def test_csv_sniffer(self):
# GH-2785
reader = CSVReader(test_filename('test_asn_data_working.csv'))
data = reader.read()
self.assertEqual(len(data), 8)
self.assertEqual(
len(data.domain.variables) + len(data.domain.metas), 15)
def test_header_call(self, writer):
CSVReader.write_headers(writer, Table("iris"), False)
self.assertEqual(len(writer.call_args_list), 1)
writer.reset_mock()
CSVReader.write_headers(writer, Table("iris"), True)
self.assertEqual(len(writer.call_args_list), 3)
def test_header_call(self, writer):
CSVReader.write_headers(writer, Table("iris"), False)
self.assertEqual(len(writer.call_args_list), 1)
writer.reset_mock()
CSVReader.write_headers(writer, Table("iris"), True)
self.assertEqual(len(writer.call_args_list), 3)
def test_csv_format_missing_values(self):
f = CSVReader()
data = Table("../../tests/test9.tab")
self.assertTrue(any([x is "" for instance in
data.metas for x in instance]))
f.write_file(self.CSV_FILE, data)
new_data = f.read_file(self.CSV_FILE)
for new_instance, old_instance in zip(new_data, data):
self.assertEqual(new_instance, old_instance)
def test_noncontinous_marked_continuous(self):
file = NamedTemporaryFile("wt", delete=False)
file.write(noncont_marked_cont)
file.close()
with self.assertRaises(ValueError) as cm:
table = CSVReader(file.name).read()
self.assertIn('line 5, column 2', cm.exception.args[0])
def test_readwrite_timevariable(self):
output_csv = StringIO()
input_csv = StringIO("""\
Date,Feature
time,continuous
,
1920-12-12,1.0
1920-12-13,3.0
1920-12-14,5.5
""")
for stream in (output_csv, input_csv):
stream.close = lambda: None # HACK: Prevent closing of streams
table = CSVReader(input_csv).read()
self.assertIsInstance(table.domain['Date'], TimeVariable)
self.assertEqual(table[0, 'Date'], '1920-12-12')
# Dates before 1970 are negative
self.assertTrue(all(inst['Date'] < 0 for inst in table))
CSVReader.write_file(output_csv, table)
self.assertEqual(input_csv.getvalue().splitlines(),
output_csv.getvalue().splitlines())
def test_readwrite_timevariable(self):
output_csv = StringIO()
input_csv = StringIO("""\
Date,Feature
time,continuous
,
1920-12-12,1.0
1920-12-13,3.0
1920-12-14,5.5
""")
for stream in (output_csv, input_csv):
stream.close = lambda: None # HACK: Prevent closing of streams
table = CSVReader(input_csv).read()
self.assertIsInstance(table.domain['Date'], TimeVariable)
self.assertEqual(table[0, 'Date'], '1920-12-12')
# Dates before 1970 are negative
self.assertTrue(all(inst['Date'] < 0 for inst in table))
CSVReader.write_file(output_csv, table)
self.assertEqual(input_csv.getvalue().splitlines(),
output_csv.getvalue().splitlines())
def test_empty_columns(self):
"""Can't read files with more columns then headers. GH-1417"""
samplefile = """\
a, b
1, 0,
1, 2,
"""
c = io.StringIO(samplefile)
with self.assertWarns(UserWarning) as cm:
table = CSVReader(c).read()
self.assertEqual(len(table.domain.attributes), 2)
self.assertEqual(cm.warning.args[0],
"Columns with no headers were removed.")
def test_pr1734(self):
ContinuousVariable('foo')
file = NamedTemporaryFile("wt", delete=False)
filename = file.name
try:
file.write('''\
foo
time
123123123
''')
file.close()
CSVReader(filename).read()
finally:
os.remove(filename)
def read_easy(self, s, name):
file = NamedTemporaryFile("wt", delete=False)
filename = file.name
try:
file.write(s)
file.close()
table = CSVReader(filename).read()
f1, f2, f3 = table.domain
self.assertIsInstance(f1, DiscreteVariable)
self.assertEqual(f1.name, name + "1")
self.assertIsInstance(f2, ContinuousVariable)
self.assertEqual(f2.name, name + "2")
self.assertIsInstance(f3, ContinuousVariable)
self.assertEqual(f3.name, name + "3")
finally:
os.remove(filename)
def test_csv_sniffer(self):
# GH-2785
reader = CSVReader(test_filename('test_asn_data_working.csv'))
data = reader.read()
self.assertEqual(len(data), 8)
self.assertEqual(len(data.domain.variables) + len(data.domain.metas), 15)
def test_read_csv_with_na(self):
c = io.StringIO(csv_file_missing)
table = CSVReader(c).read()
f1, f2 = table.domain.variables
self.assertIsInstance(f1, ContinuousVariable)
self.assertIsInstance(f2, DiscreteVariable)