amps.py

#!/usr/bin/env python

"""
Look for odd ampersands in URIs in O'Reilly data and print out the 
context URI (where the assertions came from). 

Output looks something like:
ed@inkdroid:~/bzr/oreilly-crawler$ ./amps.py 
<http://oreilly.com/catalog/0636920001744/> a rdfg:Graph;rdflib:storage [a rdflib:Store;rdfs:label 'Sleepycat'].
<http://oreilly.com/catalog/9781593271749/> a rdfg:Graph;rdflib:storage [a rdflib:Store;rdfs:label 'Sleepycat'].
<http://oreilly.com/catalog/9781933952246/> a rdfg:Graph;rdflib:storage [a rdflib:Store;rdfs:label 'Sleepycat'].
"""

import rdflib.graph
import rdflib.term

g = rdflib.graph.ConjunctiveGraph('Sleepycat')
g.open('store')

for s, p, o in g:
    has_amp = False
    if ' & ' in s:
        has_amp = True
    elif type(o) == rdflib.term.URIRef and ' & ' in o:
        has_amp = True

    if has_amp:
        for context in g.contexts((s, p, o)):
            print context

g.close()