1 ;;; nnir.el --- search mail with various search engines -*- coding: iso-8859-1 -*-
3 ;; Copyright (C) 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005, 2006,
4 ;; 2007, 2008, 2009, 2010 Free Software Foundation, Inc.
6 ;; Author: Kai Großjohann <grossjohann@ls6.cs.uni-dortmund.de>
7 ;; Swish-e and Swish++ backends by:
8 ;; Christoph Conrad <christoph.conrad@gmx.de>.
9 ;; IMAP backend by: Simon Josefsson <jas@pdc.kth.se>.
10 ;; IMAP search by: Torsten Hilbrich <torsten.hilbrich <at> gmx.net>
11 ;; IMAP search improved by Daniel Pittman <daniel@rimspace.net>.
12 ;; nnmaildir support for Swish++ and Namazu backends by:
13 ;; Justus Piater <Justus <at> Piater.name>
14 ;; Keywords: news mail searching ir
16 ;; This file is part of GNU Emacs.
18 ;; GNU Emacs is free software: you can redistribute it and/or modify
19 ;; it under the terms of the GNU General Public License as published by
20 ;; the Free Software Foundation, either version 3 of the License, or
21 ;; (at your option) any later version.
23 ;; GNU Emacs is distributed in the hope that it will be useful,
24 ;; but WITHOUT ANY WARRANTY; without even the implied warranty of
25 ;; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
26 ;; GNU General Public License for more details.
28 ;; You should have received a copy of the GNU General Public License
29 ;; along with GNU Emacs. If not, see <http://www.gnu.org/licenses/>.
33 ;; TODO: Documentation in the Gnus manual
36 ;; Subject: Re: Including nnir.el
37 ;; Newsgroups: gmane.emacs.gnus.general
38 ;; Message-ID: <v9d5dnp6aq.fsf@marauder.physik.uni-ulm.de>
39 ;; Date: 2006-06-05 22:49:01 GMT
41 ;; On Sun, Jun 04 2006, Sascha Wilde wrote:
43 ;; > The one thing most hackers like to forget: Documentation. By now the
44 ;; > documentation is only in the comments at the head of the source, I
45 ;; > would use it as basis to cook up some minimal texinfo docs.
47 ;; > Where in the existing gnus manual would this fit best?
49 ;; Maybe (info "(gnus)Combined Groups") for a general description.
50 ;; `gnus-group-make-nnir-group' might be described in (info
51 ;; "(gnus)Foreign Groups") as well.
54 ;; The most recent version of this can always be fetched from the Gnus
55 ;; repository. See http://www.gnus.org/ for more information.
57 ;; This code is still in the development stage but I'd like other
58 ;; people to have a look at it. Please do not hesitate to contact me
61 ;; What does it do? Well, it allows you to index your mail using some
62 ;; search engine (freeWAIS-sf, swish-e and others -- see later),
63 ;; then type `G G' in the Group buffer and issue a query to the search
64 ;; engine. You will then get a buffer which shows all articles
65 ;; matching the query, sorted by Retrieval Status Value (score).
67 ;; When looking at the retrieval result (in the Summary buffer) you
68 ;; can type `G T' (aka M-x gnus-summary-nnir-goto-thread RET) on an
69 ;; article. You will be teleported into the group this article came
70 ;; from, showing the thread this article is part of. (See below for
73 ;; The Lisp installation is simple: just put this file on your
74 ;; load-path, byte-compile it, and load it from ~/.gnus or something.
75 ;; This will install a new command `G G' in your Group buffer for
76 ;; searching your mail. Note that you also need to configure a number
77 ;; of variables, as described below.
81 ;; * If you don't use HyREX as your search engine, this expects that
82 ;; you use nnml or another one-file-per-message backend, because the
83 ;; others doesn't support nnfolder.
84 ;; * It can only search the mail backend's which are supported by one
85 ;; search engine, because of different query languages.
86 ;; * There are restrictions to the Wais setup.
87 ;; * There are restrictions to the imap setup.
88 ;; * gnus-summary-nnir-goto-thread: Fetches whole group first, before
89 ;; limiting to the right articles. This is much too slow, of
90 ;; course. May issue a query for number of articles to fetch; you
91 ;; must accept the default of all articles at this point or things
94 ;; The Lisp setup involves setting a few variables and setting up the
95 ;; search engine. You can define the variables in the server definition
97 ;; (setq gnus-secondary-select-methods '(
98 ;; (nnimap "" (nnimap-address "localhost")
99 ;; (nnir-search-engine hyrex)
100 ;; (nnir-hyrex-additional-switches ("-d" "ddl-nnimap.xml"))
102 ;; Or you can define the global ones. The variables set in the mailer-
103 ;; definition will be used first.
104 ;; The variable to set is `nnir-search-engine'. Choose one of the engines
105 ;; listed in `nnir-engines'. (Actually `nnir-engines' is an alist,
106 ;; type `C-h v nnir-engines RET' for more information; this includes
107 ;; examples for setting `nnir-search-engine', too.)
109 ;; The variable nnir-mail-backend isn't used anymore.
112 ;; You must also set up a search engine. I'll tell you about the two
113 ;; search engines currently supported:
117 ;; As always with freeWAIS-sf, you need a so-called `format file'. I
118 ;; use the following file:
121 ;; | # Kai's format file for freeWAIS-sf for indexing mails.
122 ;; | # Each mail is in a file, much like the MH format.
124 ;; | # Document separator should never match -- each file is a document.
125 ;; | record-sep: /^@this regex should never match@$/
127 ;; | # Searchable fields specification.
129 ;; | region: /^[sS]ubject:/ /^[sS]ubject: */
130 ;; | subject "Subject header" stemming TEXT BOTH
133 ;; | region: /^([tT][oO]|[cC][cC]):/ /^([tT][oO]|[cC][cC]): */
134 ;; | to "To and Cc headers" SOUNDEX BOTH
137 ;; | region: /^[fF][rR][oO][mM]:/ /^[fF][rR][oO][mM]: */
138 ;; | from "From header" SOUNDEX BOTH
142 ;; | stemming TEXT GLOBAL
143 ;; | end: /^@this regex should never match@$/
146 ;; 1998-07-22: waisindex would dump core on me for large articles with
147 ;; the above settings. I used /^$/ as the end regex for the global
148 ;; field. That seemed to work okay.
150 ;; There is a Perl module called `WAIS.pm' which is available from
151 ;; CPAN as well as ls6-ftp.cs.uni-dortmund.de:/pub/wais/Perl. This
152 ;; module comes with a nifty tool called `makedb', which I use for
153 ;; indexing. Here's my `makedb.conf':
156 ;; | # Config file for makedb
158 ;; | # Global options
159 ;; | waisindex = /usr/local/bin/waisindex
160 ;; | wais_opt = -stem -t fields
161 ;; | # `-stem' option necessary when `stemming' is specified for the
162 ;; | # global field in the *.fmt file
165 ;; | homedir = /home/kai
167 ;; | # The mail database.
169 ;; | files = `find $homedir/Mail -name \*[0-9] -print`
170 ;; | dbdir = $homedir/.wais
174 ;; The Lisp setup involves the `nnir-wais-*' variables. The most
175 ;; difficult to understand variable is probably
176 ;; `nnir-wais-remove-prefix'. Here's what it does: the output of
177 ;; `waissearch' basically contains the file name and the (full)
178 ;; directory name. As Gnus works with group names rather than
179 ;; directory names, the directory name is transformed into a group
180 ;; name as follows: first, a prefix is removed from the (full)
181 ;; directory name, then all `/' are replaced with `.'. The variable
182 ;; `nnir-wais-remove-prefix' should contain a regex matching exactly
183 ;; this prefix. It defaults to `$HOME/Mail/' (note the trailing
188 ;; The Namazu backend requires you to have one directory containing all
189 ;; index files, this is controlled by the `nnir-namazu-index-directory'
190 ;; variable. To function the `nnir-namazu-remove-prefix' variable must
191 ;; also be correct, see the documentation for `nnir-wais-remove-prefix'
194 ;; It is particularly important not to pass any any switches to namazu
195 ;; that will change the output format. Good switches to use include
196 ;; `--sort', `--ascending', `--early' and `--late'. Refer to the Namazu
197 ;; documentation for further information on valid switches.
199 ;; To index my mail with the `mknmz' program I use the following
200 ;; configuration file:
203 ;; | package conf; # Don't remove this line!
205 ;; | # Paths which will not be indexed. Don't use `^' or `$' anchors.
206 ;; | $EXCLUDE_PATH = "spam|sent";
208 ;; | # Header fields which should be searchable. case-insensitive
209 ;; | $REMAIN_HEADER = "from|date|message-id|subject";
211 ;; | # Searchable fields. case-insensitive
212 ;; | $SEARCH_FIELD = "from|date|message-id|subject";
214 ;; | # The max length of a word.
215 ;; | $WORD_LENG_MAX = 128;
217 ;; | # The max length of a field.
218 ;; | $MAX_FIELD_LENGTH = 256;
221 ;; My mail is stored in the directories ~/Mail/mail/, ~/Mail/lists/ and
222 ;; ~/Mail/archive/, so to index them I go to the directory set in
223 ;; `nnir-namazu-index-directory' and issue the following command.
225 ;; mknmz --mailnews ~/Mail/archive/ ~/Mail/mail/ ~/Mail/lists/
227 ;; For maximum searching efficiency I have a cron job set to run this
228 ;; command every four hours.
232 ;; The HyREX backend requires you to have one directory from where all
233 ;; your relative paths are to, if you use them. This directory must be
234 ;; set in the `nnir-hyrex-index-directory' variable, which defaults to
235 ;; your home directory. You must also pass the base, class and
236 ;; directory options or simply your dll to the `nnir-hyrex-programm' by
237 ;; setting the `nnir-hyrex-additional-switches' variable accordently.
238 ;; To function the `nnir-hyrex-remove-prefix' variable must also be
239 ;; correct, see the documentation for `nnir-wais-remove-prefix' above.
243 ;; The find-grep engine simply runs find(1) to locate eligible
244 ;; articles and searches them with grep(1). This, of course, is much
245 ;; slower than using a proper search engine but OTOH doesn't require
246 ;; maintenance of an index and is still faster than using any built-in
247 ;; means for searching. The method specification of the server to
248 ;; search must include a directory for this engine to work (E.g.,
249 ;; `nnml-directory'). The tools must be POSIX compliant. GNU Find
250 ;; prior to version 4.2.12 (4.2.26 on Linux due to incorrect ARG_MAX
251 ;; handling) does not work.
253 ;; | ;; find-grep configuration for searching the Gnus Cache
256 ;; | (nnml-get-new-mail nil)
257 ;; | (nnir-search-engine find-grep)
258 ;; | (nnml-directory "~/News/cache/")
259 ;; | (nnml-active-file "~/News/cache/active"))
262 ;; Developer information:
264 ;; I have tried to make the code expandable. Basically, it is divided
265 ;; into two layers. The upper layer is somewhat like the `nnvirtual'
266 ;; backend: given a specification of what articles to show from
267 ;; another backend, it creates a group containing exactly those
268 ;; articles. The lower layer issues a query to a search engine and
269 ;; produces such a specification of what articles to show from the
272 ;; The interface between the two layers consists of the single
273 ;; function `nnir-run-query', which just selects the appropriate
274 ;; function for the search engine one is using. The input to
275 ;; `nnir-run-query' is a string, representing the query as input by
276 ;; the user. The output of `nnir-run-query' is supposed to be a
277 ;; vector, each element of which should in turn be a three-element
278 ;; vector. The first element should be full group name of the article,
279 ;; the second element should be the article number, and the third
280 ;; element should be the Retrieval Status Value (RSV) as returned from
281 ;; the search engine. An RSV is the score assigned to the document by
282 ;; the search engine. For Boolean search engines, the
283 ;; RSV is always 1000 (or 1 or 100, or whatever you like).
285 ;; The sorting order of the articles in the summary buffer created by
286 ;; nnir is based on the order of the articles in the above mentioned
287 ;; vector, so that's where you can do the sorting you'd like. Maybe
288 ;; it would be nice to have a way of displaying the search result
289 ;; sorted differently?
291 ;; So what do you need to do when you want to add another search
292 ;; engine? You write a function that executes the query. Temporary
293 ;; data from the search engine can be put in `nnir-tmp-buffer'. This
294 ;; function should return the list of articles as a vector, as
295 ;; described above. Then, you need to register this backend in
296 ;; `nnir-engines'. Then, users can choose the backend by setting
297 ;; `nnir-search-engine'.
299 ;; Todo, or future ideas:
301 ;; * It should be possible to restrict search to certain groups.
303 ;; * There is currently no error checking.
305 ;; * The summary buffer display is currently really ugly, with all the
306 ;; added information in the subjects. How could I make this
309 ;; * A function which can be called from an nnir summary buffer which
310 ;; teleports you into the group the current article came from and
311 ;; shows you the whole thread this article is part of.
312 ;; Implementation suggestions?
313 ;; (1998-07-24: There is now a preliminary implementation, but
314 ;; it is much too slow and quite fragile.)
316 ;; * Support other mail backends. In particular, probably quite a few
317 ;; people use nnfolder. How would one go about searching nnfolders
318 ;; and producing the right data needed? The group name and the RSV
319 ;; are simple, but what about the article number?
320 ;; - The article number is encoded in the `X-Gnus-Article-Number'
321 ;; header of each mail.
322 ;; - The HyREX engine supports nnfolder.
324 ;; * Support compressed mail files. Probably, just stripping off the
325 ;; `.gz' or `.Z' file name extension is sufficient.
327 ;; * At least for imap, the query is performed twice.
330 ;; Have you got other ideas?
335 (require 'gnus-group)
344 (autoload 'nnimap-buffer "nnimap")
345 (autoload 'nnimap-command "nnimap")
346 (autoload 'nnimap-possibly-change-group "nnimap"))
349 (nnoo-define-basics nnir)
351 (gnus-declare-backend "nnir" 'mail)
353 (defvar nnir-imap-default-search-key "Whole message"
354 "The default IMAP search key for an nnir search. Must be one of
355 the keys in nnir-imap-search-arguments. To use raw imap queries
356 by default set this to \"Imap\"")
358 (defvar nnir-imap-search-arguments
359 '(("Whole message" . "TEXT")
360 ("Subject" . "SUBJECT")
364 "Mapping from user readable keys to IMAP search items for use in nnir")
366 (defvar nnir-imap-search-other "HEADER %S"
367 "The IMAP search item to use for anything other than
368 nnir-imap-search-arguments. By default this is the name of an
371 (defvar nnir-imap-search-argument-history ()
372 "The history for querying search options in nnir")
374 (defvar nnir-get-article-nov-override-function nil
375 "If non-nil, a function that will be passed each search result. This
376 should return a message's headers in NOV format.
378 If this variable is nil, or if the provided function returns nil for a search
379 result, `gnus-retrieve-headers' will be called instead.")
381 (defvar nnir-method-default-engines
384 "Alist of default search engines by server method")
386 ;;; Developer Extension Variable:
389 `((wais nnir-run-waissearch
394 ,(mapcar 'car nnir-imap-search-arguments) ; alist for completing
395 nil ; allow any user input
397 nnir-imap-search-argument-history ; the history to use
398 ,nnir-imap-default-search-key ; default
400 (swish++ nnir-run-swish++
401 ((group . "Group spec: ")))
402 (swish-e nnir-run-swish-e
403 ((group . "Group spec: ")))
404 (namazu nnir-run-namazu
406 (hyrex nnir-run-hyrex
407 ((group . "Group spec: ")))
408 (find-grep nnir-run-find-grep
409 ((grep-options . "Grep options: "))))
410 "Alist of supported search engines.
411 Each element in the alist is a three-element list (ENGINE FUNCTION ARGS).
412 ENGINE is a symbol designating the searching engine. FUNCTION is also
413 a symbol, giving the function that does the search. The third element
414 ARGS is a list of cons pairs (PARAM . PROMPT). When issuing a query,
415 the FUNCTION will issue a query for each of the PARAMs, using PROMPT.
417 The value of `nnir-search-engine' must be one of the ENGINE symbols.
418 For example, use the following line for searching using freeWAIS-sf:
419 (setq nnir-search-engine 'wais)
420 Use the following line if you read your mail via IMAP and your IMAP
421 server supports searching:
422 (setq nnir-search-engine 'imap)
423 Note that you have to set additional variables for most backends. For
424 example, the `wais' backend needs the variables `nnir-wais-program',
425 `nnir-wais-database' and `nnir-wais-remove-prefix'.
427 Add an entry here when adding a new search engine.")
429 ;;; User Customizable Variables:
432 "Search nnmh and nnml groups in Gnus with swish-e, freeWAIS-sf, or EWS."
438 ;; If `nil', use server parameters to find out which server to search. CCC
440 (defcustom nnir-mail-backend '(nnml "")
441 "*Specifies which backend should be searched.
442 More precisely, this is used to determine from which backend to fetch the
445 This must be equal to an existing server, so maybe it is best to use
446 something like the following:
447 (setq nnir-mail-backend (nth 0 gnus-secondary-select-methods))
448 The above line works fine if the mail backend you want to search is
449 the first element of gnus-secondary-select-methods (`nth' starts counting
454 ;; Search engine to use.
456 (defcustom nnir-search-engine 'wais
457 "*The search engine to use. Must be a symbol.
458 See `nnir-engines' for a list of supported engines, and for example
459 settings of `nnir-search-engine'."
465 (defcustom nnir-wais-program "waissearch"
466 "*Name of waissearch executable."
470 (defcustom nnir-wais-database (expand-file-name "~/.wais/mail")
471 "*Name of Wais database containing the mail.
473 Note that this should be a file name without extension. For example,
474 if you have a file /home/john/.wais/mail.fmt, use this:
475 (setq nnir-wais-database \"/home/john/.wais/mail\")
476 The string given here is passed to `waissearch -d' as-is."
480 (defcustom nnir-wais-remove-prefix (concat (getenv "HOME") "/Mail/")
481 "*The prefix to remove from each directory name returned by waissearch
482 in order to get a group name (albeit with / instead of .). This is a
485 For example, suppose that Wais returns file names such as
486 \"/home/john/Mail/mail/misc/42\". For this example, use the following
487 setting: (setq nnir-wais-remove-prefix \"/home/john/Mail/\")
488 Note the trailing slash. Removing this prefix gives \"mail/misc/42\".
489 `nnir' knows to remove the \"/42\" and to replace \"/\" with \".\" to
490 arrive at the correct group name, \"mail.misc\"."
494 (defcustom nnir-swish++-configuration-file
495 (expand-file-name "~/Mail/swish++.conf")
496 "*Configuration file for swish++."
500 (defcustom nnir-swish++-program "search"
501 "*Name of swish++ search executable."
505 (defcustom nnir-swish++-additional-switches '()
506 "*A list of strings, to be given as additional arguments to swish++.
508 Note that this should be a list. Ie, do NOT use the following:
509 (setq nnir-swish++-additional-switches \"-i -w\") ; wrong
511 (setq nnir-swish++-additional-switches '(\"-i\" \"-w\"))"
512 :type '(repeat (string))
515 (defcustom nnir-swish++-remove-prefix (concat (getenv "HOME") "/Mail/")
516 "*The prefix to remove from each file name returned by swish++
517 in order to get a group name (albeit with / instead of .). This is a
520 This variable is very similar to `nnir-wais-remove-prefix', except
521 that it is for swish++, not Wais."
526 ;; URL: http://swish-e.org/
527 ;; Variables `nnir-swish-e-index-file', `nnir-swish-e-program' and
528 ;; `nnir-swish-e-additional-switches'
530 (make-obsolete-variable 'nnir-swish-e-index-file
531 'nnir-swish-e-index-files "Emacs 23.1")
532 (defcustom nnir-swish-e-index-file
533 (expand-file-name "~/Mail/index.swish-e")
534 "*Index file for swish-e.
535 This could be a server parameter.
536 It is never consulted once `nnir-swish-e-index-files', which should be
537 used instead, has been customized."
541 (defcustom nnir-swish-e-index-files
542 (list nnir-swish-e-index-file)
543 "*List of index files for swish-e.
544 This could be a server parameter."
545 :type '(repeat (file))
548 (defcustom nnir-swish-e-program "swish-e"
549 "*Name of swish-e search executable.
550 This cannot be a server parameter."
554 (defcustom nnir-swish-e-additional-switches '()
555 "*A list of strings, to be given as additional arguments to swish-e.
557 Note that this should be a list. Ie, do NOT use the following:
558 (setq nnir-swish-e-additional-switches \"-i -w\") ; wrong
560 (setq nnir-swish-e-additional-switches '(\"-i\" \"-w\"))
562 This could be a server parameter."
563 :type '(repeat (string))
566 (defcustom nnir-swish-e-remove-prefix (concat (getenv "HOME") "/Mail/")
567 "*The prefix to remove from each file name returned by swish-e
568 in order to get a group name (albeit with / instead of .). This is a
571 This variable is very similar to `nnir-wais-remove-prefix', except
572 that it is for swish-e, not Wais.
574 This could be a server parameter."
578 ;; HyREX engine, see <URL:http://ls6-www.cs.uni-dortmund.de/>
580 (defcustom nnir-hyrex-program "nnir-search"
581 "*Name of the nnir-search executable."
585 (defcustom nnir-hyrex-additional-switches '()
586 "*A list of strings, to be given as additional arguments for nnir-search.
587 Note that this should be a list. Ie, do NOT use the following:
588 (setq nnir-hyrex-additional-switches \"-ddl ddl.xml -c nnir\") ; wrong !
590 (setq nnir-hyrex-additional-switches '(\"-ddl\" \"ddl.xml\" \"-c\" \"nnir\"))"
591 :type '(repeat (string))
594 (defcustom nnir-hyrex-index-directory (getenv "HOME")
595 "*Index directory for HyREX."
599 (defcustom nnir-hyrex-remove-prefix (concat (getenv "HOME") "/Mail/")
600 "*The prefix to remove from each file name returned by HyREX
601 in order to get a group name (albeit with / instead of .).
603 For example, suppose that HyREX returns file names such as
604 \"/home/john/Mail/mail/misc/42\". For this example, use the following
605 setting: (setq nnir-hyrex-remove-prefix \"/home/john/Mail/\")
606 Note the trailing slash. Removing this prefix gives \"mail/misc/42\".
607 `nnir' knows to remove the \"/42\" and to replace \"/\" with \".\" to
608 arrive at the correct group name, \"mail.misc\"."
612 ;; Namazu engine, see <URL:http://www.namazu.org/>
614 (defcustom nnir-namazu-program "namazu"
615 "*Name of Namazu search executable."
619 (defcustom nnir-namazu-index-directory (expand-file-name "~/Mail/namazu/")
620 "*Index directory for Namazu."
624 (defcustom nnir-namazu-additional-switches '()
625 "*A list of strings, to be given as additional arguments to namazu.
626 The switches `-q', `-a', and `-s' are always used, very few other switches
627 make any sense in this context.
629 Note that this should be a list. Ie, do NOT use the following:
630 (setq nnir-namazu-additional-switches \"-i -w\") ; wrong
632 (setq nnir-namazu-additional-switches '(\"-i\" \"-w\"))"
633 :type '(repeat (string))
636 (defcustom nnir-namazu-remove-prefix (concat (getenv "HOME") "/Mail/")
637 "*The prefix to remove from each file name returned by Namazu
638 in order to get a group name (albeit with / instead of .).
640 This variable is very similar to `nnir-wais-remove-prefix', except
641 that it is for Namazu, not Wais."
645 ;;; Internal Variables:
647 (defvar nnir-current-query nil
648 "Internal: stores current query (= group name).")
650 (defvar nnir-current-server nil
651 "Internal: stores current server (does it ever change?).")
653 (defvar nnir-current-group-marked nil
654 "Internal: stores current list of process-marked groups.")
656 (defvar nnir-artlist nil
657 "Internal: stores search result.")
659 (defvar nnir-tmp-buffer " *nnir*"
660 "Internal: temporary buffer.")
666 (defun gnus-group-make-nnir-group (extra-parms query)
667 "Create an nnir group. Asks for query."
668 (interactive "P\nsQuery: ")
669 (setq nnir-current-query nil
670 nnir-current-server nil
671 nnir-current-group-marked nil
675 (setq parms (nnir-read-parms query))
676 (setq parms (list (cons 'query query))))
677 (add-to-list 'parms (cons 'unique-id (message-unique-id)) t)
678 (gnus-group-read-ephemeral-group
679 (concat "nnir:" (prin1-to-string parms)) '(nnir "") t
680 (cons (current-buffer)
681 gnus-current-window-configuration)
684 ;; Why is this needed? Is this for compatibility with old/new gnusae? Using
685 ;; gnus-group-server instead works for me. -- Justus Piater
686 (defmacro nnir-group-server (group)
687 "Return the server for a newsgroup GROUP.
688 The returned format is as `gnus-server-to-method' needs it. See
689 `gnus-group-real-prefix' and `gnus-group-real-name'."
690 `(let ((gname ,group))
691 (if (string-match "^\\([^:]+\\):" gname)
693 (setq gname (match-string 1 gname))
694 (if (string-match "^\\([^+]+\\)\\+\\(.+\\)$" gname)
695 (format "%s:%s" (match-string 1 gname) (match-string 2 gname))
697 (format "%s:%s" (car gnus-select-method) (cadr gnus-select-method)))))
699 ;; Summary mode commands.
701 (defun gnus-summary-nnir-goto-thread ()
702 "Only applies to nnir groups. Go to group this article came from
703 and show thread that contains this article."
705 (unless (eq 'nnir (car (gnus-find-method-for-group gnus-newsgroup-name)))
706 (error "Can't execute this command unless in nnir group"))
707 (let* ((cur (gnus-summary-article-number))
708 (group (nnir-artlist-artitem-group nnir-artlist cur))
709 (backend-number (nnir-artlist-artitem-number nnir-artlist cur))
710 (id (mail-header-id (gnus-summary-article-header)))
712 (mail-header-references (gnus-summary-article-header)))))
713 (if (eq (car (gnus-group-method group)) 'nnimap)
714 (progn (nnimap-possibly-change-group (gnus-group-short-name group) nil)
715 (with-current-buffer (nnimap-buffer)
716 (let* ((cmd (let ((value (format
717 "(OR HEADER REFERENCES %s HEADER Message-Id %s)"
719 (dolist (refid refs value)
721 "(OR (OR HEADER Message-Id %s HEADER REFERENCES %s) %s)"
722 refid refid value)))))
723 (result (nnimap-command
724 "UID SEARCH %s" cmd)))
725 (gnus-summary-read-group-1 group t t gnus-summary-buffer nil
727 (delete 0 (mapcar #'string-to-number
728 (cdr (assoc "SEARCH" (cdr result))))))))))
729 (gnus-summary-read-group-1 group t t gnus-summary-buffer
730 nil (list backend-number))
731 (gnus-summary-limit (list backend-number))
732 (gnus-summary-refer-thread))))
735 (if (fboundp 'eval-after-load)
736 (eval-after-load "gnus-sum"
737 '(define-key gnus-summary-goto-map
738 "T" 'gnus-summary-nnir-goto-thread))
739 (add-hook 'gnus-summary-mode-hook
741 (define-key gnus-summary-goto-map
742 "T" 'gnus-summary-nnir-goto-thread)))))
746 ;; Gnus backend interface functions.
748 (deffoo nnir-open-server (server &optional definitions)
749 ;; Just set the server variables appropriately.
750 (nnoo-change-server 'nnir server definitions))
752 (deffoo nnir-request-group (group &optional server fast info)
753 "GROUP is the query string."
754 (nnir-possibly-change-server server)
755 ;; Check for cache and return that if appropriate.
756 (if (and (equal group nnir-current-query)
757 (equal gnus-group-marked nnir-current-group-marked)
759 (equal server nnir-current-server)))
762 (setq nnir-artlist (nnir-run-query group)))
763 (with-current-buffer nntp-server-buffer
764 (if (zerop (length nnir-artlist))
766 (setq nnir-current-query nil
767 nnir-current-server nil
768 nnir-current-group-marked nil
770 (nnheader-report 'nnir "Search produced empty results."))
771 ;; Remember data for cache.
772 (setq nnir-current-query group)
773 (when server (setq nnir-current-server server))
774 (setq nnir-current-group-marked gnus-group-marked)
775 (nnheader-insert "211 %d %d %d %s\n"
776 (nnir-artlist-length nnir-artlist) ; total #
778 (nnir-artlist-length nnir-artlist) ; last #
779 group)))) ; group name
781 (deffoo nnir-retrieve-headers (articles &optional group server fetch-old)
783 (let ((artlist (copy-sequence articles))
784 art artitem artgroup artno artrsv artfullgroup
785 novitem novdata foo server)
786 (while (not (null artlist))
787 (setq art (car artlist))
791 "nnir-retrieve-headers doesn't grok message ids: %s"
793 (setq artitem (nnir-artlist-article nnir-artlist art))
794 (setq artrsv (nnir-artitem-rsv artitem))
795 (setq artfullgroup (nnir-artitem-group artitem))
796 (setq artno (nnir-artitem-number artitem))
797 (setq artgroup (gnus-group-real-name artfullgroup))
798 (setq server (nnir-group-server artfullgroup))
799 ;; retrieve NOV or HEAD data for this article, transform into
800 ;; NOV data and prepend to `novdata'
801 (set-buffer nntp-server-buffer)
802 (nnir-possibly-change-server server)
803 (let ((gnus-override-method
804 (gnus-server-to-method server)))
805 ;; if nnir-get-article-nov-override-function is set, use it
806 (if nnir-get-article-nov-override-function
807 (setq novitem (funcall nnir-get-article-nov-override-function
809 ;; else, set novitem through nnheader-parse-nov/nnheader-parse-head
810 (case (setq foo (gnus-retrieve-headers (list artno)
813 (goto-char (point-min))
814 (setq novitem (nnheader-parse-nov)))
816 (goto-char (point-min))
817 (setq novitem (nnheader-parse-head)))
818 (t (error "Unknown header type %s while requesting article %s of group %s"
819 foo artno artfullgroup)))))
820 ;; replace article number in original group with article number
823 (mail-header-set-number novitem art)
824 (mail-header-set-from novitem
825 (mail-header-from novitem))
826 (mail-header-set-subject
828 (format "[%d: %s/%d] %s"
829 artrsv artgroup artno
830 (mail-header-subject novitem)))
831 (push novitem novdata)
832 (setq artlist (cdr artlist))))
833 (setq novdata (nreverse novdata))
834 (set-buffer nntp-server-buffer) (erase-buffer)
835 (mapc 'nnheader-insert-nov novdata)
838 (deffoo nnir-request-article (article
839 &optional group server to-buffer)
840 (if (stringp article)
843 "nnir-retrieve-headers doesn't grok message ids: %s"
846 (let* ((artitem (nnir-artlist-article nnir-artlist
848 (artfullgroup (nnir-artitem-group artitem))
849 (artno (nnir-artitem-number artitem))
851 ;; Why must we bind nntp-server-buffer here? It won't
852 ;; work if `buf' is used, say. (Of course, the set-buffer
853 ;; line below must then be updated, too.)
854 (nntp-server-buffer (or to-buffer nntp-server-buffer)))
855 (set-buffer nntp-server-buffer)
857 (message "Requesting article %d from group %s"
859 (gnus-request-article artno artfullgroup nntp-server-buffer)
860 (cons artfullgroup artno)))))
863 (nnoo-define-skeleton nnir)
866 (defmacro nnir-add-result (dirnam artno score prefix server artlist)
867 "Ask `nnir-compose-result' to construct a result vector,
868 and if it is non-nil, add it to artlist."
869 `(let ((result (nnir-compose-result ,dirnam ,artno ,score ,prefix ,server)))
870 (when (not (null result))
871 (push result ,artlist))))
873 (autoload 'nnmaildir-base-name-to-article-number "nnmaildir")
875 ;; Helper function currently used by the Swish++ and Namazu backends;
876 ;; perhaps useful for other backends as well
877 (defun nnir-compose-result (dirnam article score prefix server)
878 "Extract the group from dirnam, and create a result vector
879 ready to be added to the list of search results."
881 ;; remove nnir-*-remove-prefix from beginning of dirnam filename
882 (when (string-match (concat "^" prefix) dirnam)
883 (setq dirnam (replace-match "" t t dirnam)))
885 (when (file-readable-p (concat prefix dirnam article))
886 ;; remove trailing slash and, for nnmaildir, cur/new/tmp
889 (if (string= (gnus-group-server server) "nnmaildir")
892 ;; Set group to dirnam without any leading dots or slashes,
893 ;; and with all subsequent slashes replaced by dots
894 (let ((group (gnus-replace-in-string
895 (gnus-replace-in-string dirnam "^[./\\]" "" t)
898 (vector (nnir-group-full-name group server)
899 (if (string= (gnus-group-server server) "nnmaildir")
900 (nnmaildir-base-name-to-article-number
901 (substring article 0 (string-match ":" article))
903 (string-to-number article))
904 (string-to-number score)))))
906 ;;; Search Engine Interfaces:
908 ;; freeWAIS-sf interface.
909 (defun nnir-run-waissearch (query server &optional group)
910 "Run given query agains waissearch. Returns vector of (group name, file name)
911 pairs (also vectors, actually)."
913 (error "The freeWAIS-sf backend cannot search specific groups"))
915 (let ((qstring (cdr (assq 'query query)))
916 (prefix (nnir-read-server-parm 'nnir-wais-remove-prefix server))
917 artlist score artno dirnam)
918 (set-buffer (get-buffer-create nnir-tmp-buffer))
920 (message "Doing WAIS query %s..." query)
921 (call-process nnir-wais-program
922 nil ; input from /dev/null
923 t ; output to current buffer
924 nil ; don't redisplay
925 "-d" (nnir-read-server-parm 'nnir-wais-database server) ; database to search
927 (message "Massaging waissearch output...")
928 ;; remove superfluous lines
929 (keep-lines "Score:")
930 ;; extract data from result lines
931 (goto-char (point-min))
932 (while (re-search-forward
933 "Score: +\\([0-9]+\\).*'\\([0-9]+\\) +\\([^']+\\)/'" nil t)
934 (setq score (match-string 1)
935 artno (match-string 2)
936 dirnam (match-string 3))
937 (unless (string-match prefix dirnam)
938 (nnheader-report 'nnir "Dir name %s doesn't contain prefix %s"
940 (setq group (gnus-replace-in-string
941 (replace-match "" t t dirnam) "/" "."))
942 (push (vector (nnir-group-full-name group server)
943 (string-to-number artno)
944 (string-to-number score))
946 (message "Massaging waissearch output...done")
949 (function (lambda (x y)
950 (> (nnir-artitem-rsv x)
951 (nnir-artitem-rsv y)))))))))
955 ;; send queries as literals
959 (defun nnir-run-imap (query srv &optional group-option)
960 "Run a search against an IMAP back-end server.
961 This uses a custom query language parser; see `nnir-imap-make-query' for
962 details on the language and supported extensions"
964 (let ((qstring (cdr (assq 'query query)))
965 (server (cadr (gnus-server-to-method srv)))
966 (group (or group-option (gnus-group-group-name)))
967 (defs (caddr (gnus-server-to-method srv)))
968 (criteria (or (cdr (assq 'criteria query))
969 (cdr (assoc nnir-imap-default-search-key
970 nnir-imap-search-arguments))))
971 (gnus-inhibit-demon t)
973 (message "Opening server %s" server)
975 (when (nnimap-possibly-change-group (gnus-group-short-name group) server)
976 (with-current-buffer (nnimap-buffer)
977 (message "Searching %s..." group)
980 (nnimap-command "UID SEARCH %s"
981 (if (string= criteria "")
983 (nnir-imap-make-query criteria qstring)
987 (push (vector group artnum 1) artlist)
988 (setq arts (1+ arts)))
990 (delete 0 (mapcar #'string-to-number
991 (cdr (assoc "SEARCH" (cdr result)))))))
992 (message "Searching %s... %d matches" group arts)))
993 (message "Searching %s...done" group))
997 (defun nnir-imap-make-query (criteria qstring)
998 "Parse the query string and criteria into an appropriate IMAP search
999 expression, returning the string query to make.
1001 This implements a little language designed to return the expected results
1002 to an arbitrary query string to the end user.
1004 The search is always case-insensitive, as defined by RFC2060, and supports
1005 the following features (inspired by the Google search input language):
1007 Automatic \"and\" queries
1008 If you specify multiple words then they will be treated as an \"and\"
1009 expression intended to match all components.
1012 If you wrap your query in double-quotes then it will be treated as a
1016 If you precede a term with \"-\" then it will negate that.
1019 If you include an upper-case \"OR\" in your search it will cause the
1020 term before it and the term after it to be treated as alternatives.
1022 In future the following will be added to the language:
1023 * support for date matches
1024 * support for location of text matching within the query
1025 * from/to/etc headers
1026 * additional search terms
1027 * flag based searching
1028 * anything else that the RFC supports, basically."
1029 ;; Walk through the query and turn it into an IMAP query string.
1030 (nnir-imap-query-to-imap criteria (nnir-imap-parse-query qstring)))
1033 (defun nnir-imap-query-to-imap (criteria query)
1034 "Turn a s-expression format query into IMAP."
1036 ;; Turn the expressions into IMAP text
1038 (nnir-imap-expr-to-imap criteria item))
1039 ;; The query, already in s-expr format.
1041 ;; Append a space between each expression
1045 (defun nnir-imap-expr-to-imap (criteria expr)
1046 "Convert EXPR into an IMAP search expression on CRITERIA"
1047 ;; What sort of expression is this, eh?
1049 ;; Simple string term
1051 (format "%s %S" criteria expr))
1052 ;; Trivial term: and
1053 ((eq expr 'and) nil)
1054 ;; Composite term: or expression
1055 ((eq (car-safe expr) 'or)
1057 (nnir-imap-expr-to-imap criteria (second expr))
1058 (nnir-imap-expr-to-imap criteria (third expr))))
1059 ;; Composite term: just the fax, mam
1060 ((eq (car-safe expr) 'not)
1061 (format "NOT (%s)" (nnir-imap-query-to-imap criteria (rest expr))))
1062 ;; Composite term: just expand it all.
1063 ((and (not (null expr)) (listp expr))
1064 (format "(%s)" (nnir-imap-query-to-imap criteria expr)))
1065 ;; Complex value, give up for now.
1066 (t (error "Unhandled input: %S" expr))))
1069 (defun nnir-imap-parse-query (string)
1070 "Turn STRING into an s-expression based query based on the IMAP
1071 query language as defined in `nnir-imap-make-query'.
1073 This involves turning individual tokens into higher level terms
1074 that the search language can then understand and use."
1076 ;; Set up the parsing environment.
1078 (goto-char (point-min))
1079 ;; Now, collect the output terms and return them.
1081 (while (not (nnir-imap-end-of-input))
1082 (push (nnir-imap-next-expr) out))
1086 (defun nnir-imap-next-expr (&optional count)
1087 "Return the next expression from the current buffer."
1088 (let ((term (nnir-imap-next-term count))
1089 (next (nnir-imap-peek-symbol)))
1090 ;; Are we looking at an 'or' expression?
1092 ;; Handle 'expr or expr'
1094 (list 'or term (nnir-imap-next-expr 2)))
1099 (defun nnir-imap-next-term (&optional count)
1100 "Return the next TERM from the current buffer."
1101 (let ((term (nnir-imap-next-symbol count)))
1102 ;; What sort of term is this?
1104 ;; and -- just ignore it
1105 ((eq term 'and) 'and)
1107 ((eq term 'not) (list 'not (nnir-imap-next-expr)))
1112 (defun nnir-imap-peek-symbol ()
1113 "Return the next symbol from the current buffer, but don't consume it."
1115 (nnir-imap-next-symbol)))
1117 (defun nnir-imap-next-symbol (&optional count)
1118 "Return the next symbol from the current buffer, or nil if we are
1119 at the end of the buffer. If supplied COUNT skips some symbols before
1120 returning the one at the supplied position."
1121 (when (and (numberp count) (> count 1))
1122 (nnir-imap-next-symbol (1- count)))
1123 (let ((case-fold-search t))
1124 ;; end of input stream?
1125 (unless (nnir-imap-end-of-input)
1126 ;; No, return the next symbol from the stream.
1128 ;; negated expression -- return it and advance one char.
1129 ((looking-at "-") (forward-char 1) 'not)
1131 ((looking-at "\"") (nnir-imap-delimited-string "\""))
1132 ;; list expression -- we parse the content and return this as a list.
1134 (nnir-imap-parse-query (nnir-imap-delimited-string ")")))
1135 ;; keyword input -- return a symbol version
1136 ((looking-at "\\band\\b") (forward-char 3) 'and)
1137 ((looking-at "\\bor\\b") (forward-char 2) 'or)
1138 ((looking-at "\\bnot\\b") (forward-char 3) 'not)
1139 ;; Simple, boring keyword
1140 (t (let ((start (point))
1141 (end (if (search-forward-regexp "[[:blank:]]" nil t)
1144 ;; unskip if we hit a non-blank terminal character.
1145 (when (string-match "[^[:blank:]]" (match-string 0))
1147 (goto-char (point-max)))))
1148 (buffer-substring start end)))))))
1150 (defun nnir-imap-delimited-string (delimiter)
1151 "Return a delimited string from the current buffer."
1152 (let ((start (point)) end)
1153 (forward-char 1) ; skip the first delimiter.
1155 (unless (search-forward delimiter nil t)
1156 (error "Unmatched delimited input with %s in query" delimiter))
1157 (let ((here (point)))
1158 (unless (equal (buffer-substring (- here 2) (- here 1)) "\\")
1159 (setq end (point)))))
1160 (buffer-substring (1+ start) (1- end))))
1162 (defun nnir-imap-end-of-input ()
1163 "Are we at the end of input?"
1164 (skip-chars-forward "[[:blank:]]")
1168 ;; Swish++ interface.
1177 (defun nnir-run-swish++ (query server &optional group)
1178 "Run QUERY against swish++.
1179 Returns a vector of (group name, file name) pairs (also vectors,
1182 Tested with swish++ 4.7 on GNU/Linux and with swish++ 5.0b2 on
1186 (error "The swish++ backend cannot search specific groups"))
1189 (let ( (qstring (cdr (assq 'query query)))
1190 (groupspec (cdr (assq 'group query)))
1191 (prefix (nnir-read-server-parm 'nnir-swish++-remove-prefix server))
1193 ;; nnml-use-compressed-files might be any string, but probably this
1194 ;; is sufficient. Note that we can't only use the value of
1195 ;; nnml-use-compressed-files because old articles might have been
1196 ;; saved with a different value.
1197 (article-pattern (if (string= (gnus-group-server server) "nnmaildir")
1199 "^[0-9]+\\(\\.[a-z0-9]+\\)?$"))
1200 score artno dirnam filenam)
1202 (when (equal "" qstring)
1203 (error "swish++: You didn't enter anything"))
1205 (set-buffer (get-buffer-create nnir-tmp-buffer))
1209 (message "Doing swish++ query %s on %s..." qstring groupspec)
1210 (message "Doing swish++ query %s..." qstring))
1212 (let* ((cp-list `( ,nnir-swish++-program
1213 nil ; input from /dev/null
1215 nil ; don't redisplay
1216 "--config-file" ,(nnir-read-server-parm 'nnir-swish++-configuration-file server)
1217 ,@(nnir-read-server-parm 'nnir-swish++-additional-switches server)
1218 ,qstring ; the query, in swish++ format
1222 (message "%s args: %s" nnir-swish++-program
1223 (mapconcat 'identity (cddddr cp-list) " ")) ;; ???
1224 (apply 'call-process cp-list))))
1225 (unless (or (null exitstatus)
1227 (nnheader-report 'nnir "Couldn't run swish++: %s" exitstatus)
1228 ;; swish++ failure reason is in this buffer, show it if
1229 ;; the user wants it.
1230 (when (> gnus-verbose 6)
1231 (display-buffer nnir-tmp-buffer))))
1233 ;; The results are output in the format of:
1235 ;; rank relative-path-name file-size file-title
1237 ;; rank relative-path-name file-size topic??
1238 ;; where rank is an integer from 1 to 100.
1239 (goto-char (point-min))
1240 (while (re-search-forward
1241 "\\(^[0-9]+\\) \\([^ ]+\\) [0-9]+ \\(.*\\)$" nil t)
1242 (setq score (match-string 1)
1243 filenam (match-string 2)
1244 artno (file-name-nondirectory filenam)
1245 dirnam (file-name-directory filenam))
1247 ;; don't match directories
1248 (when (string-match article-pattern artno)
1249 (when (not (null dirnam))
1251 ;; maybe limit results to matching groups.
1252 (when (or (not groupspec)
1253 (string-match groupspec dirnam))
1254 (nnir-add-result dirnam artno score prefix server artlist)))))
1256 (message "Massaging swish++ output...done")
1261 (function (lambda (x y)
1262 (> (nnir-artitem-rsv x)
1263 (nnir-artitem-rsv y)))))))))
1265 ;; Swish-E interface.
1266 (defun nnir-run-swish-e (query server &optional group)
1267 "Run given query against swish-e.
1268 Returns a vector of (group name, file name) pairs (also vectors,
1271 Tested with swish-e-2.0.1 on Windows NT 4.0."
1273 ;; swish-e crashes with empty parameter to "-w" on commandline...
1275 (error "The swish-e backend cannot search specific groups"))
1278 (let ((qstring (cdr (assq 'query query)))
1280 (or (nnir-read-server-parm 'nnir-swish-e-remove-prefix server)
1281 (error "Missing parameter `nnir-swish-e-remove-prefix'")))
1282 artlist score artno dirnam group )
1284 (when (equal "" qstring)
1285 (error "swish-e: You didn't enter anything"))
1287 (set-buffer (get-buffer-create nnir-tmp-buffer))
1290 (message "Doing swish-e query %s..." query)
1292 (or (nnir-read-server-parm
1293 'nnir-swish-e-index-files server)
1294 (error "Missing parameter `nnir-swish-e-index-files'")))
1295 (additional-switches
1296 (nnir-read-server-parm
1297 'nnir-swish-e-additional-switches server))
1298 (cp-list `(,nnir-swish-e-program
1299 nil ; input from /dev/null
1301 nil ; don't redisplay
1303 ,@additional-switches
1305 ,qstring ; the query, in swish-e format
1309 (message "%s args: %s" nnir-swish-e-program
1310 (mapconcat 'identity (cddddr cp-list) " "))
1311 (apply 'call-process cp-list))))
1312 (unless (or (null exitstatus)
1314 (nnheader-report 'nnir "Couldn't run swish-e: %s" exitstatus)
1315 ;; swish-e failure reason is in this buffer, show it if
1316 ;; the user wants it.
1317 (when (> gnus-verbose 6)
1318 (display-buffer nnir-tmp-buffer))))
1320 ;; The results are output in the format of:
1321 ;; rank path-name file-title file-size
1322 (goto-char (point-min))
1323 (while (re-search-forward
1324 "\\(^[0-9]+\\) \\([^ ]+\\) \"\\([^\"]+\\)\" [0-9]+$" nil t)
1325 (setq score (match-string 1)
1326 artno (match-string 3)
1327 dirnam (file-name-directory (match-string 2)))
1329 ;; don't match directories
1330 (when (string-match "^[0-9]+$" artno)
1331 (when (not (null dirnam))
1333 ;; remove nnir-swish-e-remove-prefix from beginning of dirname
1334 (when (string-match (concat "^" prefix) dirnam)
1335 (setq dirnam (replace-match "" t t dirnam)))
1337 (setq dirnam (substring dirnam 0 -1))
1338 ;; eliminate all ".", "/", "\" from beginning. Always matches.
1339 (string-match "^[./\\]*\\(.*\\)$" dirnam)
1341 (setq group (gnus-replace-in-string (match-string 1 dirnam) "/" "."))
1342 ;; Windows "\\" -> "."
1343 (setq group (gnus-replace-in-string group "\\\\" "."))
1345 (push (vector (nnir-group-full-name group server)
1346 (string-to-number artno)
1347 (string-to-number score))
1350 (message "Massaging swish-e output...done")
1355 (function (lambda (x y)
1356 (> (nnir-artitem-rsv x)
1357 (nnir-artitem-rsv y)))))))))
1360 (defun nnir-run-hyrex (query server &optional group)
1363 (groupspec (cdr (assq 'group query)))
1364 (qstring (cdr (assq 'query query)))
1365 (prefix (nnir-read-server-parm 'nnir-hyrex-remove-prefix server))
1367 (when (and group groupspec)
1368 (error (concat "It does not make sense to use a group spec"
1369 " with process-marked groups.")))
1371 (setq groupspec (gnus-group-real-name group)))
1372 (when (and group (not (equal group (nnir-group-full-name groupspec server))))
1373 (message "%s vs. %s" group (nnir-group-full-name groupspec server))
1374 (error "Server with groupspec doesn't match group !"))
1375 (set-buffer (get-buffer-create nnir-tmp-buffer))
1378 (message "Doing hyrex-search query %s on %s..." query groupspec)
1379 (message "Doing hyrex-search query %s..." query))
1381 `( ,nnir-hyrex-program
1382 nil ; input from /dev/null
1384 nil ; don't redisplay
1385 "-i",(nnir-read-server-parm 'nnir-hyrex-index-directory server) ; index directory
1386 ,@(nnir-read-server-parm 'nnir-hyrex-additional-switches server)
1387 ,qstring ; the query, in hyrex-search format
1391 (message "%s args: %s" nnir-hyrex-program
1392 (mapconcat 'identity (cddddr cp-list) " "))
1393 (apply 'call-process cp-list))))
1394 (unless (or (null exitstatus)
1396 (nnheader-report 'nnir "Couldn't run hyrex-search: %s" exitstatus)
1397 ;; nnir-search failure reason is in this buffer, show it if
1398 ;; the user wants it.
1399 (when (> gnus-verbose 6)
1400 (display-buffer nnir-tmp-buffer)))) ;; FIXME: Dont clear buffer !
1402 (message "Doing hyrex-search query \"%s\" on %s...done" qstring groupspec)
1403 (message "Doing hyrex-search query \"%s\"...done" qstring))
1405 ;; nnir-search returns:
1406 ;; for nnml/nnfolder: "filename mailid weigth"
1407 ;; for nnimap: "group mailid weigth"
1408 (goto-char (point-min))
1409 (delete-non-matching-lines "^\\S + [0-9]+ [0-9]+$")
1410 ;; HyREX couldn't search directly in groups -- so filter out here.
1412 (keep-lines groupspec))
1413 ;; extract data from result lines
1414 (goto-char (point-min))
1415 (while (re-search-forward
1416 "\\(\\S +\\) \\([0-9]+\\) \\([0-9]+\\)" nil t)
1417 (setq dirnam (match-string 1)
1418 artno (match-string 2)
1419 score (match-string 3))
1420 (when (string-match prefix dirnam)
1421 (setq dirnam (replace-match "" t t dirnam)))
1422 (push (vector (nnir-group-full-name
1423 (gnus-replace-in-string dirnam "/" ".") server)
1424 (string-to-number artno)
1425 (string-to-number score))
1427 (message "Massaging hyrex-search output...done.")
1430 (function (lambda (x y)
1431 (if (string-lessp (nnir-artitem-group x)
1432 (nnir-artitem-group y))
1434 (< (nnir-artitem-number x)
1435 (nnir-artitem-number y)))))))
1439 (defun nnir-run-namazu (query server &optional group)
1440 "Run given query against Namazu. Returns a vector of (group name, file name)
1441 pairs (also vectors, actually).
1443 Tested with Namazu 2.0.6 on a GNU/Linux system."
1445 (error "The Namazu backend cannot search specific groups"))
1447 (let ((article-pattern (if (string= (gnus-group-server server) "nnmaildir")
1451 (qstring (cdr (assq 'query query)))
1452 (prefix (nnir-read-server-parm 'nnir-namazu-remove-prefix server))
1454 (process-environment (copy-sequence process-environment)))
1455 (setenv "LC_MESSAGES" "C")
1456 (set-buffer (get-buffer-create nnir-tmp-buffer))
1459 `( ,nnir-namazu-program
1460 nil ; input from /dev/null
1462 nil ; don't redisplay
1463 "-q" ; don't be verbose
1464 "-a" ; show all matches
1465 "-s" ; use short format
1466 ,@(nnir-read-server-parm 'nnir-namazu-additional-switches server)
1467 ,qstring ; the query, in namazu format
1468 ,(nnir-read-server-parm 'nnir-namazu-index-directory server) ; index directory
1472 (message "%s args: %s" nnir-namazu-program
1473 (mapconcat 'identity (cddddr cp-list) " "))
1474 (apply 'call-process cp-list))))
1475 (unless (or (null exitstatus)
1477 (nnheader-report 'nnir "Couldn't run namazu: %s" exitstatus)
1478 ;; Namazu failure reason is in this buffer, show it if
1479 ;; the user wants it.
1480 (when (> gnus-verbose 6)
1481 (display-buffer nnir-tmp-buffer))))
1483 ;; Namazu output looks something like this:
1484 ;; 2. Re: Gnus agent expire broken (score: 55)
1485 ;; /home/henrik/Mail/mail/sent/1310 (4,138 bytes)
1487 (goto-char (point-min))
1488 (while (re-search-forward
1489 "^\\([0-9]+\\.\\).*\\((score: \\([0-9]+\\)\\))\n\\([^ ]+\\)"
1491 (setq score (match-string 3)
1492 group (file-name-directory (match-string 4))
1493 article (file-name-nondirectory (match-string 4)))
1495 ;; make sure article and group is sane
1496 (when (and (string-match article-pattern article)
1498 (nnir-add-result group article score prefix server artlist)))
1500 ;; sort artlist by score
1503 (function (lambda (x y)
1504 (> (nnir-artitem-rsv x)
1505 (nnir-artitem-rsv y)))))))))
1507 (defun nnir-run-find-grep (query server &optional group)
1508 "Run find and grep to obtain matching articles."
1509 (let* ((method (gnus-server-to-method server))
1511 (concat (symbol-name (car method)) "-directory")))
1512 (directory (cadr (assoc sym (cddr method))))
1513 (regexp (cdr (assoc 'query query)))
1514 (grep-options (cdr (assoc 'grep-options query)))
1517 (error "No directory found in method specification of server %s"
1519 (message "Searching %s using find-grep..." (or group server))
1520 (save-window-excursion
1521 (set-buffer (get-buffer-create nnir-tmp-buffer))
1523 (if (> gnus-verbose 6)
1524 (pop-to-buffer (current-buffer)))
1525 (cd directory) ; Using relative paths simplifies postprocessing.
1529 ;; Try accessing the group literally as well as
1530 ;; interpreting dots as directory separators so the
1531 ;; engine works with plain nnml as well as the Gnus Cache.
1532 (let ((group (gnus-group-real-name group)))
1533 ;; Replace cl-func find-if.
1534 (if (file-directory-p group)
1536 (if (file-directory-p
1537 (setq group (gnus-replace-in-string group "\\." "/" t)))
1540 (error "Cannot locate directory for group"))
1543 'call-process "find" nil t
1544 "find" group "-type" "f" "-name" "[0-9]*" "-exec"
1546 `("-l" ,@(and grep-options
1547 (split-string grep-options "\\s-" t))
1548 "-e" ,regexp "{}" "+"))))
1550 ;; Translate relative paths to group names.
1552 (let* ((path (split-string
1553 (buffer-substring (point) (line-end-position)) "/" t))
1554 (art (string-to-number (car (last path)))))
1555 (while (string= "." (car path))
1556 (setq path (cdr path)))
1557 (let ((group (mapconcat 'identity
1558 ;; Replace cl-func: (subseq path 0 -1)
1559 (let ((end (1- (length path)))
1561 (while (>= (setq end (1- end)) 0)
1562 (push (pop path) res))
1565 (push (vector (nnir-group-full-name group server) art 0)
1568 (message "Searching %s using find-grep...done" (or group server))
1573 (defun nnir-read-parms (query)
1574 "Reads additional search parameters according to `nnir-engines'."
1575 (let ((parmspec (caddr (assoc nnir-search-engine nnir-engines))))
1576 (cons (cons 'query query)
1577 (mapcar 'nnir-read-parm parmspec))))
1579 (defun nnir-read-parm (parmspec)
1580 "Reads a single search parameter.
1581 `parmspec' is a cons cell, the car is a symbol, the cdr is a prompt."
1582 (let ((sym (car parmspec))
1583 (prompt (cdr parmspec)))
1585 (let* ((result (apply 'gnus-completing-read prompt))
1586 (mapping (or (assoc result nnir-imap-search-arguments)
1587 (cons nil nnir-imap-search-other))))
1588 (cons sym (format (cdr mapping) result)))
1589 (cons sym (read-string prompt)))))
1591 (defun nnir-run-query (query)
1592 "Invoke appropriate search engine function (see `nnir-engines').
1593 If some groups were process-marked, run the query for each of the groups
1594 and concat the results."
1595 (let ((q (car (read-from-string query))))
1596 (if gnus-group-marked
1599 (let* ((server (nnir-group-server x))
1601 (or (nnir-read-server-parm 'nnir-search-engine
1604 (assoc (car (gnus-server-to-method server))
1605 nnir-method-default-engines))))
1607 (setq search-func (cadr
1612 (funcall search-func q server x)
1617 (if (and (equal (cadr x) 'ok) (not (equal (cadar x) "-ephemeral")))
1618 (let* ((server (format "%s:%s" (caar x) (cadar x)))
1620 (or (nnir-read-server-parm 'nnir-search-engine
1623 (assoc (car (gnus-server-to-method server))
1624 nnir-method-default-engines))))
1626 (setq search-func (cadr
1631 (funcall search-func q server nil)
1634 gnus-opened-servers)
1638 (defun nnir-read-server-parm (key server)
1639 "Returns the parameter value of for the given server, where server is of
1640 form 'backend:name'."
1641 (let ((method (gnus-server-to-method server)))
1642 (cond ((and method (assq key (cddr method)))
1643 (nth 1 (assq key (cddr method))))
1644 ((and nnir-mail-backend
1645 (gnus-server-equal method nnir-mail-backend))
1649 ;; (if (assq key (cddr method))
1650 ;; (nth 1 (assq key (cddr method)))
1651 ;; (symbol-value key))
1652 ;; (symbol-value key))
1655 (defun nnir-group-full-name (shortname server)
1656 "For the given group name, return a full Gnus group name.
1657 The Gnus backend/server information is added."
1658 (gnus-group-prefixed-name shortname (gnus-server-to-method server)))
1660 (defun nnir-possibly-change-server (server)
1661 (unless (and server (nnir-server-opened server))
1662 (nnir-open-server server)))
1665 ;; Data type article list.
1667 (defun nnir-artlist-length (artlist)
1668 "Returns number of articles in artlist."
1671 (defun nnir-artlist-article (artlist n)
1672 "Returns from ARTLIST the Nth artitem (counting starting at 1)."
1673 (elt artlist (1- n)))
1675 (defun nnir-artitem-group (artitem)
1676 "Returns the group from the ARTITEM."
1679 (defun nnir-artlist-artitem-group (artlist n)
1680 "Returns from ARTLIST the group of the Nth artitem (counting from 1)."
1681 (nnir-artitem-group (nnir-artlist-article artlist n)))
1683 (defun nnir-artitem-number (artitem)
1684 "Returns the number from the ARTITEM."
1687 (defun nnir-artlist-artitem-number (artlist n)
1688 "Returns from ARTLIST the number of the Nth artitem (counting from 1)."
1689 (nnir-artitem-number (nnir-artlist-article artlist n)))
1691 (defun nnir-artitem-rsv (artitem)
1692 "Returns the Retrieval Status Value (RSV, score) from the ARTITEM."
1695 (defun nnir-artlist-artitem-rsv (artlist n)
1696 "Returns from ARTLIST the Retrieval Status Value of the Nth artitem
1697 \(counting from 1)."
1698 (nnir-artitem-rsv (nnir-artlist-article artlist n)))
1701 (defun nnir-artlist-groups (artlist)
1702 "Returns a list of all groups in the given ARTLIST."
1705 ;; from each artitem, extract group component
1706 (setq with-dups (mapcar 'nnir-artitem-group artlist))
1707 ;; remove duplicates from above
1708 (mapc (function (lambda (x) (add-to-list 'res x)))
1716 ;;; nnir.el ends here