lisp/gnus/nnweb.el

   1 ;;; nnweb.el --- retrieving articles via web search engines
   2
   3 ;; Copyright (C) 1996-2012 Free Software Foundation, Inc.
   4
   5 ;; Author: Lars Magne Ingebrigtsen <larsi@gnus.org>
   6 ;; Keywords: news
   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 ;; Note: You need to have `w3' installed for some functions to work.
  26
  27 ;;; Code:
  28
  29 (eval-when-compile (require 'cl))
  30
  31 (require 'nnoo)
  32 (require 'message)
  33 (require 'gnus-util)
  34 (require 'gnus)
  35 (require 'nnmail)
  36 (require 'mm-util)
  37 (require 'mm-url)
  38 (eval-and-compile
  39   (ignore-errors
  40     (require 'url)))
  41 (autoload 'w3-parse-buffer "w3-parse")
  42
  43 (nnoo-declare nnweb)
  44
  45 (defvoo nnweb-directory (nnheader-concat gnus-directory "nnweb/")
  46   "Where nnweb will save its files.")
  47
  48 (defvoo nnweb-type 'google
  49   "What search engine type is being used.
  50 Valid types include `google', `dejanews', and `gmane'.")
  51
  52 (defvar nnweb-type-definition
  53   '((google
  54      (id . "http://www.google.com/groups?as_umsgid=%s&hl=en&dmode=source")
  55      (result . "http://groups.google.com/group/%s/msg/%s?dmode=source")
  56      (article . nnweb-google-wash-article)
  57      (reference . identity)
  58      (map . nnweb-google-create-mapping)
  59      (search . nnweb-google-search)
  60      (address . "http://groups.google.com/groups")
  61      (base    . "http://groups.google.com")
  62      (identifier . nnweb-google-identity))
  63     (dejanews ;; alias of google
  64      (id . "http://www.google.com/groups?as_umsgid=%s&hl=en&dmode=source")
  65      (result . "http://groups.google.com/group/%s/msg/%s?dmode=source")
  66      (article . nnweb-google-wash-article)
  67      (reference . identity)
  68      (map . nnweb-google-create-mapping)
  69      (search . nnweb-google-search)
  70      (address . "http://groups.google.com/groups")
  71      (base    . "http://groups.google.com")
  72      (identifier . nnweb-google-identity))
  73     (gmane
  74      (article . nnweb-gmane-wash-article)
  75      (id . "http://gmane.org/view.php?group=%s")
  76      (reference . identity)
  77      (map . nnweb-gmane-create-mapping)
  78      (search . nnweb-gmane-search)
  79      (address . "http://search.gmane.org/nov.php")
  80      (identifier . nnweb-gmane-identity)))
  81   "Type-definition alist.")
  82
  83 (defvoo nnweb-search nil
  84   "Search string to feed to Google.")
  85
  86 (defvoo nnweb-max-hits 999
  87   "Maximum number of hits to display.")
  88
  89 (defvoo nnweb-ephemeral-p nil
  90   "Whether this nnweb server is ephemeral.")
  91
  92 ;;; Internal variables
  93
  94 (defvoo nnweb-articles nil)
  95 (defvoo nnweb-buffer nil)
  96 (defvoo nnweb-group-alist nil)
  97 (defvoo nnweb-group nil)
  98 (defvoo nnweb-hashtb nil)
  99
 100 ;;; Interface functions
 101
 102 (nnoo-define-basics nnweb)
 103
 104 (deffoo nnweb-retrieve-headers (articles &optional group server fetch-old)
 105   (nnweb-possibly-change-server group server)
 106   (with-current-buffer nntp-server-buffer
 107     (erase-buffer)
 108     (let (article header)
 109       (mm-with-unibyte-current-buffer
 110         (while (setq article (pop articles))
 111           (when (setq header (cadr (assq article nnweb-articles)))
 112             (nnheader-insert-nov header))))
 113       'nov)))
 114
 115 (deffoo nnweb-request-scan (&optional group server)
 116   (nnweb-possibly-change-server group server)
 117   (if nnweb-ephemeral-p
 118       (setq nnweb-hashtb (gnus-make-hashtable 4095))
 119     (unless nnweb-articles
 120       (nnweb-read-overview group)))
 121   (funcall (nnweb-definition 'map))
 122   (unless nnweb-ephemeral-p
 123     (nnweb-write-active)
 124     (nnweb-write-overview group)))
 125
 126 (deffoo nnweb-request-group (group &optional server dont-check info)
 127   (nnweb-possibly-change-server group server)
 128   (unless (or nnweb-ephemeral-p
 129               dont-check
 130               nnweb-articles)
 131     (nnweb-read-overview group))
 132   (cond
 133    ((not nnweb-articles)
 134     (nnheader-report 'nnweb "No matching articles"))
 135    (t
 136     (let ((active (if nnweb-ephemeral-p
 137                       (cons (caar nnweb-articles)
 138                             (caar (last nnweb-articles)))
 139                     (cadr (assoc group nnweb-group-alist)))))
 140       (nnheader-report 'nnweb "Opened group %s" group)
 141       (nnheader-insert
 142        "211 %d %d %d %s\n" (length nnweb-articles)
 143        (car active) (cdr active) group)))))
 144
 145 (deffoo nnweb-close-group (group &optional server)
 146   (nnweb-possibly-change-server group server)
 147   (when (gnus-buffer-live-p nnweb-buffer)
 148     (with-current-buffer nnweb-buffer
 149       (set-buffer-modified-p nil)
 150       (kill-buffer nnweb-buffer)))
 151   t)
 152
 153 (deffoo nnweb-request-article (article &optional group server buffer)
 154   (nnweb-possibly-change-server group server)
 155   (with-current-buffer (or buffer nntp-server-buffer)
 156     (let* ((header (cadr (assq article nnweb-articles)))
 157            (url (and header (mail-header-xref header))))
 158       (when (or (and url
 159                      (mm-with-unibyte-current-buffer
 160                        (mm-url-insert url)))
 161                 (and (stringp article)
 162                      (nnweb-definition 'id t)
 163                      (let ((fetch (nnweb-definition 'id))
 164                            art active)
 165                        (when (string-match "^<\\(.*\\)>$" article)
 166                          (setq art (match-string 1 article)))
 167                        (when (and fetch art)
 168                          (setq url (format fetch
 169                                            (mm-url-form-encode-xwfu art)))
 170                          (mm-with-unibyte-current-buffer
 171                            (mm-url-insert url))
 172                          (if (nnweb-definition 'reference t)
 173                              (setq article
 174                                    (funcall (nnweb-definition
 175                                              'reference) article)))))))
 176         (unless nnheader-callback-function
 177           (funcall (nnweb-definition 'article)))
 178         (nnheader-report 'nnweb "Fetched article %s" article)
 179         (cons group (and (numberp article) article))))))
 180
 181 (deffoo nnweb-close-server (&optional server)
 182   (when (and (nnweb-server-opened server)
 183              (gnus-buffer-live-p nnweb-buffer))
 184     (with-current-buffer nnweb-buffer
 185       (set-buffer-modified-p nil)
 186       (kill-buffer nnweb-buffer)))
 187   (nnoo-close-server 'nnweb server))
 188
 189 (deffoo nnweb-request-list (&optional server)
 190   (nnweb-possibly-change-server nil server)
 191   (with-current-buffer nntp-server-buffer
 192     (nnmail-generate-active (list (assoc server nnweb-group-alist)))
 193     t))
 194
 195 (deffoo nnweb-request-update-info (group info &optional server))
 196
 197 (deffoo nnweb-asynchronous-p ()
 198   nil)
 199
 200 (deffoo nnweb-request-create-group (group &optional server args)
 201   (nnweb-possibly-change-server nil server)
 202   (nnweb-request-delete-group group)
 203   (push `(,group ,(cons 1 0)) nnweb-group-alist)
 204   (nnweb-write-active)
 205   t)
 206
 207 (deffoo nnweb-request-delete-group (group &optional force server)
 208   (nnweb-possibly-change-server group server)
 209   (gnus-alist-pull group nnweb-group-alist t)
 210   (nnweb-write-active)
 211   (gnus-delete-file (nnweb-overview-file group))
 212   t)
 213
 214 (nnoo-define-skeleton nnweb)
 215
 216 ;;; Internal functions
 217
 218 (defun nnweb-read-overview (group)
 219   "Read the overview of GROUP and build the map."
 220   (when (file-exists-p (nnweb-overview-file group))
 221     (mm-with-unibyte-buffer
 222       (nnheader-insert-file-contents (nnweb-overview-file group))
 223       (goto-char (point-min))
 224       (let (header)
 225         (while (not (eobp))
 226           (setq header (nnheader-parse-nov))
 227           (forward-line 1)
 228           (push (list (mail-header-number header)
 229                       header (mail-header-xref header))
 230                 nnweb-articles)
 231           (nnweb-set-hashtb header (car nnweb-articles)))))))
 232
 233 (defun nnweb-write-overview (group)
 234   "Write the overview file for GROUP."
 235   (with-temp-file (nnweb-overview-file group)
 236     (let ((articles nnweb-articles))
 237       (while articles
 238         (nnheader-insert-nov (cadr (pop articles)))))))
 239
 240 (defun nnweb-set-hashtb (header data)
 241   (gnus-sethash (nnweb-identifier (mail-header-xref header))
 242                 data nnweb-hashtb))
 243
 244 (defun nnweb-get-hashtb (url)
 245   (gnus-gethash (nnweb-identifier url) nnweb-hashtb))
 246
 247 (defun nnweb-identifier (ident)
 248   (funcall (nnweb-definition 'identifier) ident))
 249
 250 (defun nnweb-overview-file (group)
 251   "Return the name of the overview file of GROUP."
 252   (nnheader-concat nnweb-directory group ".overview"))
 253
 254 (defun nnweb-write-active ()
 255   "Save the active file."
 256   (gnus-make-directory nnweb-directory)
 257   (with-temp-file (nnheader-concat nnweb-directory "active")
 258     (prin1 `(setq nnweb-group-alist ',nnweb-group-alist) (current-buffer))))
 259
 260 (defun nnweb-read-active ()
 261   "Read the active file."
 262   (load (nnheader-concat nnweb-directory "active") t t t))
 263
 264 (defun nnweb-definition (type &optional noerror)
 265   "Return the definition of TYPE."
 266   (let ((def (cdr (assq type (assq nnweb-type nnweb-type-definition)))))
 267     (when (and (not def)
 268                (not noerror))
 269       (error "Undefined definition %s" type))
 270     def))
 271
 272 (defun nnweb-possibly-change-server (&optional group server)
 273   (when server
 274     (unless (nnweb-server-opened server)
 275       (nnweb-open-server server))
 276     (nnweb-init server))
 277   (unless nnweb-group-alist
 278     (nnweb-read-active))
 279   (unless nnweb-hashtb
 280     (setq nnweb-hashtb (gnus-make-hashtable 4095)))
 281   (when group
 282     (setq nnweb-group group)))
 283
 284 (defun nnweb-init (server)
 285   "Initialize buffers and such."
 286   (unless (gnus-buffer-live-p nnweb-buffer)
 287     (setq nnweb-buffer
 288           (save-current-buffer
 289             (nnheader-set-temp-buffer
 290              (format " *nnweb %s %s %s*"
 291                      nnweb-type nnweb-search server))
 292             (mm-disable-multibyte)
 293             (current-buffer)))))
 294
 295 ;;;
 296 ;;; groups.google.com
 297 ;;;
 298
 299 (defun nnweb-google-wash-article ()
 300   ;; We have Google's masked e-mail addresses here.  :-/
 301   (let ((case-fold-search t)
 302         (start-re "<pre>[\r\n ]*")
 303         (end-re "[\r\n ]*</pre>"))
 304     (goto-char (point-min))
 305     (if (save-excursion
 306           (or (re-search-forward "The requested message.*could not be found."
 307                                  nil t)
 308               (not (and (re-search-forward start-re nil t)
 309                         (re-search-forward end-re nil t)))))
 310         ;; FIXME: Don't know how to indicate "not found".
 311         ;; Should this function throw an error?  --rsteib
 312         (progn
 313           (gnus-message 3 "Requested article not found")
 314           (erase-buffer))
 315       (delete-region (point-min)
 316                      (re-search-forward start-re))
 317       (goto-char (point-min))
 318       (delete-region (progn
 319                        (re-search-forward end-re)
 320                        (match-beginning 0))
 321                      (point-max))
 322       (mm-url-decode-entities))))
 323
 324 (defun nnweb-google-parse-1 (&optional Message-ID)
 325   "Parse search result in current buffer."
 326   (let ((i 0)
 327         (case-fold-search t)
 328         (active (cadr (assoc nnweb-group nnweb-group-alist)))
 329         Subject Score Date Newsgroups From
 330         map url mid)
 331     (unless active
 332       (push (list nnweb-group (setq active (cons 1 0)))
 333             nnweb-group-alist))
 334     ;; Go through all the article hits on this page.
 335     (goto-char (point-min))
 336     (while
 337         (re-search-forward
 338          "a +href=\"/group/\\([^>\"]+\\)/browse_thread/[^>]+#\\([0-9a-f]+\\)"
 339          nil t)
 340       (setq Newsgroups (match-string-no-properties 1)
 341             ;; Note: Starting with Google Groups 2, `mid' is a Google-internal
 342             ;; ID, not a proper Message-ID.
 343             mid (match-string-no-properties 2)
 344             url (format
 345                  (nnweb-definition 'result) Newsgroups mid))
 346       (narrow-to-region (search-forward ">" nil t)
 347                         (search-forward "</a>" nil t))
 348       (mm-url-remove-markup)
 349       (mm-url-decode-entities)
 350       (setq Subject (buffer-string))
 351       (goto-char (point-max))
 352       (widen)
 353       (narrow-to-region (point)
 354                         (search-forward "</table" nil t))
 355
 356       (mm-url-remove-markup)
 357       (mm-url-decode-entities)
 358       (goto-char (point-max))
 359       (when
 360           (re-search-backward
 361            "^\\(?:\\(\\w+\\) \\([0-9]+\\)\\|\\S-+\\)\\(?: \\([0-9]\\{4\\}\\)\\)? by ?\\(.*\\)"
 362            nil t)
 363         (setq Date (if (match-string 1)
 364                        (format "%s %s 00:00:00 %s"
 365                                (match-string 1)
 366                                (match-string 2)
 367                                (or (match-string 3)
 368                                    (format-time-string "%Y")))
 369                      (current-time-string)))
 370         (setq From (match-string 4)))
 371       (widen)
 372       (incf i)
 373       (unless (nnweb-get-hashtb url)
 374         (push
 375          (list
 376           (incf (cdr active))
 377           (make-full-mail-header
 378            (cdr active) (if Newsgroups
 379                             (concat  "(" Newsgroups ") " Subject)
 380                           Subject)
 381            From Date (or Message-ID mid)
 382            nil 0 0 url))
 383          map)
 384         (nnweb-set-hashtb (cadar map) (car map))))
 385     map))
 386
 387 (defun nnweb-google-reference (id)
 388   (let ((map (nnweb-google-parse-1 id)) header)
 389     (setq nnweb-articles
 390           (nconc nnweb-articles map))
 391     (when (setq header (cadar map))
 392       (mm-with-unibyte-current-buffer
 393         (mm-url-insert (mail-header-xref header)))
 394       (caar map))))
 395
 396 (defun nnweb-google-create-mapping ()
 397   "Perform the search and create a number-to-url alist."
 398   (with-current-buffer nnweb-buffer
 399     (erase-buffer)
 400     (nnheader-message 7 "Searching google...")
 401     (when (funcall (nnweb-definition 'search) nnweb-search)
 402         (let ((more t)
 403               (i 0))
 404           (while more
 405             (setq nnweb-articles
 406                   (nconc nnweb-articles (nnweb-google-parse-1)))
 407             ;; Check if there are more articles to fetch
 408             (goto-char (point-min))
 409             (incf i 100)
 410             (if (or (not (re-search-forward
 411                           "<a [^>]+href=\"\n?\\([^>\" \n\t]+\\)[^<]*<img[^>]+src=[^>]+next"
 412                           nil t))
 413                     (>= i nnweb-max-hits))
 414                 (setq more nil)
 415               ;; Yup, there are more articles
 416               (setq more (concat (nnweb-definition 'base) (match-string 1)))
 417             (when more
 418               (erase-buffer)
 419               (nnheader-message 7 "Searching google...(%d)" i)
 420               (mm-url-insert more))))
 421           ;; Return the articles in the right order.
 422           (nnheader-message 7 "Searching google...done")
 423           (setq nnweb-articles
 424                 (sort nnweb-articles 'car-less-than-car))))))
 425
 426 (defun nnweb-google-search (search)
 427   (mm-url-insert
 428    (concat
 429     (nnweb-definition 'address)
 430     "?"
 431     (mm-url-encode-www-form-urlencoded
 432      `(("q" . ,search)
 433        ("num" . ,(number-to-string
 434                   (min 100 nnweb-max-hits)))
 435        ("hq" . "")
 436        ("hl" . "en")
 437        ("lr" . "")
 438        ("safe" . "off")
 439        ("sites" . "groups")
 440        ("filter" . "0")))))
 441   t)
 442
 443 (defun nnweb-google-identity (url)
 444   "Return an unique identifier based on URL."
 445   (if (string-match "selm=\\([^ &>]+\\)" url)
 446       (match-string 1 url)
 447     url))
 448
 449 ;;;
 450 ;;; gmane.org
 451 ;;;
 452 (defun nnweb-gmane-create-mapping ()
 453   "Perform the search and create a number-to-url alist."
 454   (with-current-buffer nnweb-buffer
 455     (let ((case-fold-search t)
 456           (active (or (cadr (assoc nnweb-group nnweb-group-alist))
 457                       (cons 1 0)))
 458           map)
 459       (erase-buffer)
 460       (nnheader-message 7 "Searching Gmane..." )
 461       (when (funcall (nnweb-definition 'search) nnweb-search)
 462         (goto-char (point-min))
 463         ;; Skip the status line
 464         (forward-line 1)
 465         ;; Thanks to Olly Betts we now have NOV lines in our buffer!
 466         (while (not (eobp))
 467           (unless (or (eolp) (looking-at "\x0d"))
 468             (let ((header (nnheader-parse-nov)))
 469               (let ((xref (mail-header-xref header))
 470                     (from (mail-header-from header))
 471                     (subject (mail-header-subject header))
 472                     (rfc2047-encoding-type 'mime))
 473                 (when (string-match " \\([^:]+\\)[:/]\\([0-9]+\\)" xref)
 474                   (mail-header-set-xref
 475                    header
 476                    (format "http://article.gmane.org/%s/%s/raw"
 477                            (match-string 1 xref)
 478                            (match-string 2 xref))))
 479
 480                 ;; Add host part to gmane-encrypted addresses
 481                 (when (string-match "@$" from)
 482                   (mail-header-set-from header
 483                                         (concat from "public.gmane.org")))
 484
 485                 (mail-header-set-subject header
 486                                          (rfc2047-encode-string subject))
 487
 488                 (unless (nnweb-get-hashtb (mail-header-xref header))
 489                   (mail-header-set-number header (incf (cdr active)))
 490                   (push (list (mail-header-number header) header) map)
 491                   (nnweb-set-hashtb (cadar map) (car map))))))
 492           (forward-line 1)))
 493       (nnheader-message 7 "Searching Gmane...done")
 494       (setq nnweb-articles
 495             (sort (nconc nnweb-articles map) 'car-less-than-car)))))
 496
 497 (defun nnweb-gmane-wash-article ()
 498   (let ((case-fold-search t))
 499     (goto-char (point-min))
 500     (when (search-forward "<!--X-Head-of-Message-->" nil t)
 501       (delete-region (point-min) (point))
 502       (goto-char (point-min))
 503       (while (looking-at "^<li><em>\\([^ ]+\\)</em>.*</li>")
 504         (replace-match "\\1\\2" t)
 505         (forward-line 1))
 506       (mm-url-remove-markup))))
 507
 508 (defun nnweb-gmane-search (search)
 509   (mm-url-insert
 510    (concat
 511     (nnweb-definition 'address)
 512     "?"
 513     (mm-url-encode-www-form-urlencoded
 514      `(("query" . ,search)
 515        ("HITSPERPAGE" . ,(number-to-string nnweb-max-hits))
 516        ;;("TOPDOC" . "1000")
 517        ))))
 518   (setq buffer-file-name nil)
 519   (unless (featurep 'xemacs) (set-buffer-multibyte t))
 520   (mm-decode-coding-region (point-min) (point-max) 'utf-8)
 521   t)
 522
 523 (defun nnweb-gmane-identity (url)
 524   "Return a unique identifier based on URL."
 525   (if (string-match "group=\\(.+\\)" url)
 526       (match-string 1 url)
 527     url))
 528
 529 ;;;
 530 ;;; General web/w3 interface utility functions
 531 ;;;
 532
 533 (defun nnweb-insert-html (parse)
 534   "Insert HTML based on a w3 parse tree."
 535   (if (stringp parse)
 536       ;; We used to call nnheader-string-as-multibyte here, but it cannot
 537       ;; be right, so I removed it.  If a bug shows up because of this change,
 538       ;; please do not blindly revert the change, but help me find the real
 539       ;; cause of the bug instead.  --Stef
 540       (insert parse)
 541     (insert "<" (symbol-name (car parse)) " ")
 542     (insert (mapconcat
 543              (lambda (param)
 544                (concat (symbol-name (car param)) "="
 545                        (prin1-to-string
 546                         (if (consp (cdr param))
 547                             (cadr param)
 548                           (cdr param)))))
 549              (nth 1 parse)
 550              " "))
 551     (insert ">\n")
 552     (mapc 'nnweb-insert-html (nth 2 parse))
 553     (insert "</" (symbol-name (car parse)) ">\n")))
 554
 555 (defun nnweb-parse-find (type parse &optional maxdepth)
 556   "Find the element of TYPE in PARSE."
 557   (catch 'found
 558     (nnweb-parse-find-1 type parse maxdepth)))
 559
 560 (defun nnweb-parse-find-1 (type contents maxdepth)
 561   (when (or (null maxdepth)
 562             (not (zerop maxdepth)))
 563     (when (consp contents)
 564       (when (eq (car contents) type)
 565         (throw 'found contents))
 566       (when (listp (cdr contents))
 567         (dolist (element contents)
 568           (when (consp element)
 569             (nnweb-parse-find-1 type element
 570                                 (and maxdepth (1- maxdepth)))))))))
 571
 572 (defun nnweb-parse-find-all (type parse)
 573   "Find all elements of TYPE in PARSE."
 574   (catch 'found
 575     (nnweb-parse-find-all-1 type parse)))
 576
 577 (defun nnweb-parse-find-all-1 (type contents)
 578   (let (result)
 579     (when (consp contents)
 580       (if (eq (car contents) type)
 581           (push contents result)
 582         (when (listp (cdr contents))
 583           (dolist (element contents)
 584             (when (consp element)
 585               (setq result
 586                     (nconc result (nnweb-parse-find-all-1 type element))))))))
 587     result))
 588
 589 (defvar nnweb-text)
 590 (defun nnweb-text (parse)
 591   "Return a list of text contents in PARSE."
 592   (let ((nnweb-text nil))
 593     (nnweb-text-1 parse)
 594     (nreverse nnweb-text)))
 595
 596 (defun nnweb-text-1 (contents)
 597   (dolist (element contents)
 598     (if (stringp element)
 599         (push element nnweb-text)
 600       (when (and (consp element)
 601                  (listp (cdr element)))
 602         (nnweb-text-1 element)))))
 603
 604 (provide 'nnweb)
 605
 606 ;;; nnweb.el ends here