def test_submitting_list(label1, label2):
q = SimpleLabellingQueue()
q.enqueue(1)
with pytest.raises(ValueError):
q.submit(0, label1)
id_, val = q.pop()
q.submit(id_, [label1, label2])
assert q.list_labels() == {label1, label2}
def test_progress():
inps = [str(i) for i in range(50)]
q = SimpleLabellingQueue()
# assert my little hack for dividing by zero
assert q.progress == 0
q.enqueue_many(inps)
for i, (id_, val) in enumerate(q):
assert q.progress == i / len(inps)
q.submit(id_, str(i))
def test_undo():
inp = "input 1"
q = SimpleLabellingQueue()
q.enqueue(inp)
id_, val = q.pop()
q.submit(id_, "label 1")
# ensure the queue is empty now:
with pytest.raises(IndexError):
q.pop()
q.undo()
# see if it's possible to pop now:
id_, val = q.pop()
assert val == "input 1"
def test_list_uncompleted(inputs, labels):
q = SimpleLabellingQueue()
q.enqueue_many(inputs)
popped_ids = []
for i in range(5):
id_, val = q.pop()
q.submit(id_, labels[i])
popped_ids.append(id_)
ids, x = q.list_uncompleted()
assert len(ids) == (len(inputs) - 5)
# test that the popped IDs and completed IDs don't share members
assert pytest.helpers.no_shared_members(ids, popped_ids)
assert pytest.helpers.same_elements(x, inputs[5:])
def test_list_completed(inputs, labels):
q = SimpleLabellingQueue()
q.enqueue_many(inputs)
popped_ids = []
for i in range(5):
id_, val = q.pop()
q.submit(id_, labels[i])
popped_ids.append(id_)
ids, x, y = q.list_completed()
assert len(ids) == 5
# test that the popped IDs and completed IDs have the same members
assert pytest.helpers.same_elements(ids, popped_ids)
assert pytest.helpers.same_elements(y, labels[:5])
def test_list_all(inputs, labels):
q = SimpleLabellingQueue()
q.enqueue_many(inputs)
popped_ids = []
for i in range(5):
id_, val = q.pop()
q.submit(id_, labels[i])
popped_ids.append(id_)
ids, x, y = q.list_all()
assert len(ids) == len(inputs)
assert all(label in labels for label in y if label is not None)
assert all(label is None or id_ in popped_ids
for id_, label in zip(ids, y))
assert Counter(y)[None] == (len(inputs) - 5)
assert pytest.helpers.same_elements(ids, range(len(inputs)))
def test_enqueue_dataframe(inputs):
n = len(inputs)
q = SimpleLabellingQueue()
q.enqueue_many(inputs)
assert len(q.data) == n
# assert we can pop everything:
for _ in range(n):
id_, val = q.pop()
assert isinstance(val, pd.Series)
# assert there's nothing else to pop:
with pytest.raises(IndexError):
q.pop()
# assert it re-constructs a df on list all
if n > 0:
ids, X, y = q.list_all()
assert isinstance(X, pd.DataFrame) or len(X) == 0
# assert it re-constructs a df on list uncomplete
q.submit(ids[0], "hello")
ids, X = q.list_uncompleted()
assert isinstance(X, pd.DataFrame) or len(X) == 0
# assert it re-constructs a df on list uncomplete
ids, X, y = q.list_completed()
assert isinstance(X, pd.DataFrame) or len(X) == 0
def test_submitting_text(label1, label2):
q = SimpleLabellingQueue()
q.enqueue(1)
q.enqueue(2)
with pytest.raises(ValueError):
q.submit(0, label1)
id_, val = q.pop()
q.submit(id_, label1)
assert q.progress == 0.5
id_, val = q.pop()
q.submit(id_, label2)
assert q.progress == 1
assert q.list_labels() == {label1, label2}
