Convert network with string node ids to integer ids with node names in the Pajek format

convert_network_string_ids_to_integer.py

import pandas as pd

def convert_to_integer_pajek(input_file, output_file, columns=['from', 'to', 'weight']):
    """
        Convert network in link list format with string node ids to Pajek format with integer node ids
        Assumes first row is a header with names matching `columns`
    """
    net = pd.read_csv(input_file, delimiter=' ').dropna()
    nodes = set(net[columns[0]])
    nodes.update(net[columns[1]])
    nodes = dict(zip(nodes, range(1, len(nodes)+1)))
    print(f"Convert network with {len(nodes)} nodes and {net.shape[0]} links...")
    with open(output_file, 'w') as fp:
        fp.write(f"*Vertices {len(nodes)}\n")
        for name, node_id in nodes.items():
            fp.write(f"{node_id} \"{name}\"\n")
        fp.write(f"*Edges\n")
        for index, row in net.iterrows():
            if index % 1000:
                print(f"{index/net.shape[0]:.2%}", end='\r')
            fp.write(f"{nodes[row[columns[0]]]} {nodes[row[columns[1]]]} {row[columns[2]]}\n")
    print(f"Done! Wrote '{output_file}'.")