def _to_data_matrix(self):
from collections import OrderedDict
data_matrix = []
for row_idx, row in enumerate(self._ltsv_input_stream):
if typepy.is_empty_sequence(row):
continue
ltsv_record = OrderedDict()
for col_idx, ltsv_item in enumerate(row.strip().split("\t")):
try:
label, value = ltsv_item.split(":")
except ValueError:
raise DataError(
"invalid lstv item found: line={}, col={}, item='{}'".
format(row_idx, col_idx, ltsv_item))
label = label.strip('"')
try:
pv.validate_ltsv_label(label)
except (pv.NullNameError, pv.InvalidCharError):
raise InvalidHeaderNameError(
"invalid label found (acceptable chars are [0-9A-Za-z_.-]): "
"line={}, col={}, label='{}'".format(
row_idx, col_idx, label))
ltsv_record[label] = value
data_matrix.append(ltsv_record)
# using generator to prepare for future enhancement to support
# iterative load.
yield data_matrix
def test_exception_invalid_char(self, value):
with pytest.raises(ValidationError) as e:
validate_ltsv_label(value)
assert e.value.reason == ErrorReason.INVALID_CHARACTER
def test_normal(self, value):
validate_ltsv_label(value)
def test_exception_invalid_char(self, value):
with pytest.raises(InvalidCharError):
validate_ltsv_label(value)
def test_normal(self, value):
validate_ltsv_label(value)
def test_exception_invalid_char(self, value):
with pytest.raises(InvalidCharError):
validate_ltsv_label(value)
