def execute(self, transform_manager, input=None):
inputs = [input] if input else []
inputs += [other(transform_manager) for other in self.others]
transform_manager.start(self, inputs)
inputs = [parse(open(fn)) for fn in inputs]
triples = itertools.chain(*inputs)
with transform_manager('nt') as output:
serialize(triples, output)
return output.name
def execute(self, transform_manager, input=None):
inputs = [input] if input else []
inputs += [other(transform_manager) for other in self.others]
transform_manager.start(self, inputs)
inputs = [parse(open(fn)) for fn in inputs]
triples = itertools.chain(*inputs)
with transform_manager('nt') as output:
serialize(triples, output)
return output.name
def archive(self):
notation = self.notation or hashlib.sha1(self.dataset).hexdigest()
archive_path = os.path.join(SOURCE_DIRECTORY, 'archive', self.store.slug, notation.replace('/', '-'))
archive_graph_name = rdflib.URIRef('{0}archive/{1}'.format(settings.GRAPH_BASE, notation))
data_dump_url = rdflib.URIRef('{0}archive/{1}/{2}/latest.rdf'.format(SOURCE_URL, self.store.slug, notation.replace('/', '-')))
if not os.path.exists(archive_path):
os.makedirs(archive_path, 0755)
nt_fd, nt_name = tempfile.mkstemp('.nt')
rdf_fd, rdf_name = tempfile.mkstemp('.rdf')
try:
nt_out, rdf_out = os.fdopen(nt_fd, 'w'), os.fdopen(rdf_fd, 'w')
for graph_name in self.graph_names:
self._graph_triples(nt_out, graph_name)
nt_out.close()
sort = subprocess.Popen(['sort', '-u', nt_name], stdout=subprocess.PIPE)
try:
triples = itertools.chain(self._get_metadata(rdflib.URIRef(''),
archive_graph_name),
parse(sort.stdout, 'nt').get_triples())
serialize(triples, rdf_out, rdf_name)
finally:
# Make sure stdout gets closed so that if the try block raises
# an exception we don't keep a sort process hanging around.
sort.stdout.close()
sort.wait()
rdf_out.close()
previous_name = os.path.join(archive_path, 'latest.rdf')
# Only update if the file has changed, or hasn't been archived before.
if not os.path.exists(previous_name) or not filecmp._do_cmp(previous_name, rdf_name):
new_name = os.path.join(archive_path,
self.updated.astimezone(pytz.utc).isoformat() + '.rdf')
shutil.move(rdf_name, new_name)
os.chmod(new_name, 0644)
if os.path.exists(previous_name):
os.unlink(previous_name)
os.symlink(new_name, previous_name)
# Upload the metadata to the store using an absolute URI.
metadata = self._get_metadata(data_dump_url, archive_graph_name)
Uploader.upload([self.store], archive_graph_name, graph=metadata)
finally:
os.unlink(nt_name)
if os.path.exists(rdf_name):
os.unlink(rdf_name)
self.filter_old_archives(archive_path)
def run_normalization(self, normalization, triples):
try:
in_file, out_file = [tempfile.NamedTemporaryFile(suffix='.rdf', delete=False) for i in range(2)]
serialize(triples, in_file)
while not normalization.done:
in_file.seek(0)
out_file.seek(0)
pipeline = normalization(parse(in_file).get_triples())
serialize(pipeline, out_file)
out_file.truncate()
in_file, out_file = out_file, in_file
in_file.seek(0)
graph = rdflib.ConjunctiveGraph()
graph.parse(in_file, preserve_bnode_ids=True)
return graph
finally:
in_file.close()
out_file.close()
def execute(self, transform_manager, input):
transform_manager.start(self, [])
endpoint = Endpoint(transform_manager.store.query_endpoint)
for normalization in self.normalizations:
normalization.endpoint = endpoint
normalization.store = transform_manager.store
while self.normalizations:
with open(input, 'r') as source:
pipeline = parse(source).get_triples()
for normalization in self.normalizations:
pipeline = normalization(pipeline)
with open(transform_manager('rdf'), 'w') as target:
serialize(pipeline, target)
input = target.name
self.normalizations = [n for n in self.normalizations if not n.done]
return input
def execute(self, transform_manager, input):
transform_manager.start(self, [])
endpoint = Endpoint(transform_manager.store.query_endpoint)
for normalization in self.normalizations:
normalization.endpoint = endpoint
normalization.store = transform_manager.store
while self.normalizations:
with open(input, 'r') as source:
pipeline = parse(source).get_triples()
for normalization in self.normalizations:
pipeline = normalization(pipeline)
with open(transform_manager('rdf'), 'w') as target:
serialize(pipeline, target)
input = target.name
self.normalizations = [
n for n in self.normalizations if not n.done
]
return input
def run_normalization(self, normalization, triples):
try:
in_file, out_file = [
tempfile.NamedTemporaryFile(suffix='.rdf', delete=False)
for i in range(2)
]
serialize(triples, in_file)
while not normalization.done:
in_file.seek(0)
out_file.seek(0)
pipeline = normalization(parse(in_file).get_triples())
serialize(pipeline, out_file)
out_file.truncate()
in_file, out_file = out_file, in_file
in_file.seek(0)
graph = rdflib.ConjunctiveGraph()
graph.parse(in_file, preserve_bnode_ids=True)
return graph
finally:
in_file.close()
out_file.close()
def archive(self):
notation = self.notation or hashlib.sha1(self.dataset).hexdigest()
archive_path = os.path.join(SOURCE_DIRECTORY, 'archive', self.store.slug, notation.replace('/', '-'))
archive_graph_name = rdflib.URIRef('{0}archive/{1}'.format(settings.GRAPH_BASE, notation))
data_dump_url = rdflib.URIRef('{0}archive/{1}/{2}/latest.rdf'.format(SOURCE_URL, self.store.slug, notation.replace('/', '-')))
data_dump_with_labels_url = rdflib.URIRef('{0}archive/{1}/{2}/latest-with-labels.rdf'.format(SOURCE_URL, self.store.slug, notation.replace('/', '-')))
if not os.path.exists(archive_path):
os.makedirs(archive_path, 0755)
nt_fd, nt_name = tempfile.mkstemp('.nt')
rdf_fd, rdf_name = tempfile.mkstemp('.rdf')
rdf_with_labels_fd, rdf_with_labels_name = tempfile.mkstemp('.rdf')
try:
nt_out, rdf_out = os.fdopen(nt_fd, 'w'), os.fdopen(rdf_fd, 'w')
rdf_with_labels_out = os.fdopen(rdf_with_labels_fd, 'w')
for graph_name in self.graph_names:
self._graph_triples(nt_out, graph_name)
nt_out.close()
with tempfile.TemporaryFile() as sorted_triples:
subprocess.call(['sort', '-u', nt_name], stdout=sorted_triples)
sorted_triples.seek(0)
triples = itertools.chain(self._get_metadata(rdflib.URIRef(''),
data_dump_with_labels_url,
archive_graph_name),
parse(sorted_triples, 'nt').get_triples())
serialize(triples, rdf_out, 'rdf')
rdf_out.close()
sorted_triples.seek(0)
triples = itertools.chain(self._get_metadata(rdflib.URIRef(''),
data_dump_with_labels_url,
archive_graph_name),
self.with_labels(parse(sorted_triples, 'nt').get_triples()))
serialize(triples, rdf_with_labels_out, 'rdf')
rdf_with_labels_out.close()
previous_name = os.path.join(archive_path, 'latest.rdf')
# Only update if the file has changed, or hasn't been archived before.
if not os.path.exists(previous_name) or not filecmp._do_cmp(previous_name, rdf_name):
new_name = os.path.join(archive_path,
self.updated.astimezone(pytz.utc).isoformat() + '.rdf')
shutil.move(rdf_name, new_name)
os.chmod(new_name, 0644)
if os.path.exists(previous_name):
os.unlink(previous_name)
os.symlink(new_name, previous_name)
new_with_labels_name = os.path.join(archive_path, 'latest-with-labels.rdf')
shutil.move(rdf_with_labels_name, new_with_labels_name)
os.chmod(new_with_labels_name, 0644)
# Upload the metadata to the store using an absolute URI.
metadata = self._get_metadata(data_dump_url, data_dump_with_labels_url, archive_graph_name)
Uploader.upload([self.store], archive_graph_name, graph=metadata)
finally:
os.unlink(nt_name)
if os.path.exists(rdf_name):
os.unlink(rdf_name)
self.filter_old_archives(archive_path)
