def assign_sampled_ballots(
jurisdiction: Jurisdiction,
round: Round,
audit_boards: List[AuditBoard],
):
# Collect the physical ballots for each batch that were sampled for this
# jurisdiction for this round
sampled_ballots = (
SampledBallot.query.join(Batch).filter_by(
jurisdiction_id=jurisdiction.id).join(
SampledBallot.draws).filter_by(round_id=round.id).order_by(
Batch.name) # group_by prefers a sorted list
.options(contains_eager(SampledBallot.batch)).all())
ballots_by_batch = group_by(sampled_ballots, key=lambda sb: sb.batch.name)
# Divvy up batches of ballots between the audit boards.
# Note: BalancedBucketList doesn't care which buckets have which batches to
# start, so we add all the batches to the first bucket before balancing.
buckets = [Bucket(audit_board.id) for audit_board in audit_boards]
for batch_name, sampled_ballots in ballots_by_batch.items():
buckets[0].add_batch(batch_name, len(sampled_ballots))
balanced_buckets = BalancedBucketList(buckets)
for bucket in balanced_buckets.buckets:
ballots_in_bucket = [
ballot for batch_name in bucket.batches
for ballot in ballots_by_batch[batch_name]
]
for ballot in ballots_in_bucket:
ballot.audit_board_id = bucket.name
db.session.add(ballot)
def sample_ballots(session: Session, election: Election, round: Round):
# assume only one contest
round_contest = round.round_contests[0]
jurisdiction = election.jurisdictions[0]
num_sampled = (session.query(SampledBallotDraw).join(SampledBallot).join(
SampledBallot.batch).filter_by(
jurisdiction_id=jurisdiction.id).count())
if not num_sampled:
num_sampled = 0
chosen_sample_size = round_contest.sample_size
# the sampler needs to have the same inputs given the same manifest
# so we use the batch name, rather than the batch id
# (because the batch ID is an internally generated uuid
# that changes from one run to the next.)
manifest = {}
batch_id_from_name = {}
for batch in jurisdiction.batches:
manifest[batch.name] = batch.num_ballots
batch_id_from_name[batch.name] = batch.id
sample = sampler.draw_sample(
election.random_seed,
manifest,
chosen_sample_size,
num_sampled=num_sampled,
)
audit_boards = jurisdiction.audit_boards
batch_sizes: Dict[str, int] = {}
batches_to_ballots: Dict[str, List[Tuple[int, str, int]]] = {}
# Build batch - batch_size map
for (ticket_number, (batch_name, ballot_position),
sample_number) in sample:
if batch_name in batch_sizes:
if (sample_number == 1
): # if we've already seen it, it doesn't affect batch size
batch_sizes[batch_name] += 1
batches_to_ballots[batch_name].append(
(ballot_position, ticket_number, sample_number))
else:
batch_sizes[batch_name] = 1
batches_to_ballots[batch_name] = [(ballot_position, ticket_number,
sample_number)]
# Create the buckets and initially assign batches
buckets = [Bucket(audit_board.name) for audit_board in audit_boards]
for i, batch in enumerate(batch_sizes):
buckets[i % len(audit_boards)].add_batch(batch, batch_sizes[batch])
# Now assign batchest fairly
bucket_list = BalancedBucketList(buckets)
# read audit board and batch info out
for audit_board_num, bucket in enumerate(bucket_list.buckets):
audit_board = audit_boards[audit_board_num]
for batch_name in bucket.batches:
for (ballot_position, ticket_number,
sample_number) in batches_to_ballots[batch_name]:
batch_id = batch_id_from_name[batch_name]
if sample_number == 1:
sampled_ballot = SampledBallot(
id=str(uuid.uuid4()),
batch_id=batch_id,
ballot_position=ballot_position,
audit_board_id=audit_board.id,
status=BallotStatus.NOT_AUDITED,
)
session.add(sampled_ballot)
else:
sampled_ballot = SampledBallot.query.filter_by(
batch_id=batch_id,
ballot_position=ballot_position,
).one()
sampled_ballot_draw = SampledBallotDraw(
ballot_id=sampled_ballot.id,
round_id=round.id,
ticket_number=ticket_number,
)
session.add(sampled_ballot_draw)
session.commit()
def sample_ballots(election, round):
# assume only one contest
round_contest = round.round_contests[0]
jurisdiction = election.jurisdictions[0]
num_sampled = db.session.query(db.func.sum(SampledBallot.times_sampled)).filter_by(jurisdiction_id=jurisdiction.id).one()[0]
if not num_sampled:
num_sampled = 0
chosen_sample_size = round_contest.sample_size
sampler = get_sampler(election)
# the sampler needs to have the same inputs given the same manifest
# so we use the batch name, rather than the batch id
# (because the batch ID is an internally generated uuid
# that changes from one run to the next.)
manifest = {}
batch_id_from_name = {}
for batch in jurisdiction.batches:
manifest[batch.name] = batch.num_ballots
batch_id_from_name[batch.name] = batch.id
sample = sampler.draw_sample(manifest, chosen_sample_size, num_sampled=num_sampled)
audit_boards = jurisdiction.audit_boards
last_sample = None
last_sampled_ballot = None
batch_sizes = {}
batches_to_ballots = {}
seen_ballot_positions = set()
# Build batch - batch_size map
for batch_name, ballot_position in sample:
batch_id = batch_id_from_name[batch_name]
lookup = (batch_id, ballot_position)
# Only count ballots once here since it's only pulled once
if lookup in seen_ballot_positions:
batches_to_ballots[batch_id].append(ballot_position)
continue
seen_ballot_positions.add(lookup)
if batch_id in batch_sizes:
batch_sizes[batch_id] += 1
batches_to_ballots[batch_id].append(ballot_position)
else:
batch_sizes[batch_id] = 1
batches_to_ballots[batch_id] = [ballot_position]
# Create the buckets and initially assign batches
buckets = []
for audit_board in audit_boards:
buckets.append(Bucket(audit_board.name))
for i, batch in enumerate(batch_sizes):
buckets[i%len(audit_boards)].add_batch(batch, batch_sizes[batch])
# Now assign batchest fairly
bl = BalancedBucketList(buckets)
# read audit board and batch info out
for bucket in bl.buckets:
audit_board_num = bl.buckets.index(bucket)
audit_board = audit_boards[audit_board_num]
for batch_id in bucket.batches:
for ballot_position in batches_to_ballots[batch_id]:
if last_sample == (batch_id, ballot_position):
last_sampled_ballot.times_sampled += 1
continue
sampled_ballot = SampledBallot(
round_id = round.id,
jurisdiction_id = jurisdiction.id,
batch_id = batch_id,
ballot_position = ballot_position + 1, # sampler is 0-indexed, we're 1-indexing here
times_sampled = 1,
audit_board_id = audit_board.id)
# keep track for doubly-sampled ballots
last_sample = (batch_id, ballot_position)
last_sampled_ballot = sampled_ballot
db.session.add(sampled_ballot)
db.session.commit()
def bucketlist():
buckets = []
b = Bucket('1')
b.add_batch('1', 100)
b.add_batch('2', 50)
buckets.append(b)
b = Bucket('2')
b.add_batch('3', 100)
b.add_batch('4', 150)
buckets.append(b)
b = Bucket('3')
b.add_batch('5', 50)
b.add_batch('6', 50)
buckets.append(b)
b = Bucket('4')
b.add_batch('7', 100)
b.add_batch('8', 200)
buckets.append(b)
yield BucketList(buckets)
def bucket():
yield Bucket('1')