memgraph/tools/plot_througput

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
#
# Example usage:
# ./plot_througput --vendor-references neo4j memgraph           \
#                  --vendor-titles neo4j memgraph               \
#                  --results neo_to_plot memgraph_to_plot       \
#                  --plot-title "Pokec small" --window-size 10


import json
import os
import matplotlib.pyplot as plt
from matplotlib.cbook import get_sample_data
from argparse import ArgumentParser

COLORS = {
    'memgraph': '#ff7300',
    'neo4j': '#008cc2'
}


def parse_args():
    argp = ArgumentParser(description=__doc__)
    argp.add_argument("--vendor-references", nargs="+",
                      help="Short references that represent all the "
                           "vendors that are going to be "
                           "visualized on the plot.")
    argp.add_argument("--vendor-titles", nargs="+",
                      help="Vender titles that are going to appear "
                           "on the plot, e.g. legend titles.")
    argp.add_argument("--results", nargs="+",
                      help="Result files for each vendor")
    argp.add_argument("--plot-title", default="{{Plot title placeholder}}",
                      help="Plot title.")
    argp.add_argument("--window-size", type=int, default=1)
    return argp.parse_args()


def main():
    # Read the arguments.
    args = parse_args()

    # Prepare the datastructure.
    vendors = {}
    for vendor_reference, vendor_title, vendor_results in \
            zip(args.vendor_references, args.vendor_titles, args.results):
        vendors[vendor_reference] = {}
        vendors[vendor_reference]['title'] = vendor_title
        vendors[vendor_reference]['results_path'] = vendor_results
        vendors[vendor_reference]['color'] = COLORS[vendor_reference]
        vendors[vendor_reference]['t'] = [0.0]
        vendors[vendor_reference]['q'] = [0]
        vendors[vendor_reference]['dq/dt'] = [0.0]

    fig, ax = plt.subplots()
    ax.set_ylabel('Throughput (queries per second)')
    ax.set_xlabel('Time (seconds)')
    ax.set_title(args.plot_title)
    ax.set_aspect(0.01)

    # Collect the benchmark data and plot lines.
    print("Pokec throughput")
    for vendor_reference, vendor_data in vendors.items():
        print("Vendor: %s" % vendor_reference)
        with open(vendor_data['results_path']) as results_file:
            # Skip first line which contains titles.
            prev_time, prev_num_queries  = 0.0, 0
            for line in results_file.readlines()[1:]:
                data = line.split()
                if data == []: break
                assert len(data) == 2, "Invalid data"
                vendor_data['t'].append(float(data[0]))
                vendor_data['q'].append(int(data[1]))
            for i in range(1, len(vendor_data['t'])):
                j = max(0, i - args.window_size)
                vendor_data['dq/dt'].append(
                        (vendor_data['q'][i] - vendor_data['q'][j]) /
                        (vendor_data['t'][i] - vendor_data['t'][j]))

        line1, = ax.plot(vendor_data['t'], vendor_data['dq/dt'], '-', linewidth=2,
                         label=vendor_data['title'], color=vendor_data['color'])

    ax.legend(loc='lower right')
    plt.grid()
    plt.show()

if __name__ == '__main__':
    main()