import os import glob import time import simplejson as json import click import operator from util import * import random @click.command() @click.option('--path', '-d', default='report/institutions_geocoded', help='Path to CSVs') def s2_merge_csv(path): print(path) lookup = {} for fn in glob.iglob('{}/*.csv'.format(path)): rows = read_csv(fn, keys=False) for row in rows: key = row[0] if key not in lookup: lookup[key] = row keys = sorted(lookup.keys()) deduped = [] for key in keys: deduped.append(lookup[key]) write_csv('{}.csv'.format(path), keys=None, rows=deduped) if __name__ == '__main__': s2_merge_csv()