Parser for network data

parse_net.py

#!/bin/python3

import re
from pprint import pprint

# Parameters
in_file = "nw-metrics"
out_file = "nw-parsed"
delimiters = '["\n ]'

# Dictionaries to store sums
source_data = {}
dest_data = {}

# Open file and calculate sums for destination and source
with open(in_file, "r") as rfile:
    for line in rfile:
        # Abuse the fact that the data is structured and split according to
        # the delimiters defined above
        parsed = re.split(delimiters, line)

        # The values in parsed after splitting are like:
        # ['pcap_bytes_transfered{DestinationAddress=', '103.174.252.216',
        # ',SourceAddress=', '158.146.78.104', '}', '179028', '']
        # index 1 --> Destination IP
        # index 3 --> Source IP
        # index 5 --> Data
        dest_ip = parsed[1]
        source_ip = parsed[3]
        data = int(parsed[5])

        # Calculate sum of destination ip's data
        if dest_ip in dest_data:
            dest_data[dest_ip] += data
        else:
            dest_data[dest_ip] = data

        # Calculate sum of source ip's data
        if source_ip in source_data:
            source_data[source_ip] += data
        else:
            source_data[source_ip] = data

# Open file and print parsed data to it
with open(out_file, "w") as ofile:
    for ip, data in dest_data.items():
        out_data = f'pcap_bytes_transfered{{DestinationAddress="{ip}"}} {data}\n'
        ofile.write(out_data)

    for ip, data in source_data.items():
        out_data = f'pcap_bytes_transfered{{SourceAddress="{ip}"}} {data}\n'
        ofile.write(out_data)