def test_durable_table(durable_s3_table):
from historical.s3.models import DurableS3Model
# We are explicit about our eventTimes because as RANGE_KEY it will need to be unique.
S3_BUCKET['eventTime'] = datetime(2017, 5, 11, 23, 30)
DurableS3Model(**S3_BUCKET).save()
items = list(DurableS3Model.query('arn:aws:s3:::testbucket1'))
assert len(items) == 1
assert not getattr(items[0], "ttl", None)
S3_BUCKET['eventTime'] = datetime(2017, 5, 12, 23, 30)
DurableS3Model(**S3_BUCKET).save()
items = list(DurableS3Model.query('arn:aws:s3:::testbucket1'))
assert len(items) == 2
def test_differ(durable_s3_table, mock_lambda_environment):
from historical.s3.models import DurableS3Model
from historical.s3.differ import handler
from historical.models import TTL_EXPIRY
ttl = int(time.time() + TTL_EXPIRY)
new_bucket = S3_BUCKET.copy()
new_bucket['eventTime'] = datetime(
year=2017, month=5, day=12, hour=10, minute=30,
second=0).isoformat() + 'Z'
new_bucket["ttl"] = ttl
new_item = DynamoDBRecordsFactory(records=[
DynamoDBRecordFactory(dynamodb=DynamoDBDataFactory(
NewImage=new_bucket, Keys={'arn': new_bucket['arn']}),
eventName='INSERT')
])
data = json.loads(json.dumps(new_item, default=serialize))
handler(data, None)
assert DurableS3Model.count() == 1
# Test duplicates don't change anything:
data = json.loads(json.dumps(new_item, default=serialize))
handler(data, None)
assert DurableS3Model.count() == 1
# Test ephemeral changes don't add new models:
ephemeral_changes = S3_BUCKET.copy()
ephemeral_changes["eventTime"] = \
datetime(year=2017, month=5, day=12, hour=11, minute=30, second=0).isoformat() + 'Z'
ephemeral_changes["configuration"]["_version"] = 99999
ephemeral_changes["ttl"] = ttl
data = DynamoDBRecordsFactory(records=[
DynamoDBRecordFactory(dynamodb=DynamoDBDataFactory(
NewImage=ephemeral_changes, Keys={'arn': ephemeral_changes['arn']
}),
eventName='MODIFY')
])
data = json.loads(json.dumps(data, default=serialize))
handler(data, None)
assert DurableS3Model.count() == 1
# Add an update:
new_changes = S3_BUCKET.copy()
new_date = datetime(
year=2017, month=5, day=12, hour=11, minute=30,
second=0).isoformat() + 'Z'
new_changes["eventTime"] = new_date
new_changes["Tags"] = {"ANew": "Tag"}
new_changes["configuration"]["Tags"] = {"ANew": "Tag"}
new_changes["ttl"] = ttl
data = DynamoDBRecordsFactory(records=[
DynamoDBRecordFactory(dynamodb=DynamoDBDataFactory(
NewImage=new_changes, Keys={'arn': new_changes['arn']}),
eventName='MODIFY')
])
data = json.loads(json.dumps(data, default=serialize))
handler(data, None)
results = list(DurableS3Model.query("arn:aws:s3:::testbucket1"))
assert len(results) == 2
assert results[1].Tags["ANew"] == results[
1].configuration.attribute_values["Tags"]["ANew"] == "Tag"
assert results[1].eventTime == new_date
# And deletion (ensure new record -- testing TTL):
delete_bucket = S3_BUCKET.copy()
delete_bucket["eventTime"] = datetime(
year=2017, month=5, day=12, hour=12, minute=30,
second=0).isoformat() + 'Z'
delete_bucket["ttl"] = ttl
data = DynamoDBRecordsFactory(records=[
DynamoDBRecordFactory(dynamodb=DynamoDBDataFactory(
OldImage=delete_bucket, Keys={'arn': delete_bucket['arn']}),
eventName='REMOVE',
userIdentity=UserIdentityFactory(
type='Service',
principalId='dynamodb.amazonaws.com'))
])
data = json.loads(json.dumps(data, default=serialize))
handler(data, None)
assert DurableS3Model.count() == 3
def durable_s3_table(dynamodb):
from historical.s3.models import DurableS3Model
yield DurableS3Model.create_table(read_capacity_units=1,
write_capacity_units=1,
wait=True)
def test_snsproxy_dynamodb_differ(historical_role, current_s3_table,
durable_s3_table, mock_lambda_environment,
buckets):
"""
This mostly checks that the differ is able to properly load the reduced dataset from the SNSProxy.
"""
# Create the item in the current table:
from historical.s3.collector import handler as current_handler
from historical.s3.differ import handler as diff_handler
from historical.s3.models import CurrentS3Model, DurableS3Model
from historical.common.sns import shrink_sns_blob
# Mock out the loggers:
import historical.common.dynamodb
old_logger = historical.common.dynamodb.log
mocked_logger = MagicMock()
historical.common.dynamodb.log = mocked_logger
now = datetime.utcnow().replace(tzinfo=None, microsecond=0)
create_event = CloudwatchEventFactory(
detail=DetailFactory(requestParameters={"bucketName": "testbucket1"},
eventSource="aws.s3",
eventName="CreateBucket",
eventTime=now))
data = json.dumps(create_event, default=serialize)
data = RecordsFactory(records=[SQSDataFactory(body=data)])
data = json.dumps(data, default=serialize)
data = json.loads(data)
current_handler(data, mock_lambda_environment)
result = list(CurrentS3Model.query("arn:aws:s3:::testbucket1"))
assert len(result) == 1
# Mock out the DDB Stream for this creation and for an item that is NOT in the current table::
ttl = int(time.time() + TTL_EXPIRY)
new_bucket = S3_BUCKET.copy()
new_bucket['eventTime'] = datetime(
year=2017, month=5, day=12, hour=10, minute=30,
second=0).isoformat() + 'Z'
new_bucket['ttl'] = ttl
ddb_existing_item = DynamoDBRecordFactory(dynamodb=DynamoDBDataFactory(
NewImage=new_bucket,
Keys={'arn': new_bucket['arn']},
OldImage=new_bucket),
eventName='INSERT')
missing_bucket = S3_BUCKET.copy()
missing_bucket['eventTime'] = datetime(
year=2017, month=5, day=12, hour=10, minute=30,
second=0).isoformat() + 'Z'
missing_bucket['ttl'] = ttl
missing_bucket['BucketName'] = 'notinthecurrenttable'
missing_bucket['arn'] = 'arn:aws:s3:::notinthecurrenttable'
missing_bucket['configuration']['Name'] = 'notinthecurrenttable'
ddb_missing_item = DynamoDBRecordFactory(dynamodb=DynamoDBDataFactory(
NewImage=missing_bucket,
Keys={'arn': 'arn:aws:s3:::notinthecurrenttable'},
OldImage=new_bucket),
eventName='INSERT')
# Get the shrunken blob:
shrunken_existing = json.dumps(
shrink_sns_blob(
json.loads(json.dumps(ddb_existing_item, default=serialize))))
shrunken_missing = json.dumps(
shrink_sns_blob(
json.loads(json.dumps(ddb_missing_item, default=serialize))))
records = RecordsFactory(records=[
SQSDataFactory(body=json.dumps(
SnsDataFactory(Message=shrunken_existing), default=serialize)),
SQSDataFactory(body=json.dumps(
SnsDataFactory(Message=shrunken_missing), default=serialize))
])
records_event = json.loads(json.dumps(records, default=serialize))
# Run the differ:
diff_handler(records_event, mock_lambda_environment)
# Verify that the existing bucket in the Current table is in the Durable table with the correct configuration:
result = list(DurableS3Model.query("arn:aws:s3:::testbucket1"))
assert len(result) == 1
assert result[0].configuration.attribute_values['Name'] == 'testbucket1'
# Verify that the missing bucket is ignored -- as it will be processed presumably later:
result = list(DurableS3Model.query("arn:aws:s3:::notinthecurrenttable"))
assert not result
# Verify that the proper log statements were reached:
assert mocked_logger.debug.called
assert mocked_logger.error.called
debug_calls = [
'[-->] Item with ARN: arn:aws:s3:::notinthecurrenttable was too big for SNS '
'-- fetching it from the Current table...',
'[+] Saving new revision to durable table.',
'[-->] Item with ARN: arn:aws:s3:::testbucket1 was too big for SNS -- fetching it from the Current table...'
]
for dc in debug_calls:
mocked_logger.debug.assert_any_call(dc)
mocked_logger.error.assert_called_once_with(
'[?] Received item too big for SNS, and was not able to '
'find the original item with ARN: arn:aws:s3:::notinthecurrenttable')
# Unmock the logger:
historical.common.dynamodb.log = old_logger