def test_collector(historical_role, buckets, mock_lambda_environment,
swag_accounts, current_s3_table):
from historical.s3.collector import handler
now = datetime.utcnow().replace(tzinfo=None, microsecond=0)
create_event = CloudwatchEventFactory(
detail=DetailFactory(requestParameters={"bucketName": "testbucket1"},
source="aws.s3",
eventName="CreateBucket",
eventTime=now))
data = json.dumps(create_event, default=serialize)
data = KinesisRecordsFactory(
records=[KinesisRecordFactory(kinesis=KinesisDataFactory(data=data))])
data = json.dumps(data, default=serialize)
data = json.loads(data)
handler(data, None)
result = list(CurrentS3Model.query("arn:aws:s3:::testbucket1"))
assert len(result) == 1
# Verify that the tags are duplicated in the top level and configuration:
assert len(result[0].Tags.attribute_values) == len(
result[0].configuration.attribute_values["Tags"]) == 1
assert result[0].Tags.attribute_values["theBucketName"] == \
result[0].configuration.attribute_values["Tags"]["theBucketName"] == "testbucket1" # noqa
# Polling (make sure the date is included):
polling_event = CloudwatchEventFactory(
detail=DetailFactory(requestParameters={
"bucketName": "testbucket1",
"creationDate": now
},
source="aws.s3",
eventName="DescribeBucket",
eventTime=now))
data = json.dumps(polling_event, default=serialize)
data = KinesisRecordsFactory(
records=[KinesisRecordFactory(kinesis=KinesisDataFactory(data=data))])
data = json.dumps(data, default=serialize)
data = json.loads(data)
handler(data, None)
assert CurrentS3Model.count() == 1
# Load the config and verify the polling timestamp is in there:
result = list(CurrentS3Model.query("arn:aws:s3:::testbucket1"))
assert result[0].configuration["CreationDate"] == now.isoformat() + "Z"
# And deletion:
delete_event = CloudwatchEventFactory(
detail=DetailFactory(requestParameters={"bucketName": "testbucket1"},
source="aws.s3",
eventName="DeleteBucket",
eventTime=now))
data = json.dumps(delete_event, default=serialize)
data = KinesisRecordsFactory(
records=[KinesisRecordFactory(kinesis=KinesisDataFactory(data=data))])
data = json.dumps(data, default=serialize)
data = json.loads(data)
handler(data, None)
assert CurrentS3Model.count() == 0
def test_collector(historical_role, buckets, mock_lambda_environment,
swag_accounts, current_s3_table):
"""Test the Collector."""
from historical.s3.models import CurrentS3Model
from historical.s3.collector import handler
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)
handler(data, mock_lambda_environment)
result = list(CurrentS3Model.query("arn:aws:s3:::testbucket1"))
assert len(result) == 1
assert result[0].Tags.attribute_values["theBucketName"] == "testbucket1"
assert result[0].eventSource == "aws.s3"
# Polling (make sure the date is included):
polling_event = CloudwatchEventFactory(
detail=DetailFactory(requestParameters={
"bucketName": "testbucket1",
"creationDate": now
},
eventSource="historical.s3.poller",
eventName="PollS3",
eventTime=now))
data = json.dumps(polling_event, default=serialize)
data = RecordsFactory(records=[SQSDataFactory(body=data)])
data = json.dumps(data, default=serialize)
data = json.loads(data)
handler(data, mock_lambda_environment)
assert CurrentS3Model.count() == 1
# Load the config and verify the polling timestamp is in there:
result = list(CurrentS3Model.query("arn:aws:s3:::testbucket1"))
assert result[0].configuration["CreationDate"] == now.isoformat() + "Z"
assert result[0].eventSource == "historical.s3.poller"
# And deletion:
delete_event = CloudwatchEventFactory(
detail=DetailFactory(requestParameters={"bucketName": "testbucket1"},
eventSource="aws.s3",
eventName="DeleteBucket",
eventTime=now))
data = json.dumps(delete_event, default=serialize)
data = RecordsFactory(records=[SQSDataFactory(body=data)])
data = json.dumps(data, default=serialize)
data = json.loads(data)
handler(data, mock_lambda_environment)
assert CurrentS3Model.count() == 0
def test_current_table(current_s3_table):
from historical.s3.models import CurrentS3Model
CurrentS3Model(**S3_BUCKET).save()
items = list(CurrentS3Model.query('arn:aws:s3:::testbucket1'))
assert len(items) == 1
assert isinstance(items[0].ttl, int)
assert items[0].ttl > 0
def test_current_table(current_s3_table): # pylint: disable=W0613
"""Tests for the Current PynamoDB model."""
from historical.s3.models import CurrentS3Model
CurrentS3Model(**S3_BUCKET).save()
items = list(CurrentS3Model.query('arn:aws:s3:::testbucket1'))
assert len(items) == 1
assert isinstance(items[0].ttl, int)
assert items[0].ttl > 0
def process_durable_event(record, s3_report):
"""Processes a group of Historical Durable Table events."""
if record.get(EVENT_TOO_BIG_FLAG):
result = list(CurrentS3Model.query(record['arn']))
# Is the record too big and also not found in the Current Table? Then delete it:
if not result:
record['item'] = {'configuration': {}, 'BucketName': record['arn'].split('arn:aws:s3:::')[1]}
else:
record['item'] = dict(result[0])
if not record['item']['configuration']:
log.debug(f"[ ] Processing deletion for: {record['item']['BucketName']}")
s3_report["buckets"].pop(record['item']['BucketName'], None)
else:
log.debug(f"[ ] Processing: {record['item']['BucketName']}")
s3_report["all_buckets"].append(record['item'])
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
