def pjk_writer(entries, output_file, bench_freq):
pjk = PajekFactory()
b = Benchmark()
for entry_list in iter(entries.get, 'STOP'):
for entry in entry_list:
for citation in entry["citations"]:
pjk.add_edge(entry["id"], citation)
b.increment()
entries.task_done()
b.print_freq()
with open_file(output_file, "w") as f:
pjk.write(f)
entries.task_done()
help="Directory to store temporary files in",
default=None)
parser.add_argument(
'--subject',
'-s',
help=
"For WoS, subject must include this. Can be a comma seperated list.")
parser.add_argument(
'--wos-only',
help="For WoS, exclude any citations or ids that contain a dot (.)",
action="store_true")
parser.add_argument('infile', nargs='+')
arguments = parser.parse_args()
b = Benchmark()
pjk = PajekFactory(temp_dir=arguments.temp_dir)
subjects = None
if arguments.subject:
subjects = set(arguments.subject.split(","))
for filename in arguments.infile:
with open_file(filename) as f:
for line in f:
entry = ujson.loads(line)
b.increment()
if arguments.wos_only and '.' in entry["id"]:
continue
if subjects:
def setUp(self):
self.output = io.StringIO()
self._node_stream = io.StringIO()
self.pjk = PajekFactory()
