def create_metadata_table(table, cols, addition):
db.create_table(table, cols, ifnotexists=True)
db.execute(f"GRANT INSERT, SELECT ON {table} TO ROLE {SA_ROLE}")
table_names = (row['name'] for row in db.fetch(f'desc table {table}'))
if any(name == addition[0].upper() for name in table_names):
return
db.execute(f'ALTER TABLE {table} ADD COLUMN {addition[0]} {addition[1]}')
def connect(connection_name, options):
base_name = f"azure_subscription_{connection_name}"
tenant_id = options['tenant_id']
client_id = options['client_id']
client_secret = options['client_secret']
comment = f'''
---
module: azure_subscription
client_id: {client_id}
tenant_id: {tenant_id}
client_secret: {client_secret}
'''
db.create_table(name=f'data.{base_name}_connection',
cols=LANDING_TABLE_COLUMNS,
comment=comment)
db.execute(
f'GRANT INSERT, SELECT ON data.{base_name}_connection TO ROLE {SA_ROLE}'
)
return {
'newStage': 'finalized',
'newMessage': 'Landing table created for collectors to populate.'
}
def record_ticket_id(ticket_id, alert_id):
query = f"UPDATE results.alerts SET ticket='{ticket_id}' WHERE alert:ALERT_ID='{alert_id}'"
print('Updating alert table:', query)
try:
db.execute(query)
except Exception as e:
log.error(e, f"Failed to update alert {alert_id} with ticket id {ticket_id}")
def create_alerts(rule_name: str) -> Dict[str, Any]:
metadata: Dict[str, Any] = {
'QUERY_NAME': rule_name,
'RUN_ID': RUN_ID,
'ATTEMPTS': 1,
'START_TIME': datetime.datetime.utcnow(),
'ROW_COUNT': {
'INSERTED': 0,
'UPDATED': 0,
}
}
try:
db.execute(RUN_ALERT_QUERY.format(
query_name=rule_name,
from_time_sql="DATEADD(minute, -90, CURRENT_TIMESTAMP())",
to_time_sql="CURRENT_TIMESTAMP()",
), fix_errors=False)
insert_count, update_count = merge_alerts(rule_name, GROUPING_CUTOFF)
metadata['ROW_COUNT'] = {
'INSERTED': insert_count,
'UPDATED': update_count,
}
db.execute(f"DROP TABLE results.RUN_{RUN_ID}_{rule_name}")
except Exception as e:
db.record_metadata(metadata, table=QUERY_METADATA_TABLE, e=e)
return metadata
db.record_metadata(metadata, table=QUERY_METADATA_TABLE)
log.info(f"{rule_name} done.")
return metadata
def bail_out(alert_id):
query = f"UPDATE results.alerts SET handled='no handler' WHERE alert:ALERT_ID='{alert_id}'"
print('Updating alert table:', query)
try:
db.execute(query)
except Exception as e:
log.error(e, f"Failed to update alert {alert_id} with handler status")
def connect(connection_name, options):
base_name = f'AZURE_VM_{connection_name}'
client_id = options['client_id']
client_secret = options['client_secret']
tenant_id = options['tenant_id']
subscription_connection_name = options['subscription_connection_name']
comment = (
f'---',
f'module: azure_vm',
f'client_id: {client_id}',
f'client_secret: {client_secret}',
f'tenant_id: {tenant_id}',
f'subscription_connection_name: {subscription_connection_name}',
)
db.create_table(name=f'data.{base_name}_CONNECTION',
cols=LANDING_TABLE_COLUMNS,
comment=comment)
db.execute(
f'GRANT INSERT, SELECT ON data.{base_name}_CONNECTION TO ROLE {SA_ROLE}'
)
cols = [
('SNAPSHOT_AT', 'TIMESTAMP_LTZ'),
('RUN_ID', 'STRING(100)'),
('SUBSCRIPTION_ID', 'STRING(500)'),
('VM_INSTANCE_COUNT', 'NUMBER'),
]
create_metadata_table(AZURE_COLLECTION_METADATA, cols, cols[3])
return {
'newStage':
'finalized',
'newMessage':
'Landing and metadata tables created for collectors to populate.',
}
def connect(connection_name, options):
table_prefix = f'aws_collect' + ('' if connection_name in ('', 'default')
else connection_name)
table_name = f'{table_prefix}_organizations_list_accounts_connection'
landing_table = f'data.{table_name}'
audit_assumer_arn = options['audit_assumer_arn']
org_account_ids = options['org_account_ids']
audit_reader_role = options['audit_reader_role']
reader_eid = options.get('reader_eid', '')
comment = yaml_dump(
module='aws_collect',
audit_assumer_arn=audit_assumer_arn,
org_account_ids=org_account_ids,
audit_reader_role=audit_reader_role,
reader_eid=reader_eid,
collect_apis='all',
)
db.create_table(name=landing_table,
cols=LANDING_TABLE_COLUMNS,
comment=comment)
db.execute(f'GRANT INSERT, SELECT ON {landing_table} TO ROLE {SA_ROLE}')
for table_postfix, cols in SUPPLEMENTARY_TABLES.items():
supp_table = f'data.{table_prefix}_{table_postfix}'
db.create_table(name=supp_table, cols=cols)
db.execute(f'GRANT INSERT, SELECT ON {supp_table} TO ROLE {SA_ROLE}')
return {
'newStage': 'finalized',
'newMessage': "AWS Collect connector tables created.",
}
def record_status(results, alert_id):
try:
db.execute(
f"UPDATE results.alerts "
f"SET handled=PARSE_JSON(%s) "
f"WHERE alert:ALERT_ID='{alert_id}'",
params=[json_dumps(results)])
except Exception as e:
log.error(e,
f"Failed to update alert {alert_id} with status {results}")
def create_user_table(connection_name, options):
table_name = f'data.TENABLE_SETTINGS_{connection_name}_USER_CONNECTION'
token = options['token']
secret = options['secret']
comment = f"""
---
module: tenable_settings
token: {token}
secret: {secret}
"""
db.create_table(table_name, cols=USER_LANDING_TABLE, comment=comment)
db.execute(f'GRANT INSERT, SELECT ON {table_name} TO ROLE {SA_ROLE}')
def connect(connection_name, options):
connection_type = options['connection_type']
base_name = f'gsuite_logs_{connection_name}_{connection_type}'
landing_table = f'data.{base_name}_connection'
comment = yaml_dump(module='gsuite_logs', **options)
db.create_table(
name=landing_table, cols=LANDING_TABLES_COLUMNS['login'], comment=comment
)
db.execute(f'GRANT INSERT, SELECT ON data.{base_name}_connection TO ROLE {SA_ROLE}')
return {
'newStage': 'finalized',
'newMessage': 'Landing table created for collectors to populate.',
}
def connect(connection_name, options):
table_name = f'crowdstrike_devices_{connection_name}_connection'
landing_table = f'data.{table_name}'
comment = yaml_dump(module='crowdstrike_devices', **options)
db.create_table(name=landing_table,
cols=LANDING_TABLE_COLUMNS,
comment=comment)
db.execute(f'GRANT INSERT, SELECT ON {landing_table} TO ROLE {SA_ROLE}')
return {
'newStage': 'finalized',
'newMessage': "Crowdstrike Devices ingestion table created!",
}
def connect(connection_name, options):
base_name = f"azure_subscription_{connection_name}"
comment = yaml_dump(module='azure_subscription', **options)
db.create_table(
name=f'data.{base_name}_connection', cols=LANDING_TABLE_COLUMNS, comment=comment
)
db.execute(f'GRANT INSERT, SELECT ON data.{base_name}_connection TO ROLE {SA_ROLE}')
return {
'newStage': 'finalized',
'newMessage': 'Landing table created for collectors to populate.',
}
def connect(connection_name, options):
landing_table = f'data.assetpanda_{connection_name}_connection '
comment = yaml_dump(module='assetpanda', **options)
db.create_table(name=landing_table,
cols=LANDING_TABLE_COLUMNS,
comment=comment)
db.execute(f'GRANT INSERT, SELECT ON {landing_table} TO ROLE {SA_ROLE}')
return {
'newStage': 'finalized',
'newMessage': "AssetPanda ingestion tables created!",
}
def connect(connection_name, options):
ctype = options['connection_type']
ctable = f'data.tenable_io_{connection_name}_{ctype}_connection'
cols = {
'user': USER_LANDING_TABLE,
'agent': AGENT_LANDING_TABLE,
'vuln': VULN_LANDING_TABLE,
}[ctype]
comment = yaml_dump(module='tenable_io', **options)
db.create_table(ctable, cols=cols, comment=comment)
db.execute(f'GRANT INSERT, SELECT ON {ctable} TO ROLE {SA_ROLE}')
return {
'newStage': 'finalized',
'newMessage': 'Landing table created for collectors to populate.',
}
def create_asset_table(connection_name, asset_type, columns, options):
# create the tables, based on the config type (i.e. SG, EC2, ELB)
table_name = f'aws_asset_inv_{asset_type}_{connection_name}_connection'
landing_table = f'data.{table_name}'
aws_access_key = options['aws_access_key']
aws_secret_key = options['aws_secret_key']
comment = f'''
---
module: aws_inventory
aws_access_key: {aws_access_key}
aws_secret_key: {aws_secret_key}
'''
db.create_table(name=landing_table, cols=columns, comment=comment)
db.execute(f'GRANT INSERT, SELECT ON {landing_table} TO ROLE {SA_ROLE}')
return f"AWS {asset_type} asset ingestion table created!"
def connect(connection_name, options):
table_name = f'salesforce_events_{connection_name}'
landing_log_table = f'data.{table_name}_connection'
comment = yaml_dump(module='salesforce_event_log', **options)
db.create_table(
name=landing_log_table,
cols=LANDING_TABLE_COLUMNS,
comment=comment,
stage_file_format='TYPE = JSON STRIP_OUTER_ARRAY = TRUE',
stage_copy_options='PURGE = TRUE',
)
db.execute(
f'GRANT INSERT, SELECT ON {landing_log_table} TO ROLE {SA_ROLE}')
return {
'newStage': 'finalized',
'newMessage': "Salesforce Event Log ingestion table created!",
}
def ingest(table_name, options):
base_name = re.sub(r'_CONNECTION$', '', table_name)
storage_account = options['storage_account']
sas_token = vault.decrypt_if_encrypted(options['sas_token'])
suffix = options['suffix']
container_name = options['container_name']
snowflake_account = options['snowflake_account']
sa_user = options['sa_user']
database = options['database']
block_blob_service = BlockBlobService(account_name=storage_account,
sas_token=sas_token,
endpoint_suffix=suffix)
db.execute(f"select SYSTEM$PIPE_FORCE_RESUME('DATA.{base_name}_PIPE');")
last_loaded = db.fetch_latest(f'data.{table_name}', 'loaded_on')
log.info(f"Last loaded time is {last_loaded}")
blobs = block_blob_service.list_blobs(container_name)
new_files = [
StagedFile(b.name, None) for b in blobs
if (last_loaded is None or b.properties.creation_time > last_loaded)
]
log.info(f"Found {len(new_files)} files to ingest")
# Proxy object that abstracts the Snowpipe REST API
ingest_manager = SimpleIngestManager(
account=snowflake_account,
host=f'{snowflake_account}.snowflakecomputing.com',
user=sa_user,
pipe=f'{database}.data.{base_name}_PIPE',
private_key=load_pkb_rsa(PRIVATE_KEY, PRIVATE_KEY_PASSWORD))
if len(new_files) > 0:
for file_group in groups_of(4999, new_files):
response = ingest_manager.ingest_files(file_group)
log.info(response)
yield len(file_group)
def merge_alerts(query_name, from_time_sql):
log.info(f"{query_name} processing...")
sql = MERGE_ALERTS.format(
query_name=query_name,
from_time_sql=from_time_sql,
new_alerts_table=f"RUN_{RUN_ID}_{query_name}",
)
result = db.execute(sql, fix_errors=False).fetchall()
created_count, updated_count = result[0]
log.info(f"{query_name} created {created_count}, updated {updated_count} rows.")
return created_count, updated_count
def connect(connection_name, options):
landing_table_client = f'data.meraki_devices_{connection_name}_client_connection'
landing_table_device = f'data.meraki_devices_{connection_name}_device_connection'
comment = yaml_dump(module='meraki_devices', **options)
db.create_table(name=landing_table_client,
cols=LANDING_TABLE_COLUMNS_CLIENT,
comment=comment)
db.execute(
f'GRANT INSERT, SELECT ON {landing_table_client} TO ROLE {SA_ROLE}')
db.create_table(name=landing_table_device,
cols=LANDING_TABLE_COLUMNS_DEVICE,
comment=comment)
db.execute(
f'GRANT INSERT, SELECT ON {landing_table_device} TO ROLE {SA_ROLE}')
return {
'newStage': 'finalized',
'newMessage': "Meraki ingestion tables created!"
}
def connect(connection_name, options):
table_name = f'okta_{connection_name}_connection'
landing_table = f'data.{table_name}'
api_key = options['api_key']
subdomain = options['subdomain']
comment = f'''
---
module: okta
api_key: {api_key}
subdomain: {subdomain}
'''
db.create_table(name=landing_table,
cols=LANDING_TABLE_COLUMNS,
comment=comment)
db.execute(f'GRANT INSERT, SELECT ON {landing_table} TO ROLE {SA_ROLE}')
return {
'newStage': 'finalized',
'newMessage': "Okta ingestion table created!",
}
def connect(connection_name, options):
table_name = f'okta_{connection_name}'
landing_log_table = f'data.{table_name}_connection'
landing_user_table = f'data.{table_name}_users_connection'
landing_group_table = f'data.{table_name}_groups_connection'
comment = yaml_dump(module='okta', **options)
db.create_table(name=landing_log_table,
cols=LANDING_LOG_TABLE_COLUMNS,
comment=comment)
db.execute(
f'GRANT INSERT, SELECT ON {landing_log_table} TO ROLE {SA_ROLE}')
db.create_table(name=landing_user_table,
cols=LANDING_USER_TABLE_COLUMNS,
comment=comment)
db.execute(
f'GRANT INSERT, SELECT ON {landing_user_table} TO ROLE {SA_ROLE}')
db.create_table(name=landing_group_table,
cols=LANDING_GROUP_TABLE_COLUMNS,
comment=comment)
db.execute(
f'GRANT INSERT, SELECT ON {landing_group_table} TO ROLE {SA_ROLE}')
return {
'newStage': 'finalized',
'newMessage': "Okta ingestion table, user table, group table created!",
}
def create_asset_table(connection_name, asset_type, columns, options):
# create the tables, based on the config type (i.e. SG, EC2, ELB)
table_name = f'aws_asset_inv_{asset_type}_{connection_name}_connection'
landing_table = f'data.{table_name}'
comment = yaml_dump(module='aws_inventory', **options)
db.create_table(name=landing_table, cols=columns, comment=comment)
metadata_cols = [
('snapshot_at', 'TIMESTAMP_LTZ'),
('run_id', 'VARCHAR(100)'),
('account_id', 'VARCHAR(100)'),
('account_alias', 'VARCHAR(100)'),
(f'{asset_type}_count', 'NUMBER'),
('error', 'VARCHAR'),
]
create_metadata_table(table=AWS_ACCOUNTS_METADATA,
cols=metadata_cols,
addition=metadata_cols[4])
db.execute(f'GRANT INSERT, SELECT ON {landing_table} TO ROLE {SA_ROLE}')
return f"AWS {asset_type} asset ingestion table created!"
def connect(connection_name, options):
table_name = f'aws_accounts_{connection_name}_connection'
landing_table = f'data.{table_name}'
source_role_arn = options['source_role_arn']
destination_role_arn = options['destination_role_arn']
destination_role_external_id = options['destination_role_external_id']
comment = yaml_dump(
module='aws_accounts',
source_role_arn=source_role_arn,
destination_role_arn=destination_role_arn,
destination_role_external_id=destination_role_external_id,
)
db.create_table(name=landing_table,
cols=LANDING_TABLE_COLUMNS,
comment=comment)
db.execute(f'GRANT INSERT, SELECT ON {landing_table} TO ROLE {SA_ROLE}')
return {
'newStage': 'finalized',
'newMessage': "AWS Account ingestion table created!",
}
def update_recent_alerts(ctx, alert_map):
update_array = []
update_array_length = 0
for key in alert_map:
if alert_map[key][2]:
update_array.extend(
[alert_map[key][0]['ALERT_ID'], alert_map[key][1]])
update_array_length = update_array_length + 1
if update_array_length:
format_string = ", ".join(["(%s, %s)"] * update_array_length)
db.execute(ctx,
f"DROP TABLE IF EXISTS {RESULTS_SCHEMA}.counter_table;")
db.execute(
ctx,
f"CREATE TEMPORARY TABLE {RESULTS_SCHEMA}.counter_table(ALERT_ID string, COUNTER number);"
)
ctx.cursor().execute(
f"INSERT INTO {RESULTS_SCHEMA}.counter_table (ALERT_ID, COUNTER) VALUES {format_string};",
update_array)
ctx.cursor().execute(
f"MERGE INTO {ALERTS_TABLE} s"
f" USING {RESULTS_SCHEMA}.counter_table t"
f" ON s.alert:ALERT_ID = t.ALERT_ID WHEN MATCHED THEN UPDATE"
f" SET s.COUNTER = t.COUNTER;")
def delete_results():
yield
db.execute(f"DELETE FROM results.alerts")
db.execute(f"DELETE FROM results.violations")
db.execute(f"DELETE FROM results.run_metadata")
db.execute(f"DELETE FROM results.query_metadata")
def connect(connection_name, options):
connection_type = options['connection_type']
base_name = f"azure_log_{connection_name}_{connection_type}"
account_name = options['account_name']
container_name = options['container_name']
suffix = options['suffix']
cloud_type = options['cloud_type']
sas_token = options['sas_token']
comment = yaml_dump(module='azure_log')
db.create_stage(
name=f'data.{base_name}_stage',
url=f"azure://{account_name}.blob.{suffix}/{container_name}",
cloud='azure',
prefix='',
credentials=sas_token,
file_format=FILE_FORMAT,
)
db.execute(
f'GRANT USAGE ON STAGE data.{base_name}_stage TO ROLE {SA_ROLE}')
db.create_table(
name=f'data.{base_name}_connection',
cols=LANDING_TABLES_COLUMNS[connection_type],
comment=comment,
ifnotexists=True,
)
db.execute(
f'GRANT INSERT, SELECT ON data.{base_name}_connection TO ROLE {SA_ROLE}'
)
external_table_columns = [(
'timestamp_part',
'TIMESTAMP_LTZ',
GET_TIMESTAMP_FROM_FILENAME_SQL[connection_type],
)]
db.create_external_table(
name=f'data.{base_name}_external',
location=f'@data.{base_name}_stage',
cols=external_table_columns,
partition='timestamp_part',
file_format=db.TypeOptions(type='JSON'),
)
db.execute(f'GRANT SELECT ON data.{base_name}_external TO ROLE {SA_ROLE}')
stored_proc_def = f"""
var sql_command = "ALTER EXTERNAL TABLE data.{base_name}_external REFRESH";
try {{
snowflake.execute ({{sqlText: sql_command}});
return "Succeeded.";
}} catch (err) {{
return "Failed: " + err;
}}
"""
db.create_stored_procedure(
name=f'data.{base_name}_procedure',
args=[],
return_type='string',
executor='OWNER',
definition=stored_proc_def,
)
refresh_task_sql = f'CALL data.{base_name}_procedure()'
db.create_task(
name=f'data.{base_name}_refresh_task',
warehouse=WAREHOUSE,
schedule='5 minutes',
sql=refresh_task_sql,
)
select_statement_sql = {
'reg':
(f"SELECT value "
f"FROM data.{base_name}_external "
f"WHERE timestamp_part >= DATEADD(HOUR, -2, CURRENT_TIMESTAMP())"),
'gov':
(f"SELECT value FROM ("
f" SELECT value AS a "
f" FROM data.{base_name}_external"
f" WHERE timestamp_part >= DATEADD(HOUR, -2, CURRENT_TIMESTAMP())"
f"), LATERAL FLATTEN (INPUT => a:records)"),
}
insert_task_sql = {
'operation':
f"""
INSERT (
RAW, HASH_RAW, CALLER_IP_ADDRESS, CATEGORY, CORRELATION_ID, DURATION_MS,
IDENTITY, IDENTITY_AUTHORIZATION, IDENTITY_CLAIMS, LEVEL, LOCATION,
OPERATION_NAME, PROPERTIES, PROPERTIES_ANCESTORS, PROPERTIES_IS_COMPLIANCE_CHECK,
PROPERTIES_POLICIES, PROPERTIES_RESOURCE_LOCATION, RESOURCE_ID, RESULT_SIGNATURE,
RESULT_TYPE, EVENT_TIME, LOADED_ON
) VALUES (
VALUE, HASH(VALUE), VALUE:callerIpAddress::STRING, VALUE:category::STRING, VALUE:correlationId::STRING,
VALUE:durationMs::NUMBER, VALUE:identity::VARIANT, VALUE:identity.authorization::VARIANT,
VALUE:identity.claims::VARIANT, VALUE:level::STRING, VALUE:location::STRING, VALUE:operationName::STRING,
VALUE:properties::VARIANT, VALUE:properties.ancestors::STRING, VALUE:properties.isComplianceCheck::STRING,
PARSE_JSON(VALUE:properties.policies),VALUE:properties.resourceLocation::STRING, VALUE:resourceId::STRING,
VALUE:resultSignature::STRING,VALUE:resultType::STRING, value:time::TIMESTAMP_LTZ, CURRENT_TIMESTAMP()
)
""",
'audit':
f"""
INSERT (
RAW, HASH_RAW, CALLER_IP_ADDRESS, CATEGORY, CORRELATION_ID,
DURATION_MS, LEVEL, OPERATION_NAME, OPERATION_VERSION, PROPERTIES,
PROPERTIES_ACTIVITY_DATE_TIME, PROPERTIES_ACTIVITY_DISPLAY_NAME,
PROPERTIES_ADDITIONAL_DETAILS, PROPERTIES_CATEGORY, PROPERTIES_ID,
PROPERTIES_INITIATED_BY, PROPERTIES_LOGGED_BY_SERVICE, PROPERTIES_OPERATION_TYPE,
PROPERTIES_RESULT, PROPERTIES_RESULT_REASON, PROPERTIES_TARGET_RESOURCES,
RESOURCE_ID, RESULT_SIGNATURE, TENANT_ID, EVENT_TIME, LOADED_ON
) VALUES (
VALUE, HASH(VALUE), VALUE:callerIpAddress::STRING, VALUE:category::STRING, VALUE:correlationId::STRING,
VALUE:durationMs::NUMBER, VALUE:level::STRING, VALUE:operationName::STRING, VALUE:operationVersion::STRING,
VALUE:properties::VARIANT, VALUE:properties.activityDateTime::TIMESTAMP_LTZ,
VALUE:properties.activityDisplayName::STRING, VALUE:properties.additionalDetails::VARIANT,
VALUE:properties.category::STRING, VALUE:properties.id::STRING, VALUE:properties.initiatedBy::VARIANT,
VALUE:properties.loggedByService::STRING, VALUE:properties.operationType::STRING, VALUE:properties.result::STRING,
VALUE:resultReason::STRING, VALUE:properties.targetResources::VARIANT, VALUE:resourceId::STRING,
VALUE:resultSignature::STRING, VALUE:tenantId::STRING, VALUE:time::TIMESTAMP_LTZ, CURRENT_TIMESTAMP()
)
""",
'signin':
f"""
INSERT (
RAW, HASH_RAW, LEVEL, CALLER_IP_ADDRESS, CATEGORY, CORRELATION_ID, DURATION_MS,
IDENTITY, LOCATION, OPERATION_NAME, OPERATION_VERSION, PROPERTIES,
PROPERTIES_APP_DISPLAY_NAME, PROPERTIES_APP_ID,
PROPERTIES_APPLIED_CONDITIONAL_ACESS_POLICIES, PROPERTIES_AUTHENTICATION_METHODS_USED,
PROPERTIES_AUTHENTICATION_PROCESSING_DETAILS, PROPERTIES_CLIENT_APP_USED,
PROPERTIES_CONDITIONAL_ACCESS_STATUS, PROPERTIES_CREATED_DATE_TIME,
PROPERTIES_DEVICE_DETAIL, PROPERTIES_ID, PROPERTIES_IP_ADDRESS, PROPERTIES_IS_INTERACTIVE, PROPERTIES_LOCATION,
PROPERTIES_MFA_DETAIL, PROPERTIES_NETWORK_LOCATION, PROPERTIES_PROCESSING_TIME_IN_MILLISECONDS,
PROPERTIES_RESOURCE_DISPLAY_NAME, PROPERTIES_RESOURCE_ID, PROPERTIES_RISK_DETAIL,
PROPERTIES_RISK_EVENT_TYPES, PROPERTIES_RISK_LEVEL_AGGREGATED, PROPERTIES_RISK_LEVEL_DURING_SIGNIN,
PROPERTIES_RISK_STATE, PROPERTIES_STATUS, PROPERTIES_TOKEN_ISSUER_TYPE, PROPERTIES_USER_DISPLAY_NAME,
PROPERTIES_USER_ID, PROPERTIES_USER_PRINCIPAL_NAME, RESOURCE_ID, RESULT_DESCRIPTION, RESULT_SIGNATURE,
RESULT_TYPE, TENANT_ID, EVENT_TIME, LOADED_ON
) VALUES (
VALUE, HASH(VALUE), VALUE:Level::NUMBER, VALUE:callerIpAddress::STRING, VALUE:category::STRING,
VALUE:correlationId::STRING, VALUE:durationMs, VALUE:identity::STRING, VALUE:location::STRING,
VALUE:operationName::STRING, VALUE:operationVersion::STRING, VALUE:properties::VARIANT,
VALUE:properties.appDisplayName::STRING, VALUE:properties.appId::STRING,
VALUE:properties.appliedConditionalAccessPolicies::VARIANT, VALUE:properties.authenticationMethodsUsed::VARIANT,
VALUE:properties.authenticationProcessingDetails::VARIANT, VALUE:properties.clientAppUsed::STRING,
VALUE:properties.conditionalAccessStatus::STRING, VALUE:properties.createdDateTime::TIMESTAMP_LTZ,
VALUE:properties.deviceDetail::VARIANT, VALUE:properties.id::STRING, VALUE:properties.ipAddress::STRING,
VALUE:properties.isInteractive::BOOLEAN, VALUE:properties.location::VARIANT,
VALUE:properties.mfaDetail::VARIANT, VALUE:properties.networkLocationDetails::VARIANT,
VALUE:properties.processingTimeInMilliseconds::NUMBER, VALUE:properties.resourceDisplayName::STRING,
VALUE:properties.resourceId::STRING, VALUE:properties.riskDetail::STRING,
VALUE:properties.riskEventTypes::VARIANT, VALUE:properties.riskLevelAggregated::STRING,
VALUE:properties.riskLevelDuringSignIn::STRING, VALUE:properties.riskState::VARIANT,
VALUE:properties.status::VARIANT, VALUE:properties.tokenIssuerType::STRING,
VALUE:properties.userDisplayName::STRING, VALUE:properties.userId::STRING,
VALUE:properties.userPrincipalName::STRING, VALUE:resourceId::STRING, VALUE:resultDescription::STRING,
VALUE:resultSignature::STRING, VALUE:resultType::STRING, VALUE:tenantId::STRING, VALUE:time::TIMESTAMP_LTZ,
CURRENT_TIMESTAMP()
)
""",
}
ingest_task_sql = f"""
MERGE INTO data.{base_name}_connection a
USING (
{select_statement_sql[cloud_type]}
) b
ON a.raw = b.value
WHEN NOT MATCHED THEN
{insert_task_sql[connection_type]}
"""
db.create_task(
name=f'data.{base_name}_ingest_task',
warehouse=WAREHOUSE,
schedule=f'AFTER data.{base_name}_refresh_task',
sql=ingest_task_sql,
)
return {
'newStage': 'finalized',
'newMessage': 'Created Stage, Tables, Stored Procedure, and Tasks.',
}
def sample_alert_rule(db_schemas):
db.execute(TEST_ALERT)
yield
db.execute("DROP VIEW rules.simple_alert_query")
def finalize(connection_name):
base_name = f'AWS_CLOUDTRAIL_{connection_name}_EVENTS'.upper()
pipe = f'data.{base_name}_PIPE'
landing_table = f'data.{base_name}_CONNECTION'
# Step two: Configure the remainder once the role is properly configured.
cloudtrail_ingest_task = f'''
INSERT INTO {landing_table} (
insert_time, raw, hash_raw, event_time, aws_region, event_id, event_name, event_source, event_type,
event_version, recipient_account_id, request_id, request_parameters, response_elements, source_ip_address,
user_agent, user_identity, user_identity_type, user_identity_principal_id, user_identity_arn,
user_identity_accountid, user_identity_invokedby, user_identity_access_key_id, user_identity_username,
user_identity_session_context_attributes_mfa_authenticated, user_identity_session_context_attributes_creation_date,
user_identity_session_context_session_issuer_type, user_identity_session_context_session_issuer_principal_id,
user_identity_session_context_session_issuer_arn, user_identity_session_context_session_issuer_account_id,
user_identity_session_context_session_issuer_user_name, error_code, error_message, additional_event_data,
api_version, read_only, resources, service_event_details, shared_event_id, vpc_endpoint_id
)
SELECT CURRENT_TIMESTAMP() insert_time
, value raw
, HASH(value) hash_raw
--- In the rare event of an unparsable timestamp, the following COALESCE keeps the pipeline from failing.
--- Compare event_time to TRY_TO_TIMESTAMP(raw:eventTime::STRING) to establish if the timestamp was parsed.
, COALESCE(
TRY_TO_TIMESTAMP(value:eventTime::STRING)::TIMESTAMP_LTZ(9),
CURRENT_TIMESTAMP()
) event_time
, value:awsRegion::STRING aws_region
, value:eventID::STRING event_id
, value:eventName::STRING event_name
, value:eventSource::STRING event_source
, value:eventType::STRING event_type
, value:eventVersion::STRING event_version
, value:recipientAccountId::STRING recipient_account_id
, value:requestID::STRING request_id
, value:requestParameters::VARIANT request_parameters
, value:responseElements::VARIANT response_elements
, value:sourceIPAddress::STRING source_ip_address
, value:userAgent::STRING user_agent
, value:userIdentity::VARIANT user_identity
, value:userIdentity.type::STRING user_identity_type
, value:userIdentity.principalId::STRING user_identity_principal_id
, value:userIdentity.arn::STRING user_identity_arn
, value:userIdentity.accountId::STRING user_identity_accountid
, value:userIdentity.invokedBy::STRING user_identity_invokedby
, value:userIdentity.accessKeyId::STRING user_identity_access_key_id
, value:userIdentity.userName::STRING user_identity_username
, value:userIdentity.sessionContext.attributes.mfaAuthenticated::STRING user_identity_session_context_attributes_mfa_authenticated
, value:userIdentity.sessionContext.attributes.creationDate::STRING user_identity_session_context_attributes_creation_date
, value:userIdentity.sessionContext.sessionIssuer.type::STRING user_identity_session_context_session_issuer_type
, value:userIdentity.sessionContext.sessionIssuer.principalId::STRING user_identity_session_context_session_issuer_principal_id
, value:userIdentity.sessionContext.sessionIssuer.arn::STRING user_identity_session_context_session_issuer_arn
, value:userIdentity.sessionContext.sessionIssuer.accountId::STRING user_identity_session_context_session_issuer_account_id
, value:userIdentity.sessionContext.sessionIssuer.userName::STRING user_identity_session_context_session_issuer_user_name
, value:errorCode::STRING error_code
, value:errorMessage::STRING error_message
, value:additionalEventData::VARIANT additional_event_data
, value:apiVersion::STRING api_version
, value:readOnly::BOOLEAN read_only
, value:resources::VARIANT resources
, value:serviceEventDetails::STRING service_event_details
, value:sharedEventId::STRING shared_event_id
, value:vpcEndpointId::STRING vpc_endpoint_id
FROM data.{base_name}_STREAM, table(flatten(input => v:Records))
WHERE ARRAY_SIZE(v:Records) > 0
'''
db.create_stream(
name=f'data.{base_name}_STREAM',
target=f'data.{base_name}_STAGING'
)
# IAM change takes 5-15 seconds to take effect
sleep(5)
db.retry(
lambda: db.create_pipe(
name=pipe,
sql=f"COPY INTO data.{base_name}_STAGING(v) FROM @data.{base_name}_STAGE/",
replace=True,
autoingest=True
),
n=10,
sleep_seconds_btw_retry=1
)
db.create_task(name=f'data.{base_name}_TASK', schedule='1 minute',
warehouse=WAREHOUSE, sql=cloudtrail_ingest_task)
db.execute(f"ALTER PIPE {pipe} REFRESH")
pipe_description = list(db.fetch(f'DESC PIPE {pipe}'))
if len(pipe_description) < 1:
return {
'newStage': 'error',
'newMessage': f"{pipe} doesn't exist; please reach out to Snowflake Security for assistance."
}
else:
sqs_arn = pipe_description[0]['notification_channel']
return {
'newStage': 'finalized',
'newMessage': (
f"Please add this SQS Queue ARN to the bucket event notification"
f"channel for all object create events: {sqs_arn}"
)
}
def sample_alert_rules(db_schemas):
db.execute(TEST_ALERT)
db.execute(TEST_SUPPRESSED_ALERT)
db.execute(TEST_SUPPRESSION)
db.execute(TEST_CORRELATED_ALERT)
db.execute(TEST_ALERT_WITH_SLACK_HANDLER)
db.execute(
f"""
CREATE OR REPLACE VIEW rules.__suppress_sample_alerts_alert_suppression COPY GRANTS
COMMENT='this should suppress anything not a test alert'
AS
SELECT id
FROM data.alerts
WHERE suppressed IS NULL
AND query_name NOT ILIKE '_TEST%'
"""
)
yield
db.execute(f"DROP VIEW rules._test1_alert_query")
db.execute(f"DROP VIEW rules._test2_alert_query")
db.execute(f"DROP VIEW rules._test2_alert_suppression")
db.execute(f"DROP VIEW rules._test3_alert_query")
db.execute(f"DROP VIEW rules._test4_alert_query")
db.execute(f"DROP VIEW rules.__suppress_sample_alerts_alert_suppression")
def connect(connection_name, options):
connection_type = options['connection_type']
base_name = f"azure_{connection_name}_{connection_type}"
account_name = options['account_name']
container_name = options['container_name']
suffix = options['suffix']
sas_token = options['sas_token']
sas_token_ct = vault.encrypt(sas_token)
comment = f'''
---
module: azure
storage_account: {account_name}
container_name: {container_name}
suffix: {suffix}
sas_token: {sas_token_ct}
sa_user: {USER}
snowflake_account: {ACCOUNT}
database: {DATABASE}
'''
db.create_stage(
name=f'data.{base_name}_STAGE',
url=f"azure://{account_name}.blob.{suffix}/{container_name}",
cloud='azure',
prefix='',
credentials=sas_token,
file_format=FILE_FORMAT)
db.execute(
f'GRANT USAGE ON STAGE data.{base_name}_STAGE TO ROLE {SA_ROLE}')
db.create_table(name=f'data.{base_name}_CONNECTION',
cols=LANDING_TABLES_COLUMNS[connection_type],
comment=comment)
db.execute(
f'GRANT INSERT, SELECT ON data.{base_name}_CONNECTION TO ROLE {SA_ROLE}'
)
pipe_sql = {
'operation':
f'''
COPY INTO DATA.{base_name}_CONNECTION(RAW, HASH_RAW, CALLER_IP_ADDRESS, CATEGORY, CORRELATION_ID, DURATION_MS,
IDENTITY, IDENTITY_AUTHORIZATION, IDENTITY_CLAIMS, LEVEL, LOCATION,
OPERATION_NAME, PROPERTIES, PROPERTIES_ANCESTORS, PROPERTIES_IS_COMPLIANCE_CHECK,
PROPERTIES_POLICIES, PROPERTIES_RESOURCE_LOCAATION, RESOURCE_ID, RESULT_SIGNATURE,
RESULT_TYPE, EVENT_TIME, LOADED_ON)
FROM (
SELECT $1, HASH($1), $1:callerIpAddress::STRING, $1:category::STRING, $1:correlationId::STRING,
$1:durationMs::NUMBER, $1:identity::VARIANT, $1:identity.authorization::VARIANT, $1:identity.claims::VARIANT,
$1:level::STRING, $1:location::STRING, $1:operationName::STRING, $1:properties::VARIANT,
$1:properties.ancestors::STRING, $1:properties.isComplianceCheck::STRING, PARSE_JSON($1:properties.policies),
$1:properties.resourceLocation::STRING, $1:resourceId::STRING, $1:resultSignature::STRING,
$1:resultType::STRING, $1:time::TIMESTAMP_LTZ, CURRENT_TIMESTAMP()
FROM @DATA.{base_name}_STAGE)
''',
'audit':
f'''
COPY INTO data.{base_name}_CONNECTION (RAW, HASH_RAW, CALLER_IP_ADDRESS, CATEGORY, CORRELATION_ID,
DURATION_MS, LEVEL, OPERATION_NAME, OPERATION_VERSION, PROPERTIES,
PROPERTIES_ACTIVITY_DATE_TIME, PROPERTIES_ACTIVITY_DISPLAY_NAME,
PROPERTIES_ADDITIONAL_DETAILS, PROPERTIES_CATEGORY, PROPERTIES_ID,
PROPERTIES_INITIATED_BY, PROPERTIES_LOGGED_BY_SERVICE, PROPERTIES_OPERATION_TYPE,
PROPERTIES_RESULT, PROPERTIES_RESULT_REASON, PROPERTIES_TARGET_RESOURCES,
RESOURCE_ID, RESULT_SIGNATURE, TENANT_ID, EVENT_TIME, LOADED_ON)
FROM (
SELECT $1, HASH($1), $1:callerIpAddress::STRING, $1:category::STRING, $1:correlationId::STRING,
$1:durationMs::NUMBER, $1:level::STRING, $1:operationName::STRING, $1:operationVersion::STRING,
$1:properties::VARIANT, $1:properties.activityDateTime::TIMESTAMP_LTZ,
$1:properties.activityDisplayName::STRING, $1:properties.additionalDetails::VARIANT,
$1:properties.category::STRING, $1:properties.id::STRING, $1:properties.initiatedBy::VARIANT,
$1:properties.loggedByService::STRING, $1:properties.operationType::STRING, $1:properties.result::STRING,
$1:resultReason::STRING, $1:properties.targetResources::VARIANT, $1:resourceId::STRING,
$1:resultSignature::STRING, $1:tenantId::STRING, $1:time::TIMESTAMP_LTZ, CURRENT_TIMESTAMP()
FROM @data.{base_name}_STAGE
)
''',
'signin':
f'''
COPY INTO DATA.{base_name}_CONNECTION (
RAW, HASH_RAW, LEVEL, CALLER_IP_ADDRESS, CATEGORY, CORRELATION_ID, DURATION_MS,
IDENTITY, LOCATION, OPERATION_NAME, OPERATION_VERSION, PROPERTIES,
PROPERTIES_APP_DISPLAY_NAME, PROPERTIES_APP_ID,
PROPERTIES_APPLIED_CONDITIONAL_ACESS_POLICIES, PROPERTIES_AUTHENTICATION_METHODS_USED,
PROPERTIES_AUTHENTICATION_PROCESSING_DETAILS, PROPERTIES_CLIENT_APP_USED,
PROPERTIES_CONDITIONAL_ACCESS_STATUS, PROPERTIES_CREATED_DATE_TIME,
PROPERTIES_DEVICE_DETAIL, PROPERTIES_ID, PROPERTIES_IP_ADDRESS, PROPERTIES_IS_INTERACTIVE, PROPERTIES_LOCATION,
PROPERTIES_MFA_DETAIL, PROPERTIES_NETWORK_LOCATION, PROPERTIES_PROCESSING_TIME_IN_MILLISECONDS,
PROPERTIES_RESOURCE_DISPLAY_NAME, PROPERTIES_RESOURCE_ID, PROPERTIES_RISK_DETAIL,
PROPERTIES_RISK_EVENT_TYPES, PROPERTIES_RISK_LEVEL_AGGREGATED, PROPERTIES_RISK_LEVEL_DURING_SIGNIN,
PROPERTIES_RISK_STATE, PROPERTIES_STATUS, PROPERTIES_TOKEN_ISSUER_TYPE, PROPERTIES_USER_DISPLAY_NAME,
PROPERTIES_USER_ID, PROPERTIES_USER_PRINCIPAL_NAME, RESOURCE_ID, RESULT_DESCRIPTION, RESULT_SIGNATURE,
RESULT_TYPE, TENANT_ID, EVENT_TIME, LOADED_ON
)
FROM (
SELECT $1, HASH($1), $1:Level::NUMBER, $1:callerIpAddress::STRING, $1:category::STRING, $1:correlationId::STRING,
$1:durationMs, $1:identity::STRING, $1:location::STRING, $1:operationName::STRING,
$1:operationVersion::STRING, $1:properties::VARIANT, $1:properties.appDisplayName::STRING,
$1:properties.appId::STRING, $1:properties.appliedConditionalAccessPolicies::VARIANT,
$1:properties.authenticationMethodsUsed::VARIANT, $1:properties.authenticationProcessingDetails::VARIANT,
$1:properties.clientAppUsed::STRING, $1:properties.conditionalAccessStatus::STRING,
$1:properties.createdDateTime::TIMESTAMP_LTZ, $1:properties.deviceDetail::VARIANT, $1:properties.id::STRING,
$1:properties.ipAddress::STRING, $1:properties.isInteractive::BOOLEAN, $1:properties.location::VARIANT,
$1:properties.mfaDetail::VARIANT, $1:properties.networkLocationDetails::VARIANT,
$1:properties.processingTimeInMilliseconds::NUMBER, $1:properties.resourceDisplayName::STRING,
$1:properties.resourceId::STRING, $1:properties.riskDetail::STRING, $1:properties.riskEventTypes::VARIANT,
$1:properties.riskLevelAggregated::STRING, $1:properties.riskLevelDuringSignIn::STRING,
$1:properties.riskState::VARIANT, $1:properties.status::VARIANT, $1:properties.tokenIssuerType::STRING,
$1:properties.userDisplayName::STRING, $1:properties.userId::STRING, $1:properties.userPrincipalName::STRING,
$1:resourceId::STRING, $1:resultDescription::STRING, $1:resultSignature::STRING, $1:resultType::STRING,
$1:tenantId::STRING, $1:time::TIMESTAMP_LTZ,
CURRENT_TIMESTAMP()
FROM @DATA.{base_name}_STAGE
)
'''
}
db.create_pipe(name=f"data.{base_name}_PIPE",
sql=pipe_sql[options['connection_type']],
replace=True)
db.execute(
f'ALTER PIPE data.{base_name}_PIPE SET PIPE_EXECUTION_PAUSED=true')
db.execute(
f'GRANT OWNERSHIP ON PIPE data.{base_name}_PIPE TO ROLE {SA_ROLE}')
return {
'newStage': 'finalized',
'newMessage': 'Table, Stage, and Pipe created'
}
