def test_removing_idx(idx): munge.remove_col_big_data( sample_file, sample_file_delcol, [idx], delimiter='\t') data1 = [] with open(sample_file) as f: for line in f: row = line.strip().split() del row[idx] data1.append(row) data2 = [] with open(sample_file_delcol) as f: for line in f: data2.append(line.strip().split()) self.assertEqual(data1, data2)
def test_remove_col_big_data(self): sample_file = os.path.join(self.curdir, 'res', 'sample1.txt') sample_file_delcol = os.path.join( self.curdir, 'res', 'sample1.delcol.txt') def test_removing_idx(idx): munge.remove_col_big_data( sample_file, sample_file_delcol, [idx], delimiter='\t') data1 = [] with open(sample_file) as f: for line in f: row = line.strip().split() del row[idx] data1.append(row) data2 = [] with open(sample_file_delcol) as f: for line in f: data2.append(line.strip().split()) self.assertEqual(data1, data2) for col in range(3): test_removing_idx(col) # Delete file created os.remove(sample_file_delcol) # Test removing two columns sample_file = os.path.join(self.curdir, 'res', 'sample1.txt') sample_file_delcol = os.path.join(self.curdir, 'res', 'sample1.delcol.txt') munge.remove_col_big_data(sample_file, sample_file_delcol, [0,2], delimiter='\t') data1 = [] with open(sample_file) as f: for line in f: row = line.strip().split() data1.append([row[1]]) data2 = [] with open(sample_file_delcol) as f: for line in f: data2.append(line.strip().split()) self.assertEqual(data1, data2) os.remove(sample_file_delcol)