tw2created_at_iso.py

-

#!/usr/bin/env python
import ujson as json
import time,sys
from datetime import datetime

def parse_date(twitter_lame_datetime_string):
  # e.g. the 'created_at' field
  ts = time.strptime(twitter_lame_datetime_string, "%a %b %d %H:%M:%S +0000 %Y")
  return datetime(*ts[:7])

for line in sys.stdin:
    tw = line.split('\t')[-1]
    tw = json.loads(tw)
    d = parse_date(tw['created_at'])
    created_at_iso = d.strftime("%Y-%m-%dT%H:%M:%S")
    print "%s\t%s" % (created_at_iso, line.rstrip('\n'))