Re: [Tracker] Text extraction on text formats

[Jamie McCracken <jamiemcc blueyonder co uk>]

Laurent Aguerreche wrote:
> Le vendredi 17 novembre 2006 Ã 14:37 +0100, Luca Ferretti a Ãcrit :
> > Il giorno gio, 16/11/2006 alle 21.36 +0100, Laurent Aguerreche ha
> > scritto:
> > > Le jeudi 16 novembre 2006 Ã 18:55 +0000, Jamie McCracken a Ãcrit :
> > > > Luca Ferretti wrote:
> > > > > I'm trying to check and eventually expand info in
> > > > > http://live.gnome.org/Tracker/SupportedFormats.
> > > > >
> > > > >
> > > > > OTT (OpenDocument Text Template)
> > > > >   content:              no (????)
> > > Now yes.
> > Lauren, maybe a similar addition is needed for other OO.o 1&2 *template
> > mimetypes? 
>
> So here a patch to add them. I do not add more support for Star Division
> files because I just cannot make a file of that type!
> I remove calls to "nice" because children of a processus inherit its
> priority... so there are at 19.
> I saw that MS Word filter uses vwText. According to the site of wvWare
> ( http://wvware.sourceforge.net/ ), Abiword is now preferred to this
> tool. But I wonder if we could use libGSF directly to extract content of
> Word files... If I remember correctly, Wv just uses libGSF.
>
> I also propose a patch to:
> * extract text content only
> in /tmp/Tracker-user.pid/tmp_text_file_XXXXXX so now everything happens
> in /tmp/Tracker-user.pid and it should ensure privacy of files,
> * not make a not useful hierarchy like /home/user
> in /tmp/Tracker-user.pid to store cache of SQLite.
>
>
> Laurent.

have applied new templates and tmp usage patches

I have left out the changes to existing filters as they are not needed. 
WvText *must* only be used in /tmp only as some versions of it 
incorrectly touches the doc file which can cause looping in trackerd 
with the doc file being constantly reindexed

--
Mr Jamie McCracken
http://jamiemcc.livejournal.com/