\)" url)
 452       (match-string 1 url)
 453     url))
 454
 455 ;;;
 456 ;;; gmane.org
 457 ;;;
 458 (defun nnweb-gmane-create-mapping ()
 459   "Perform the search and create a number-to-url alist."
 460   (save-excursion
 461     (set-buffer nnweb-buffer)
 462     (erase-buffer)
 463     (when (funcall (nnweb-definition 'search) nnweb-search)
 464       (let ((more t)
 465             (case-fold-search t)
 466             (active (or (cadr (assoc nnweb-group nnweb-group-alist))
 467                         (cons 1 0)))
 468             subject group url
 469             map)
 470           ;; Remove stuff from the beginning of results
 471         (goto-char (point-min))
 472         (search-forward "Search Results</h1><ul>" nil t)
 473         (delete-region (point-min) (point))
 474         (goto-char (point-min))
 475         ;; Iterate over the actual hits
 476         (while (re-search-forward ".*href=\"\\([^\"]+\\)\">\\(.*\\)" nil t)
 477             (setq url (concat "http://gmane.org/" (match-string 1)))
 478             (setq subject (match-string 2))
 479           (unless (nnweb-get-hashtb url)
 480             (push
 481              (list
 482               (incf (cdr active))
 483               (make-full-mail-header
 484                (cdr active) (concat  "(" group ") " subject) nil nil
 485                nil nil 0 0 url))
 486              map)
 487             (nnweb-set-hashtb (cadar map) (car map))))
 488         ;; Return the articles in the right order.
 489         (setq nnweb-articles
 490               (sort (nconc nnweb-articles map) 'car-less-than-car))))))
 491
 492 (defun nnweb-gmane-wash-article ()
 493   (let ((case-fold-search t))
 494     (goto-char (point-min))
 495     (re-search-forward "<!--X-Head-of-Message-->" nil t)
 496     (delete-region (point-min) (point))
 497     (goto-char (point-min))
 498     (while (looking-at "^<li><em>\\([^ ]+\\)</em>.*</li>")
 499       (replace-match "\\1\\2" t)
 500       (forward-line 1))
 501     (mm-url-remove-markup)))
 502
 503 (defun nnweb-gmane-search (search)
 504   (mm-url-insert
 505    (concat
 506     (nnweb-definition 'address)
 507     "?"
 508     (mm-url-encode-www-form-urlencoded
 509      `(("query" . ,search)))))
 510   (setq buffer-file-name nil)
 511   t)
 512
 513
 514 (defun nnweb-gmane-identity (url)
 515   "Return a unique identifier based on URL."
 516   (if (string-match "group=\\(.+\\)" url)
 517       (match-string 1 url)
 518     url))
 519
 520 ;;;
 521 ;;; General web/w3 interface utility functions
 522 ;;;
 523
 524 (defun nnweb-insert-html (parse)
 525   "Insert HTML based on a w3 parse tree."
 526   (if (stringp parse)
 527       (insert (nnheader-string-as-multibyte parse))
 528     (insert "<" (symbol-name (car parse)) " ")
 529     (insert (mapconcat
 530              (lambda (param)
 531                (concat (symbol-name (car param)) "="
 532                        (prin1-to-string
 533                         (if (consp (cdr param))
 534                             (cadr param)
 535                           (cdr param)))))
 536              (nth 1 parse)
 537              " "))
 538     (insert ">\n")
 539     (mapcar 'nnweb-insert-html (nth 2 parse))
 540     (insert "</" (symbol-name (car parse)) ">\n")))
 541
 542 (defun nnweb-parse-find (type parse &optional maxdepth)
 543   "Find the element of TYPE in PARSE."
 544   (catch 'found
 545     (nnweb-parse-find-1 type parse maxdepth)))
 546
 547 (defun nnweb-parse-find-1 (type contents maxdepth)
 548   (when (or (null maxdepth)
 549             (not (zerop maxdepth)))
 550     (when (consp contents)
 551       (when (eq (car contents) type)
 552         (throw 'found contents))
 553       (when (listp (cdr contents))
 554         (dolist (element contents)
 555           (when (consp element)
 556             (nnweb-parse-find-1 type element
 557                                 (and maxdepth (1- maxdepth)))))))))
 558
 559 (defun nnweb-parse-find-all (type parse)
 560   "Find all elements of TYPE in PARSE."
 561   (catch 'found
 562     (nnweb-parse-find-all-1 type parse)))
 563
 564 (defun nnweb-parse-find-all-1 (type contents)
 565   (let (result)
 566     (when (consp contents)
 567       (if (eq (car contents) type)
 568           (push contents result)
 569         (when (listp (cdr contents))
 570           (dolist (element contents)
 571             (when (consp element)
 572               (setq result
 573                     (nconc result (nnweb-parse-find-all-1 type element))))))))
 574     result))
 575
 576 (defvar nnweb-text)
 577 (defun nnweb-text (parse)
 578   "Return a list of text contents in PARSE."
 579   (let ((nnweb-text nil))
 580     (nnweb-text-1 parse)
 581     (nreverse nnweb-text)))
 582
 583 (defun nnweb-text-1 (contents)
 584   (dolist (element contents)
 585     (if (stringp element)
 586         (push element nnweb-text)
 587       (when (and (consp element)
 588                  (listp (cdr element)))
 589         (nnweb-text-1 element)))))
 590
 591 (provide 'nnweb)
 592
 593 ;;; arch-tag: f59307eb-c90f-479f-b7d2-dbd8bf51b697
 594 ;;; nnweb.el ends here