What?
A simple guide usable by any Python developer seeking to exploit
SPARQL without hassles.
Why?
SPARQL is a powerful query language, results serialization
format, and an HTTP based data access protocol from
the W3C. It provides a mechanism for accessing and integrating data
across Deductive Database Systems (colloquially
referred to as triple or quad stores in Semantic Web and Linked Data circles) -- database systems
(or data spaces) that manage proposition oriented records in
3-tuple (triples) or 4-tuple (quads) form.
How?
SPARQL queries are actually HTTP payloads (typically). Thus,
using a RESTful client-server interaction pattern, you can dispatch
calls to a SPARQL compliant data server and receive a payload for
local processing e.g. local object binding re. Python.
Steps:
- From your command line execute: aptitude search '^python26', to
verify Python is in place
- Determine which SPARQL endpoint you want to access e.g.
DBpedia or a local Virtuoso instance (typically:
http://localhost:8890/sparql).
- If using Virtuoso, and you want to populate its quad store
using SPARQL, assign "SPARQL_SPONGE" privileges to user
"SPARQL" (this is basic control, more sophisticated WebID based
ACLs are available for controlling SPARQL access).
Script:
#!/usr/bin/env python
#
# Demonstrating use of a single query to populate a # Virtuoso Quad Store via Python.
#
import urllib, json
# HTTP URL is constructed accordingly with JSON query results format in mind.
def sparqlQuery(query, baseURL, format="application/json"):
params={
"default-graph": "",
"should-sponge": "soft",
"query": query,
"debug": "on",
"timeout": "",
"format": format,
"save": "display",
"fname": ""
}
querypart=urllib.urlencode(params)
response = urllib.urlopen(baseURL,querypart).read()
return json.loads(response)
# Setting Data Source Name (DSN)
dsn="http://dbpedia.org/resource/DBpedia"
# Virtuoso pragmas for instructing SPARQL engine to perform an HTTP GET
# using the IRI in FROM clause as Data Source URL
query="""DEFINE get:soft "replace"
SELECT DISTINCT * FROM <%s> WHERE {?s ?p ?o}""" % dsn
data=sparqlQuery(query, "http://localhost:8890/sparql/")
print "Retrieved data:\n" + json.dumps(data, sort_keys=True, indent=4)
#
# End
Output
Retrieved data:
{
"head": {
"link": [],
"vars": [
"s",
"p",
"o"
]
},
"results": {
"bindings": [
{
"o": {
"type": "uri",
"value": "http://www.w3.org/2002/07/owl#Thing"
},
"p": {
"type": "uri",
"value": "http://www.w3.org/1999/02/22-rdf-syntax-ns#type"
},
"s": {
"type": "uri",
"value": "http://dbpedia.org/resource/DBpedia"
}
},
...
Conclusion
JSON was chosen over XML (re. output format) since this is about
a "no-brainer installation and utilization" guide for a Python
developer that already knows how to use Python for HTTP based data
access. SPARQL just provides an added bonus to URL dexterity
(delivered via URI abstraction) with regards to constructing Data
Source Names or Addresses.
Related