Mercurial > emacs
annotate lisp/gnus/nnweb.el @ 19860:c17fd465ea95 libc-970911 libc-970912 libc-970913 libc-970914 libc-970915 libc-970916 libc-970917 libc-970918 libc-970919 libc-970920 libc-970921 libc-970922 libc-970923 libc-970924 libc-970925 libc-970926 libc-970927 libc-970928 libc-970929 libc-970930 libc-971001 libc-971018 libc-971019 libc-971020 libc-971021 libc-971022 libc-971023 libc-971024 libc-971025 libc-971026 libc-971027 libc-971028 libc-971029 libc-971030 libc-971031 libc-971101 libc-971102 libc-971103 libc-971104 libc-971105 libc-971106 libc-971107 libc-971108 libc-971109 libc-971110 libc-971111 libc-971112 libc-971113 libc-971114 libc-971115 libc-971116 libc-971117 libc-971118 libc-971120 libc-971121 libc-971122 libc-971123 libc-971124 libc-971125 libc-971126 libc-971127 libc-971128 libc-971129 libc-971130 libc-971201 libc-971203 libc-971204 libc-971205 libc-971206 libc-971207 libc-971208 libc-971209 libc-971210 libc-971211 libc-971212 libc-971213 libc-971214 libc-971217 libc-971218 libc-971219 libc-971220 libc-971221 libc-971222 libc-971223 libc-971224 libc-971225 libc-971226 libc-971227 libc-971228 libc-971229 libc-971230 libc-971231 libc-980103 libc-980104 libc-980105 libc-980106 libc-980107 libc-980108 libc-980109 libc-980110 libc-980111 libc-980112 libc-980114 libc-980115 libc-980116 libc-980117 libc-980118 libc-980119 libc-980120 libc-980121 libc-980122 libc-980123 libc-980124 libc-980125 libc-980126 libc-980127 libc-980128
typos.
author | Jeff Law <law@redhat.com> |
---|---|
date | Wed, 10 Sep 1997 21:16:20 +0000 |
parents | 6f6cf9184e93 |
children | 5f1ab3dd344d |
rev | line source |
---|---|
17493 | 1 ;;; nnweb.el --- retrieving articles via web search engines |
2 ;; Copyright (C) 1996,97 Free Software Foundation, Inc. | |
3 | |
4 ;; Author: Lars Magne Ingebrigtsen <larsi@ifi.uio.no> | |
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 | |
19521
6f6cf9184e93
Require cl at compile time.
Richard M. Stallman <rms@gnu.org>
parents:
17493
diff
changeset
|
31 (eval-when-compile (require 'cl)) |
6f6cf9184e93
Require cl at compile time.
Richard M. Stallman <rms@gnu.org>
parents:
17493
diff
changeset
|
32 |
17493 | 33 (require 'nnoo) |
34 (require 'message) | |
35 (require 'gnus-util) | |
36 (require 'gnus) | |
37 (require 'w3) | |
38 (require 'url) | |
39 (require 'nnmail) | |
40 (ignore-errors | |
41 (require 'w3-forms)) | |
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 'dejanews | |
49 "What search engine type is being used.") | |
50 | |
51 (defvar nnweb-type-definition | |
52 '((dejanews | |
53 (article . nnweb-dejanews-wash-article) | |
54 (map . nnweb-dejanews-create-mapping) | |
55 (search . nnweb-dejanews-search) | |
56 (address . "http://xp9.dejanews.com/dnquery.xp") | |
57 (identifier . nnweb-dejanews-identity)) | |
58 (reference | |
59 (article . nnweb-reference-wash-article) | |
60 (map . nnweb-reference-create-mapping) | |
61 (search . nnweb-reference-search) | |
62 (address . "http://www.reference.com/cgi-bin/pn/go") | |
63 (identifier . identity)) | |
64 (altavista | |
65 (article . nnweb-altavista-wash-article) | |
66 (map . nnweb-altavista-create-mapping) | |
67 (search . nnweb-altavista-search) | |
68 (address . "http://www.altavista.digital.com/cgi-bin/query") | |
69 (id . "/cgi-bin/news?id@%s") | |
70 (identifier . identity))) | |
71 "Type-definition alist.") | |
72 | |
73 (defvoo nnweb-search nil | |
74 "Search string to feed to DejaNews.") | |
75 | |
76 (defvoo nnweb-max-hits 100 | |
77 "Maximum number of hits to display.") | |
78 | |
79 (defvoo nnweb-ephemeral-p nil | |
80 "Whether this nnweb server is ephemeral.") | |
81 | |
82 ;;; Internal variables | |
83 | |
84 (defvoo nnweb-articles nil) | |
85 (defvoo nnweb-buffer nil) | |
86 (defvoo nnweb-group-alist nil) | |
87 (defvoo nnweb-group nil) | |
88 (defvoo nnweb-hashtb nil) | |
89 | |
90 ;;; Interface functions | |
91 | |
92 (nnoo-define-basics nnweb) | |
93 | |
94 (deffoo nnweb-retrieve-headers (articles &optional group server fetch-old) | |
95 (nnweb-possibly-change-server group server) | |
96 (save-excursion | |
97 (set-buffer nntp-server-buffer) | |
98 (erase-buffer) | |
99 (let (article header) | |
100 (while (setq article (pop articles)) | |
101 (when (setq header (cadr (assq article nnweb-articles))) | |
102 (nnheader-insert-nov header))) | |
103 'nov))) | |
104 | |
105 (deffoo nnweb-request-scan (&optional group server) | |
106 (nnweb-possibly-change-server group server) | |
107 (setq nnweb-hashtb (gnus-make-hashtable 4095)) | |
108 (funcall (nnweb-definition 'map)) | |
109 (unless nnweb-ephemeral-p | |
110 (nnweb-write-active) | |
111 (nnweb-write-overview group))) | |
112 | |
113 (deffoo nnweb-request-group (group &optional server dont-check) | |
114 (nnweb-possibly-change-server nil server) | |
115 (when (and group | |
116 (not (equal group nnweb-group)) | |
117 (not nnweb-ephemeral-p)) | |
118 (let ((info (assoc group nnweb-group-alist))) | |
119 (setq nnweb-group group) | |
120 (setq nnweb-type (nth 2 info)) | |
121 (setq nnweb-search (nth 3 info)) | |
122 (unless dont-check | |
123 (nnweb-read-overview group)))) | |
124 (cond | |
125 ((not nnweb-articles) | |
126 (nnheader-report 'nnweb "No matching articles")) | |
127 (t | |
128 (let ((active (if nnweb-ephemeral-p | |
129 (cons (caar nnweb-articles) | |
130 (caar (last nnweb-articles))) | |
131 (cadr (assoc group nnweb-group-alist))))) | |
132 (nnheader-report 'nnweb "Opened group %s" group) | |
133 (nnheader-insert | |
134 "211 %d %d %d %s\n" (length nnweb-articles) | |
135 (car active) (cdr active) group))))) | |
136 | |
137 (deffoo nnweb-close-group (group &optional server) | |
138 (nnweb-possibly-change-server group server) | |
139 (when (gnus-buffer-live-p nnweb-buffer) | |
140 (save-excursion | |
141 (set-buffer nnweb-buffer) | |
142 (set-buffer-modified-p nil) | |
143 (kill-buffer nnweb-buffer))) | |
144 t) | |
145 | |
146 (deffoo nnweb-request-article (article &optional group server buffer) | |
147 (nnweb-possibly-change-server group server) | |
148 (save-excursion | |
149 (set-buffer (or buffer nntp-server-buffer)) | |
150 (let* ((header (cadr (assq article nnweb-articles))) | |
151 (url (and header (mail-header-xref header)))) | |
152 (when (or (and url | |
153 (nnweb-fetch-url url)) | |
154 (and (stringp article) | |
155 (nnweb-definition 'id t) | |
156 (let ((fetch (nnweb-definition 'id)) | |
157 art) | |
158 (when (string-match "^<\\(.*\\)>$" article) | |
159 (setq art (match-string 1 article))) | |
160 (and fetch | |
161 art | |
162 (nnweb-fetch-url | |
163 (format fetch article)))))) | |
164 (unless nnheader-callback-function | |
165 (funcall (nnweb-definition 'article)) | |
166 (nnweb-decode-entities)) | |
167 (nnheader-report 'nnweb "Fetched article %s" article) | |
168 t)))) | |
169 | |
170 (deffoo nnweb-close-server (&optional server) | |
171 (when (and (nnweb-server-opened server) | |
172 (gnus-buffer-live-p nnweb-buffer)) | |
173 (save-excursion | |
174 (set-buffer nnweb-buffer) | |
175 (set-buffer-modified-p nil) | |
176 (kill-buffer nnweb-buffer))) | |
177 (nnoo-close-server 'nnweb server)) | |
178 | |
179 (deffoo nnweb-request-list (&optional server) | |
180 (nnweb-possibly-change-server nil server) | |
181 (save-excursion | |
182 (set-buffer nntp-server-buffer) | |
183 (nnmail-generate-active nnweb-group-alist) | |
184 t)) | |
185 | |
186 (deffoo nnweb-request-update-info (group info &optional server) | |
187 (nnweb-possibly-change-server group server) | |
188 ;;(setcar (cddr info) nil) | |
189 ) | |
190 | |
191 (deffoo nnweb-asynchronous-p () | |
192 t) | |
193 | |
194 (deffoo nnweb-request-create-group (group &optional server args) | |
195 (nnweb-possibly-change-server nil server) | |
196 (nnweb-request-delete-group group) | |
197 (push `(,group ,(cons 1 0) ,@args) nnweb-group-alist) | |
198 (nnweb-write-active) | |
199 t) | |
200 | |
201 (deffoo nnweb-request-delete-group (group &optional force server) | |
202 (nnweb-possibly-change-server group server) | |
203 (gnus-delete-assoc group nnweb-group-alist) | |
204 (gnus-delete-file (nnweb-overview-file group)) | |
205 t) | |
206 | |
207 (nnoo-define-skeleton nnweb) | |
208 | |
209 ;;; Internal functions | |
210 | |
211 (defun nnweb-read-overview (group) | |
212 "Read the overview of GROUP and build the map." | |
213 (when (file-exists-p (nnweb-overview-file group)) | |
214 (nnheader-temp-write nil | |
215 (nnheader-insert-file-contents (nnweb-overview-file group)) | |
216 (goto-char (point-min)) | |
217 (let (header) | |
218 (while (not (eobp)) | |
219 (setq header (nnheader-parse-nov)) | |
220 (forward-line 1) | |
221 (push (list (mail-header-number header) | |
222 header (mail-header-xref header)) | |
223 nnweb-articles) | |
224 (nnweb-set-hashtb header (car nnweb-articles))))))) | |
225 | |
226 (defun nnweb-write-overview (group) | |
227 "Write the overview file for GROUP." | |
228 (nnheader-temp-write (nnweb-overview-file group) | |
229 (let ((articles nnweb-articles)) | |
230 (while articles | |
231 (nnheader-insert-nov (cadr (pop articles))))))) | |
232 | |
233 (defun nnweb-set-hashtb (header data) | |
234 (gnus-sethash (nnweb-identifier (mail-header-xref header)) | |
235 data nnweb-hashtb)) | |
236 | |
237 (defun nnweb-get-hashtb (url) | |
238 (gnus-gethash (nnweb-identifier url) nnweb-hashtb)) | |
239 | |
240 (defun nnweb-identifier (ident) | |
241 (funcall (nnweb-definition 'identifier) ident)) | |
242 | |
243 (defun nnweb-overview-file (group) | |
244 "Return the name of the overview file of GROUP." | |
245 (nnheader-concat nnweb-directory group ".overview")) | |
246 | |
247 (defun nnweb-write-active () | |
248 "Save the active file." | |
249 (nnheader-temp-write (nnheader-concat nnweb-directory "active") | |
250 (prin1 `(setq nnweb-group-alist ',nnweb-group-alist) (current-buffer)))) | |
251 | |
252 (defun nnweb-read-active () | |
253 "Read the active file." | |
254 (load (nnheader-concat nnweb-directory "active") t t t)) | |
255 | |
256 (defun nnweb-definition (type &optional noerror) | |
257 "Return the definition of TYPE." | |
258 (let ((def (cdr (assq type (assq nnweb-type nnweb-type-definition))))) | |
259 (when (and (not def) | |
260 (not noerror)) | |
261 (error "Undefined definition %s" type)) | |
262 def)) | |
263 | |
264 (defun nnweb-possibly-change-server (&optional group server) | |
265 (nnweb-init server) | |
266 (when server | |
267 (unless (nnweb-server-opened server) | |
268 (nnweb-open-server server))) | |
269 (unless nnweb-group-alist | |
270 (nnweb-read-active)) | |
271 (when group | |
272 (when (and (not nnweb-ephemeral-p) | |
273 (not (equal group nnweb-group))) | |
274 (nnweb-request-group group nil t)))) | |
275 | |
276 (defun nnweb-init (server) | |
277 "Initialize buffers and such." | |
278 (unless (gnus-buffer-live-p nnweb-buffer) | |
279 (setq nnweb-buffer | |
280 (save-excursion | |
281 (nnheader-set-temp-buffer | |
282 (format " *nnweb %s %s %s*" nnweb-type nnweb-search server)))))) | |
283 | |
284 (defun nnweb-fetch-url (url) | |
285 (save-excursion | |
286 (if (not nnheader-callback-function) | |
287 (let ((buf (current-buffer))) | |
288 (save-excursion | |
289 (set-buffer nnweb-buffer) | |
290 (erase-buffer) | |
291 (prog1 | |
292 (url-insert-file-contents url) | |
293 (copy-to-buffer buf (point-min) (point-max))))) | |
294 (nnweb-url-retrieve-asynch | |
295 url 'nnweb-callback (current-buffer) nnheader-callback-function) | |
296 t))) | |
297 | |
298 (defun nnweb-callback (buffer callback) | |
299 (when (gnus-buffer-live-p url-working-buffer) | |
300 (save-excursion | |
301 (set-buffer url-working-buffer) | |
302 (funcall (nnweb-definition 'article)) | |
303 (nnweb-decode-entities) | |
304 (set-buffer buffer) | |
305 (goto-char (point-max)) | |
306 (insert-buffer-substring url-working-buffer)) | |
307 (funcall callback t) | |
308 (gnus-kill-buffer url-working-buffer))) | |
309 | |
310 (defun nnweb-url-retrieve-asynch (url callback &rest data) | |
311 (let ((url-request-method "GET") | |
312 (old-asynch url-be-asynchronous) | |
313 (url-request-data nil) | |
314 (url-request-extra-headers nil) | |
315 (url-working-buffer (generate-new-buffer-name " *nnweb*"))) | |
316 (setq-default url-be-asynchronous t) | |
317 (save-excursion | |
318 (set-buffer (get-buffer-create url-working-buffer)) | |
319 (setq url-current-callback-data data | |
320 url-be-asynchronous t | |
321 url-current-callback-func callback) | |
322 (url-retrieve url)) | |
323 (setq-default url-be-asynchronous old-asynch))) | |
324 | |
325 (defun nnweb-encode-www-form-urlencoded (pairs) | |
326 "Return PAIRS encoded for forms." | |
327 (mapconcat | |
328 (function | |
329 (lambda (data) | |
330 (concat (w3-form-encode-xwfu (car data)) "=" | |
331 (w3-form-encode-xwfu (cdr data))))) | |
332 pairs "&")) | |
333 | |
334 (defun nnweb-fetch-form (url pairs) | |
335 (let ((url-request-data (nnweb-encode-www-form-urlencoded pairs)) | |
336 (url-request-method "POST") | |
337 (url-request-extra-headers | |
338 '(("Content-type" . "application/x-www-form-urlencoded")))) | |
339 (url-insert-file-contents url) | |
340 (setq buffer-file-name nil)) | |
341 t) | |
342 | |
343 (defun nnweb-decode-entities () | |
344 (goto-char (point-min)) | |
345 (while (re-search-forward "&\\([a-z]+\\);" nil t) | |
346 (replace-match (char-to-string (or (cdr (assq (intern (match-string 1)) | |
347 w3-html-entities )) | |
348 ?#)) | |
349 t t))) | |
350 | |
351 (defun nnweb-remove-markup () | |
352 (goto-char (point-min)) | |
353 (while (search-forward "<!--" nil t) | |
354 (delete-region (match-beginning 0) | |
355 (or (search-forward "-->" nil t) | |
356 (point-max)))) | |
357 (goto-char (point-min)) | |
358 (while (re-search-forward "<[^>]+>" nil t) | |
359 (replace-match "" t t))) | |
360 | |
361 ;;; | |
362 ;;; DejaNews functions. | |
363 ;;; | |
364 | |
365 (defun nnweb-dejanews-create-mapping () | |
366 "Perform the search and create an number-to-url alist." | |
367 (save-excursion | |
368 (set-buffer nnweb-buffer) | |
369 (erase-buffer) | |
370 (when (funcall (nnweb-definition 'search) nnweb-search) | |
371 (let ((i 0) | |
372 (more t) | |
373 (case-fold-search t) | |
374 (active (or (cadr (assoc nnweb-group nnweb-group-alist)) | |
375 (cons 1 0))) | |
376 Subject Score Date Newsgroup Author | |
377 map url) | |
378 (while more | |
379 ;; Go through all the article hits on this page. | |
380 (goto-char (point-min)) | |
381 (nnweb-decode-entities) | |
382 (goto-char (point-min)) | |
383 (while (re-search-forward "^ +[0-9]+\\." nil t) | |
384 (narrow-to-region | |
385 (point) | |
386 (cond ((re-search-forward "^ +[0-9]+\\." nil t) | |
387 (match-beginning 0)) | |
388 ((search-forward "\n\n" nil t) | |
389 (point)) | |
390 (t | |
391 (point-max)))) | |
392 (goto-char (point-min)) | |
393 (when (looking-at ".*HREF=\"\\([^\"]+\\)\"") | |
394 (setq url (match-string 1))) | |
395 (nnweb-remove-markup) | |
396 (goto-char (point-min)) | |
397 (while (search-forward "\t" nil t) | |
398 (replace-match " ")) | |
399 (goto-char (point-min)) | |
400 (while (re-search-forward "^ +\\([^:]+\\): +\\(.*\\)$" nil t) | |
401 (set (intern (match-string 1)) (match-string 2))) | |
402 (widen) | |
403 (when (string-match "#[0-9]+/[0-9]+ *$" Subject) | |
404 (setq Subject (substring Subject 0 (match-beginning 0)))) | |
405 (incf i) | |
406 (unless (nnweb-get-hashtb url) | |
407 (push | |
408 (list | |
409 (incf (cdr active)) | |
410 (make-full-mail-header | |
411 (cdr active) (concat "(" Newsgroup ") " Subject) Author Date | |
412 (concat "<" (nnweb-identifier url) "@dejanews>") | |
413 nil 0 (string-to-int Score) url)) | |
414 map) | |
415 (nnweb-set-hashtb (cadar map) (car map)))) | |
416 ;; See whether there is a "Get next 20 hits" button here. | |
417 (if (or (not (re-search-forward | |
418 "HREF=\"\\([^\"]+\\)\">Get next" nil t)) | |
419 (>= i nnweb-max-hits)) | |
420 (setq more nil) | |
421 ;; Yup -- fetch it. | |
422 (setq more (match-string 1)) | |
423 (erase-buffer) | |
424 (url-insert-file-contents more))) | |
425 ;; Return the articles in the right order. | |
426 (setq nnweb-articles | |
427 (sort (nconc nnweb-articles map) | |
428 (lambda (s1 s2) (< (car s1) (car s2))))))))) | |
429 | |
430 (defun nnweb-dejanews-wash-article () | |
431 (let ((case-fold-search t)) | |
432 (goto-char (point-min)) | |
433 (re-search-forward "<PRE>" nil t) | |
434 (delete-region (point-min) (point)) | |
435 (re-search-forward "</PRE>" nil t) | |
436 (delete-region (point) (point-max)) | |
437 (nnweb-remove-markup) | |
438 (goto-char (point-min)) | |
439 (while (and (looking-at " *$") | |
440 (not (eobp))) | |
441 (gnus-delete-line)) | |
442 (while (looking-at "\\(^[^ ]+:\\) *") | |
443 (replace-match "\\1 " t) | |
444 (forward-line 1)) | |
445 (when (re-search-forward "\n\n+" nil t) | |
446 (replace-match "\n" t t)))) | |
447 | |
448 (defun nnweb-dejanews-search (search) | |
449 (nnweb-fetch-form | |
450 (nnweb-definition 'address) | |
451 `(("query" . ,search) | |
452 ("defaultOp" . "AND") | |
453 ("svcclass" . "dncurrent") | |
454 ("maxhits" . "100") | |
455 ("format" . "verbose") | |
456 ("threaded" . "0") | |
457 ("showsort" . "score") | |
458 ("agesign" . "1") | |
459 ("ageweight" . "1"))) | |
460 t) | |
461 | |
462 (defun nnweb-dejanews-identity (url) | |
463 "Return an unique identifier based on URL." | |
464 (if (string-match "recnum=\\([0-9]+\\)" url) | |
465 (match-string 1 url) | |
466 url)) | |
467 | |
468 ;;; | |
469 ;;; InReference | |
470 ;;; | |
471 | |
472 (defun nnweb-reference-create-mapping () | |
473 "Perform the search and create an number-to-url alist." | |
474 (save-excursion | |
475 (set-buffer nnweb-buffer) | |
476 (erase-buffer) | |
477 (when (funcall (nnweb-definition 'search) nnweb-search) | |
478 (let ((i 0) | |
479 (more t) | |
480 (case-fold-search t) | |
481 (active (or (cadr (assoc nnweb-group nnweb-group-alist)) | |
482 (cons 1 0))) | |
483 Subject Score Date Newsgroups From Message-ID | |
484 map url) | |
485 (while more | |
486 ;; Go through all the article hits on this page. | |
487 (goto-char (point-min)) | |
488 (search-forward "</pre><hr>" nil t) | |
489 (delete-region (point-min) (point)) | |
490 ;(nnweb-decode-entities) | |
491 (goto-char (point-min)) | |
492 (while (re-search-forward "^ +[0-9]+\\." nil t) | |
493 (narrow-to-region | |
494 (point) | |
495 (if (re-search-forward "^$" nil t) | |
496 (match-beginning 0) | |
497 (point-max))) | |
498 (goto-char (point-min)) | |
499 (when (looking-at ".*href=\"\\([^\"]+\\)\"") | |
500 (setq url (match-string 1))) | |
501 (nnweb-remove-markup) | |
502 (goto-char (point-min)) | |
503 (while (search-forward "\t" nil t) | |
504 (replace-match " ")) | |
505 (goto-char (point-min)) | |
506 (while (re-search-forward "^\\([^:]+\\): \\(.*\\)$" nil t) | |
507 (set (intern (match-string 1)) (match-string 2))) | |
508 (widen) | |
509 (search-forward "</pre>" nil t) | |
510 (incf i) | |
511 (unless (nnweb-get-hashtb url) | |
512 (push | |
513 (list | |
514 (incf (cdr active)) | |
515 (make-full-mail-header | |
516 (cdr active) (concat "(" Newsgroups ") " Subject) From Date | |
517 Message-ID | |
518 nil 0 (string-to-int Score) url)) | |
519 map) | |
520 (nnweb-set-hashtb (cadar map) (car map)))) | |
521 (setq more nil)) | |
522 ;; Return the articles in the right order. | |
523 (setq nnweb-articles | |
524 (sort (nconc nnweb-articles map) | |
525 (lambda (s1 s2) (< (car s1) (car s2))))))))) | |
526 | |
527 (defun nnweb-reference-wash-article () | |
528 (let ((case-fold-search t)) | |
529 (goto-char (point-min)) | |
530 (re-search-forward "^</center><hr>" nil t) | |
531 (delete-region (point-min) (point)) | |
532 (search-forward "<pre>" nil t) | |
533 (forward-line -1) | |
534 (let ((body (point-marker))) | |
535 (search-forward "</pre>" nil t) | |
536 (delete-region (point) (point-max)) | |
537 (nnweb-remove-markup) | |
538 (goto-char (point-min)) | |
539 (while (looking-at " *$") | |
540 (gnus-delete-line)) | |
541 (narrow-to-region (point-min) body) | |
542 (while (and (re-search-forward "^$" nil t) | |
543 (not (eobp))) | |
544 (gnus-delete-line)) | |
545 (goto-char (point-min)) | |
546 (while (looking-at "\\(^[^ ]+:\\) *") | |
547 (replace-match "\\1 " t) | |
548 (forward-line 1)) | |
549 (goto-char (point-min)) | |
550 (when (re-search-forward "^References:" nil t) | |
551 (narrow-to-region | |
552 (point) (if (re-search-forward "^$\\|^[^:]+:" nil t) | |
553 (match-beginning 0) | |
554 (point-max))) | |
555 (goto-char (point-min)) | |
556 (while (not (eobp)) | |
557 (unless (looking-at "References") | |
558 (insert "\t") | |
559 (forward-line 1))) | |
560 (goto-char (point-min)) | |
561 (while (search-forward "," nil t) | |
562 (replace-match " " t t))) | |
563 (widen) | |
564 (set-marker body nil)))) | |
565 | |
566 (defun nnweb-reference-search (search) | |
567 (prog1 | |
568 (url-insert-file-contents | |
569 (concat | |
570 (nnweb-definition 'address) | |
571 "?" | |
572 (nnweb-encode-www-form-urlencoded | |
573 `(("search" . "advanced") | |
574 ("querytext" . ,search) | |
575 ("subj" . "") | |
576 ("name" . "") | |
577 ("login" . "") | |
578 ("host" . "") | |
579 ("organization" . "") | |
580 ("groups" . "") | |
581 ("keywords" . "") | |
582 ("choice" . "Search") | |
583 ("startmonth" . "Jul") | |
584 ("startday" . "25") | |
585 ("startyear" . "1996") | |
586 ("endmonth" . "Aug") | |
587 ("endday" . "24") | |
588 ("endyear" . "1996") | |
589 ("mode" . "Quick") | |
590 ("verbosity" . "Verbose") | |
591 ("ranking" . "Relevance") | |
592 ("first" . "1") | |
593 ("last" . "25") | |
594 ("score" . "50"))))) | |
595 (setq buffer-file-name nil)) | |
596 t) | |
597 | |
598 ;;; | |
599 ;;; Alta Vista | |
600 ;;; | |
601 | |
602 (defun nnweb-altavista-create-mapping () | |
603 "Perform the search and create an number-to-url alist." | |
604 (save-excursion | |
605 (set-buffer nnweb-buffer) | |
606 (erase-buffer) | |
607 (let ((part 0)) | |
608 (when (funcall (nnweb-definition 'search) nnweb-search part) | |
609 (let ((i 0) | |
610 (more t) | |
611 (case-fold-search t) | |
612 (active (or (cadr (assoc nnweb-group nnweb-group-alist)) | |
613 (cons 1 0))) | |
614 subject date from id group | |
615 map url) | |
616 (while more | |
617 ;; Go through all the article hits on this page. | |
618 (goto-char (point-min)) | |
619 (search-forward "<dt>" nil t) | |
620 (delete-region (point-min) (match-beginning 0)) | |
621 (goto-char (point-min)) | |
622 (while (search-forward "<dt>" nil t) | |
623 (replace-match "\n<blubb>")) | |
624 (nnweb-decode-entities) | |
625 (goto-char (point-min)) | |
626 (while (re-search-forward "<blubb>.*href=\"\\([^\"]+\\)\"><strong>\\([^>]*\\)</strong></a><dd>\\([^-]+\\)- <b>\\([^<]+\\)<.*href=\"news:\\([^\"]+\\)\">.*\">\\(.+\\)</a><P>" | |
627 nil t) | |
628 (setq url (match-string 1) | |
629 subject (match-string 2) | |
630 date (match-string 3) | |
631 group (match-string 4) | |
632 id (concat "<" (match-string 5) ">") | |
633 from (match-string 6)) | |
634 (incf i) | |
635 (unless (nnweb-get-hashtb url) | |
636 (push | |
637 (list | |
638 (incf (cdr active)) | |
639 (make-full-mail-header | |
640 (cdr active) (concat "(" group ") " subject) from date | |
641 id nil 0 0 url)) | |
642 map) | |
643 (nnweb-set-hashtb (cadar map) (car map)))) | |
644 ;; See if we want more. | |
645 (when (or (not nnweb-articles) | |
646 (>= i nnweb-max-hits) | |
647 (not (funcall (nnweb-definition 'search) | |
648 nnweb-search (incf part)))) | |
649 (setq more nil))) | |
650 ;; Return the articles in the right order. | |
651 (setq nnweb-articles | |
652 (sort (nconc nnweb-articles map) | |
653 (lambda (s1 s2) (< (car s1) (car s2)))))))))) | |
654 | |
655 (defun nnweb-altavista-wash-article () | |
656 (goto-char (point-min)) | |
657 (let ((case-fold-search t)) | |
658 (when (re-search-forward "^<strong>" nil t) | |
659 (delete-region (point-min) (match-beginning 0))) | |
660 (goto-char (point-min)) | |
661 (while (looking-at "<strong>\\([^ ]+\\) +</strong> +\\(.*\\)$") | |
662 (replace-match "\\1: \\2" t) | |
663 (forward-line 1)) | |
664 (when (re-search-backward "^References:" nil t) | |
665 (narrow-to-region (point) (progn (forward-line 1) (point))) | |
666 (goto-char (point-min)) | |
667 (while (re-search-forward "<A.*\\?id@\\([^\"]+\\)\">[0-9]+</A>" nil t) | |
668 (replace-match "<\\1> " t))) | |
669 (widen) | |
670 (nnweb-remove-markup))) | |
671 | |
672 (defun nnweb-altavista-search (search &optional part) | |
673 (prog1 | |
674 (url-insert-file-contents | |
675 (concat | |
676 (nnweb-definition 'address) | |
677 "?" | |
678 (nnweb-encode-www-form-urlencoded | |
679 `(("pg" . "aq") | |
680 ("what" . "news") | |
681 ,@(when part `(("stq" . ,(int-to-string (* part 30))))) | |
682 ("fmt" . "d") | |
683 ("q" . ,search) | |
684 ("r" . "") | |
685 ("d0" . "") | |
686 ("d1" . ""))))) | |
687 (setq buffer-file-name nil))) | |
688 | |
689 (provide 'nnweb) | |
690 | |
691 ;;; nnweb.el ends here |