cfr_decorate_sort.py

"""
from fileinput or stdin, decorate lines with userid, ts for later sort.

Full usage:

hadoop dfs -text /bagheera/testpilot_contextfeaturerecommender/*/* | python dsu.py  | sort -k 1,2 |  cut -f 3-  | less -S

(decorate-sort-undecorate pattern  sorting)
"""

import fileinput
import json

for l in fileinput.input():
     try:
          payload = l.split("\t",1)[1]
          payload = json.loads(json.loads(payload)["dp"])
          print "\t".join(map(str,[payload["userid"], payload["ts"], l.rstrip()]))
     except:
          continue