SPARQL endpoint

Introduction

The filesystem indexer SPARQL endpoint can be contacted at the org.freedesktop.Tracker3.Miner.Files D-Bus name, use the Tracker SPARQL library to set up a connection to it.

#!/bin/env python3
import gi
gi.require_version('Tracker', '3.0')
from gi.repository import Tracker
conn = Tracker.SparqlConnection.bus_new('org.freedesktop.Tracker3.Miner.Files')

$ tracker3 sparql -b org.freedesktop.Tracker3.Miner.Files -q \
    "SELECT ('hello world' AS ?str) { }"

You should not provide access within flatpaks to this D-Bus name. Access from within the sandbox is handled through the Tracker portal.

Data is offered in the format of the Nepomuk ontology, see the Nepomuk documentation to learn about the different classes and properties that may be handled through this ontology.

Graphs

The filesystem indexer uses the following graphs to store data:

  • tracker:FileSystem
  • tracker:Documents
  • tracker:Audio
  • tracker:Video
  • tracker:Pictures
  • tracker:Software

See how the Tracker portal uses graphs as the data units to perform isolation between clients.

The tracker:FileSystem graph contains all nfo:FileDataObject objects representing the filesystem structure, and vertebrates all the remaining "content" graphs. All other graphs contain the nie:InformationElement objects that represent the various media (documents, videos, ...), each of those graphs gets populated only with the relevant media, and miscellaneous data (e.g. audio albums and performers)

This way to partition the data is enabled by the split between interpretation and content present in the Nepomuk ontology, read more about its design.