File: nnir.el.html

What does it do? Well, it searches your mail using some search
engine (imap, namazu, swish-e, gmane and others -- see later).

The Lisp setup may involve setting a few variables and setting up the search engine. You can define the variables in the server definition like this :
  (setq gnus-secondary-select-methods '(
      (nnimap "" (nnimap-address "localhost")
                 (nnir-search-engine namazu)
      )))
The main variable to set is nnir-search-engine. Choose one of the engines listed in nnir-engines. (Actually nnir-engines is an alist, type C-h v nnir-engines RET for more information; this includes examples for setting nnir-search-engine, too.)

The entry to searching is the single function nnir-run-query, which dispatches the search to the proper search function. The argument of nnir-run-query is an alist with two keys:
'nnir-query-spec and 'nnir-group-spec. The value for
'nnir-query-spec is an alist. The only required key/value pair is
(query . "query") specifying the search string to pass to the query
engine. Individual engines may have other elements. The value of
'nnir-group-spec is a list with the specification of the
groups/servers to search. The format of the 'nnir-group-spec is
(("server1" ("group11" "group12")) ("server2" ("group21"
"group22"))). If any of the group lists is absent then all groups
on that server are searched.

The output of nnir-run-query is a vector, each element of which should in turn be a three-element vector with the form: [fully prefixed group-name of the article; the article number; the Retrieval Status Value (RSV)] as returned from the search engine. An RSV is the score assigned to the document by the search engine. For Boolean search engines, the RSV is always 1000 (or 1 or 100, or whatever you like).

A vector of this form is used by the nnselect backend to create virtual groups. So nnir-run-query is a suitable function to use in nnselect groups.

The default sorting order of articles in an nnselect summary buffer is based on the order of the articles in the above mentioned vector, so that's where you can do the sorting you'd like. Maybe it would be nice to have a way of displaying the search result sorted differently?

So what do you need to do when you want to add another search
engine? You write a function that executes the query. Temporary
data from the search engine can be put in nnir-tmp-buffer. This function should return the list of articles as a vector, as described above. Then, you need to register this backend in nnir-engines. Then, users can choose the backend by setting nnir-search-engine as a server variable.

If you use one of the local indices (namazu, find-grep, swish) you must also set up a search engine backend.

1. Namazu

The Namazu backend requires you to have one directory containing all index files, this is controlled by the nnir-namazu-index-directory variable. To function the nnir-namazu-remove-prefix variable must also be correct, see the documentation for nnir-namazu-remove-prefix above.

It is particularly important not to pass any switches to namazu that will change the output format. Good switches to use include
--sort, --ascending, --early and --late. Refer to the Namazu
documentation for further information on valid switches.

To index my mail with the mknmz program I use the following configuration file:

,----
| package conf; # Don't remove this line!
|
| # Paths which will not be indexed. Don't use ^ or $ anchors.
| $EXCLUDE_PATH = "spam|sent";
|
| # Header fields which should be searchable. case-insensitive
| $REMAIN_HEADER = "from|date|message-id|subject";
|
| # Searchable fields. case-insensitive
| $SEARCH_FIELD = "from|date|message-id|subject";
|
| # The max length of a word.
| $WORD_LENG_MAX = 128;
|
| # The max length of a field.
| $MAX_FIELD_LENGTH = 256;
`----

My mail is stored in the directories ~/Mail/mail/, ~/Mail/lists/ and
~/Mail/archive/, so to index them I go to the directory set in
nnir-namazu-index-directory and issue the following command.

     mknmz --mailnews ~/Mail/archive/ ~/Mail/mail/ ~/Mail/lists/

For maximum searching efficiency I have a cron job set to run this command every four hours.

2. find-grep

The find-grep engine simply runs find(1) to locate eligible articles and searches them with grep(1). This, of course, is much slower than using a proper search engine but OTOH doesn't require maintenance of an index and is still faster than using any built-in means for searching. The method specification of the server to search must include a directory for this engine to work (E.g., nnml-directory). The tools must be POSIX compliant. GNU Find prior to version 4.2.12 (4.2.26 on Linux due to incorrect ARG_MAX handling) does not work.
,----
| ;; find-grep configuration for searching the Gnus Cache
|
| (nnml "cache"
| (nnml-get-new-mail nil)
| (nnir-search-engine find-grep)
| (nnml-directory "~/News/cache/")
| (nnml-active-file "~/News/cache/active"))
`----

Defined variables (30)

Defined functions (29)

Defined faces (0)