network-sessionto-uml.py

"""
Network sequence geneator from network session
"""

import pandas as pd
import numpy as np

from urllib.parse import urlparse

csv_file_path = "xxxx.csv"
output_filename = "output-uml.txt"
client_addess = '/192.168.x.y'

data = pd.read_csv(
    csv_file_path,
    header=None,
    names=['URL', 'Status', 'ResponseCode', 'Protocol', 'Method', 'ContentType', 'ClientAddress', 'ClientPort', 'RemoteAddress', 'RemotePort', 'Exception', 'RequestStartTime', 'RequestEndTime', 'ResponseStartTime', 'ResponseEndTime', 'Duration', 'DNSDuration', 'ConnectDuration', 'SSLDuration', 'RequestDuration', 'ResponseDuration', 'Latency', 'Speed', 'RequestSpeed', 'ResponseSpeed', 'RequestHandshakeSize', 'RequestHeaderSize', 'RequestBodySize', 'ResponseHandshakeSize', 'ResponseHeaderSize', 'ResponseBodySize', 'RequestCompression', 'ResponseCompression'],
    skiprows=1
).fillna(value=0)

data = data[data['ClientAddress'] == client_addess]
data['RequestStartTime'] = pd.to_datetime(data['RequestStartTime'])
data['ResponseEndTime'] = pd.to_datetime(data['ResponseEndTime'])
data['RemoteAddress'] = data['RemoteAddress'].apply(lambda x: x.split('/')[0])

data.index = data['RequestStartTime']
domain_groups = list(data.groupby('RemoteAddress').groups.keys())

document = ""

for domain in domain_groups:
    target = data[data['RemoteAddress'] == domain]
    document += f"\r## {domain}"

    completed_request = target[target['Status'] == 'COMPLETE']
    availability = completed_request[completed_request['ResponseCode'] < 500].shape[0] / (1 if completed_request.shape[0] <= 0 else 0 + completed_request.shape[0]) * 100
    avg_latency = np.mean(completed_request['Duration'])
    avg_request_size = np.mean(completed_request['RequestHeaderSize'] + completed_request['RequestBodySize'] + completed_request['RequestHandshakeSize'])
    avg_response_size = np.mean(completed_request['ResponseHeaderSize'] + completed_request['ResponseBodySize'] + completed_request['ResponseHandshakeSize'])

    document += f"""
    \r<details><summary>シーケンスを表示</summary>\n
    \rAvailability: `{round(availability, 3)} %`
    \rAverage Latency: `{round(avg_latency, 1)} ms`
    \rAverage Request Size: `{round(avg_request_size, 1)} bytes`
    \rAverage Response Size: `{round(avg_response_size, 1)} bytes`\n
    \r```uml
    \rskinparam monochrome false
    """

    prev_record = None
    prev_concurent = False
    records = target.to_dict('records')
    records_size = len(records)

    if records_size < 2:
        continue

    for index, record in enumerate(records):
        if record['Status'] != 'COMPLETE' or record['Protocol'] != 'https':
            continue

        # 前回リクエストが完了する前に今回リクエストが開始されている場合は少なくとも直列ではない
        if prev_record:
            is_concurrent = (record['RequestStartTime'] <= prev_record['ResponseEndTime']) or (False if index >= records_size - 1 else records[index + 1]['RequestStartTime'] <= record['ResponseEndTime'])
        else:
            is_concurrent = records[index+1]['RequestStartTime'] <= record['ResponseEndTime']

        header = "group Parallel\n" if not prev_concurent and is_concurrent else ""
        footer = "end" if prev_concurent and not is_concurrent else ""

        prev_record = record
        prev_concurent = is_concurrent

        document += f'''
        {header}
        "Client" -> "{domain}": {urlparse(record['URL']).path} {record['Method']}
        Note over of "{domain}": {record['RequestStartTime']} takes {record['Duration']} ms
        "Client" <-- "{domain}": {record['ResponseCode']}
        {footer}
        '''

    document += "\r```\n</details>\n\n"

with open(output_filename, "w") as f:
    f.write(document)