def generate_data_table_schema(config, table, schema_override=None):
"""Generate the schema for data table in terraform
Args:
config (CLIConfig): Loaded StreamAlert config
table (string): The name of data table
Returns:
athena_schema (dict): Equivalent Athena schema used for generating create table statement
"""
enabled_logs = FirehoseClient.load_enabled_log_sources(
config['global']['infrastructure']['firehose'], config['logs'])
# Convert special characters in schema name to underscores
sanitized_table_name = FirehoseClient.sanitized_value(table)
# Check that the log type is enabled via Firehose
if sanitized_table_name not in enabled_logs:
LOGGER.error(
'Table name %s missing from configuration or '
'is not enabled.', sanitized_table_name)
return None
log_info = config['logs'][enabled_logs.get(sanitized_table_name)]
schema = dict(log_info['schema'])
sanitized_schema = FirehoseClient.sanitize_keys(schema)
athena_schema = logs_schema_to_athena_schema(sanitized_schema, False)
# Add envelope keys to Athena Schema
configuration_options = log_info.get('configuration')
if configuration_options:
envelope_keys = configuration_options.get('envelope_keys')
if envelope_keys:
sanitized_envelope_key_schema = FirehoseClient.sanitize_keys(
envelope_keys)
# Note: this key is wrapped in backticks to be Hive compliant
athena_schema[
'streamalert:envelope_keys'] = logs_schema_to_athena_schema(
sanitized_envelope_key_schema, False)
# Handle Schema overrides
# This is useful when an Athena schema needs to differ from the normal log schema
if schema_override:
for override in schema_override:
column_name, column_type = override.split('=')
# Columns are escaped to avoid Hive issues with special characters
column_name = '{}'.format(column_name)
if column_name in athena_schema:
athena_schema[column_name] = column_type
LOGGER.info('Applied schema override: %s:%s', column_name,
column_type)
else:
LOGGER.error(
'Schema override column %s not found in Athena Schema, skipping',
column_name)
return format_schema_tf(athena_schema)
def test_sanitize_keys(self):
"""FirehoseClient - Sanitize Keys"""
test_event = {
'date': 'January 01, 3005',
'data': {
'super-duper': 'secret',
'do_not_sanitize_me': 1,
'example-key': 2,
'moar**data': 3,
'even.more': 4
}
}
expected_sanitized_event = {
'date': 'January 01, 3005',
'data': {
'super_duper': 'secret',
'do_not_sanitize_me': 1,
'example_key': 2,
'moar__data': 3,
'even_more': 4
}
}
sanitized_event = FirehoseClient.sanitize_keys(test_event)
assert_equal(sanitized_event, expected_sanitized_event)
def create_table(table, bucket, config, schema_override=None):
"""Create a 'streamalert' Athena table
Args:
table (str): The name of the table being rebuilt
bucket (str): The s3 bucket to be used as the location for Athena data
table_type (str): The type of table being refreshed
config (CLIConfig): Loaded StreamAlert config
schema_override (set): An optional set of key=value pairs to be used for
overriding the configured column_name=value_type.
Returns:
bool: False if errors occurred, True otherwise
"""
enabled_logs = FirehoseClient.load_enabled_log_sources(
config['global']['infrastructure']['firehose'], config['logs'])
# Convert special characters in schema name to underscores
sanitized_table_name = FirehoseClient.sanitized_value(table)
# Check that the log type is enabled via Firehose
if sanitized_table_name != 'alerts' and sanitized_table_name not in enabled_logs:
LOGGER.error(
'Table name %s missing from configuration or '
'is not enabled.', sanitized_table_name)
return False
athena_client = get_athena_client(config)
# Check if the table exists
if athena_client.check_table_exists(sanitized_table_name):
LOGGER.info('The \'%s\' table already exists.', sanitized_table_name)
return True
if table == 'alerts':
# get a fake alert so we can get the keys needed and their types
alert = Alert('temp_rule_name', {}, {})
output = alert.output_dict()
schema = record_to_schema(output)
athena_schema = helpers.logs_schema_to_athena_schema(schema)
# Use the bucket if supplied, otherwise use the default alerts bucket
bucket = bucket or firehose_alerts_bucket(config)
query = _construct_create_table_statement(
schema=athena_schema,
table_name=table,
bucket=bucket,
file_format=get_data_file_format(config))
else: # all other tables are log types
config_data_bucket = firehose_data_bucket(config)
if not config_data_bucket:
LOGGER.warning(
'The \'firehose\' module is not enabled in global.json')
return False
# Use the bucket if supplied, otherwise use the default data bucket
bucket = bucket or config_data_bucket
log_info = config['logs'][enabled_logs.get(sanitized_table_name)]
schema = dict(log_info['schema'])
sanitized_schema = FirehoseClient.sanitize_keys(schema)
athena_schema = helpers.logs_schema_to_athena_schema(sanitized_schema)
# Add envelope keys to Athena Schema
configuration_options = log_info.get('configuration')
if configuration_options:
envelope_keys = configuration_options.get('envelope_keys')
if envelope_keys:
sanitized_envelope_key_schema = FirehoseClient.sanitize_keys(
envelope_keys)
# Note: this key is wrapped in backticks to be Hive compliant
athena_schema[
'`streamalert:envelope_keys`'] = helpers.logs_schema_to_athena_schema(
sanitized_envelope_key_schema)
# Handle Schema overrides
# This is useful when an Athena schema needs to differ from the normal log schema
if schema_override:
for override in schema_override:
column_name, column_type = override.split('=')
# Columns are escaped to avoid Hive issues with special characters
column_name = '`{}`'.format(column_name)
if column_name in athena_schema:
athena_schema[column_name] = column_type
LOGGER.info('Applied schema override: %s:%s', column_name,
column_type)
else:
LOGGER.error(
'Schema override column %s not found in Athena Schema, skipping',
column_name)
query = _construct_create_table_statement(
schema=athena_schema,
table_name=sanitized_table_name,
bucket=bucket,
file_format=get_data_file_format(config))
success = athena_client.run_query(query=query)
if not success:
LOGGER.error('The %s table could not be created', sanitized_table_name)
return False
# Update the CLI config
if table != 'alerts' and bucket != config_data_bucket:
# Only add buckets to the config if they are not one of the default/configured buckets
# Ensure 'buckets' exists in the config (since it is not required)
config['lambda']['athena_partitioner_config']['buckets'] = (
config['lambda']['athena_partitioner_config'].get('buckets', {}))
if bucket not in config['lambda']['athena_partitioner_config'][
'buckets']:
config['lambda']['athena_partitioner_config']['buckets'][
bucket] = 'data'
config.write()
LOGGER.info('The %s table was successfully created!', sanitized_table_name)
return True
