histogram-year

#!/usr/bin/env python

"""

Output year histogram for movies in IMDB, for use when creating a graph.

"""

import argparse
import json

def load(h, l, stats, path, verbose=False, cutyear=None):
    #print("Loading %s" % path)
    with open(path, 'rt') as input:
    	 j = json.load(input)

	 for imdburl in j.keys():
             if imdburl in l:
                 continue
	     if 0 > imdburl.find("imdb.com"):
	         continue
             if 'status' not in j[imdburl] or 'free' != j[imdburl]['status']:
                 continue
             if 'year' not in j[imdburl]:
                 year = 'n/a'
             else:
                 #print j[imdburl]
                 year = int(j[imdburl]['year'])
                 if cutyear:
                     if year < cutyear:
                         stats['before'] = stats['before'] + 1
                     else:
                         stats['after'] = stats['after'] + 1
             if not year in h:
                     h[year] = 0
             if verbose:
                 print imdburl, j[imdburl]['title']
             h[year] = h[year] + 1
             l[imdburl] = j[imdburl]

def main():
    parser = argparse.ArgumentParser()
    parser.add_argument('files', metavar='N', type=str, nargs='*',
                        default=['free-movies-manual.json'],
                        help='input file(s)')
    parser.add_argument('--cutyear', type=int, default=None,
                        help='Report movie count before and after this year')
    parser.add_argument('--total', action='store_true', default=False)
    args = parser.parse_args()

    h = {}
    l = {}
    stats = {
        'before' : 0,
        'after' : 0
    }
    for file in args.files:
        load(h, l, stats, file, cutyear=args.cutyear)

    for y in sorted(h.keys()):
        #print y
        print("%s %s" % (y, h[y]))

    if args.total:
        print "Total:", len(l.keys())
    if args.cutyear:
        print("Cutyear %d.  %d before, %d after" %
              (args.cutyear, stats['before'], stats['after']))
if __name__ == '__main__':
    main()