osm_editor_graphs.py

#!/usr/bin/env python3

import argparse
import os
import re

from matplotlib import pyplot


def parse_num(num_str):
    """
    Convert a string into a number.
    The string may contain spaces.
    """
    return int(num_str.replace(' ', ''))


def read_files(editor, files):
    """
    Read *.lst files and extract relevant usage information about editor.
    Returns lists of tuples (date, number) for changesets, user, and edits statistics.
    """
    by_changesets = []
    by_users = []
    by_edits = []
    for filename in files:
        with open(filename) as f:
            text = f.read()
            results = re.findall(editor + r": ([\d\s]+)\(\d+.\d+%\)", text)
            date = os.path.splitext(os.path.basename(filename))[0]
            by_changesets.append((date, parse_num(results[0])))
            by_users.append((date, parse_num(results[1])))
            by_edits.append((date, parse_num(results[2])))
    by_changesets.sort(key=lambda x: x[0])
    by_users.sort(key=lambda x: x[0])
    by_edits.sort(key=lambda x: x[0])
    return by_changesets, by_users, by_edits


def plot_data(data):
    """
    Plot and return a graph out of a list of tuples (x-label, y-point).
    """
    fig, ax = pyplot.subplots(figsize=(13, 5))
    x_data = range(len(data))
    y_data = [x[1] for x in data]
    ax.plot(x_data, y_data, marker='o', linestyle='dashed')
    for i, j in zip(x_data, y_data):
        ax.annotate(str(j), xy=(i, j), xytext=(-10, 4), textcoords='offset points')
    ax.set_xticks(range(len(data)))
    ax.set_xticklabels([x[0] for x in data])
    return fig, ax


def main(args):
    """
    Main function of editor graph creation script.
    Read some .lst files, plot graphs, export to files.
    """
    by_changesets, by_users, by_edits = read_files(args.editor, args.files)

    pyplot.rc('font', size=11)

    fig, ax = plot_data(by_changesets)
    fig.suptitle(f"Number of Changesets done with {args.editor}")
    ax.set_xlabel("Date")
    ax.set_ylabel("Number of Changesets")
    fig.savefig(os.path.join(args.outdir, f"{args.editor}_changesets.{args.format}"))

    fig, ax = plot_data(by_users)
    fig.suptitle(f"Distinct Users of {args.editor}")
    ax.set_xlabel("Date")
    ax.set_ylabel("Number of Distinct Users")
    fig.savefig(os.path.join(args.outdir, f"{args.editor}_users.{args.format}"))

    fig, ax = plot_data(by_edits)
    fig.suptitle(f"Number of Edits done with {args.editor}")
    ax.set_xlabel("Date")
    ax.set_ylabel("Number of Edits")
    fig.savefig(os.path.join(args.outdir, f"{args.editor}_edits.{args.format}"))


if __name__ == '__main__':
    argparser = argparse.ArgumentParser(description="Create usage graphs for a given OpenStreetMap editor")
    argparser.add_argument('-e', '--editor', required=True, help="OSM editor to create graphs for")
    argparser.add_argument('-f', '--format', choices=['png', 'pdf', 'svg'], default='png', help="graph file format")
    argparser.add_argument('-o', '--outdir', default='./', help="output folder for graphs")
    argparser.add_argument('files', nargs='+', help="the .lst files generated by the editor-stats tool")
    parsed_args = argparser.parse_args()
    main(parsed_args)