Re: Opera backend for Beagle

["Andrey Melentyev" <rikz yandex ru>]

No, I have no such file.
I found why I didn't get any +url messages. Its because of Gentoo
portage and my stupidness. There was a patch in gentoo to make debug
output less (I don't know for what) and it was automatically applied
when I rewrote ebuild to use subversion instead of old release
tarballs. So, finally now I can get all the messages about indexed
pages.

Now about the problem with indexing itself. As I've already said, I'm
using Opera 9.50 Beta builds. After downgrading to latest stable
release of old 9.2 branch (9.24) the situation with indexing changed
slightly. Now almost all pages are indexed successfully. I say
"almost" because I found at least one page which is not indexed on my
system even with Opera 9.24. Unfortunately, the site is on Russian,
but there's a lot of English words. Here's the link:
http://kino.local.pp.ru/index.php?module=Kino&func=Alph It may ask use
for a login and password, use beagle:beagle On that page there's a
long list of films with their names on Russian and English. I can't
search them neither by their English name nor by Russian. When I
search for "kino.local.pp.ru" I can see that the page were somehow
indexed. But the list of films itself is not searchable. Here's the
log from console:

Debug: Starting thread EHT 07697 [07452 IndexHelper]
Beagle.Daemon.UnixConnectionHandler:HandleConnection
Debug: -http://mail.google.com/mail/?ui=1&view=page&name=gp&ver=sh3fib53pgpk
Debug: -http://kino.local.pp.ru/index.php?module=Kino&func=Alph
Debug: -http://kino.local.pp.ru/user.php
Debug: +http://mail.google.com/mail/?ui=1&view=page&name=gp&ver=sh3fib53pgpk
(file:///home/rikz/.opera/cache4/opr000JW.htm)
Debug: +http://kino.local.pp.ru/index.php?module=Kino&func=Alph
(file:///tmp/tmp866c7a2.tmp)
Debug: +http://kino.local.pp.ru/user.php (file:///tmp/tmp8213cbb.tmp)
Debug: Sending response of type Beagle.FinishedResponse
Debug: Finished thread EHT 07697 [07452 IndexHelper]
Beagle.Daemon.UnixConnectionHandler:HandleConnection
Debug: Helper Size: VmRSS=28,3 MB, size=2,25, 31,3%

So, for me:
Opera 9.24 + beagle = almost everything is ok, except the example with
kino.local.pp.ru where the main part of the page is not indexed
Opera 9.50 + beagle = many sites are not indexed.

Maybe Opera devs has changed something about disk caching? I definely
know that one of the changes in 9.5 branch of Opera was removing of
file extensions from cache files. Some info on this change:
http://groups.google.com/group/opera.beta/browse_thread/thread/302c8493cf673ab2/8df43fb3401b538e

Debajyoti, I think that you can easily reproduce the problem by
upgrading Opera to 9.5 beta.
http://my.opera.com/desktopteam/blog/2007/11/09/work-in-progress

PS: feel free to contact me via Jabber at rikz jabber ru or
andrey melentyev gmail com if it is enough flood in mailing list.

> The "Warn: Error in filtering /home/rikz/.opera/cache4/opr00VAB with
> Beagle.Filters.FilterHtml, falling back" is why you are not seeing
> anything more.
> But there should still be
>
> -http://...
> +http://...
>
> Do you have this file /home/rikz/.opera/cache4/opr00VAB ? Is it an
> HTML file ? Does it have write permission ? I dont see the reason for
> "Warn: Couldn't store file attributes for
> /home/rikz/.opera/cache4/opr00VAB".