def test_build_node_dict(file_r_name, file_weights_name, path_w):
# default
with open(file_r_name, 'r') as file_json, \
open(path_w + 'nodes.json', 'w') as file_w:
node_dict = lem.build_node_dict(file_json)
pretty_json(node_dict, file_w)
# different options
path_w += 'options/'
if not os.path.exists(path_w):
os.makedirs(path_w)
weights_options = (None, True)
bastard_options = (True, False)
non_cyrillic_options = (True, False)
for weights in weights_options:
for bastard in bastard_options:
for non_cyrillic in non_cyrillic_options:
with open(file_r_name, 'r') as file_json:
if weights:
file_weights = open(file_weights_name, 'r')
else:
file_weights = None
file_w_name = '{0}nodes (' \
'weights={1}, ' \
'bastard={2}, ' \
'non_cyrillic={3}' \
').json'.format(path_w, weights,
bastard, non_cyrillic)
file_w = open(file_w_name, 'w')
node_dict = lem.build_node_dict(file_json, file_weights,
bastard, non_cyrillic)
pretty_json(node_dict, file_w)
print('Done test_build_node_dict')
def test_write_node_dict(file_r_name, path_w):
# build default node_dict
with open(file_r_name, 'r') as file_json:
node_dict = lem.build_node_dict(file_json)
# default
with open(path_w + 'nodes.csv', 'w') as file_w:
lem.write_node_dict(file_w, node_dict)
# different options
path_w += 'options/'
if not os.path.exists(path_w):
os.makedirs(path_w)
# with cut
with open(path_w + 'nodes (cut=10).csv', 'w') as file_w:
lem.write_node_dict(file_w, node_dict, cut=10)
# with '\t'
with open(path_w + 'nodes.tsv', 'w') as file_w:
lem.write_node_dict(file_w, node_dict, sep='\t')
print('Done test_write_node_dict')
# рассмотренные ранее. Если не указать файл с весами, то каждой строке будет
# присвоен единичный вес.
file_nodes_name = path_w + 'nodes.csv'
file_edges_name = path_w + 'edges.csv'
file_nodes_json_name = path_w + 'nodes.json'
file_edges_json_name = path_w + 'edges.json'
# Создаём узлы
with open(file_json_name, 'r') as file_json, \
open(file_weight_name, 'r') as file_weight, \
open(file_nodes_name, 'w') as file_nodes, \
open(file_nodes_json_name, 'w') as file_nodes_json:
# создаём словарь узлов
node_dict = lem.build_node_dict(file_json, weights=file_weight,
include_bastard=bastard,
include_non_cyrillic=non_cyrillic)
# записываем узлы в файл для Gephi
lem.write_node_dict(file_nodes, node_dict)
# отдельно сохраняем json-версию словаря узлов (для отладки)
pretty_json(node_dict, file_nodes_json)
# Создаём рёбра
with open(file_json_name, 'r') as file_json, \
open(file_weight_name, 'r') as file_weight, \
open(file_edges_name, 'w') as file_edges, \
open(file_edges_json_name, 'w') as file_edges_json:
# создаём словарь рёбер
edge_dict = lem.build_edge_dict(file_json, weights=file_weight,
include_bastard=bastard,
include_non_cyrillic=non_cyrillic)