[Nepomuk] Nepomuk HTTP server

[Martin Klapetek]

Hey,

On Sat, Sep 8, 2012 at 12:28 PM, Bogdan Bivolaru
<bogdan.bivolaru at gmail.com>wrote:

> Hello,
> I'm looking for a way to integrate my browsers (Firefox, Chormium and
> sometimes Rekonq) and my torrent client (Ktorrent) with Nepomuk.
> I want to link a file downloaded to the site I got it from and if it is a
> movie or a song I would like to extract author information from the
> website. It is likely that the site contains in the download page some
> information about the performers in the downloaded movie/song and I would
> like to save that information to the file.
>
> I have thought of two solutions to this problem:
>  1) in Firefox, make a native plugin using NPAPI (
> https://developer.mozilla.org/en-US/docs/Gecko_Plugin_API_Reference); I
> suppose other browsers also have such a native code API.
> This is hard to do because someone has to write and maintain the plugin
> code for every browser in use.
>

I was looking at this few months ago with Vishesh (unfortunately we missed
the one by Sebastian), writing a NPAPI plugin is imho too much effort, so I
went with standard extension[1]. Also Chrome has this new PAPI and they are
going to(?) ditch the NPAPI, in turn Firefox is not going to support PAPI.
Anyways, Chrome/Chromium missed the proper downloads API for extensions, so
it was not possible to implement it in the same way as the firefox one.


>  2) An easier task would be to make a http server that listens on
> localhost and receives and answers JSON documents.
> I would then be able to write extensions in HTML which query and add new
> properties to resources. Most/all of the HTML/JS code for the extensions
> would run without changes on all browsers.
>

Basically what we did was the extension calling a simple binary file that
writes into Nepomuk, which is imho way easier than implementing/keeping
running a http server. See [1].


>
> UPDATE:
>  3) From what I remember reading in an article about a year ago, Zeitgeist
> is to be integrated with Nepomuk and zeitgeist-datasources are to provide
> such functionality. What is the status of the integration of Zeitgeist in
> KDE?
> Zeitgeist data providers (
> http://bazaar.launchpad.net/~zeitgeist-dataproviders/zeitgeist-datasources/git/files) support
> a lot of applications!
> /UPDATE
>

The main problem here is the Chrome/Chromium API, which didn't provide any
info on downloaded files, so there you'd have to go with the PAPI plugin I
guess. There was also a way to write Qt NPAPI plugins, but I haven't looked
much into that.

This second solution seems pretty easy, but I have some questions:
>  4) has anyone else started such a project?
> I have googled "Nepomuk JSON" on the net and I came out with a GWT
> application (https://bitbucket.org/pombredanne/webannotator/), an Eclipse
> application (
> http://dev.nepomuk.semanticdesktop.org/wiki/EclipseDevelopment) and a
> server called SWIM. I have also found a dead link to the home page of a
> Mandriva software. So in my opinion, the existing http server software is
> unmantained, undocumented and/or does not do what I need.
>
>  5) putting on a network a server which knows about most of the files in
> my user's home directory might pose a security risk, doesn't it?
>

Sort of, but if it would be a write-only server, then probably a low-level
one (I'm no security expert).


>  6) Do you have more up to date documentation? Do you have any advice?
>
>  7) Querying the Nepomuk database also queries the Zeitgeist DB?
>

AFAIK Zeitgeist pushes all data to Nepomuk and Nepomuk does not query
Zeitgeist back, but I'm not sure about the current state, so might be wrong.

[1] -
http://martys.typepad.com/blog/2012/02/so-you-want-to-keep-the-url-of-downloaded-file-eh.html

-- 
Martin Klapetek | KDE Developer
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.kde.org/pipermail/nepomuk/attachments/20120909/7a59c77d/attachment.html>