def get_valids(groups, train_last_timestamps, test_summary, questions, chains):
supercollisions = grouped_find_overlapping_segments(
test_summary, questions, groups)
test_first_timestamps = {
x: summary[-2]
for x, summary in test_summary.iteritems()
}
initial_valids = {
question: {
p
for p in groups[proposed]
if is_valid(infra.shift_train_timestamp(train_last_timestamps[p]),
test_first_timestamps[question])
}
for question, proposed in questions.iteritems()
}
valids = get_immediate_valids(groups, questions, initial_valids,
supercollisions)
singletons = {k for k, v in valids.iteritems() if len(v) == 1}
while True:
last = len(singletons)
print last
tainted_chains = [c for c in chains if set(c) & singletons]
for chain in tainted_chains:
x = list(set(chain) & singletons)[0]
color = list(valids[x])[0]
for x in chain:
valids[x] = {color}
valids = get_immediate_valids(groups, questions, valids,
supercollisions)
singletons = {k for k, v in valids.iteritems() if len(v) == 1}
if len(singletons) == last:
break
return valids
def validate_questions(questions, train_last_timestamps,
test_first_timestamps):
return {
test_id:
is_valid(infra.shift_train_timestamp(train_last_timestamps[train_id]),
test_first_timestamps[test_id])
for test_id, train_id in questions.iteritems()
}
def validate_chain(chain, questions, train_last_timestamps,
test_first_timestamps):
first_timestamp = test_first_timestamps[chain[0]]
return {
test_id: (infra.shift_train_timestamp(
is_validtrain_last_timestamps[questions[test_id]]),
first_timestamp)
for test_id in chain
}
def match_test_to_train(train_last_timestamps, test_first_timestamps):
samples = [(timestamp, devid)
for devid, timestamp in test_first_timestamps.iteritems()]
timeindex = TimeIndex(samples)
train_to_test = {}
for train_id, last_timestamp in train_last_timestamps.iteritems():
last_timestamp = infra.shift_train_timestamp(last_timestamp)
available = [
devid for ts, devid in timeindex[last_timestamp +
0:last_timestamp + 1000]
]
if len(available) == 1:
train_to_test[train_id] = available[0]
counts = Counter(train_to_test.itervalues())
result = {
test_id: train_id
for train_id, test_id in train_to_test.iteritems()
if counts[test_id] == 1
}
return result
