def _aggregate(args):
filename, (hours, ) = args
uid = filename + "~" +str(hours)
if r.sismember(PROCESSED_KEY, uid):
print "{} with {} hours already processed. Carrying on...".format(filename, hours)
return
gcc = CrashCounts("global", hours)
ccs = {}
for crash in crashes(filename):
bucket = buckify(hours, crash.time)
gcc.incr(bucket, 1)
cc = ccs.get(crash.signature_id)
if not cc:
ccs[crash.signature_id] = cc = CrashCounts(crash.signature_id, hours)
cc.incr(bucket, 1)
for cc in ccs.itervalues():
cc.finalize()
gcc.finalize()
r.sadd(PROCESSED_KEY, uid)
def _top_crashes(args):
filename, (start, end) = args
pipe = r.pipeline()
for crash in crashes(filename):
pipe.hincrby(top_crashes_key(start, end), crash.signature_id, 1)
pipe.execute()
