Differences between revisions 10 and 11
Revision 10 as of 2013-02-02 14:18:59
Size: 3829
Comment: RDF/XML to Turtle
Revision 11 as of 2013-02-19 10:27:03
Size: 3998
Comment: Add reference paper
Deletions are marked like this. Additions are marked like this.
Line 37: Line 37:
See http://www-public.it-sudparis.eu/~berger_o/weblog/2012/08/29/debian-package-tracking-system-now-produces-rdf-description-of-source-packages/ for more details and a pointer to a (clickable) diagram illustrating this. See the following resources for more details and a pointer to a (clickable) diagram illustrating this :
 * http://www-public.it-sudparis.eu/~berger_o/weblog/2012/08/29/debian-package-tracking-system-now-produces-rdf-description-of-source-packages/
 * [[http://www-public.telecom-sudparis.eu/~berger_o/papier-swese2012/|Linked Data descriptions of Debian source packages using ADMS.SW]]

The PTS has a Linked Data / RDF interface.

See http://packages.qa.debian.org/common/RDF.html for the "RDF homepage" of the PTS.

The PTS produces RDF descriptions (as Turtle and RDF/XML) of Debian source packages (based on the same reference version as the PTS, i.e. usually the "latest" version in unstable).

Every package has a "stable" reference RDF document available (through content-negociation) at the "same" place as the human-readable variant, at http://packages.qa.debian.org/SRC-PACKAGE-NAME.

In practive, the HTML variant is generated (statically) by the PTS at http://packages.qa.debian.org/PREFIX/SRC-PACKAGE-NAME.html and the RDF variant at http://packages.qa.debian.org/PREFIX/SRC-PACKAGE-NAME.[ttl|rdf] (the PREFIX being computed depending on the source package name first letters.

Content-negociation for RDF/XML is classically done with the 'text/turtle' 'application/rdf+xml' content type requested (in the HTTP 'Accept' header.

To view the Turtle document, issue :  $ curl -L -H 'Accept: text/turtle'  http://packages.qa.debian.org/SRC-PACKAGE-NAME

For other format, use 'rapper', for instance (from package 'raptor2-utils') :  $ rapper -o nquads http://packages.qa.debian.org/SRC-PACKAGE-NAME

The RDF data uses the ADMS.SW 1.0 ontology, which includes DOAP properties and relations.

The structure of the RDF model is the following :

  • every source package is modeled as interlinked resources :
    • a "source packaging project" (SoftwareProject) resource : <http://packages.qa.debian.org/SRC-PACKAGE-NAME>

    • "source package versions" (SoftwareRelease) resources for every version of the source package known by the PTS (present in the different suites) : <http://packages.qa.debian.org/SRC-PACKAGE-NAME#SRC-PACKAGE-NAME_DEBIAN-PKG-VERSION>

      • For one of these only (the "latest" one known by the PTS) there's, in that SoftwareRelease 2 includedAsset SoftwareReleases :

        • one for the upstream sources <http://packages.qa.debian.org/SRC-PACKAGE-NAME#upstreamsrc_UPSTREAM-VERSION>

          • with resources for all archives of upstream sources as SoftwarePackages (typically SRC-PACKAGE-NAME_UPSTREAM-VERSION.orig.tar.gz) at URIs like <http://packages.qa.debian.org/SRC-PACKAGE-NAME#FILENAME>

        • one for the Debian packaging files <http://packages.qa.debian.org/SRC-PACKAGE-NAME#debiansrc_DEBIAN-PKG-VERSION>

          • with resources for all files comprising the Debian packaging source archive (typically SRC-PACKAGE-NAME_DEBIAN-PKG-VERSION.debian.tar.gz) at URIs like <http://packages.qa.debian.org/SRC-PACKAGE-NAME#FILENAME>

        • and a resource pointing to the Upstream release (SoftwareRelease) for that version with a URI in the form <http://packages.qa.debian.org/SRC-PACKAGE-NAME#upstreamsrc_UPSTREAM-VERSION>

    • one resource for the .dsc file at a URI like <http://packages.qa.debian.org/SRC-PACKAGE-NAME#SRC-PACKAGE-NAME_DEBIAN-PKG-VERSION.dsc>

    • one resource for the Upstream project (SoftwareProject) <http://packages.qa.debian.org/SRC-PACKAGE-NAME#upstream> whith a doap:homepage if it's known by the PTS

    • resources (blank nodes at the moment) for the contributors (maintainers)
    • resources for the Ubuntu packaging project and latest release known by the PTS as SoftwareProject and SoftwareRelease

See the following resources for more details and a pointer to a (clickable) diagram illustrating this :

A full dump of all the meta-data is available (to Debian members) on packages.qa.debian.org:/srv/packages.qa.debian.org/www/web/full-dump.ttl

This is all work in progress...