def process_file(file):
root, filename = os.path.split(file)
rdf_file = os.path.splitext(filename)[0]+'.nt'
rdf_path = root + '/' + rdf_file
clean_file(file)
try:
xml = etree.parse(file)
rdf = transform(xml)
g = Graph()
g.parse(StringInputSource(rdf),"xml")
# If the graph already exists then we want to generate some diffs before overwriting it: these can be used generating changesets when uploading to a datastore
existing = False
if os.path.exists(rdf_path):
print "Comparing graphs"
go = Graph()
go.parse(rdf_path,format='nt')
existing = True
elif os.path.exists(root+'/archive/'+rdf_file):
print "Comparing with archived graph"
go = Graph()
go.parse(root+'/archive/'+rdf_file,format='nt')
existing = True
if existing:
both, old, new = graph_diff(go,g)
if(len(old)):
# old.serialize(rdf_path+'_old',format='nt') #Uncomment if you want a non-reified version of the statements
cs = BatchChangeSet()
cs.setCreatorName('IATI Update Scripts')
cs.setChangeReason('Statements to remove from'+file)
for (s,p,o) in old.triples((None, None, None)):
cs.remove(s,p,o)
cs.getGraph().serialize(rdf_path+'_csremove',format='nt')
if(len(new)):
# new.serialize(rdf_path+'_new',format='nt') #Uncomment if you want a non-reified version of the statements
cs = BatchChangeSet()
cs.setCreatorName('IATI Update Scripts')
cs.setChangeReason('Statements to add from '+file)
for (s,p,o) in new.triples((None, None, None)):
cs.remove(s,p,o)
cs.getGraph().serialize(rdf_path+'_csadd',format='nt')
g.serialize(rdf_path,format='nt')
except Exception, e:
print "Error processing file "+ file
print e
try:
og.parse(rdf_path,format='nt')
except Exception, e:
print "Failed reading archived online data"
print e
print "Running graph diff - new data against archived data"
both, old, new = graph_diff(og,ng)
if(len(old)):
cs = BatchChangeSet()
cs.setCreatorName('R4D Update Scripts')
cs.setChangeReason('Statements to remove from'+dirList[0])
for (s,p,o) in old.triples((None, None, None)):
cs.remove(s,p,o)
print "Saving triples for removal to changeset"
cs.getGraph().serialize(rdf_path+'_csremove',format='nt')
if(len(new)):
cs = BatchChangeSet()
cs.setCreatorName('IATI Update Scripts')
cs.setChangeReason('Statements to add from '+dirList[0])
for (s,p,o) in new.triples((None, None, None)):
cs.remove(s,p,o)
print "Saving new triples to changeset"
cs.getGraph().serialize(rdf_path+'_csadd',format='nt')
archive(exec_path+"/"+dirList[0],True)
ng.serialize(rdf_path,format='nt')
except Exception, e:
print "Could not read RDFXML file: "
print e