Downloader for NASA's DSCOVR EPIC images

epic.py

import argparse
import datetime
import json
import os
import re

import requests


parser = argparse.ArgumentParser()
parser.add_argument('-r', '--root', default=os.path.abspath(os.path.join(__file__, '..', 'epic')))
parser.add_argument('-d', '--dir-fmt', default='%Y/%m/%d')
parser.add_argument('date', nargs='+')
args = parser.parse_args()

for raw_date in args.date:

    date_parts = re.split(r'\D+', raw_date)
    date_parts = tuple(map(int, date_parts))
    assert len(date_parts) == 3

    res = requests.get('http://epic.gsfc.nasa.gov/api/images.php?date=%s-%s-%s' % date_parts)
    images = res.json()

    for meta in images:

        date = datetime.datetime.strptime(meta['date'], '%Y-%m-%d %H:%M:%S')

        dir_ = os.path.join(args.root, date.strftime(args.dir_fmt))
        if not os.path.exists(dir_):
            os.makedirs(dir_)

        meta_path = os.path.join(dir_, meta['image'] + '.json')
        if not os.path.exists(meta_path):
            with open(meta_path, 'w') as fh:
                fh.write(json.dumps(meta, sort_keys=True, indent=4))

        png_path = os.path.join(dir_, meta['image'] + '.png')
        if not os.path.exists(png_path) or not os.path.getsize(png_path):
            png_url = 'http://epic.gsfc.nasa.gov/epic-archive/png/%s.png' % meta['image']
            print png_url
            png_res = requests.get(png_url)
            png_data = png_res.content # Force it to read it all.
            with open(png_path, 'w') as fh:
                fh.write(png_data)