def test_iterating_over_queue():
inps = [str(i) for i in range(50)]
q = SimpleLabellingQueue()
q.enqueue_many(inps)
for i, (id_, val) in enumerate(q):
assert i == id_
def test_shuffling():
inps = [str(i) for i in range(50)]
q = SimpleLabellingQueue()
q.enqueue_many(inps)
q.shuffle()
# assert the order is not the same:
assert any(val != inp for inp, (id_, val) in zip(inps, q))
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_enqueue_many(inputs):
n = len(inputs)
q = SimpleLabellingQueue()
q.enqueue_many(inputs)
assert len(q.data) == n
# assert we can pop everything:
for _ in range(n):
q.pop()
# assert there's nothing else to pop:
with pytest.raises(IndexError):
q.pop()
def test_reordering():
inp = ["b", "a", "d", "c"]
q = SimpleLabellingQueue()
q.enqueue_many(inp)
q.reorder(OrderedDict([(0, 1), (1, 0), (2, 3), (3, 2)]))
id_, val = q.pop()
assert val == "a" and id_ == 1
id_, val = q.pop()
assert val == "b" and id_ == 0
id_, val = q.pop()
assert val == "c" and id_ == 3
id_, val = q.pop()
assert val == "d" and id_ == 2
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_enqueue_array(inputs):
n = inputs.shape[0]
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, np.ndarray)
assert len(val.shape) == 1
assert val.size == inputs.shape[-1]
# 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, np.ndarray)
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_with_labels(inputs, labels):
if len(labels) > len(inputs):
labels = labels[:len(inputs)]
elif len(labels) < len(inputs):
labels += [None] * (len(inputs) - len(labels))
n = len(inputs) - (len(labels) - Counter(labels)[None])
q = SimpleLabellingQueue()
q.enqueue_many(inputs, labels)
assert len(q.data) == len(inputs)
# assert we can pop everything where the label was not none:
for _ in range(n):
q.pop()
# assert there's nothing else to pop:
with pytest.raises(IndexError):
q.pop()
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_length_of_queue():
inps = [str(i) for i in range(50)]
q = SimpleLabellingQueue()
assert len(q) == 0
q.enqueue_many(inps)
assert len(q) == len(inps)