class JobSchemaV2(ApiStrictSchema):
id = fields.Int()
name = fields.Str()
user = fields.Str()
reported_by_user = fields.Str()
is_archived = fields.Boolean()
ui_hidden = fields.Boolean()
is_airflow_synced = fields.Boolean()
project_id = fields.Int()
project_name = fields.Str()
# computed
run_states = fields.Dict()
source_link = fields.Str()
# joined
latest_run_start_time = fields.DateTime()
latest_run_state = fields.Str()
latest_run_uid = fields.UUID()
latest_run_root_task_run_uid = fields.UUID()
latest_run_trigger = fields.Str()
latest_run_env = fields.Str()
source_instance_name = fields.Str()
source_type = fields.Str()
airflow_instance_name = fields.Str() # TODO_SERVER: API: Deprecate
scheduled_job_count = fields.Number()
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 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 TaskRunAttemptSchema(ApiStrictSchema):
task_run_uid = fields.UUID()
task_run_attempt_uid = fields.UUID()
state = EnumField(TaskRunState, allow_none=True)
timestamp = fields.DateTime(allow_none=True)
first_error = fields.Nested(ErrorInfoSchema, allow_none=True)
latest_error = fields.Nested(ErrorInfoSchema, allow_none=True)
attempt_number = fields.Number(allow_none=True)
source = fields.Str(allow_none=True)
start_date = fields.DateTime(allow_none=True)
end_date = fields.DateTime(allow_none=True)
external_links_dict = fields.Dict(allow_none=True)
class TaskRunAttemptUpdateArgsSchema(ApiObjectSchema):
task_run_uid = fields.UUID()
task_run_attempt_uid = fields.UUID()
state = EnumField(TaskRunState, allow_none=True)
timestamp = fields.DateTime(allow_none=True)
error = fields.Nested(ErrorInfoSchema, allow_none=True)
attempt_number = fields.Number(allow_none=True)
source = fields.Str(allow_none=True)
start_date = fields.DateTime(allow_none=True)
@post_load
def make_object(self, data, **kwargs):
return TaskRunAttemptUpdateArgs(**data)
class ScheduledJobSchemaV2(Schema):
class Meta:
strict = True
uid = fields.Str(attribute="DbndScheduledJob.uid", allow_none=True)
name = fields.Str(attribute="DbndScheduledJob.name", required=True)
cmd = fields.Str(attribute="DbndScheduledJob.cmd", required=True)
schedule_interval = fields.Str(
attribute="DbndScheduledJob.schedule_interval", required=True)
start_date = fields.DateTime(allow_none=True,
attribute="DbndScheduledJob.start_date",
format="iso")
end_date = fields.DateTime(allow_none=True,
attribute="DbndScheduledJob.end_date",
format="iso")
readable_schedule_interval = fields.Str(
attribute="DbndScheduledJob.readable_schedule_interval",
allow_none=True)
scheduled_interval_in_seconds = fields.Integer(
attribute="DbndScheduledJob.scheduled_interval_in_seconds",
allow_none=True)
catchup = fields.Boolean(allow_none=True,
attribute="DbndScheduledJob.catchup")
depends_on_past = fields.Boolean(
allow_none=True, attribute="DbndScheduledJob.depends_on_past")
retries = fields.Int(allow_none=True, attribute="DbndScheduledJob.retries")
active = fields.Boolean(allow_none=True,
attribute="DbndScheduledJob.active")
create_user = fields.Str(allow_none=True,
attribute="DbndScheduledJob.create_user")
create_time = fields.DateTime(allow_none=True,
attribute="DbndScheduledJob.create_time")
update_user = fields.Str(allow_none=True,
attribute="DbndScheduledJob.update_user")
update_time = fields.DateTime(allow_none=True,
attribute="DbndScheduledJob.update_time")
from_file = fields.Boolean(allow_none=True,
attribute="DbndScheduledJob.from_file")
deleted_from_file = fields.Boolean(
allow_none=True, attribute="DbndScheduledJob.deleted_from_file")
next_job_date = fields.DateTime(attribute="DbndScheduledJob.next_job_date",
allow_none=True)
alerts = fields.List(
fields.Nested(AlertEventSchema),
attribute="DbndScheduledJob.alerts",
allow_none=True,
)
job_name = fields.Str(dump_only=True,
attribute="DbndScheduledJob.job_name")
last_run_uid = fields.UUID(dump_only=True)
last_run_job = fields.Str(dump_only=True)
last_job_date = fields.DateTime(dump_only=True)
last_run_state = fields.Str(dump_only=True)
is_airflow_synced = fields.Bool(dump_only=True)
list_order = fields.Integer(attribute="DbndScheduledJob.list_order",
allow_none=True)
validation_errors = fields.Str(
allow_none=True, attribute="DbndScheduledJob.validation_errors")
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 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 InitRunArgsSchema(ApiObjectSchema):
run_uid = fields.UUID()
root_run_uid = fields.UUID()
driver_task_uid = fields.UUID(allow_none=True)
task_run_env = fields.Nested(TaskRunEnvInfoSchema)
task_runs_info = fields.Nested(TaskRunsInfoSchema)
new_run_info = fields.Nested(RunInfoSchema, allow_none=True)
scheduled_run_info = fields.Nested(ScheduledRunInfoSchema, allow_none=True)
update_existing = fields.Boolean()
source = fields.Str(allow_none=True)
@post_load
def make_init_run_args(self, data, **kwargs):
return InitRunArgs(**data)
class TaskRunInfoSchema(ApiObjectSchema):
task_run_uid = fields.UUID()
task_run_attempt_uid = fields.UUID()
task_definition_uid = fields.UUID()
run_uid = fields.UUID()
task_id = fields.String()
task_signature = fields.String()
task_signature_source = fields.String()
task_af_id = fields.String()
execution_date = fields.DateTime()
name = fields.String()
env = fields.String()
command_line = fields.String()
functional_call = fields.String()
has_downstreams = fields.Boolean()
has_upstreams = fields.Boolean()
is_reused = fields.Boolean()
is_dynamic = fields.Boolean()
is_system = fields.Boolean()
is_skipped = fields.Boolean()
is_root = fields.Boolean()
output_signature = fields.String()
state = EnumField(TaskRunState)
target_date = fields.Date(allow_none=True)
log_local = fields.String(allow_none=True)
log_remote = fields.String(allow_none=True)
version = fields.String()
task_run_params = fields.Nested(TaskRunParamSchema, many=True)
external_links = fields.Dict(allow_none=True)
@post_load
def make_task_run(self, data, **kwargs):
return TaskRunInfo(**data)
class RunInfoSchema(ApiStrictSchema):
root_run_uid = fields.UUID()
run_uid = fields.UUID()
job_name = fields.String()
project_name = fields.String(allow_none=True)
user = fields.String()
name = fields.String()
description = fields.String(allow_none=True)
state = EnumField(RunState)
start_time = fields.DateTime()
end_time = fields.DateTime(allow_none=True)
# deprecate
dag_id = fields.String()
cmd_name = fields.String(allow_none=True)
execution_date = fields.DateTime()
# move to task
target_date = fields.Date(allow_none=True)
version = fields.String(allow_none=True)
driver_name = fields.String()
is_archived = fields.Boolean()
env_name = fields.String(allow_none=True)
cloud_type = fields.String()
trigger = fields.String()
task_executor = fields.String(allow_none=True)
root_run = fields.Nested(RootRunInfoSchema)
scheduled_run = fields.Nested(ScheduledRunInfoSchema, allow_none=True)
sends_heartbeat = fields.Boolean(default=False, allow_none=True)
scheduled_job_name = fields.String(allow_none=True)
scheduled_date = fields.DateTime(allow_none=True)
external_links = fields.Dict(allow_none=True)
@post_load
def make_run_info(self, data, **kwargs):
return _as_dotted_dict(**data)
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 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 AirflowTaskContextSchema(ApiObjectSchema):
dag_id = fields.String()
execution_date = fields.String()
task_id = fields.String()
try_number = fields.Integer(allow_none=True)
airflow_instance_uid = fields.UUID()
@post_load
def make_run_info(self, data, **kwargs):
return AirflowTaskContext(**data)
class TaskRunsInfoSchema(ApiObjectSchema):
run_uid = fields.UUID()
root_run_uid = fields.UUID()
task_run_env_uid = fields.UUID()
parent_child_map = fields.List(fields.List(fields.UUID()))
task_runs = fields.Nested(TaskRunInfoSchema,
many=True,
exclude=("task_signature_source", ))
upstreams_map = fields.List(fields.List(fields.UUID()))
dynamic_task_run_update = fields.Boolean()
targets = fields.Nested(TargetInfoSchema, many=True)
task_definitions = fields.Nested(TaskDefinitionInfoSchema, many=True)
af_context = fields.Nested(AirflowTaskContextSchema, allow_none=True)
@post_load
def make_run_info(self, data, **kwargs):
return _as_dotted_dict(**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 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 JobSchemaV2(ApiObjectSchema):
id = fields.Int()
name = fields.Str()
user = fields.Str()
ui_hidden = fields.Boolean()
is_airflow_synced = fields.Boolean()
# computed
run_states = fields.Dict()
airflow_link = fields.Str()
# joined
latest_run_start_time = fields.DateTime()
latest_run_state = fields.Str()
latest_run_uid = fields.UUID()
latest_run_root_task_run_uid = fields.UUID()
latest_run_trigger = fields.Str()
latest_run_env = fields.Str()
scheduled_job_count = fields.Number()
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 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 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 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 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 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)
class HeartbeatSchema(_ApiCallSchema):
run_uid = fields.UUID()
class SetUnfinishedTasksStateShcmea(_ApiCallSchema):
run_uid = fields.UUID()
state = EnumField(TaskRunState)
timestamp = fields.DateTime()
class LogArtifactSchema(_ApiCallSchema):
task_run_attempt_uid = fields.UUID(required=True)
name = fields.String()
path = fields.String()
class LogMetricSchema(_ApiCallSchema):
task_run_attempt_uid = fields.UUID(required=True)
target_meta_uid = fields.UUID(allow_none=True)
metric = fields.Nested(MetricSchema, allow_none=True)
source = fields.String(allow_none=True)
class SaveExternalLinksSchema(_ApiCallSchema):
task_run_attempt_uid = fields.UUID(required=True)
external_links_dict = fields.Dict(
name=fields.Str(), url=fields.Str(), required=True
)
class SaveTaskRunLogSchema(_ApiCallSchema):
task_run_attempt_uid = fields.UUID(required=True)
log_body = fields.String(allow_none=True)
