lisp/gnus/nnweb.el

   1 ;;; nnweb.el --- retrieving articles via web search engines
   2 ;; Copyright (C) 1996,97,98 Free Software Foundation, Inc.
   3
   4 ;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
   5 ;; Keywords: news
   6
   7 ;; This file is part of GNU Emacs.
   8
   9 ;; GNU Emacs is free software; you can redistribute it and/or modify
  10 ;; it under the terms of the GNU General Public License as published by
  11 ;; the Free Software Foundation; either version 2, or (at your option)
  12 ;; any later version.
  13
  14 ;; GNU Emacs is distributed in the hope that it will be useful,
  15 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
  16 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  17 ;; GNU General Public License for more details.
  18
  19 ;; You should have received a copy of the GNU General Public License
  20 ;; along with GNU Emacs; see the file COPYING.  If not, write to the
  21 ;; Free Software Foundation, Inc., 59 Temple Place - Suite 330,
  22 ;; Boston, MA 02111-1307, USA.
  23
  24 ;;; Commentary:
  25
  26 ;; Note: You need to have `url' and `w3' installed for this
  27 ;; backend to work.
  28
  29 ;;; Code:
  30
  31 (eval-when-compile (require 'cl))
  32
  33 (eval-when-compile (require 'cl))
  34
  35 (require 'nnoo)
  36 (require 'message)
  37 (require 'gnus-util)
  38 (require 'gnus)
  39 (require 'nnmail)
  40 (eval-when-compile
  41   (ignore-errors
  42    (require 'w3)
  43    (require 'url)
  44    (require 'w3-forms)))
  45 ;; Report failure to find w3 at load time if appropriate.
  46 (eval '(progn
  47          (require 'w3)
  48          (require 'url)
  49          (require 'w3-forms)))
  50
  51 (nnoo-declare nnweb)
  52
  53 (defvoo nnweb-directory (nnheader-concat gnus-directory "nnweb/")
  54   "Where nnweb will save its files.")
  55
  56 (defvoo nnweb-type 'dejanews
  57   "What search engine type is being used.
  58 Valid types include `dejanews', `dejanewsold', `reference',
  59 and `altavista'.")
  60
  61 (defvoo nnweb-type-definition
  62   '((dejanews
  63      (article . nnweb-dejanews-wash-article)
  64      (map . nnweb-dejanews-create-mapping)
  65      (search . nnweb-dejanews-search)
  66      (address . "http://x8.dejanews.com/dnquery.xp")
  67      (identifier . nnweb-dejanews-identity))
  68     (dejanewsold
  69      (article . nnweb-dejanews-wash-article)
  70      (map . nnweb-dejanews-create-mapping)
  71      (search . nnweb-dejanewsold-search)
  72      (address . "http://x8.dejanews.com/dnquery.xp")
  73      (identifier . nnweb-dejanews-identity))
  74     (reference
  75      (article . nnweb-reference-wash-article)
  76      (map . nnweb-reference-create-mapping)
  77      (search . nnweb-reference-search)
  78      (address . "http://www.reference.com/cgi-bin/pn/go")
  79      (identifier . identity))
  80     (altavista
  81      (article . nnweb-altavista-wash-article)
  82      (map . nnweb-altavista-create-mapping)
  83      (search . nnweb-altavista-search)
  84      (address . "http://www.altavista.digital.com/cgi-bin/query")
  85      (id . "/cgi-bin/news?id@%s")
  86      (identifier . identity)))
  87   "Type-definition alist.")
  88
  89 (defvoo nnweb-search nil
  90   "Search string to feed to DejaNews.")
  91
  92 (defvoo nnweb-max-hits 999
  93   "Maximum number of hits to display.")
  94
  95 (defvoo nnweb-ephemeral-p nil
  96   "Whether this nnweb server is ephemeral.")
  97
  98 ;;; Internal variables
  99
 100 (defvoo nnweb-articles nil)
 101 (defvoo nnweb-buffer nil)
 102 (defvoo nnweb-group-alist nil)
 103 (defvoo nnweb-group nil)
 104 (defvoo nnweb-hashtb nil)
 105
 106 ;;; Interface functions
 107
 108 (nnoo-define-basics nnweb)
 109
 110 (deffoo nnweb-retrieve-headers (articles &optional group server fetch-old)
 111   (nnweb-possibly-change-server group server)
 112   (save-excursion
 113     (set-buffer nntp-server-buffer)
 114     (erase-buffer)
 115     (let (article header)
 116       (while (setq article (pop articles))
 117         (when (setq header (cadr (assq article nnweb-articles)))
 118           (nnheader-insert-nov header)))
 119       'nov)))
 120
 121 (deffoo nnweb-request-scan (&optional group server)
 122   (nnweb-possibly-change-server group server)
 123   (setq nnweb-hashtb (gnus-make-hashtable 4095))
 124   (funcall (nnweb-definition 'map))
 125   (unless nnweb-ephemeral-p
 126     (nnweb-write-active)
 127     (nnweb-write-overview group)))
 128
 129 (deffoo nnweb-request-group (group &optional server dont-check)
 130   (nnweb-possibly-change-server nil server)
 131   (when (and group
 132              (not (equal group nnweb-group))
 133              (not nnweb-ephemeral-p))
 134     (let ((info (assoc group nnweb-group-alist)))
 135       (setq nnweb-group group)
 136       (setq nnweb-type (nth 2 info))
 137       (setq nnweb-search (nth 3 info))
 138       (unless dont-check
 139         (nnweb-read-overview group))))
 140   (cond
 141    ((not nnweb-articles)
 142     (nnheader-report 'nnweb "No matching articles"))
 143    (t
 144     (let ((active (if nnweb-ephemeral-p
 145                       (cons (caar nnweb-articles)
 146                             (caar (last nnweb-articles)))
 147                     (cadr (assoc group nnweb-group-alist)))))
 148       (nnheader-report 'nnweb "Opened group %s" group)
 149       (nnheader-insert
 150        "211 %d %d %d %s\n" (length nnweb-articles)
 151        (car active) (cdr active) group)))))
 152
 153 (deffoo nnweb-close-group (group &optional server)
 154   (nnweb-possibly-change-server group server)
 155   (when (gnus-buffer-live-p nnweb-buffer)
 156     (save-excursion
 157       (set-buffer nnweb-buffer)
 158       (set-buffer-modified-p nil)
 159       (kill-buffer nnweb-buffer)))
 160   t)
 161
 162 (deffoo nnweb-request-article (article &optional group server buffer)
 163   (nnweb-possibly-change-server group server)
 164   (save-excursion
 165     (set-buffer (or buffer nntp-server-buffer))
 166     (let* ((header (cadr (assq article nnweb-articles)))
 167            (url (and header (mail-header-xref header))))
 168       (when (or (and url
 169                      (nnweb-fetch-url url))
 170                 (and (stringp article)
 171                      (nnweb-definition 'id t)
 172                      (let ((fetch (nnweb-definition 'id))
 173                            art)
 174                        (when (string-match "^<\\(.*\\)>$" article)
 175                          (setq art (match-string 1 article)))
 176                        (and fetch
 177                             art
 178                             (nnweb-fetch-url
 179                              (format fetch article))))))
 180         (unless nnheader-callback-function
 181           (funcall (nnweb-definition 'article))
 182           (nnweb-decode-entities))
 183         (nnheader-report 'nnweb "Fetched article %s" article)
 184         t))))
 185
 186 (deffoo nnweb-close-server (&optional server)
 187   (when (and (nnweb-server-opened server)
 188              (gnus-buffer-live-p nnweb-buffer))
 189     (save-excursion
 190       (set-buffer nnweb-buffer)
 191       (set-buffer-modified-p nil)
 192       (kill-buffer nnweb-buffer)))
 193   (nnoo-close-server 'nnweb server))
 194
 195 (deffoo nnweb-request-list (&optional server)
 196   (nnweb-possibly-change-server nil server)
 197   (save-excursion
 198     (set-buffer nntp-server-buffer)
 199     (nnmail-generate-active nnweb-group-alist)
 200     t))
 201
 202 (deffoo nnweb-request-update-info (group info &optional server)
 203   (nnweb-possibly-change-server group server)
 204   ;;(setcar (cddr info) nil)
 205   )
 206
 207 (deffoo nnweb-asynchronous-p ()
 208   t)
 209
 210 (deffoo nnweb-request-create-group (group &optional server args)
 211   (nnweb-possibly-change-server nil server)
 212   (nnweb-request-delete-group group)
 213   (push `(,group ,(cons 1 0) ,@args) nnweb-group-alist)
 214   (nnweb-write-active)
 215   t)
 216
 217 (deffoo nnweb-request-delete-group (group &optional force server)
 218   (nnweb-possibly-change-server group server)
 219   (gnus-pull group nnweb-group-alist)
 220   (gnus-delete-file (nnweb-overview-file group))
 221   t)
 222
 223 (nnoo-define-skeleton nnweb)
 224
 225 ;;; Internal functions
 226
 227 (defun nnweb-read-overview (group)
 228   "Read the overview of GROUP and build the map."
 229   (when (file-exists-p (nnweb-overview-file group))
 230     (nnheader-temp-write nil
 231       (nnheader-insert-file-contents (nnweb-overview-file group))
 232       (goto-char (point-min))
 233       (let (header)
 234         (while (not (eobp))
 235           (setq header (nnheader-parse-nov))
 236           (forward-line 1)
 237           (push (list (mail-header-number header)
 238                       header (mail-header-xref header))
 239                 nnweb-articles)
 240           (nnweb-set-hashtb header (car nnweb-articles)))))))
 241
 242 (defun nnweb-write-overview (group)
 243   "Write the overview file for GROUP."
 244   (nnheader-temp-write (nnweb-overview-file group)
 245     (let ((articles nnweb-articles))
 246       (while articles
 247         (nnheader-insert-nov (cadr (pop articles)))))))
 248
 249 (defun nnweb-set-hashtb (header data)
 250   (gnus-sethash (nnweb-identifier (mail-header-xref header))
 251                 data nnweb-hashtb))
 252
 253 (defun nnweb-get-hashtb (url)
 254   (gnus-gethash (nnweb-identifier url) nnweb-hashtb))
 255
 256 (defun nnweb-identifier (ident)
 257   (funcall (nnweb-definition 'identifier) ident))
 258
 259 (defun nnweb-overview-file (group)
 260   "Return the name of the overview file of GROUP."
 261   (nnheader-concat nnweb-directory group ".overview"))
 262
 263 (defun nnweb-write-active ()
 264   "Save the active file."
 265   (nnheader-temp-write (nnheader-concat nnweb-directory "active")
 266     (prin1 `(setq nnweb-group-alist ',nnweb-group-alist) (current-buffer))))
 267
 268 (defun nnweb-read-active ()
 269   "Read the active file."
 270   (load (nnheader-concat nnweb-directory "active") t t t))
 271
 272 (defun nnweb-definition (type &optional noerror)
 273   "Return the definition of TYPE."
 274   (let ((def (cdr (assq type (assq nnweb-type nnweb-type-definition)))))
 275     (when (and (not def)
 276                (not noerror))
 277       (error "Undefined definition %s" type))
 278     def))
 279
 280 (defun nnweb-possibly-change-server (&optional group server)
 281   (nnweb-init server)
 282   (when server
 283     (unless (nnweb-server-opened server)
 284       (nnweb-open-server server)))
 285   (unless nnweb-group-alist
 286     (nnweb-read-active))
 287   (when group
 288     (when (and (not nnweb-ephemeral-p)
 289                (not (equal group nnweb-group)))
 290       (nnweb-request-group group nil t))))
 291
 292 (defun nnweb-init (server)
 293   "Initialize buffers and such."
 294   (unless (gnus-buffer-live-p nnweb-buffer)
 295     (setq nnweb-buffer
 296           (save-excursion
 297             (nnheader-set-temp-buffer
 298              (format " *nnweb %s %s %s*" nnweb-type nnweb-search server))))))
 299
 300 (defun nnweb-fetch-url (url)
 301   (save-excursion
 302     (if (not nnheader-callback-function)
 303         (let ((buf (current-buffer)))
 304           (save-excursion
 305             (set-buffer nnweb-buffer)
 306             (erase-buffer)
 307             (url-insert-file-contents url)
 308             (copy-to-buffer buf (point-min) (point-max))
 309             t))
 310       (nnweb-url-retrieve-asynch
 311        url 'nnweb-callback (current-buffer) nnheader-callback-function)
 312       t)))
 313
 314 (defun nnweb-callback (buffer callback)
 315   (when (gnus-buffer-live-p url-working-buffer)
 316     (save-excursion
 317       (set-buffer url-working-buffer)
 318       (funcall (nnweb-definition 'article))
 319       (nnweb-decode-entities)
 320       (set-buffer buffer)
 321       (goto-char (point-max))
 322       (insert-buffer-substring url-working-buffer))
 323     (funcall callback t)
 324     (gnus-kill-buffer url-working-buffer)))
 325
 326 (defun nnweb-url-retrieve-asynch (url callback &rest data)
 327   (let ((url-request-method "GET")
 328         (old-asynch url-be-asynchronous)
 329         (url-request-data nil)
 330         (url-request-extra-headers nil)
 331         (url-working-buffer (generate-new-buffer-name " *nnweb*")))
 332     (setq-default url-be-asynchronous t)
 333     (save-excursion
 334       (set-buffer (get-buffer-create url-working-buffer))
 335       (setq url-current-callback-data data
 336             url-be-asynchronous t
 337             url-current-callback-func callback)
 338       (url-retrieve url))
 339     (setq-default url-be-asynchronous old-asynch)))
 340
 341 (defun nnweb-encode-www-form-urlencoded (pairs)
 342   "Return PAIRS encoded for forms."
 343   (mapconcat
 344    (function
 345     (lambda (data)
 346       (concat (w3-form-encode-xwfu (car data)) "="
 347               (w3-form-encode-xwfu (cdr data)))))
 348    pairs "&"))
 349
 350 (defun nnweb-fetch-form (url pairs)
 351   (let ((url-request-data (nnweb-encode-www-form-urlencoded pairs))
 352         (url-request-method "POST")
 353         (url-request-extra-headers
 354          '(("Content-type" . "application/x-www-form-urlencoded"))))
 355     (url-insert-file-contents url)
 356     (setq buffer-file-name nil))
 357   t)
 358
 359 (defun nnweb-decode-entities ()
 360   (goto-char (point-min))
 361   (while (re-search-forward "&\\([a-z]+\\);" nil t)
 362     (replace-match (char-to-string (or (cdr (assq (intern (match-string 1))
 363                                                   w3-html-entities))
 364                                        ?#))
 365                    t t)))
 366
 367 (defun nnweb-remove-markup ()
 368   (goto-char (point-min))
 369   (while (search-forward "<!--" nil t)
 370     (delete-region (match-beginning 0)
 371                    (or (search-forward "-->" nil t)
 372                        (point-max))))
 373   (goto-char (point-min))
 374   (while (re-search-forward "<[^>]+>" nil t)
 375     (replace-match "" t t)))
 376
 377 ;;;
 378 ;;; DejaNews functions.
 379 ;;;
 380
 381 (defun nnweb-dejanews-create-mapping ()
 382   "Perform the search and create an number-to-url alist."
 383   (save-excursion
 384     (set-buffer nnweb-buffer)
 385     (erase-buffer)
 386     (when (funcall (nnweb-definition 'search) nnweb-search)
 387       (let ((i 0)
 388             (more t)
 389             (case-fold-search t)
 390             (active (or (cadr (assoc nnweb-group nnweb-group-alist))
 391                         (cons 1 0)))
 392             Subject (Score "0") Date Newsgroup Author
 393             map url)
 394         (while more
 395           ;; Go through all the article hits on this page.
 396           (goto-char (point-min))
 397           (nnweb-decode-entities)
 398           (goto-char (point-min))
 399           (while (re-search-forward "^ <P>\n" nil t)
 400             (narrow-to-region
 401              (point)
 402              (cond ((re-search-forward "^ <P>\n" nil t)
 403                     (match-beginning 0))
 404                    ((search-forward "\n\n" nil t)
 405                     (point))
 406                    (t
 407                     (point-max))))
 408             (goto-char (point-min))
 409             (looking-at ".*HREF=\"\\([^\"]+\\)\"\\(.*\\)")
 410             (setq url (match-string 1))
 411             (let ((begin (point)))
 412               (nnweb-remove-markup)
 413               (goto-char begin)
 414               (while (search-forward "\t" nil t)
 415                 (replace-match " "))
 416               (goto-char begin)
 417               (end-of-line)
 418               (setq Subject (buffer-substring begin (point)))
 419               (if (re-search-forward
 420                    "^ Newsgroup: \\(.*\\)\n Posted on \\([0-9/]+\\) by \\(.*\\)$" nil t)
 421                   (setq Newsgroup (match-string 1)
 422                         Date (match-string 2)
 423                         Author (match-string 3))))
 424             (widen)
 425             (incf i)
 426             (unless (nnweb-get-hashtb url)
 427               (push
 428                (list
 429                 (incf (cdr active))
 430                 (make-full-mail-header
 431                  (cdr active) Subject Author Date
 432                  (concat "<" (nnweb-identifier url) "@dejanews>")
 433                  nil 0 (string-to-int Score) url))
 434                map)
 435               (nnweb-set-hashtb (cadar map) (car map))))
 436           ;; See whether there is a "Get next 20 hits" button here.
 437           (if (or (not (re-search-forward
 438                         "HREF=\"\\([^\"]+\\)\"[<>b]+Next result" nil t))
 439                   (>= i nnweb-max-hits))
 440               (setq more nil)
 441             ;; Yup -- fetch it.
 442             (setq more (match-string 1))
 443             (erase-buffer)
 444             (url-insert-file-contents more)))
 445         ;; Return the articles in the right order.
 446         (setq nnweb-articles
 447               (sort (nconc nnweb-articles map) 'car-less-than-car))))))
 448
 449 (defun nnweb-dejanews-wash-article ()
 450   (let ((case-fold-search t))
 451     (goto-char (point-min))
 452     (re-search-forward "<PRE>" nil t)
 453     (delete-region (point-min) (point))
 454     (re-search-forward "</PRE>" nil t)
 455     (delete-region (point) (point-max))
 456     (nnweb-remove-markup)
 457     (goto-char (point-min))
 458     (while (and (looking-at " *$")
 459                 (not (eobp)))
 460       (gnus-delete-line))
 461     (while (looking-at "\\(^[^ ]+:\\) *")
 462       (replace-match "\\1 " t)
 463       (forward-line 1))
 464     (when (re-search-forward "\n\n+" nil t)
 465       (replace-match "\n" t t))
 466     (goto-char (point-min))
 467     (when (search-forward "[More Headers]" nil t)
 468       (replace-match "" t t))))
 469
 470 (defun nnweb-dejanews-search (search)
 471   (nnweb-fetch-form
 472    (nnweb-definition 'address)
 473    `(("query" . ,search)
 474      ("defaultOp" . "AND")
 475      ("svcclass" . "dncurrent")
 476      ("maxhits" . "100")
 477      ("format" . "verbose2")
 478      ("threaded" . "0")
 479      ("showsort" . "date")
 480      ("agesign" . "1")
 481      ("ageweight" . "1")))
 482   t)
 483
 484 (defun nnweb-dejanewsold-search (search)
 485   (nnweb-fetch-form
 486    (nnweb-definition 'address)
 487    `(("query" . ,search)
 488      ("defaultOp" . "AND")
 489      ("svcclass" . "dnold")
 490      ("maxhits" . "100")
 491      ("format" . "verbose2")
 492      ("threaded" . "0")
 493      ("showsort" . "date")
 494      ("agesign" . "1")
 495      ("ageweight" . "1")))
 496   t)
 497
 498 (defun nnweb-dejanews-identity (url)
 499   "Return an unique identifier based on URL."
 500   (if (string-match "recnum=\\([0-9]+\\)" url)
 501       (match-string 1 url)
 502     url))
 503
 504 ;;;
 505 ;;; InReference
 506 ;;;
 507
 508 (defun nnweb-reference-create-mapping ()
 509   "Perform the search and create an number-to-url alist."
 510   (save-excursion
 511     (set-buffer nnweb-buffer)
 512     (erase-buffer)
 513     (when (funcall (nnweb-definition 'search) nnweb-search)
 514       (let ((i 0)
 515             (more t)
 516             (case-fold-search t)
 517             (active (or (cadr (assoc nnweb-group nnweb-group-alist))
 518                         (cons 1 0)))
 519             Subject Score Date Newsgroups From Message-ID
 520             map url)
 521         (while more
 522           ;; Go through all the article hits on this page.
 523           (goto-char (point-min))
 524           (search-forward "</pre><hr>" nil t)
 525           (delete-region (point-min) (point))
 526                                         ;(nnweb-decode-entities)
 527           (goto-char (point-min))
 528           (while (re-search-forward "^ +[0-9]+\\." nil t)
 529             (narrow-to-region
 530              (point)
 531              (if (re-search-forward "^$" nil t)
 532                  (match-beginning 0)
 533                (point-max)))
 534             (goto-char (point-min))
 535             (when (looking-at ".*href=\"\\([^\"]+\\)\"")
 536               (setq url (match-string 1)))
 537             (nnweb-remove-markup)
 538             (goto-char (point-min))
 539             (while (search-forward "\t" nil t)
 540               (replace-match " "))
 541             (goto-char (point-min))
 542             (while (re-search-forward "^\\([^:]+\\): \\(.*\\)$" nil t)
 543               (set (intern (match-string 1)) (match-string 2)))
 544             (widen)
 545             (search-forward "</pre>" nil t)
 546             (incf i)
 547             (unless (nnweb-get-hashtb url)
 548               (push
 549                (list
 550                 (incf (cdr active))
 551                 (make-full-mail-header
 552                  (cdr active) (concat  "(" Newsgroups ") " Subject) From Date
 553                  Message-ID
 554                  nil 0 (string-to-int Score) url))
 555                map)
 556               (nnweb-set-hashtb (cadar map) (car map))))
 557           (setq more nil))
 558         ;; Return the articles in the right order.
 559         (setq nnweb-articles
 560               (sort (nconc nnweb-articles map) 'car-less-than-car))))))
 561
 562 (defun nnweb-reference-wash-article ()
 563   (let ((case-fold-search t))
 564     (goto-char (point-min))
 565     (re-search-forward "^</center><hr>" nil t)
 566     (delete-region (point-min) (point))
 567     (search-forward "<pre>" nil t)
 568     (forward-line -1)
 569     (let ((body (point-marker)))
 570       (search-forward "</pre>" nil t)
 571       (delete-region (point) (point-max))
 572       (nnweb-remove-markup)
 573       (goto-char (point-min))
 574       (while (looking-at " *$")
 575         (gnus-delete-line))
 576       (narrow-to-region (point-min) body)
 577       (while (and (re-search-forward "^$" nil t)
 578                   (not (eobp)))
 579         (gnus-delete-line))
 580       (goto-char (point-min))
 581       (while (looking-at "\\(^[^ ]+:\\) *")
 582         (replace-match "\\1 " t)
 583         (forward-line 1))
 584       (goto-char (point-min))
 585       (when (re-search-forward "^References:" nil t)
 586         (narrow-to-region
 587          (point) (if (re-search-forward "^$\\|^[^:]+:" nil t)
 588                      (match-beginning 0)
 589                    (point-max)))
 590         (goto-char (point-min))
 591         (while (not (eobp))
 592           (unless (looking-at "References")
 593             (insert "\t")
 594             (forward-line 1)))
 595         (goto-char (point-min))
 596         (while (search-forward "," nil t)
 597           (replace-match " " t t)))
 598       (widen)
 599       (set-marker body nil))))
 600
 601 (defun nnweb-reference-search (search)
 602   (url-insert-file-contents
 603    (concat
 604     (nnweb-definition 'address)
 605     "?"
 606     (nnweb-encode-www-form-urlencoded
 607      `(("search" . "advanced")
 608        ("querytext" . ,search)
 609        ("subj" . "")
 610        ("name" . "")
 611        ("login" . "")
 612        ("host" . "")
 613        ("organization" . "")
 614        ("groups" . "")
 615        ("keywords" . "")
 616        ("choice" . "Search")
 617        ("startmonth" . "Jul")
 618        ("startday" . "25")
 619        ("startyear" . "1996")
 620        ("endmonth" . "Aug")
 621        ("endday" . "24")
 622        ("endyear" . "1996")
 623        ("mode" . "Quick")
 624        ("verbosity" . "Verbose")
 625        ("ranking" . "Relevance")
 626        ("first" . "1")
 627        ("last" . "25")
 628        ("score" . "50")))))
 629   (setq buffer-file-name nil)
 630   t)
 631
 632 ;;;
 633 ;;; Alta Vista
 634 ;;;
 635
 636 (defun nnweb-altavista-create-mapping ()
 637   "Perform the search and create an number-to-url alist."
 638   (save-excursion
 639     (set-buffer nnweb-buffer)
 640     (erase-buffer)
 641     (let ((part 0))
 642       (when (funcall (nnweb-definition 'search) nnweb-search part)
 643         (let ((i 0)
 644               (more t)
 645               (case-fold-search t)
 646               (active (or (cadr (assoc nnweb-group nnweb-group-alist))
 647                           (cons 1 0)))
 648               subject date from id group
 649               map url)
 650           (while more
 651             ;; Go through all the article hits on this page.
 652             (goto-char (point-min))
 653             (search-forward "<dt>" nil t)
 654             (delete-region (point-min) (match-beginning 0))
 655             (goto-char (point-min))
 656             (while (search-forward "<dt>" nil t)
 657               (replace-match "\n<blubb>"))
 658             (nnweb-decode-entities)
 659             (goto-char (point-min))
 660             (while (re-search-forward "<blubb>.*href=\"\\([^\"]+\\)\"><strong>\\([^>]*\\)</strong></a><dd>\\([^-]+\\)- <b>\\([^<]+\\)<.*href=\"news:\\([^\"]+\\)\">.*\">\\(.+\\)</a><P>"
 661                                       nil t)
 662               (setq url (match-string 1)
 663                     subject (match-string 2)
 664                     date (match-string 3)
 665                     group (match-string 4)
 666                     id (concat "<" (match-string 5) ">")
 667                     from (match-string 6))
 668               (incf i)
 669               (unless (nnweb-get-hashtb url)
 670                 (push
 671                  (list
 672                   (incf (cdr active))
 673                   (make-full-mail-header
 674                    (cdr active) (concat  "(" group ") " subject) from date
 675                    id nil 0 0 url))
 676                  map)
 677                 (nnweb-set-hashtb (cadar map) (car map))))
 678             ;; See if we want more.
 679             (when (or (not nnweb-articles)
 680                       (>= i nnweb-max-hits)
 681                       (not (funcall (nnweb-definition 'search)
 682                                     nnweb-search (incf part))))
 683               (setq more nil)))
 684           ;; Return the articles in the right order.
 685           (setq nnweb-articles
 686                 (sort (nconc nnweb-articles map) 'car-less-than-car)))))))
 687
 688 (defun nnweb-altavista-wash-article ()
 689   (goto-char (point-min))
 690   (let ((case-fold-search t))
 691     (when (re-search-forward "^<strong>" nil t)
 692       (delete-region (point-min) (match-beginning 0)))
 693     (goto-char (point-min))
 694     (while (looking-at "<strong>\\([^ ]+\\) +</strong> +\\(.*\\)$")
 695       (replace-match "\\1: \\2" t)
 696       (forward-line 1))
 697     (when (re-search-backward "^References:" nil t)
 698       (narrow-to-region (point) (progn (forward-line 1) (point)))
 699       (goto-char (point-min))
 700       (while (re-search-forward "<A.*\\?id@\\([^\"]+\\)\">[0-9]+</A>" nil t)
 701         (replace-match "&lt;\\1&gt; " t)))
 702     (widen)
 703     (nnweb-remove-markup)))
 704
 705 (defun nnweb-altavista-search (search &optional part)
 706   (url-insert-file-contents
 707    (concat
 708     (nnweb-definition 'address)
 709     "?"
 710     (nnweb-encode-www-form-urlencoded
 711      `(("pg" . "aq")
 712        ("what" . "news")
 713        ,@(when part `(("stq" . ,(int-to-string (* part 30)))))
 714        ("fmt" . "d")
 715        ("q" . ,search)
 716        ("r" . "")
 717        ("d0" . "")
 718        ("d1" . "")))))
 719   (setq buffer-file-name nil)
 720   t)
 721
 722 (provide 'nnweb)
 723
 724 ;;; nnweb.el ends here