dataset_aggragation.py

import networkx as nx
import os


def get_edge_list(file_name):
    with open(file_name) as ifs:
        lines = ifs.readlines()
        edge_list = map(lambda line: map(int, line.strip().split()), filter(lambda ele: '#' not in ele, lines))
    return edge_list


def get_undirected_graph_info(file_name):
    my_edge_list = get_edge_list(file_name)
    undirected_graph = nx.Graph()
    undirected_graph.add_edges_from(my_edge_list)
    file_info = file_name.split(os.sep)[-1].split('_')[0]
    str_list = [file_info, 'nodes num:' + str(undirected_graph.number_of_nodes()), 'edges num:'
                + str(undirected_graph.number_of_edges())]
    print ' | '.join(str_list)


def get_dir_info(dir_name):
    my_walk = os.walk(dir_name)
    my_root, sub_root_list, file_list = list(my_walk)[0]
    path_list = map(lambda ele: my_root + os.sep + ele, file_list)
    for my_path in path_list:
        get_undirected_graph_info(my_path)


if __name__ == '__main__':
    get_dir_info('../Datasets')