Generates random flow log data (v2)

generate_flow_log_data.py

import csv
import random
import string

# Load protocols from protocols.csv
protocols = []
with open('protocols.csv', 'r') as protocols_file:
    reader = csv.DictReader(protocols_file)
    for row in reader:
        protocols.append({'number': row['number'], 'name': row['name']})


def generate_ip():
    # Function to generate random IP addresses
    return '.'.join(str(random.randint(0, 255)) for _ in range(4))


def generate_eni():
    # Function to generate random ENI IDs
    return 'eni-' + ''.join(random.choices('0123456789abcdef', k=8))


# Generate lookup.csv with over 10,000 mappings
with open('lookup.csv', 'w', newline='') as lookup_file:
    fieldnames = ['dstport', 'protocol', 'tag']
    writer = csv.DictWriter(lookup_file, fieldnames=fieldnames)
    writer.writeheader()

    for _ in range(10000):
        dstport = random.randint(1, 200)
        protocol = random.choice(protocols)['name']
        tag = 'tag_' + \
            ''.join(random.choices(string.ascii_uppercase + string.digits, k=5))
        writer.writerow({'dstport': dstport, 'protocol': protocol, 'tag': tag})

# Generate flowlogs.txt up to 10 MB
with open('flowlogs.txt', 'w') as flowlogs_file:
    total_size = 0
    max_size = 10 * 1024 * 1024  # 10 MB
    while total_size < max_size:
        # Randomize values
        version = 2
        account_id = ''.join(random.choices(string.digits, k=12))
        eni_id = generate_eni()
        srcaddr = generate_ip()
        dstaddr = generate_ip()
        srcport = random.randint(1, 65535)
        dstport = random.randint(1, 200)
        protocol_entry = random.choice(protocols)
        protocol_num = protocol_entry['number']
        packets = random.randint(1, 100)
        bytes_transferred = packets * random.randint(40, 1500)
        start_time = random.randint(1727104070, 1727204070)
        end_time = start_time + random.randint(1, 600)
        action = random.choice(['ACCEPT', 'REJECT'])
        log_status = 'OK'

        # Create log entry
        log_entry = f"{version} {account_id} {eni_id} {srcaddr} {dstaddr} {dstport} {srcport} {protocol_num} {packets} {bytes_transferred} {start_time} {end_time} {action} {log_status}\n"

        # Write to file and update total_size
        flowlogs_file.write(log_entry)
        total_size += len(log_entry.encode('utf-8'))