Item8629: refactor KinoSearchContrib to use StringiferContrib
Current State: Closed
Released In: n/a
Target Release: n/a
or, consider deprecating KinoSearchContrib
when the SolrPlugin
is released. we should be able to keep KinoSearchPlugin
(but plugged into the Solr backend) so that existing topic contents continue to work. if we're really ambitious, we can ship kinosearch script shims so that KinoSearch
upgraders wouldn't have to even change their system scripts.
- 26 Feb 2010
I would much rather keep the variety. There may be other reasons why Solr isn't suitable for some users.
- 27 Feb 2010
but plugged into the Solr backend
... does not compute.
Basically, kino as well as good old plucene are embedded frontends to lucene. Solr, on the other hand is a "serverization" of lucene ... and a lot more. Feature-wise, it is a true superset of kino or plucene.
Main disadvantage, lots of perl dependencies.
is meant to be a share based between both as they both need serialization of binary data. This is comparable with Tika
which I tried as well. It does have some advantages over StringifierContrib
, i.e. also able to return meta data in an xml format, like exif data of a photo. However, I abandoned Tika as basic tests on various office files showed that StringifierContrib
had a higher coverage while being more robust against weird office files.
has been externalized and improved over time. For one, it does not crash on password encoded xls files etc.
Coming soon: caching - don't re-stringify the same binary file if it has changed in the meantime. That's part of SolrPlugin
for no good reason.
There are still some more fundamental issues with StringifierContrib
inherited from its origins, foremost it recodes all strings to iso-8859-15. This might very well clash with the rest of the site's encoding. Next, its per line way of trying to detect a documents encoding might be overkill. So there's room for speed improvements here.
Other than that, it is a solid piece of work.
- 27 Feb 2010
- 06 Jun 2010
There may be something wrong with this work?
* AndreU has quit (Quit: AndreU)
<alice|wl_> any help with kinosearch here?
<alice|wl_> accoring to ks_test I miss a Foswiki/Contrib/KinoSearchContrib/StringifyBase.pm
<alice|wl_> it is not in the package
- 12 Jul 2010
I have removed the
script from the contrib, as the functionality is covered with the
script in StringifierContrib
- 14 Jul 2010
Forgot to remove config items.
- 25 Mar 2011