AM history parser for reducer skew checks

reducer-counters.py

import sys
import re

def Counter(name):
	pattern = re.compile("%s=([^,]*)" % name)
	# warning closure
	def get(l):
		m = pattern.search(l)
		if m:
			return m.group(1)
		return None
	return get

for l in sys.stdin:
	groups = Counter("REDUCE_INPUT_GROUPS")
	records = Counter("REDUCE_INPUT_RECORDS")
	time = Counter("timeTaken")
	node = Counter("nodeId")
	vertex = Counter("vertexName")
	attempt = Counter("taskAttemptId")
	start = Counter("startTime")
	print "vertex, attempt, start, time, groups, records"
	if ("TASK_ATTEMPT_FINISHED" in l):
		print ",".join(map(str, [vertex(l), attempt(l), start(l), time(l), groups(l), records(l)]))