|
| 1 | +#!/usr/bin/env python |
| 2 | + |
| 3 | +from argparse import ArgumentParser |
| 4 | +from concurrent.futures import ProcessPoolExecutor |
| 5 | +import csv |
| 6 | +from datetime import datetime, timedelta |
| 7 | +import random |
| 8 | +import sys |
| 9 | + |
| 10 | + |
| 11 | +def write_file(args): |
| 12 | + file_name, rows, curr_time, delta_time, curr_vals, delta_val = args |
| 13 | + fieldnames = ['timestamp'] |
| 14 | + fieldnames.extend(['C{0:d}'.format(i + 1) for i in range(len(curr_vals))]) |
| 15 | + with open(file_name, 'w', newline='') as csv_file: |
| 16 | + writer = csv.DictWriter(csv_file, fieldnames=fieldnames) |
| 17 | + writer.writeheader() |
| 18 | + values = [] |
| 19 | + for _ in range(rows//2): |
| 20 | + data = {'C{0:d}'.format(i + 1): curr_vals[i] |
| 21 | + for i in range(len(curr_vals))} |
| 22 | + data['timestamp'] = curr_time |
| 23 | + writer.writerow(data) |
| 24 | + values.append(data) |
| 25 | + curr_time += delta_time |
| 26 | + curr_vals = [x + random.uniform(-delta_val, delta_val) |
| 27 | + for x in curr_vals] |
| 28 | + while values: |
| 29 | + data = values.pop() |
| 30 | + data['timestamp'] = curr_time |
| 31 | + writer.writerow(data) |
| 32 | + curr_time += delta_time |
| 33 | + return file_name |
| 34 | + |
| 35 | + |
| 36 | +if __name__ == '__main__': |
| 37 | + arg_parser = ArgumentParser(description='create a set of CSV files') |
| 38 | + arg_parser.add_argument('--files', type=int, default=1, |
| 39 | + help='number of files to create') |
| 40 | + arg_parser.add_argument('base_name', help='base file name to use') |
| 41 | + arg_parser.add_argument('--cols', type=int, default=1, |
| 42 | + help='number of columns to generate') |
| 43 | + arg_parser.add_argument('--rows', type=int, default=100, |
| 44 | + help='number of rows to generate per file') |
| 45 | + arg_parser.add_argument('--workers', type=int, default=None, |
| 46 | + help='number of workersto use') |
| 47 | + options = arg_parser.parse_args() |
| 48 | + curr_time = datetime.now() |
| 49 | + delta_time = timedelta(seconds=1) |
| 50 | + curr_vals = [1.0]*options.cols |
| 51 | + delta_val = 0.01 |
| 52 | + with ProcessPoolExecutor(max_workers=options.workers) as executor: |
| 53 | + args = [('{0}_{1:04d}.csv'.format(options.base_name, i + 1), |
| 54 | + options.rows, curr_time + i*options.rows*delta_time, |
| 55 | + delta_time, curr_vals, delta_val) for i in range(options.files)] |
| 56 | + for file_name in executor.map(write_file, args): |
| 57 | + print('{0} done'.format(file_name)) |
0 commit comments