How to rescan just one group?

How to rescan just one group?

[Nevin Kapur]

Hi,

I am writing a little front end for grepmail[1] that takes creates an
nndoc group from its results.

What I want to do is rescan this group, clear its data (since I am
using the same file over and over again) and enter it.
Currently I am doing something like

(gnus-group-jump-to-group group-name)
(gnus-group-set-current-level 1 1)
(gnus-group-clear-data)
(gnus-group-get-new-news 1)
(gnus-group-jump-to-group group-name)
(gnus-group-select-group t)

which is what I would do manually. Is there a better/faster/more
elegant way to achieve this?

Thanks,

Nevin

Footnotes: 
[1]  http://grepmail.sourceforge.net

-- 
Nevin

Re: How to rescan just one group?

[Kai Großjohann]

On 23 Mar 2001, Nevin Kapur wrote:

> I am writing a little front end for grepmail[1] that takes creates
> an nndoc group from its results.

I don't like advocating my own module, but have you thought about
adding a grepmail backend to nnir.el?  Then you'd get the rescanning
for free.

nnir.el expects the searching program to somehow print the group name
and article number.  Not sure if that's easy to do with grepmail.

kai
-- 
Be indiscrete.  Do it continuously.

Re: How to rescan just one group?

[Nevin Kapur]

On Sat, 24 Mar 2001, Kai Großjohann wrote:

> On 23 Mar 2001, Nevin Kapur wrote:
> 
>> I am writing a little front end for grepmail[1] that takes creates
>> an nndoc group from its results.
> 
> I don't like advocating my own module, but have you thought about
> adding a grepmail backend to nnir.el?  Then you'd get the rescanning
> for free.

I'd heard of nnir and I did have a look at it.

> nnir.el expects the searching program to somehow print the group
> name and article number. Not sure if that's easy to do with
> grepmail.

I don't think grepmail can do this, but it should be easy enough to
write perl that does it. But,

,----[ From nnir.el ]
| ;; Restrictions:
| ;;
| ;; * Currently, this expects that you use nnml or another
| ;;   one-file-per-message backend.
| ;; * It can only search one mail backend.
`----

I archive my mail using nnfolder. I also want to search both nnml and
nnfolder. I've hacked grepmail so that it works with nnml style
folders.

,----[ Later on in nnir.el ]
| ;; The interface between the two layers consists of the single
| ;; function `nnir-run-query', which just selects the appropriate
| ;; function for the search engine one is using.  The input to
| ;; `nnir-run-query' is a string, representing the query as input by
| ;; the user.  The output of `nnir-run-query' is supposed to be a
| ;; vector, each element of which should in turn be a three-element
| ;; vector.  The first element should be group name of the article, the
| ;; second element should be the article number, and the third element
| ;; should be 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 like Glimpse, the RSV is
| ;; always 1000 (or 1 or 100, or whatever you like).
`----

So, if the output of nnir-run-query returned a vector that contained
group names that were nnfolder groups, would it be OK?

-- 
Nevin

Re: How to rescan just one group?

[Kai Großjohann]

On 24 Mar 2001, Nevin Kapur wrote:

> I don't think grepmail can do this, but it should be easy enough to
> write perl that does it. But,
> 
> ,----[ From nnir.el ]
> | ;; Restrictions:
> | ;;
> | ;; * Currently, this expects that you use nnml or another
> | ;;   one-file-per-message backend.
> | ;; * It can only search one mail backend.
> `----
> 
> I archive my mail using nnfolder. I also want to search both nnml
> and nnfolder. I've hacked grepmail so that it works with nnml style
> folders.

I wish I had the time to implement multi-backend searching for
nnir.el.  I know how to implement part of it already: just add server
parameters for nnir-search-engine and so on and have the searching
functions iterate over all the backends and invoke the backend search
function (nnir-run-query) for each backend.  What I don't know how to
implement is the group marking; you can process-mark a couple of
groups then issue a query, the query is restricted to the groups.
Probably what needs to be done is to go over all the servers, then
fetch the list of process-marked groups from each server, then invoke
nnir-run-query for that list.  Yes, that should work.

The restriction to `one file per message' backends is an artefact of
the searching mechanisms, the real restriction is in the interface, as
you cite below:

> ,----[ Later on in nnir.el ]
> | ;; The interface between the two layers consists of the single
> | ;; function `nnir-run-query', which just selects the appropriate
> | ;; function for the search engine one is using.  The input to
> | ;; `nnir-run-query' is a string, representing the query as input by
> | ;; the user.  The output of `nnir-run-query' is supposed to be a
> | ;; vector, each element of which should in turn be a three-element
> | ;; vector.  The first element should be group name of the article, the
> | ;; second element should be the article number, and the third element
> | ;; should be 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 like Glimpse, the RSV is
> | ;; always 1000 (or 1 or 100, or whatever you like).
> `----
> 
> So, if the output of nnir-run-query returned a vector that contained
> group names that were nnfolder groups, would it be OK?

Yes, that would be okay.

kai
-- 
Be indiscrete.  Do it continuously.

Re: How to rescan just one group?

[Gregory.Chernov]

Nevin Kapur <nevin@jhu.edu> writes:

> Hi,
> 
> I am writing a little front end for grepmail[1] that takes creates an
> nndoc group from its results.
> 
> What I want to do is rescan this group, clear its data (since I am
> using the same file over and over again) and enter it.
> Currently I am doing something like

Hi,

Couple of months ago I wrote "little front end for grepmail[1]".
I just print grepmail results in emacs temp buffer
and then use sort of ephemeral nndoc backend to read it.

If you interested I can post/mail it.

--
Greg