lisp/gnus/nnweb.el

   1 ;;; nnweb.el --- retrieving articles via web search engines
   2
   3 ;; Copyright (C) 1996-2016 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 ;;; Code:
  26
  27 (eval-when-compile (require 'cl))
  28
  29 (require 'nnoo)
  30 (require 'message)
  31 (require 'gnus-util)
  32 (require 'gnus)
  33 (require 'nnmail)
  34 (require 'mm-util)
  35 (require 'mm-url)
  36 (eval-and-compile
  37   (ignore-errors
  38     (require 'url)))
  39
  40 (nnoo-declare nnweb)
  41
  42 (defvoo nnweb-directory (nnheader-concat gnus-directory "nnweb/")
  43   "Where nnweb will save its files.")
  44
  45 (defvoo nnweb-type 'google
  46   "What search engine type is being used.
  47 Valid types include `google', `dejanews', and `gmane'.")
  48
  49 (defvar nnweb-type-definition
  50   '((google
  51      (id . "http://www.google.com/groups?as_umsgid=%s&hl=en&dmode=source")
  52      (result . "http://groups.google.com/group/%s/msg/%s?dmode=source")
  53      (article . nnweb-google-wash-article)
  54      (reference . identity)
  55      (map . nnweb-google-create-mapping)
  56      (search . nnweb-google-search)
  57      (address . "http://groups.google.com/groups")
  58      (base    . "http://groups.google.com")
  59      (identifier . nnweb-google-identity))
  60     (dejanews ;; alias of google
  61      (id . "http://www.google.com/groups?as_umsgid=%s&hl=en&dmode=source")
  62      (result . "http://groups.google.com/group/%s/msg/%s?dmode=source")
  63      (article . nnweb-google-wash-article)
  64      (reference . identity)
  65      (map . nnweb-google-create-mapping)
  66      (search . nnweb-google-search)
  67      (address . "http://groups.google.com/groups")
  68      (base    . "http://groups.google.com")
  69      (identifier . nnweb-google-identity))
  70     (gmane
  71      (article . nnweb-gmane-wash-article)
  72      (id . "http://gmane.org/view.php?group=%s")
  73      (reference . identity)
  74      (map . nnweb-gmane-create-mapping)
  75      (search . nnweb-gmane-search)
  76      (address . "http://search.gmane.org/nov.php")
  77      (identifier . nnweb-gmane-identity)))
  78   "Type-definition alist.")
  79
  80 (defvoo nnweb-search nil
  81   "Search string to feed to Google.")
  82
  83 (defvoo nnweb-max-hits 999
  84   "Maximum number of hits to display.")
  85
  86 (defvoo nnweb-ephemeral-p nil
  87   "Whether this nnweb server is ephemeral.")
  88
  89 ;;; Internal variables
  90
  91 (defvoo nnweb-articles nil)
  92 (defvoo nnweb-buffer nil)
  93 (defvoo nnweb-group-alist nil)
  94 (defvoo nnweb-group nil)
  95 (defvoo nnweb-hashtb nil)
  96
  97 ;;; Interface functions
  98
  99 (nnoo-define-basics nnweb)
 100
 101 (deffoo nnweb-retrieve-headers (articles &optional group server fetch-old)
 102   (nnweb-possibly-change-server group server)
 103   (with-current-buffer nntp-server-buffer
 104     (erase-buffer)
 105     (let (article header)
 106       (mm-with-unibyte-current-buffer
 107         (while (setq article (pop articles))
 108           (when (setq header (cadr (assq article nnweb-articles)))
 109             (nnheader-insert-nov header))))
 110       'nov)))
 111
 112 (deffoo nnweb-request-scan (&optional group server)
 113   (nnweb-possibly-change-server group server)
 114   (if nnweb-ephemeral-p
 115       (setq nnweb-hashtb (gnus-make-hashtable 4095))
 116     (unless nnweb-articles
 117       (nnweb-read-overview group)))
 118   (funcall (nnweb-definition 'map))
 119   (unless nnweb-ephemeral-p
 120     (nnweb-write-active)
 121     (nnweb-write-overview group)))
 122
 123 (deffoo nnweb-request-group (group &optional server dont-check info)
 124   (nnweb-possibly-change-server group server)
 125   (unless (or nnweb-ephemeral-p
 126               dont-check
 127               nnweb-articles)
 128     (nnweb-read-overview group))
 129   (cond
 130    ((not nnweb-articles)
 131     (nnheader-report 'nnweb "No matching articles"))
 132    (t
 133     (let ((active (if nnweb-ephemeral-p
 134                       (cons (caar nnweb-articles)
 135                             (caar (last nnweb-articles)))
 136                     (cadr (assoc group nnweb-group-alist)))))
 137       (nnheader-report 'nnweb "Opened group %s" group)
 138       (nnheader-insert
 139        "211 %d %d %d %s\n" (length nnweb-articles)
 140        (car active) (cdr active) group)))))
 141
 142 (deffoo nnweb-close-group (group &optional server)
 143   (nnweb-possibly-change-server group server)
 144   (when (gnus-buffer-live-p nnweb-buffer)
 145     (with-current-buffer nnweb-buffer
 146       (set-buffer-modified-p nil)
 147       (kill-buffer nnweb-buffer)))
 148   t)
 149
 150 (deffoo nnweb-request-article (article &optional group server buffer)
 151   (nnweb-possibly-change-server group server)
 152   (with-current-buffer (or buffer nntp-server-buffer)
 153     (let* ((header (cadr (assq article nnweb-articles)))
 154            (url (and header (mail-header-xref header))))
 155       (when (or (and url
 156                      (mm-with-unibyte-current-buffer
 157                        (mm-url-insert url)))
 158                 (and (stringp article)
 159                      (nnweb-definition 'id t)
 160                      (let ((fetch (nnweb-definition 'id))
 161                            art active)
 162                        (when (string-match "^<\\(.*\\)>$" article)
 163                          (setq art (match-string 1 article)))
 164                        (when (and fetch art)
 165                          (setq url (format fetch
 166                                            (mm-url-form-encode-xwfu art)))
 167                          (mm-with-unibyte-current-buffer
 168                            (mm-url-insert url))
 169                          (if (nnweb-definition 'reference t)
 170                              (setq article
 171                                    (funcall (nnweb-definition
 172                                              'reference) article)))))))
 173         (unless nnheader-callback-function
 174           (funcall (nnweb-definition 'article)))
 175         (nnheader-report 'nnweb "Fetched article %s" article)
 176         (cons group (and (numberp article) article))))))
 177
 178 (deffoo nnweb-close-server (&optional server)
 179   (when (and (nnweb-server-opened server)
 180              (gnus-buffer-live-p nnweb-buffer))
 181     (with-current-buffer nnweb-buffer
 182       (set-buffer-modified-p nil)
 183       (kill-buffer nnweb-buffer)))
 184   (nnoo-close-server 'nnweb server))
 185
 186 (deffoo nnweb-request-list (&optional server)
 187   (nnweb-possibly-change-server nil server)
 188   (with-current-buffer nntp-server-buffer
 189     (nnmail-generate-active (list (assoc server nnweb-group-alist)))
 190     t))
 191
 192 (deffoo nnweb-request-update-info (group info &optional server))
 193
 194 (deffoo nnweb-asynchronous-p ()
 195   nil)
 196
 197 (deffoo nnweb-request-create-group (group &optional server args)
 198   (nnweb-possibly-change-server nil server)
 199   (nnweb-request-delete-group group)
 200   (push `(,group ,(cons 1 0)) nnweb-group-alist)
 201   (nnweb-write-active)
 202   t)
 203
 204 (deffoo nnweb-request-delete-group (group &optional force server)
 205   (nnweb-possibly-change-server group server)
 206   (gnus-alist-pull group nnweb-group-alist t)
 207   (nnweb-write-active)
 208   (gnus-delete-file (nnweb-overview-file group))
 209   t)
 210
 211 (nnoo-define-skeleton nnweb)
 212
 213 ;;; Internal functions
 214
 215 (defun nnweb-read-overview (group)
 216   "Read the overview of GROUP and build the map."
 217   (when (file-exists-p (nnweb-overview-file group))
 218     (mm-with-unibyte-buffer
 219       (nnheader-insert-file-contents (nnweb-overview-file group))
 220       (goto-char (point-min))
 221       (let (header)
 222         (while (not (eobp))
 223           (setq header (nnheader-parse-nov))
 224           (forward-line 1)
 225           (push (list (mail-header-number header)
 226                       header (mail-header-xref header))
 227                 nnweb-articles)
 228           (nnweb-set-hashtb header (car nnweb-articles)))))))
 229
 230 (defun nnweb-write-overview (group)
 231   "Write the overview file for GROUP."
 232   (with-temp-file (nnweb-overview-file group)
 233     (let ((articles nnweb-articles))
 234       (while articles
 235         (nnheader-insert-nov (cadr (pop articles)))))))
 236
 237 (defun nnweb-set-hashtb (header data)
 238   (gnus-sethash (nnweb-identifier (mail-header-xref header))
 239                 data nnweb-hashtb))
 240
 241 (defun nnweb-get-hashtb (url)
 242   (gnus-gethash (nnweb-identifier url) nnweb-hashtb))
 243
 244 (defun nnweb-identifier (ident)
 245   (funcall (nnweb-definition 'identifier) ident))
 246
 247 (defun nnweb-overview-file (group)
 248   "Return the name of the overview file of GROUP."
 249   (nnheader-concat nnweb-directory group ".overview"))
 250
 251 (defun nnweb-write-active ()
 252   "Save the active file."
 253   (gnus-make-directory nnweb-directory)
 254   (with-temp-file (nnheader-concat nnweb-directory "active")
 255     (prin1 `(setq nnweb-group-alist ',nnweb-group-alist) (current-buffer))))
 256
 257 (defun nnweb-read-active ()
 258   "Read the active file."
 259   (load (nnheader-concat nnweb-directory "active") t t t))
 260
 261 (defun nnweb-definition (type &optional noerror)
 262   "Return the definition of TYPE."
 263   (let ((def (cdr (assq type (assq nnweb-type nnweb-type-definition)))))
 264     (when (and (not def)
 265                (not noerror))
 266       (error "Undefined definition %s" type))
 267     def))
 268
 269 (defun nnweb-possibly-change-server (&optional group server)
 270   (when server
 271     (unless (nnweb-server-opened server)
 272       (nnweb-open-server server))
 273     (nnweb-init server))
 274   (unless nnweb-group-alist
 275     (nnweb-read-active))
 276   (unless nnweb-hashtb
 277     (setq nnweb-hashtb (gnus-make-hashtable 4095)))
 278   (when group
 279     (setq nnweb-group group)))
 280
 281 (defun nnweb-init (server)
 282   "Initialize buffers and such."
 283   (unless (gnus-buffer-live-p nnweb-buffer)
 284     (setq nnweb-buffer
 285           (save-current-buffer
 286             (nnheader-set-temp-buffer
 287              (format " *nnweb %s %s %s*"
 288                      nnweb-type nnweb-search server))
 289             (mm-disable-multibyte)
 290             (current-buffer)))))
 291
 292 ;;;
 293 ;;; groups.google.com
 294 ;;;
 295
 296 (defun nnweb-google-wash-article ()
 297   ;; We have Google's masked e-mail addresses here.  :-/
 298   (let ((case-fold-search t)
 299         (start-re "<pre>[\r\n ]*")
 300         (end-re "[\r\n ]*</pre>"))
 301     (goto-char (point-min))
 302     (if (save-excursion
 303           (or (re-search-forward "The requested message.*could not be found."
 304                                  nil t)
 305               (not (and (re-search-forward start-re nil t)
 306                         (re-search-forward end-re nil t)))))
 307         ;; FIXME: Don't know how to indicate "not found".
 308         ;; Should this function throw an error?  --rsteib
 309         (progn
 310           (gnus-message 3 "Requested article not found")
 311           (erase-buffer))
 312       (delete-region (point-min)
 313                      (re-search-forward start-re))
 314       (goto-char (point-min))
 315       (delete-region (progn
 316                        (re-search-forward end-re)
 317                        (match-beginning 0))
 318                      (point-max))
 319       (mm-url-decode-entities))))
 320
 321 (defun nnweb-google-parse-1 (&optional Message-ID)
 322   "Parse search result in current buffer."
 323   (let ((i 0)
 324         (case-fold-search t)
 325         (active (cadr (assoc nnweb-group nnweb-group-alist)))
 326         Subject Score Date Newsgroups From
 327         map url mid)
 328     (unless active
 329       (push (list nnweb-group (setq active (cons 1 0)))
 330             nnweb-group-alist))
 331     ;; Go through all the article hits on this page.
 332     (goto-char (point-min))
 333     (while
 334         (re-search-forward
 335          "a +href=\"/group/\\([^>\"]+\\)/browse_thread/[^>]+#\\([0-9a-f]+\\)"
 336          nil t)
 337       (setq Newsgroups (match-string-no-properties 1)
 338             ;; Note: Starting with Google Groups 2, `mid' is a Google-internal
 339             ;; ID, not a proper Message-ID.
 340             mid (match-string-no-properties 2)
 341             url (format
 342                  (nnweb-definition 'result) Newsgroups mid))
 343       (narrow-to-region (search-forward ">" nil t)
 344                         (search-forward "</a>" nil t))
 345       (mm-url-remove-markup)
 346       (mm-url-decode-entities)
 347       (setq Subject (buffer-string))
 348       (goto-char (point-max))
 349       (widen)
 350       (narrow-to-region (point)
 351                         (search-forward "</table" nil t))
 352
 353       (mm-url-remove-markup)
 354       (mm-url-decode-entities)
 355       (goto-char (point-max))
 356       (when
 357           (re-search-backward
 358            "^\\(?:\\(\\w+\\) \\([0-9]+\\)\\|\\S-+\\)\\(?: \\([0-9]\\{4\\}\\)\\)? by ?\\(.*\\)"
 359            nil t)
 360         (setq Date (if (match-string 1)
 361                        (format "%s %s 00:00:00 %s"
 362                                (match-string 1)
 363                                (match-string 2)
 364                                (or (match-string 3)
 365                                    (format-time-string "%Y")))
 366                      (current-time-string)))
 367         (setq From (match-string 4)))
 368       (widen)
 369       (incf i)
 370       (unless (nnweb-get-hashtb url)
 371         (push
 372          (list
 373           (incf (cdr active))
 374           (make-full-mail-header
 375            (cdr active) (if Newsgroups
 376                             (concat  "(" Newsgroups ") " Subject)
 377                           Subject)
 378            From Date (or Message-ID mid)
 379            nil 0 0 url))
 380          map)
 381         (nnweb-set-hashtb (cadar map) (car map))))
 382     map))
 383
 384 (defun nnweb-google-reference (id)
 385   (let ((map (nnweb-google-parse-1 id)) header)
 386     (setq nnweb-articles
 387           (nconc nnweb-articles map))
 388     (when (setq header (cadar map))
 389       (mm-with-unibyte-current-buffer
 390         (mm-url-insert (mail-header-xref header)))
 391       (caar map))))
 392
 393 (defun nnweb-google-create-mapping ()
 394   "Perform the search and create a number-to-url alist."
 395   (with-current-buffer nnweb-buffer
 396     (erase-buffer)
 397     (nnheader-message 7 "Searching google...")
 398     (when (funcall (nnweb-definition 'search) nnweb-search)
 399         (let ((more t)
 400               (i 0))
 401           (while more
 402             (setq nnweb-articles
 403                   (nconc nnweb-articles (nnweb-google-parse-1)))
 404             ;; Check if there are more articles to fetch
 405             (goto-char (point-min))
 406             (incf i 100)
 407             (if (or (not (re-search-forward
 408                           "<a [^>]+href=\"\n?\\([^>\" \n\t]+\\)[^<]*<img[^>]+src=[^>]+next"
 409                           nil t))
 410                     (>= i nnweb-max-hits))
 411                 (setq more nil)
 412               ;; Yup, there are more articles
 413               (setq more (concat (nnweb-definition 'base) (match-string 1)))
 414             (when more
 415               (erase-buffer)
 416               (nnheader-message 7 "Searching google...(%d)" i)
 417               (mm-url-insert more))))
 418           ;; Return the articles in the right order.
 419           (nnheader-message 7 "Searching google...done")
 420           (setq nnweb-articles
 421                 (sort nnweb-articles 'car-less-than-car))))))
 422
 423 (defun nnweb-google-search (search)
 424   (mm-url-insert
 425    (concat
 426     (nnweb-definition 'address)
 427     "?"
 428     (mm-url-encode-www-form-urlencoded
 429      `(("q" . ,search)
 430        ("num" . ,(number-to-string
 431                   (min 100 nnweb-max-hits)))
 432        ("hq" . "")
 433        ("hl" . "en")
 434        ("lr" . "")
 435        ("safe" . "off")
 436        ("sites" . "groups")
 437        ("filter" . "0")))))
 438   t)
 439
 440 (defun nnweb-google-identity (url)
 441   "Return a unique identifier based on URL."
 442   (if (string-match "selm=\\([^ &>]+\\)" url)
 443       (match-string 1 url)
 444     url))
 445
 446 ;;;
 447 ;;; gmane.org
 448 ;;;
 449 (defun nnweb-gmane-create-mapping ()
 450   "Perform the search and create a number-to-url alist."
 451   (with-current-buffer nnweb-buffer
 452     (let ((case-fold-search t)
 453           (active (or (cadr (assoc nnweb-group nnweb-group-alist))
 454                       (cons 1 0)))
 455           map)
 456       (erase-buffer)
 457       (nnheader-message 7 "Searching Gmane..." )
 458       (when (funcall (nnweb-definition 'search) nnweb-search)
 459         (goto-char (point-min))
 460         ;; Skip the status line
 461         (forward-line 1)
 462         ;; Thanks to Olly Betts we now have NOV lines in our buffer!
 463         (while (not (eobp))
 464           (unless (or (eolp) (looking-at "\x0d"))
 465             (let ((header (nnheader-parse-nov)))
 466               (let ((xref (mail-header-xref header))
 467                     (from (mail-header-from header))
 468                     (subject (mail-header-subject header))
 469                     (rfc2047-encoding-type 'mime))
 470                 (when (string-match " \\([^:]+\\)[:/]\\([0-9]+\\)" xref)
 471                   (mail-header-set-xref
 472                    header
 473                    (format "http://article.gmane.org/%s/%s/raw"
 474                            (match-string 1 xref)
 475                            (match-string 2 xref))))
 476
 477                 ;; Add host part to gmane-encrypted addresses
 478                 (when (string-match "@$" from)
 479                   (mail-header-set-from header
 480                                         (concat from "public.gmane.org")))
 481
 482                 (mail-header-set-subject header
 483                                          (rfc2047-encode-string subject))
 484
 485                 (unless (nnweb-get-hashtb (mail-header-xref header))
 486                   (mail-header-set-number header (incf (cdr active)))
 487                   (push (list (mail-header-number header) header) map)
 488                   (nnweb-set-hashtb (cadar map) (car map))))))
 489           (forward-line 1)))
 490       (nnheader-message 7 "Searching Gmane...done")
 491       (setq nnweb-articles
 492             (sort (nconc nnweb-articles map) 'car-less-than-car)))))
 493
 494 (defun nnweb-gmane-wash-article ()
 495   (let ((case-fold-search t))
 496     (goto-char (point-min))
 497     (when (search-forward "<!--X-Head-of-Message-->" nil t)
 498       (delete-region (point-min) (point))
 499       (goto-char (point-min))
 500       (while (looking-at "^<li><em>\\([^ ]+\\)</em>.*</li>")
 501         (replace-match "\\1\\2" t)
 502         (forward-line 1))
 503       (mm-url-remove-markup))))
 504
 505 (defun nnweb-gmane-search (search)
 506   (mm-url-insert
 507    (concat
 508     (nnweb-definition 'address)
 509     "?"
 510     (mm-url-encode-www-form-urlencoded
 511      `(("query" . ,search)
 512        ("HITSPERPAGE" . ,(number-to-string nnweb-max-hits))
 513        ;;("TOPDOC" . "1000")
 514        ))))
 515   (setq buffer-file-name nil)
 516   (unless (featurep 'xemacs) (set-buffer-multibyte t))
 517   (mm-decode-coding-region (point-min) (point-max) 'utf-8)
 518   t)
 519
 520 (defun nnweb-gmane-identity (url)
 521   "Return a unique identifier based on URL."
 522   (if (string-match "group=\\(.+\\)" url)
 523       (match-string 1 url)
 524     url))
 525
 526 ;;;
 527 ;;; General web interface utility functions
 528 ;;;
 529
 530 (defun nnweb-insert-html (parse)
 531   "Insert HTML based on a w3 parse tree."
 532   (if (stringp parse)
 533       ;; We used to call nnheader-string-as-multibyte here, but it cannot
 534       ;; be right, so I removed it.  If a bug shows up because of this change,
 535       ;; please do not blindly revert the change, but help me find the real
 536       ;; cause of the bug instead.  --Stef
 537       (insert parse)
 538     (insert "<" (symbol-name (car parse)) " ")
 539     (insert (mapconcat
 540              (lambda (param)
 541                (concat (symbol-name (car param)) "="
 542                        (prin1-to-string
 543                         (if (consp (cdr param))
 544                             (cadr param)
 545                           (cdr param)))))
 546              (nth 1 parse)
 547              " "))
 548     (insert ">\n")
 549     (mapc 'nnweb-insert-html (nth 2 parse))
 550     (insert "</" (symbol-name (car parse)) ">\n")))
 551
 552 (defun nnweb-parse-find (type parse &optional maxdepth)
 553   "Find the element of TYPE in PARSE."
 554   (catch 'found
 555     (nnweb-parse-find-1 type parse maxdepth)))
 556
 557 (defun nnweb-parse-find-1 (type contents maxdepth)
 558   (when (or (null maxdepth)
 559             (not (zerop maxdepth)))
 560     (when (consp contents)
 561       (when (eq (car contents) type)
 562         (throw 'found contents))
 563       (when (listp (cdr contents))
 564         (dolist (element contents)
 565           (when (consp element)
 566             (nnweb-parse-find-1 type element
 567                                 (and maxdepth (1- maxdepth)))))))))
 568
 569 (defun nnweb-parse-find-all (type parse)
 570   "Find all elements of TYPE in PARSE."
 571   (catch 'found
 572     (nnweb-parse-find-all-1 type parse)))
 573
 574 (defun nnweb-parse-find-all-1 (type contents)
 575   (let (result)
 576     (when (consp contents)
 577       (if (eq (car contents) type)
 578           (push contents result)
 579         (when (listp (cdr contents))
 580           (dolist (element contents)
 581             (when (consp element)
 582               (setq result
 583                     (nconc result (nnweb-parse-find-all-1 type element))))))))
 584     result))
 585
 586 (defvar nnweb-text)
 587 (defun nnweb-text (parse)
 588   "Return a list of text contents in PARSE."
 589   (let ((nnweb-text nil))
 590     (nnweb-text-1 parse)
 591     (nreverse nnweb-text)))
 592
 593 (defun nnweb-text-1 (contents)
 594   (dolist (element contents)
 595     (if (stringp element)
 596         (push element nnweb-text)
 597       (when (and (consp element)
 598                  (listp (cdr element)))
 599         (nnweb-text-1 element)))))
 600
 601 (provide 'nnweb)
 602
 603 ;;; nnweb.el ends here