Create files with old dates

create_dated_files.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
"""Script to generate a sample set of files with a random distribution of
dates. Right now, this is very likely to be a uniform distribution so numpy
needs to be explored some more.
"""
import argparse
import atexit
import datetime
import logging
import os
import shutil
import time

from numpy import random


LOGFORMAT = '%(asctime)-15s %(levelname)s: %(message)s'
DATEFORMAT = '%m/%d/%Y %H:%M:%S'


def _make_parser():
    """Create an argument parser for this script."""
    parser = argparse.ArgumentParser()
    parser.description = "create a set of files with a random date range"
    parser.add_argument(
        "-i", "--input",
        help="path to file to duplicate",
        required=False)
    parser.add_argument(
        "-o", "--output",
        help="path to the output folder",
        default="output-dates",
        required=False)
    parser.add_argument(
        "-n", "--number",
        help="number of files to create",
        default=100,
        required=False)
    return parser


def change_time(filepath, year):
    """Create a timestamp to attach to our files based on the year provided."""
    year = year
    month = random.randint(1, 12, 1)
    day = random.randint(1, 28, 1)
    hour = 10
    minute = 25
    second = 1

    date = datetime.datetime(
        year=year, month=month, day=day, hour=hour,
        minute=minute, second=second)
    mod_time = time.mktime(date.timetuple())
    os.utime(filepath, (mod_time, mod_time))


def copy_media_location(copy_path, output_path, date_range):
    """Create as maany copies of our sample data files as required by the
    user in the argument media_n.
    """
    ext = "ext"
    try:
        file_name, ext = os.path.basename(copy_path).split(".")
    except ValueError:
        file_name = copy_path
    for count, _ in enumerate(date_range):
        new_file_name = '{}_{}.{}'.format(file_name, count, ext)
        new_path = os.path.join(output_path, new_file_name)
        shutil.copyfile(copy_path, new_path)
        change_time(new_path, date_range[count])
    logging.info("%s files created at %s", len(date_range), output_path)


def tmp_file_content():
    """Generate some content for temporary files if no other file path is
    provided to the script.
    """
    header = "# Fake File Format Study Data"
    content = "Some data for the study."
    return "{}\n\n{}".format(header, content)


def remove_tmp(file_path):
    """Function to be registered with atexit if there isn't a file supplied
    with the script args.
    """
    try:
        os.remove(file_path)
    except OSError:
        logging.info("Not removing %s at exit", file_path)


def main():
    """Primary entry point for the script."""
    logging.basicConfig(format=LOGFORMAT, datefmt=DATEFORMAT, level="INFO")

    parser = _make_parser()
    args = parser.parse_args()

    if args.output:
        try:
            os.mkdir(args.output)
        except OSError:
            pass

    if not args.input:
        tmp_file_path = "file.md"
        atexit.register(remove_tmp, file_path=tmp_file_path)
        with open(tmp_file_path, 'wb') as tmp_file:
            tmp_file.write(tmp_file_content())
        args.input = tmp_file_path

    dist = random.randint(1970, 2018, args.number)
    copy_media_location(args.input, args.output, dist)


if __name__ == "__main__":
    main()