def test_can_read_fixed_rows_with_missing_terminating_line_delimiter(self):
data_format, field_names_and_lengths = FixedRowsTest._create_fixed_data_format_and_fields_for_name_and_height()
with io.StringIO('hugo172\nsepp163') as data_io:
rows = list(rowio.fixed_rows(
data_io, data_format.encoding, field_names_and_lengths, data_format.line_delimiter))
self._assert_rows_contain_data(rows)
self.assertEqual(2, len(rows))
def test_can_read_empty_fixed_rows(self):
data_format = data.DataFormat(data.FORMAT_FIXED)
data_format.validate()
with io.StringIO('') as data_io:
rows = list(rowio.fixed_rows(
data_io, data_format.encoding, (('dummy', 1),), data_format.line_delimiter))
self.assertEqual([], rows)
def test_can_read_fixed_rows_with_mixed_line_delimiters_terminated_by_carriage_return(
self):
data_format, field_names_and_lengths = FixedRowsTest._create_fixed_data_format_and_fields_for_name_and_height(
)
with io.StringIO('john172\r\nmary163\r') as data_io:
rows = list(
rowio.fixed_rows(data_io, data_format.encoding,
field_names_and_lengths,
data_format.line_delimiter))
self.assertEqual([['john', '172'], ['mary', '163']], rows)
def test_can_read_fixed_rows_with_mixed_line_delimiters(self):
data_format, field_names_and_lengths = FixedRowsTest._create_fixed_data_format_and_fields_for_name_and_height(
)
with io.StringIO(
'john172\rmary163\nbill167\r\njane184\r\n') as data_io:
rows = list(
rowio.fixed_rows(data_io, data_format.encoding,
field_names_and_lengths,
data_format.line_delimiter))
self.assertEqual([['john', '172'], ['mary', '163'], ['bill', '167'],
['jane', '184']], rows)
def test_can_read_fixed_rows(self):
cid_path = dev_test.path_to_test_cid('customers_fixed.ods')
customer_cid = interface.Cid(cid_path)
fixed_path = dev_test.path_to_test_data('valid_customers_fixed.txt')
field_names_and_lengths = interface.field_names_and_lengths(customer_cid)
rows = list(rowio.fixed_rows(fixed_path, customer_cid.data_format.encoding, field_names_and_lengths))
self.assertNotEqual(0, len(rows))
for row_index in range(len(rows) - 1):
row = rows[row_index]
next_row = rows[row_index + 1]
self.assertNotEqual(0, len(row))
self.assertEqual(len(row), len(next_row))
def _test_can_read_fixed_rows_from_stringio(self, data_text, data_format=None):
assert (data_format is None) or data_format.is_valid
default_data_format, field_names_and_lengths = \
FixedRowsTest._create_fixed_data_format_and_fields_for_name_and_height()
actual_data_format = default_data_format if data_format is None else data_format
with io.StringIO(data_text) as data_io:
rows = list(rowio.fixed_rows(
data_io, actual_data_format.encoding, field_names_and_lengths, actual_data_format.line_delimiter))
self._assert_rows_contain_data(rows)
for row in rows:
for item in row:
self.assertTrue(item in data_text, 'item %r must be part of data %r' % (item, data_text))
def _fails_on_fixed_rows_from_stringio(self, data_text, expected_error_pattern='*', data_format=None):
assert (data_format is None) or data_format.is_valid
default_data_format, field_names_and_lengths = \
FixedRowsTest._create_fixed_data_format_and_fields_for_name_and_height()
actual_data_format = default_data_format if data_format is None else data_format
with io.StringIO(data_text) as data_io:
rows = rowio.fixed_rows(
data_io, actual_data_format.encoding, field_names_and_lengths, actual_data_format.line_delimiter)
try:
for _ in rows:
pass
self.fail()
except errors.DataFormatError as anticipated_error:
dev_test.assert_fnmatches(self, str(anticipated_error), expected_error_pattern)
def _raw_rows(self):
data_format = self.cid.data_format
format = data_format.format
if format == data.FORMAT_EXCEL:
return rowio.excel_rows(self._source_data_stream_or_path, data_format.sheet)
elif format == data.FORMAT_DELIMITED:
return rowio.delimited_rows(self._source_data_stream_or_path, data_format)
elif format == data.FORMAT_FIXED:
return rowio.fixed_rows(
self._source_data_stream_or_path, data_format.encoding, interface.field_names_and_lengths(self.cid),
data_format.line_delimiter)
elif format == data.FORMAT_ODS:
return rowio.ods_rows(self._source_data_stream_or_path, data_format.sheet)
else:
assert False, 'format=%r' % format
def _raw_rows(self):
data_format = self.cid.data_format
format = data_format.format
if format == data.FORMAT_EXCEL:
return rowio.excel_rows(self._source_data_stream_or_path, data_format.sheet)
elif format == data.FORMAT_DELIMITED:
return rowio.delimited_rows(self._source_data_stream_or_path, data_format)
elif format == data.FORMAT_FIXED:
return rowio.fixed_rows(
self._source_data_stream_or_path, data_format.encoding, interface.field_names_and_lengths(self.cid),
data_format.line_delimiter)
elif format == data.FORMAT_ODS:
return rowio.ods_rows(self._source_data_stream_or_path, data_format.sheet)
else:
assert False, 'format=%r' % format
def test_can_read_fixed_rows_with_mixed_line_delimiters_terminated_by_carriage_return(self):
data_format, field_names_and_lengths = FixedRowsTest._create_fixed_data_format_and_fields_for_name_and_height()
with io.StringIO('john172\r\nmary163\r') as data_io:
rows = list(rowio.fixed_rows(
data_io, data_format.encoding, field_names_and_lengths, data_format.line_delimiter))
self.assertEqual([['john', '172'], ['mary', '163']], rows)
def test_can_read_fixed_rows_with_mixed_line_delimiters(self):
data_format, field_names_and_lengths = FixedRowsTest._create_fixed_data_format_and_fields_for_name_and_height()
with io.StringIO('john172\rmary163\nbill167\r\njane184\r\n') as data_io:
rows = list(rowio.fixed_rows(
data_io, data_format.encoding, field_names_and_lengths, data_format.line_delimiter))
self.assertEqual([['john', '172'], ['mary', '163'], ['bill', '167'], ['jane', '184']], rows)
