This is an old library that I maintain for an old project. I updated the dependencies and modified it to run on Python 3.6+. Only rdflib and RDF4J have been tested; Allegro Graph is not working and I have no interest in making it work at this time. Also, I haven't pushed it to PyPI — please feel free to do so if interested. Credit for the library should go to the original implementor.

Sparrow is a library that provides a high-level API for RDF Databases. Any database the provides support for SPARQL Queries and has a triplestore that can handle contexts can be used as a backend.

The goal of Sparrow is to make sure all these different backends act the same, making it possible to change RDF Database backends without having to change your code.

At the moment there is support for the following backends:

• Redland / librdf
• RDFLib
• Sesame / openrdf
• Allegro Graph

The API provides support for the following basic functions:

• Parsing RDF statements from different formats into a contextual database
• Serializing the RDF statements from a specific context of a database
• Removing statements from a specific context of a database
• Performing SPARQL Queries

Sparrow does not provide a Graph API, but it can parse and serialize python dictionary objects. This uses the same format as the JSON serialization. Each TripleStore backend provides the following formats:

• RDFXML
• NTriples
• Turtle
• JSON
• Python dictionaries

Sparrow comes with buildout profiles for several databases. These profiles will install and setup the different backends for you. You don't have to use buildout, but I would recommend it.

To install you need Pipenv. Having installed the latter, type:

> pipenv -dev -e .

This will create some scripts in the bin folder like a testrunner and (depending on which profile you choose) scripts for configuring and starting the different backends.

Normally, you will only need to import the base sparrow module

>>> import sparrow

Most of the database backends will not work out of the box. Since the RDFLib backend is written in python and packaged on pypi, it is always available, and installed with Sparrow.

Let's create an in memory rdflib database

>>> db = sparrow.database('rdflib', 'memory')
>>> db
<sparrow.rdflib_backend.RDFLibTripleStore ...>

Let's add some triples to the database, we will use turtle syntax for this. We'll make some example statements where we will state that john is a person, and that his firstname is "John".

>>> data = """@prefix ex: <http://example.org#> .
... ex:john a ex:Person; ex:name "John" ."""

Now we can add this to the database. We will need to tell the database in which context to store the data. The data itself can be either a file or http based URI, a string of data, or a file-like object.

So, let's add this to the persons context.

>>> db.add_turtle(StringIO(data), 'persons')

We can now ask the database, which contexts it has:

>>> db.contexts()
['persons']

You can store data in as many different contexts as you like, or put everything in a single context.

Lets do a simple SPARQL query on the database

>>> result = db.select('SELECT ?x {?x <http://example.org#name> "John".}')

We can get the results as a list of dictionaries. This follows the SPARQL JSON result format.

>>> assert result == [{'x': {'type': 'uri', 'value': 'http://example.org#john'}}]

Besides querying, we can also get the data back from the database in any of the supported formats. We specify which format we want, and which context to use.

>>> db.get_ntriples('persons').read()
'<http://example.org#john> ...'

If the database backend supports it, you can ask how many triples are in a context.

>>> db.count('persons')
2

If you want to remove triples, you will need to supply data describing which triples to remove.

>>> data = '<http://example.org#john> a <http://example.org#Person>.'
>>> db.remove_turtle(data, 'persons')
>>> db.count('persons')
1

You can also remove all triples in a context

>>> db.clear('persons')
>>> db.count('persons')
0

Since the 'persons' context is now empty, it is also removed.

>>> db.contexts()
[]