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 AirflowServerInfoSchema(ApiStrictSchema):
airflow_version = fields.String(allow_none=True)
airflow_export_version = fields.String(allow_none=True)
airflow_monitor_version = fields.String(allow_none=True)
last_sync_time = fields.DateTime(allow_none=True)
monitor_error_message = fields.String(allow_none=True)
synced_to = fields.DateTime(allow_none=True)
api_mode = fields.String(allow_none=True)
sync_interval = fields.Integer(allow_none=True)
is_sync_enabled = fields.Boolean(allow_none=True)
system_alert_definitions = fields.Dict()
fetcher = fields.String(allow_none=True)
composer_client_id = fields.String(allow_none=True)
dag_ids = fields.String(allow_none=True)
base_url = fields.String()
external_url = fields.String(allow_none=True)
source_instance_uid = fields.String(allow_none=True)
tracking_source_uid = fields.String()
airflow_instance_uid = fields.String(
allow_none=True) # TODO_API: deprecate
name = fields.String(allow_none=True)
env = fields.String(allow_none=True)
monitor_status = fields.String(allow_none=True)
monitor_config = fields.Nested(MonitorConfigSchema, allow_none=True)
airflow_environment = fields.String(allow_none=True)
last_seen_dag_run_id = fields.Integer(allow_none=True)
last_seen_log_id = fields.Integer(allow_none=True)
@post_load
def make_object(self, data, **kwargs):
return AirflowServerInfo(**data)
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 AirflowServerInfoSchema(_ApiCallSchema):
base_url = fields.String()
external_url = fields.String(allow_none=True)
airflow_instance_uid = fields.String(allow_none=True)
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)
api_mode = fields.String(allow_none=True)
sync_interval = fields.Integer(allow_none=True)
is_sync_enabled = fields.Boolean(allow_none=True)
fetcher = fields.String(allow_none=True)
composer_client_id = fields.String(allow_none=True)
active_dags = fields.Dict(allow_none=True)
name = fields.String(allow_none=True)
env = fields.String(allow_none=True)
include_logs = fields.Boolean(allow_none=True)
include_task_args = fields.Boolean(allow_none=True)
fetch_quantity = fields.Integer(allow_none=True)
oldest_incomplete_data_in_days = fields.Integer(allow_none=True)
@post_load
def make_object(self, data, **kwargs):
return AirflowServerInfo(**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 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 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 MetricSchema(ApiObjectSchema):
key = fields.String()
value = fields.Raw(allow_none=True)
value_str = fields.String(allow_none=True)
value_json = fields.Dict(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 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 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)
external_links_dict = fields.Dict(allow_none=True)
@post_load
def make_object(self, data, **kwargs):
return TaskRunAttemptUpdateArgs(**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 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 DagSchema(ApiStrictSchema):
description = fields.String()
root_task_ids = fields.List(fields.String())
tasks = fields.Nested(TaskSchema, many=True)
owner = fields.String()
dag_id = fields.String()
schedule_interval = fields.String()
catchup = fields.Boolean(allow_none=True)
start_date = fields.DateTime(allow_none=True)
end_date = fields.DateTime(allow_none=True)
is_committed = fields.Boolean()
git_commit = fields.String()
dag_folder = fields.String()
hostname = fields.String()
source_code = fields.String(allow_none=True)
module_source_hash = fields.String(allow_none=True)
is_subdag = fields.Boolean()
tags = fields.List(fields.String(), allow_none=True)
task_type = fields.String()
task_args = fields.Dict()
is_active = fields.Boolean(allow_none=True)
is_paused = fields.Boolean(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 LogDataframeHistogramSchema(LogTargetSchema):
histograms = fields.Dict()
descriptive_stats = fields.Dict()
timestamp = fields.DateTime()
class AlertDefsSchema(ApiStrictSchema):
severity = fields.Str(required=True)
type = fields.Str(required=True)
user_metric = fields.Str()
operator = fields.Str()
is_str_value = fields.Bool()
created_at = fields.DateTime()
scheduled_job_name = fields.Str(attribute="scheduled_job.name")
source_instance_name = fields.Method("get_tracking_source_name")
env = fields.Method("get_tracking_source_env")
# TODO_CORE: API: Deprecate airflow_server_info
airflow_instance_name = fields.Method("get_tracking_source_name")
project_id = fields.Int(attribute="job.project_id")
project_name = fields.Str(attribute="job.project.name")
alert_on_historical_runs = fields.Bool()
group_uid = fields.Str(allow_none=True, load_from="alert_group_uid")
root_group_uid = fields.Str(allow_none=True,
load_from="alert_root_group_uid")
uid = fields.Str(allow_none=True)
value = fields.Str(allow_none=True)
job_id = fields.Int(allow_none=True)
summary = fields.Str(allow_none=True)
job_name = fields.Str(attribute="job.name", allow_none=True)
task_repr = fields.Str(allow_none=True)
task_name = fields.Str(allow_none=True)
custom_name = fields.Str(allow_none=True)
original_uid = fields.Str(allow_none=True)
advanced_json = fields.Str(allow_none=True)
scheduled_job_uid = fields.Str(allow_none=True)
custom_description = fields.Str(allow_none=True)
ml_alert = fields.Nested(MLAlert, allow_none=True)
# Fields for DatasetSlaAlert/DatasetSlaAdvancedAlert alert
# --------------------------------------
seconds_delta = fields.Int(
allow_none=True) # Converts to datetime.timedelta
dataset_partial_name = fields.Str(allow_none=True)
datasets_uids = fields.List(fields.Str(), allow_none=True)
# Fields for OperationColumnStatAdvancedAlert alert
# --------------------------------------
dataset_uid = fields.Str(allow_none=True)
# Operation type (e.g. "read", "write", None=any) to filter stats by
operation_type = fields.Str(allow_none=True)
# Type of MetricRule, found in dbnd_web. Used to build advanced_json
metrics_rules = fields.List(fields.Dict(), allow_none=True)
# Used only used by the UI
affected_datasets = fields.List(fields.Dict(),
allow_none=True,
dump_only=True)
assigned_jobs = fields.Method(serialize="get_assigned_jobs",
dump_only=True)
is_system = fields.Function(lambda alert_def: alert_def.owner == "system",
dump_only=True)
def get_tracking_source_name(self, obj):
return self._get_tracking_source_instance(obj).name
def get_tracking_source_env(self, obj):
return self._get_tracking_source_instance(obj).env
def _get_tracking_source_instance(self, obj):
if obj.job:
return obj.job.tracking_source
return obj.tracking_source
@pre_load
def prepere(self, data: dict, **kwargs):
value = data.get("value", None)
if value is not None:
data["value"] = str(data["value"])
return data
def get_assigned_jobs(self, alert_def):
self_job = (alert_def.job_id, alert_def.job_name)
sub_alerts_jobs = ((sub_alert.job_id, sub_alert.job_name)
for sub_alert in alert_def.sub_alert_definitions)
alert_jobs = chain(sub_alerts_jobs, [self_job])
alert_jobs = set(filter(lambda l: l != (None, None), alert_jobs))
serialized_assigned_jobs = [{
"job_id": job_id,
"job_name": job_name
} for job_id, job_name in alert_jobs]
return serialized_assigned_jobs
class AirflowExportMetaSchema(ApiStrictSchema):
airflow_version = fields.String()
plugin_version = fields.String()
request_args = fields.Dict()
metrics = fields.Nested(MetricsSchema)
class MetricsSchema(ApiStrictSchema):
performance = fields.Dict()
sizes = fields.Dict()
