def check_instance(self, data, data_kwargs={}):
if isinstance(data, np.ndarray):
self.data = np.ndaray(data, **data_kwargs)
self.data_type = 'numpy.ndarray'
elif isinstance(data, list):
self.data = data
self.data_type = 'list'
else:
# Check optional installed modules to see if data type matches
try:
from pandas import DataFrame
if isinstance(data, DataFrame):
self.data = DataFrame(data, **data_kwargs)
self.data_type = 'pandas.core.frame.DataFrame'
except ImportError:
pass
return self.data_type is None
def check_instance(self, data, data_kwargs={}):
if isinstance(data, np.ndarray):
self.data = np.ndaray(data, **data_kwargs)
self.data_type = 'numpy.ndarray'
elif isinstance(data, list):
self.data = data
self.data_type = 'list'
else:
# Check optional installed modules to see if data type matches
try:
from pandas import DataFrame
if isinstance(data, DataFrame):
self.data = DataFrame(data, **data_kwargs)
self.data_type = 'pandas.core.frame.DataFrame'
except ImportError:
pass
return self.data_type is None
def set_data(self, data=None, data_type=None, data_kwargs={}):
"""
Set the data for the given source based on the data type or a user specified type
"""
import toyz.utils.io
self.data_type = None
if data is None:
if self.paths['data']['io_module']=='':
raise ToyzDataError(
'You must supply a data object or file info to initialize a DataSource')
else:
self.data = toyz.utils.io.load_data(**self.paths['data'])
if data_type is None:
self.data_type = type(self.data).__module__+'.'+type(self.data).__name__
else:
self.data_type = data_type
else:
if data_type is None:
# Attempt to detect the data_type
if isinstance(data, np.ndarray):
self.data = np.ndaray(data, **data_kwargs)
self.data_type = 'numpy.ndarray'
elif isinstance(data, list):
self.data = data
self.data_type = 'list'
else:
# Check optional installed modules to see if data type matches
try:
from pandas import DataFrame
pandas_installed=True
except ImportError:
pandas_installed=False
if pandas_installed and isinstance(data, DataFrame):
self.data = DataFrame(data, **data_kwargs)
self.data_type = 'pandas.core.frame.DataFrame'
else:
self.data_type = data_type
if data_type == 'pandas.core.frame.DataFrame':
from pandas import DataFrame
self.data = DataFrame(data, **data_kwargs)
elif data_type == 'numpy.ndarray':
self.data = numpy.ndarray(data, **data_kwargs)
elif data_type == 'list':
self.data = list(data)
if 'columns' in data_kwargs:
self.columns = data_kwargs['columns']
else:
self.columns = ['col-'+n for n in range(len(self.data))]
else:
self.data_type = None
# If the data_type was not found in the Toyz standard data types look in
# user defined toyz
if self.data_type is None:
import toyz.utils.core
user_modules = toyz.utils.core.get_all_user_modules(
session_vars.toyz_settings, self.user_id)
data_types = {}
for module in user_modules:
toy = toyz.utils.core.get_user_toyz(
session_vars.toyz_settings, self.user_id, module)
if hasattr(toy.config.data_types):
for k,v in toy.config.data_types.items():
data_types[k] = v
for dt, obj in data_types.items():
if (data_type is None and obj.check_instance(data) or
data_type is not None and dt==data_type):
obj.set_data(data, data_type, data_kwargs)
self.data_type = dt
break
if self.data_type is None:
raise ToyzDataError(
"Could not find data type {0} in users approved modules".format(dt))
# Set the column names based on the data type
self.name_columns()