Re: [Tracker] Keywords from pdf dokuments

[Martyn Russell <martyn lanedo com>]

On 30/12/10 21:03, Joerg Beyer wrote:
> Hello,

Hi,

> I have Tracker 0.8.17 on ubuntu running - everything is fine. But I
> can't find keywords from pdf documents in the index. Pdf keywords are
> something like tags. When I index a pdf like this:
>
> /usr/lib/tracker/tracker-extract -v 3 -f example.pdf
>
> then I see this among the verbose output:
> ...
>       nao:hasTag [ a nao:Tag ;
>       nao:prefLabel "myFirstTag"] ;
>       nao:hasTag [ a nao:Tag ;
>       nao:prefLabel "mySecondTag"] ;
> ...
>
> This looks fine - but the keywords seem not to be in the index. This here
>
> tracker-info example.pdf | grep myFirstTag
>
> has no outout.

It is unlikely that you would see any output. Tags are not stored 
according to their preferred label. Using tracker-info, you would see 
the tags listed as something like:

  'nao:hasTag' = 'urn:uuid:f2f41315-51d9-df02-0dce-1448b8b24d4f'
  'nao:hasTag' = 'urn:uuid:97308668-8c4a-4b74-2ada-a65306f5e85c'

If you use tracker-info on the 'urn:uuid:...' string then you can see 
what the details of the tag are.

To make this all easier, there is tracker-tag. You can use man 
tracker-tag to see how to use it, to do what you're asking should work 
using:

  tracker-tag example.pdf

This will list all tags for that filename. Alternatively, you can use 
tracker-tag -st which will list all tags and all files per tag.

> Is there a way to search for pdf documents by their keywords?

Well, you can search for *any* documents using keywords:

  tracker-tag -st myFirstTag

For PDFs specifically, you would have to use a SPARQL query with 
tracker-sparql. Something like:

  tracker-sparql -q 'select nie:url(?file) where { ?file a 
nfo:FileDataObject ; nie:mimeType "application/pdf" ; nao:hasTag ?tag . 
?tag nao:prefLabel "myFirstTag" }'

Hope this helps.

--
Regards,
Martyn