[Nepomuk] Future Plans for the Nepomuk Query Parser

[Denis Steckelmacher]

Hi,

Here is a status report of my work on the Nepomuk query parser.

Before my holidays (start started on July 3), I implemented a 
syntax-highlighted query builder using a single-line QTextEdit and the 
syntax-highlighting facilities of Qt.

When I came back on July 13, I replaced this query builder with a new 
one, using colored background rectangles to group terms together. You 
can see how this looks like at [1]. The result is quite nice and I have 
fixed many bugs in the implementation since then.

After that, I added auto-completion facilities to the widget. This was 
quicker than I expected, and in 3 days, I had the first working 
implementation. You can see it at [2].

This week-end, I fixed several bugs in the parser and the query builder 
itself. The auto-completion of tags and contacts is also more 
intelligent, as typing the beginning of a term pre-selects it in the list.

You can find my work in the gsoc2013 branches of the nepomuk-core and 
nepomuk-widget repositories available at [3]. I still have no reply to 
my KDE Developer Account application (done on the KDE Identity website), 
who do I need to contact to see if everything is good ?

For the next week, I plan to fix bugs and document things. I've also 
added a small test case for the parser, that can be extended. Another 
feature I plan to implement is "type-dependent properties". For 
instance, if I type "mails titled "Foo"" and "documents titled "Foo"", 
the "title" property is not the same for the two queries. The first one 
is nmo:messageSubject, the second one is nie:title. Also, don't hesitate 
to point me to ways to improve my code and to make it more in-line with 
all the KDE best-practices.

The problem is that I don't think that all this will take the next 8 
weeks or so before the end of the GSoC.

So, do you have ideas of project I can work on ? My first idea was of 
course to polish the query builder widget and to integrate it into 
Dolphin or other applications, and to bridge the query parser into 
KRunner and Plasma Active. After that, there will be plenty of good ways 
to query the Nepomuk database, and the next step would be to make more 
applications Nepomuk-ready (indexing documents and history in Nepomuk, 
etc)*.

I want to do my best to have Nepomuk used and loved by the users, so I'm 
open to any suggestion that will make it more useable and interesting.

Denis Steckelmacher.

*For instance, it could be interesting to see if it is possible to 
implement a FireFox addon that extracts e-mails and contacts from the 
webmails used by the user (with his or her permission).

[1]: http://steckdenis.be/post-2013-07-16-a-nicer-query-builder-widget.html
[2]: 
http://steckdenis.be/post-2013-07-19-the-journey-to-auto-completion.html
[3]: http://public.steckdenis.be/git/