def __init__(self, engine_name: str = 'pandas',
hostname: typing.List[str] = [],
start_time: str = '', end_time: str = '',
view: str = 'latest', namespace: typing.List[str] = [],
columns: typing.List[str] = ['default'],
context=None, table: str = '', config_file=None) -> None:
if context is None:
self.ctxt = SqContext(engine_name, config_file)
else:
self.ctxt = context
if not self.ctxt:
self.ctxt = SqContext(engine_name)
self._cfg = self.ctxt.cfg
self._schema = SchemaForTable(table, self.ctxt.schemas)
self._table = table
self._sort_fields = self._schema.key_fields()
if not namespace and self.ctxt.namespace:
self.namespace = self.ctxt.namespace
else:
self.namespace = namespace
if not hostname and self.ctxt.hostname:
self.hostname = self.ctxt.hostname
else:
self.hostname = hostname
if not start_time and self.ctxt.start_time:
self.start_time = self.ctxt.start_time
else:
self.start_time = start_time
if not end_time and self.ctxt.end_time:
self.end_time = self.ctxt.end_time
else:
self.end_time = end_time
if not view and self.ctxt.view:
self.view = self.ctxt.view
else:
self.view = view
self.columns = columns
if engine_name and engine_name != '':
self.engine = get_sqengine(engine_name,
self._table)(self._table, self)
elif self.ctxt.engine:
self.engine = get_sqengine(self.ctxt.engine,
self._table)(self._table, self)
if not self.engine:
raise ValueError('Unknown analysis engine')
self._addnl_filter = None
self._addnl_fields = []
self._valid_get_args = None
self._valid_assert_args = None
self._valid_arg_vals = None
def __init__(self, engine_name: str = '',
hostname: typing.List[str] = None,
start_time: str = '', end_time: str = '',
view: str = '', namespace: typing.List[str] = None,
columns: typing.List[str] = None,
context=None, table: str = '', config_file=None) -> None:
if not context:
self.ctxt = SqContext(cfg=load_sq_config(validate=True,
config_file=config_file),
engine=engine_name)
self.ctxt.schemas = Schema(self.ctxt.cfg["schema-directory"])
else:
self.ctxt = context
if not self.ctxt.cfg:
self.ctxt.cfg = load_sq_config(validate=True,
config_file=config_file)
self.ctxt.schemas = Schema(self.ctxt.cfg["schema-directory"])
if not self.ctxt.engine:
self.ctxt.engine = engine_name
self._cfg = self.ctxt.cfg
self._schema = SchemaForTable(table, self.ctxt.schemas)
self._table = table
self._sort_fields = self._schema.key_fields()
self._convert_args = {}
self.namespace = namespace or self.ctxt.namespace or []
self.hostname = hostname or self.ctxt.hostname or []
self.start_time = start_time or self.ctxt.start_time
self.end_time = end_time or self.ctxt.end_time
view = view or self.ctxt.view
if self.start_time and self.end_time and not view:
self.view = 'all'
else:
self.view = view or 'latest'
self.columns = columns or ['default']
self._unique_def_column = ['hostname']
if engine_name and engine_name != '':
self.engine = get_sqengine(engine_name, self._table)(self)
elif self.ctxt.engine:
self.engine = get_sqengine(self.ctxt.engine, self._table)(self)
if not self.engine:
raise ValueError('Unknown analysis engine')
self.summarize_df = pd.DataFrame()
self._addnl_filter = self._addnl_fields = []
self._valid_get_args = self._valid_assert_args = []
self._valid_arg_vals = self._valid_find_args = []
self._valid_summarize_args = []
def __init__(self, engine="pandas"):
self.cfg = load_sq_config(validate=False)
self.schemas = Schema(self.cfg["schema-directory"])
self.namespace = ""
self.hostname = ""
self.start_time = ""
self.end_time = ""
self.exec_time = ""
self.engine_name = engine
self.sort_fields = []
self.engine = get_sqengine(engine)
super().__init__()
def __init__(self, engine):
self.cfg = load_sq_config(validate=False)
self.schemas = Schema(self.cfg['schema-directory'])
self.namespace = ''
self.hostname = ''
self.start_time = ''
self.end_time = ''
self.exec_time = ''
self.engine = 'pandas'
self.sort_fields = []
self.engine = get_sqengine(self.engine)
if not self.engine:
# We really should define our own error
raise ValueError
def change_engine(self, engine: str):
if engine == self.engine_name:
return
self.engine_name = engine
self.engine = get_sqengine(engine)
def __init__(self,
engine_name: str = '',
hostname: typing.List[str] = [],
start_time: str = '',
end_time: str = '',
view: str = 'latest',
namespace: typing.List[str] = [],
columns: typing.List[str] = ['default'],
context=None,
table: str = '') -> None:
if context is None:
self.ctxt = SqContext(engine_name)
else:
self.ctxt = context
if not self.ctxt:
self.ctxt = SqContext(engine_name)
self._cfg = self.ctxt.cfg
self._schemas = self.ctxt.schemas
self._table = table
self._sort_fields = []
self._cat_fields = []
self._ign_key_fields = [] # Used when keys != parquet partition cols
if not namespace and self.ctxt.namespace:
self.namespace = self.ctxt.namespace
else:
self.namespace = namespace
if not hostname and self.ctxt.hostname:
self.hostname = self.ctxt.hostname
else:
self.hostname = hostname
if not start_time and self.ctxt.start_time:
self.start_time = self.ctxt.start_time
else:
self.start_time = start_time
if not end_time and self.ctxt.end_time:
self.end_time = self.ctxt.end_time
else:
self.end_time = end_time
if not view and self.ctxt.view:
self.view = self.ctxt.view
else:
self.view = view
self.columns = columns
if engine_name and engine_name != '':
self.engine = get_sqengine(engine_name)
else:
self.engine = self.ctxt.engine
if self._table:
self.engine_obj = self.engine.get_object(self._table, self)
else:
self.engine_obj = None
self._addnl_filter = None
self._addnl_fields = []
