class StructuredDataSchema(ApiStrictSchema):
type = fields.String(allow_none=True)
columns_names = fields.List(fields.String(),
load_from="columns",
dump_to="columns",
allow_none=True)
columns_types = fields.Dict(load_from="dtypes",
dump_to="dtypes",
allow_none=True)
# Shape: Tuple of ints-> (records, columns) -> fields.Tuple() added in marshmallow 3.0.0.
shape = fields.List(fields.Integer(allow_none=True), allow_none=True)
byte_size = fields.Integer(load_from="size.bytes",
dump_to="size.bytes",
allow_none=True)
@pre_load
def pre_load(self, data: dict) -> dict:
# Support Snowflake & PostgreSQL data_schemas before dbnd-sdk-0.61.0
columns_types = data.get("column_types")
if columns_types:
data["columns_types"] = columns_types
return data
@post_load
def make_object(self, data) -> DataSchemaArgs:
return DataSchemaArgs(**data)
class TaskRunParamSchema(ApiObjectSchema):
parameter_name = fields.String()
value_origin = fields.String()
value = fields.String()
@post_load
def make_task_run_param(self, data, **kwargs):
return TaskRunParamInfo(**data)
class UpdateDagRunsRequestSchema(ApiStrictSchema):
dag_runs = fields.Nested(DagRunSchema, many=True)
task_instances = fields.Nested(TaskInstanceSchema, many=True)
last_seen_log_id = fields.Integer(allow_none=True)
airflow_export_meta = fields.Nested(AirflowExportMetaSchema, required=False)
error_message = fields.String(required=False, allow_none=True)
syncer_type = fields.String(allow_none=True)
class AirflowTaskContextSchema(ApiObjectSchema):
dag_id = fields.String()
execution_date = fields.String()
task_id = fields.String()
try_number = fields.Integer(allow_none=True)
@post_load
def make_run_info(self, data, **kwargs):
return AirflowTaskContext(**data)
class DagRunSchema(ApiStrictSchema):
dag_id = fields.String(allow_none=True)
run_id = fields.String(required=False)
dagrun_id = fields.Integer()
start_date = fields.DateTime(allow_none=True)
state = fields.String()
end_date = fields.DateTime(allow_none=True)
execution_date = fields.DateTime(allow_none=True)
task_args = fields.Dict()
class LogTargetMetricsSchema(_ApiCallSchema):
task_run_uid = fields.UUID(required=True)
task_run_attempt_uid = fields.UUID(required=True)
target_path = fields.String()
value_preview = fields.String(allow_none=True)
data_dimensions = fields.List(fields.Integer(), allow_none=True)
data_schema = fields.String(allow_none=True)
class ScheduledRunInfoSchema(ApiObjectSchema):
scheduled_job_uid = fields.UUID(allow_none=True)
scheduled_date = fields.DateTime(allow_none=True)
scheduled_job_dag_run_id = fields.String(allow_none=True)
scheduled_job_name = fields.String(allow_none=True)
@post_load
def make_object(self, data, **kwargs):
return _as_dotted_dict(**data)
class BaseSourceMonitorStateSchema(ApiStrictSchema):
monitor_version = fields.String(required=False, allow_none=True)
monitor_status = fields.String(required=False, allow_none=True)
monitor_error_message = fields.String(required=False, allow_none=True)
monitor_start_time = fields.DateTime(required=False, allow_none=True)
@post_load
def make_object(self, data):
return BaseSourceMonitorState(**data)
class MLSaveModelSchema(_ApiCallSchema):
model = fields.Dict() # pickel
job_name = fields.String()
is_enabled = fields.Boolean(allow_none=True)
data_set = fields.String(allow_none=True) # df?
@post_load
def make_object(self, data, **kwargs):
return MLSaveModel(**data)
class TaskInstanceSchema(ApiStrictSchema):
execution_date = fields.DateTime()
dag_id = fields.String()
state = fields.String(allow_none=True)
try_number = fields.Integer()
task_id = fields.String()
start_date = fields.DateTime(allow_none=True)
end_date = fields.DateTime(allow_none=True)
log_body = fields.String(allow_none=True)
xcom_dict = fields.Dict()
class TargetInfoSchema(ApiObjectSchema):
parameter_name = fields.String()
path = fields.String()
created_date = fields.DateTime(allow_none=True)
task_run_uid = fields.UUID(allow_none=True)
@post_load
def make_target(self, data, **kwargs):
return TargetInfo(**data)
class AirflowTaskInfoSchema(_ApiCallSchema):
execution_date = fields.DateTime()
last_sync = fields.DateTime(allow_none=True)
dag_id = fields.String()
task_id = fields.String()
task_run_attempt_uid = fields.UUID()
retry_number = fields.Integer(required=False, allow_none=True)
@post_load
def make_object(self, data, **kwargs):
return AirflowTaskInfo(**data)
class SyncedTransactionOperationSchema(ApiStrictSchema):
op_type = fields.String()
started_date = fields.DateTime() # type: datetime
records_count = fields.Integer()
dataset_uri = fields.String()
dataset_uid = fields.UUID(allow_none=True)
@post_load
def make_object(self, data):
return SyncedTransactionOperation(**data)
class MetricSchema(ApiStrictSchema):
key = fields.String()
value = fields.Raw(allow_none=True)
value_str = fields.String(allow_none=True)
value_json = fields.Raw(allow_none=True)
value_int = fields.Integer(allow_none=True)
value_float = fields.Float(allow_none=True)
timestamp = fields.DateTime()
@post_load
def make_object(self, data, **kwargs):
return Metric(**data)
class SyncedTransactionsReportSchema(ApiStrictSchema):
sync_event_uid = fields.UUID()
sync_event_timestamp = fields.DateTime()
source_type = fields.String(allow_none=True) # bigquery / snowflake / etc
syncer_type = fields.String(allow_none=True)
monitor_state = fields.Nested(DatasourceMonitorStateSchema)
transactions = fields.Nested(SyncedTransactionSchema, many=True)
@post_load
def make_object(self, data):
return SyncedTransactionsReport(**data)
class DatasetsReportSchema(ApiStrictSchema):
sync_event_uid = fields.UUID()
monitor_state = fields.Nested(DatasourceMonitorStateSchema)
source_type = fields.String(allow_none=True) # bigquery / snowflake / etc
syncer_type = fields.String(allow_none=True)
sync_event_timestamp = fields.DateTime(required=False, allow_none=True)
datasets = fields.Nested(SyncedDatasetSchema, many=True)
@post_load
def make_object(self, data):
return DatasetsReport(**data)
class MetricSchema(ApiObjectSchema):
key = fields.String()
value = fields.String(allow_none=True)
value_int = fields.Integer(allow_none=True)
value_float = fields.Float(allow_none=True)
timestamp = fields.DateTime()
@post_load
def make_object(self, data, **kwargs):
from dbnd._core.tracking.metrics import Metric
return Metric(**data)
class ErrorInfoSchema(ApiObjectSchema):
msg = fields.String()
help_msg = fields.String(allow_none=True)
databand_error = fields.Bool()
exc_type = fields.Function(lambda obj: str(obj.exc_type), allow_none=True)
traceback = fields.String()
nested = fields.String(allow_none=True)
user_code_traceback = fields.String()
show_exc_info = fields.Bool()
@post_load
def make_object(self, data, **kwargs):
return ErrorInfo(**data)
class TaskSchema(ApiStrictSchema):
upstream_task_ids = fields.List(fields.String())
downstream_task_ids = fields.List(fields.String())
task_type = fields.String()
task_source_code = fields.String(allow_none=True)
task_source_hash = fields.String(allow_none=True)
task_module_code = fields.String(allow_none=True)
module_source_hash = fields.String(allow_none=True)
dag_id = fields.String()
task_id = fields.String()
retries = fields.Integer()
command = fields.String(allow_none=True)
task_args = fields.Dict()
class SyncedTransactionSchema(ApiStrictSchema):
datasource_transaction_id = fields.String()
created_date = fields.DateTime()
started_date = fields.DateTime()
ended_date = fields.DateTime()
write_operation = fields.Nested(SyncedTransactionOperationSchema)
read_operations = fields.Nested(
SyncedTransactionOperationSchema, many=True, allow_none=True
)
query_string = fields.String(allow_none=True)
data_schema = fields.Nested(StructuredDataSchema, allow_none=True)
@post_load
def make_object(self, data):
return SyncedTransaction(**data)
class RootRunInfoSchema(ApiObjectSchema):
root_run_uid = fields.UUID()
root_task_run_uid = fields.UUID(allow_none=True)
root_run_url = fields.String(allow_none=True)
@post_load
def make_object(self, data, **kwargs):
return _as_dotted_dict(**data)
class ColumnStatsSchema(ApiStrictSchema):
column_name = fields.String(required=True)
column_type = fields.String(required=True)
records_count = fields.Integer(required=True)
distinct_count = fields.Integer(required=False, allow_none=True)
null_count = fields.Integer(required=False, allow_none=True)
# Metric for non-numeric column type
unique_count = fields.Integer(required=False, allow_none=True)
# Most frequent value - load_from attrs to support backward compatibility with SDK < 0.59.0
most_freq_value = fields.Raw(required=False,
allow_none=True,
load_from="top_value")
most_freq_value_count = fields.Integer(required=False,
allow_none=True,
load_from="top_freq_count")
# Metric for numeric column type
mean_value = fields.Float(required=False, allow_none=True)
min_value = fields.Float(required=False, allow_none=True)
max_value = fields.Float(required=False, allow_none=True)
std_value = fields.Float(required=False, allow_none=True)
# Percentiles
quartile_1 = fields.Float(required=False, allow_none=True)
quartile_2 = fields.Float(required=False, allow_none=True)
quartile_3 = fields.Float(required=False, allow_none=True)
# Hybrid properties - We only dump them so we don't load them back to ColumnStatsArgs when we make_object
# We don't save them in DB, and they are only computed in ColumnStatsArgs
non_null_count = fields.Integer(required=False,
allow_none=True,
dump_only=True)
null_percent = fields.Float(required=False,
allow_none=True,
dump_only=True)
@post_load
def make_object(self, data: dict) -> ColumnStatsArgs:
return ColumnStatsArgs(**data)
@post_dump
def dump_object(self, data: dict) -> dict:
return {k: v for k, v in data.items() if v is not None}
class SyncedDatasetSchema(ApiStrictSchema):
uri = fields.String() # {storage_type://region/project_id/scheme_name/table_name}
created_date = fields.DateTime()
last_modified_date = fields.DateTime(allow_none=True)
metadata = fields.Nested(SyncedDatasetMetadataSchema)
uid = fields.UUID(allow_none=True)
@post_load
def make_object(self, data):
return SyncedDataset(**data)
class TaskDefinitionParamSchema(ApiObjectSchema):
"""
Based on TaskDefinitionParam object
"""
name = fields.String()
default = fields.String(allow_none=True)
description = fields.String()
group = EnumField(ParameterGroup)
kind = EnumField(_ParameterKind)
load_on_build = fields.Boolean()
significant = fields.Boolean()
value_type = fields.String()
@post_load
def make_task_definition_param(self, data, **kwargs):
return _as_dotted_dict(**data)
class UpdateAirflowMonitorStateRequestSchema(ApiStrictSchema):
airflow_version = fields.String(required=False, allow_none=True)
airflow_export_version = fields.String(required=False, allow_none=True)
airflow_monitor_version = fields.String(required=False, allow_none=True)
monitor_status = fields.String(required=False, allow_none=True)
monitor_error_message = fields.String(required=False, allow_none=True)
airflow_instance_uid = fields.UUID(required=False, allow_none=True)
api_mode = fields.String(required=False, allow_none=True)
class AirflowServerInfoSchema(_ApiCallSchema):
base_url = fields.String()
airflow_version = fields.String(allow_none=True)
airflow_export_version = fields.String(allow_none=True)
airflow_monitor_version = fields.String(allow_none=True)
dags_path = fields.String(allow_none=True)
logs_path = fields.String(allow_none=True)
last_sync_time = fields.DateTime(allow_none=True)
monitor_status = fields.String(allow_none=True)
monitor_error_message = fields.String(allow_none=True)
monitor_start_time = fields.DateTime(allow_none=True)
synced_from = fields.DateTime(allow_none=True)
synced_to = fields.DateTime(allow_none=True)
@post_load
def make_object(self, data, **kwargs):
return AirflowServerInfo(**data)
class TaskDefinitionInfoSchema(ApiObjectSchema):
task_definition_uid = fields.UUID()
name = fields.String()
class_version = fields.String()
family = fields.String()
module_source = fields.String(allow_none=True)
module_source_hash = fields.String(allow_none=True)
source = fields.String(allow_none=True)
source_hash = fields.String(allow_none=True)
type = fields.String()
task_param_definitions = fields.Nested(TaskDefinitionParamSchema, many=True)
@post_load
def make_task_definition(self, data, **kwargs):
return TaskDefinitionInfo(**data)
class TaskRunEnvInfoSchema(ApiObjectSchema):
uid = fields.UUID()
cmd_line = fields.String()
user = fields.String()
machine = fields.String()
databand_version = fields.String()
user_code_version = fields.String()
user_code_committed = fields.Boolean()
project_root = fields.String()
user_data = fields.String()
heartbeat = fields.DateTime()
@post_load
def make_object(self, data, **kwargs):
return TaskRunEnvInfo(**data)
class LogTargetSchema(_ApiCallSchema):
run_uid = fields.UUID(required=True)
task_run_uid = fields.UUID(required=True)
task_run_name = fields.String()
task_run_attempt_uid = fields.UUID(required=True)
target_path = fields.String()
param_name = fields.String(allow_none=True)
task_def_uid = fields.UUID(allow_none=True)
operation_type = EnumField(DbndTargetOperationType)
operation_status = EnumField(DbndTargetOperationStatus)
value_preview = fields.String(allow_none=True)
data_dimensions = fields.List(fields.Integer(), allow_none=True)
data_schema = fields.String(allow_none=True)
data_hash = fields.String(allow_none=True)
class ScheduledJobInfoSchema(ApiObjectSchema):
uid = fields.UUID()
name = fields.String()
cmd = fields.String()
start_date = fields.DateTime()
create_user = fields.String()
create_time = fields.DateTime()
end_date = fields.DateTime(allow_none=True)
schedule_interval = fields.String(allow_none=True)
catchup = fields.Boolean(allow_none=True)
depends_on_past = fields.Boolean(allow_none=True)
retries = fields.Integer(allow_none=True)
active = fields.Boolean(allow_none=True)
update_user = fields.String(allow_none=True)
update_time = fields.DateTime(allow_none=True)
from_file = fields.Boolean(allow_none=True)
deleted_from_file = fields.Boolean(allow_none=True)
list_order = fields.Integer(allow_none=True)
job_name = fields.String(allow_none=True)
@post_load
def make_object(self, data, **kwargs):
return ScheduledJobInfo(**data)
