def main(self):
"""
Main method for base ArffConverter class. Several helper methods must be defined by
child classes.
"""
self.create_data_frame()
self.collect_comments()
self.output_file.write('@RELATION {} \n\n'.format(
quote_if_space(self.relation)))
output_header = self.convert_header()
for line in output_header:
self.output_file.write(line)
self.output_file.write('\n@DATA\n')
for row in self.output_rows():
self.output_file.write(row)
self.output_file.close()
if self.validate:
validator = ArffValidator(input_file=self.input_file,
arff_file=self.output_file)
validator.validate()
def test_quote_if_space(self):
expected_outcomes = {
'abcd': 'abcd',
' defg': '" defg"',
'hijk ': '"hijk "',
'lm no': '"lm no"',
}
for val in expected_outcomes.keys():
outcome = utils.quote_if_space(val)
self.assertEqual(outcome, expected_outcomes[val])
def compare_values(line, arff_line):
"""
Compares entries in lines from the input & output files and raises a ValidationError on mismatch
:param line: line from input file split into a list of entries
:param arff_line: line from output file split into a list of entries
"""
# TODO: How do I keep this from turning into one big switch statement or if, elif nightmare?
msg = 'Line mismatch between input:\n{}\nand ARFF output:\n{}'.format(
line, arff_line)
for i, entry in enumerate(line):
print(entry, arff_line[i])
# should match, but doesn't
if entry != arff_line[i]:
if ' ' in entry and quote_if_space(entry) != arff_line[i]:
raise ValidationError(msg)
# shouldn't match, but does
if entry == arff_line[i]:
if ' ' in arff_line[i] and quote_if_space(entry) != arff_line[i]:
raise ValidationError(msg)
if entry in ARFF_FIELD_MAPS['none'] and arff_line[i] != '?':
raise ValidationError(msg)
def test_arff_data(self):
self.input_file.next()
arff_line = self.output_file.readline()
while not arff_line.startswith('@DATA'):
arff_line = self.output_file.readline()
for csv_line in self.input_file:
csv_line = csv_line.split(',')
csv_line = [utils.quote_if_space(item) for item in csv_line]
arff_line = self.output_file.readline().split(',')
if '?' not in arff_line:
self.assertEqual(csv_line, arff_line)
else:
# TODO: Implement special case testing
pass
def convert_header(self):
"""
Converts header from data_frame to arff
:return: list of lines
"""
arff_header = []
for column in self.data_frame.columns:
attribute_name = column
pd_dtype = str(self.data_frame[attribute_name].dtype)
arff_dtype = self.map_data_types(pd_dtype, column)
line = '@ATTRIBUTE {} {}\n'.format(quote_if_space(attribute_name),
arff_dtype)
arff_header.append(line)
return arff_header