def main(args):
with open(args.label_path) as label_file, open(
args.training_path, 'wb') as training_file, open(
args.testing_path, 'wb') as testing_file:
label_reader = labelled_data.Reader(label_file)
training_writer = labelled_data.Writer(training_file)
testing_writer = labelled_data.Writer(testing_file)
partitioner.split_labels(label_reader, training_writer, testing_writer,
args.training_fraction, args.max_labels)
def test_writes_with_utf8_encoding(self):
mock_file = io.StringIO()
writer = labelled_data.Writer(mock_file)
writer.writerow({
'input': '2 jalape\xc3\xb1os',
'name': 'jalape\xc3\xb1os',
'qty': 2.0,
'unit': '',
'range_end': 0.0,
'comment': '',
})
self.assertMultiLineEqual(
('input,name,qty,range_end,unit,comment\n'
'2 jalape\xc3\xb1os,jalape\xc3\xb1os,2.0,0.0,,\n'),
mock_file.getvalue())
def test_writes_with_utf8_encoding(self):
mock_file = io.StringIO()
writer = labelled_data.Writer(mock_file)
writer.writerow({
"input": u"2 jalape\xc3\xb1os",
"name": u"jalape\xc3\xb1os",
"qty": 2.0,
"unit": u"",
"range_end": 0.0,
"comment": u"",
})
self.assertMultiLineEqual(
(u"input,name,qty,range_end,unit,comment\n"
u"2 jalape\xc3\xb1os,jalape\xc3\xb1os,2.0,0.0,,\n"),
mock_file.getvalue(),
)
def test_writes_valid_rows(self):
mock_file = io.StringIO()
writer = labelled_data.Writer(mock_file)
writer.writerows([
{
"input": u"4 to 6 large cloves garlic",
"qty": 4.0,
"unit": u"clove",
"name": u"garlic",
"range_end": 6.0,
"comment": u"",
},
{
"input": u"3 bananas",
"qty": 3.0,
"unit": u"",
"name": u"bananas",
"comment": u"",
"range_end": 0.0,
},
{
"input": (u"2 1/2 pounds bell peppers (about 6 peppers in "
u"assorted colors), cut into 2-inch chunks"),
"qty":
2.5,
"unit":
u"pound",
"name":
u"bell peppers",
"range_end":
0.0,
"comment": (u"(about 6 peppers in assorted colors), cut into "
u"2-inch chunks"),
},
])
self.assertMultiLineEqual(
"""\
input,name,qty,range_end,unit,comment
4 to 6 large cloves garlic,garlic,4.0,6.0,clove,
3 bananas,bananas,3.0,0.0,,
"2 1/2 pounds bell peppers (about 6 peppers in assorted colors), cut into 2-inch chunks",bell peppers,2.5,0.0,pound,"(about 6 peppers in assorted colors), cut into 2-inch chunks"\
""",
mock_file.getvalue().strip(),
)
def test_writes_valid_rows_one_by_one(self):
mock_file = io.StringIO()
writer = labelled_data.Writer(mock_file)
writer.writerow({
'input': '4 to 6 large cloves garlic',
'qty': 4.0,
'unit': 'clove',
'name': 'garlic',
'range_end': 6.0,
'comment': '',
})
writer.writerow({
'input': '3 bananas',
'qty': 3.0,
'unit': '',
'name': 'bananas',
'comment': '',
'range_end': 0.0,
})
writer.writerow({
'input': ('2 1/2 pounds bell peppers (about 6 peppers in '
'assorted colors), cut into 2-inch chunks'),
'qty':
2.5,
'unit':
'pound',
'name':
'bell peppers',
'range_end':
0.0,
'comment': ('(about 6 peppers in assorted colors), cut into '
'2-inch chunks'),
})
self.assertMultiLineEqual(
"""
input,name,qty,range_end,unit,comment
4 to 6 large cloves garlic,garlic,4.0,6.0,clove,
3 bananas,bananas,3.0,0.0,,
"2 1/2 pounds bell peppers (about 6 peppers in assorted colors), cut into 2-inch chunks",bell peppers,2.5,0.0,pound,"(about 6 peppers in assorted colors), cut into 2-inch chunks"
""".strip(),
mock_file.getvalue().strip())
def setUp(self):
self.mock_training_file = io.StringIO()
self.mock_training_writer = labelled_data.Writer(
self.mock_training_file)
self.mock_testing_file = io.StringIO()
self.mock_testing_writer = labelled_data.Writer(self.mock_testing_file)
