def test_no_irr_all_queues(db, test_project_no_irr_data):
"""A queue containing data from the test project, with length set to the global
len."""
normal_q = add_queue(test_project_no_irr_data, TEST_QUEUE_LEN)
admin_q = add_queue(test_project_no_irr_data, TEST_QUEUE_LEN, type="admin")
irr_q = add_queue(test_project_no_irr_data, MAX_DATA_LEN, type="irr")
return [normal_q, admin_q, irr_q]
def test_gnb_queue_list(db, test_project_gnb_data_tfidf):
normal_q = add_queue(test_project_gnb_data_tfidf, TEST_QUEUE_LEN)
admin_q = add_queue(test_project_gnb_data_tfidf,
TEST_QUEUE_LEN,
type="admin")
irr_q = add_queue(test_project_gnb_data_tfidf, MAX_DATA_LEN, type="irr")
return [normal_q, admin_q, irr_q]
def test_init_redis_one_empty_queue(db, test_project, test_redis):
add_queue(test_project, 10)
test_redis.flushdb()
init_redis()
assert_redis_matches_db(test_redis)
def test_add_queue_no_profile(test_project):
QUEUE_LEN = 10
add_queue(test_project, QUEUE_LEN)
assert_obj_exists(Queue, {
'project': test_project,
'length': QUEUE_LEN,
'profile': None
})
def test_add_queue_profile(test_project, test_profile):
QUEUE_LEN = 10
add_queue(test_project, QUEUE_LEN, profile=test_profile)
assert_obj_exists(Queue, {
'project': test_project,
'length': QUEUE_LEN,
'profile': test_profile
})
def test_randomforest_queue_list(db, test_project_randomforest_data_tfidf):
normal_q = add_queue(
test_project_randomforest_data_tfidf, TEST_QUEUE_LEN, type="normal"
)
admin_q = add_queue(
test_project_randomforest_data_tfidf, TEST_QUEUE_LEN, type="admin"
)
irr_q = add_queue(test_project_randomforest_data_tfidf, TEST_QUEUE_LEN, type="irr")
return [normal_q, admin_q, irr_q]
def test_init_redis_multiple_queues(db, test_project_data, test_redis):
queue = add_queue(test_project_data, 10)
fill_queue(queue, orderby="random")
add_queue(test_project_data, 10)
test_redis.flushdb()
init_redis()
assert_redis_matches_db(test_redis)
def test_get_nonempty_queue_noprofile(db, test_project_data):
queue_len = 10
queue = add_queue(test_project_data, queue_len)
queue2 = add_queue(test_project_data, queue_len)
assert get_nonempty_queue(test_project_data) is None
fill_queue(queue2, orderby="random")
assert get_nonempty_queue(test_project_data) == queue2
fill_queue(queue, orderby="random")
assert get_nonempty_queue(test_project_data) == queue
def test_get_nonempty_profile_queue(db, test_project_data, test_profile):
queue_len = 10
add_queue(test_project_data, queue_len)
profile_queue = add_queue(test_project_data, queue_len, profile=test_profile)
profile_queue2 = add_queue(test_project_data, queue_len, profile=test_profile)
assert get_nonempty_queue(test_project_data, profile=test_profile) is None
fill_queue(profile_queue2, orderby="random")
assert get_nonempty_queue(test_project_data, profile=test_profile) == profile_queue2
fill_queue(profile_queue, orderby="random")
assert get_nonempty_queue(test_project_data, profile=test_profile) == profile_queue
def seed_project(creator, name, description, data_file, label_list, perm_list,
classifier):
project = Project.objects.create(name=name,
description=description,
creator=creator,
classifier=classifier)
TrainingSet.objects.create(project=project, set_number=0)
labels = []
for name in label_list:
labels.append(Label.objects.create(name=name, project=project))
permissions = []
for perm in perm_list:
permissions.append(
ProjectPermissions.objects.create(profile=perm,
project=project,
permission="CODER"))
batch_size = 10 * len(labels)
project.batch_size = batch_size
project.save()
num_coders = len(permissions) + 1
q_length = find_queue_length(batch_size, num_coders)
queue = add_queue(project=project, length=q_length, type="normal")
# Data
f_data = read_test_data_backend(file=data_file)
data_length = len(f_data)
add_queue(project=project, length=data_length, type="admin")
irr_queue = add_queue(project=project, length=2000000, type="irr")
new_df = add_data(project, f_data)
fill_queue(queue,
irr_queue=irr_queue,
orderby="random",
batch_size=batch_size)
save_data_file(new_df, project.pk)
tasks.send_tfidf_creation_task.apply(args=[project.pk])
tasks.send_check_and_trigger_model_task.apply(args=[project.pk])
return project
def test_init_redis_one_nonempty_queue(db, test_project_data, test_redis):
queue = add_queue(test_project_data, 10)
fill_queue(queue, orderby="random")
test_redis.flushdb()
init_redis()
assert_redis_matches_db(test_redis)
def test_pop_only_affects_one_queue(db, test_project_data, test_redis):
queue_len = 10
queue = add_queue(test_project_data, queue_len)
queue2 = add_queue(test_project_data, queue_len)
fill_queue(queue, orderby="random")
fill_queue(queue2, orderby="random")
datum = pop_queue(queue)
assert isinstance(datum, Data)
assert test_redis.llen("queue:" + str(queue.pk)) == (queue_len - 1)
assert test_redis.scard("set:" + str(queue.pk)) == (queue_len)
assert queue.data.count() == queue_len
assert test_redis.llen("queue:" + str(queue2.pk)) == queue_len
assert test_redis.scard("set:" + str(queue2.pk)) == (queue_len)
assert queue2.data.count() == queue_len
def test_pop_empty_queue(db, test_project, test_redis):
queue = add_queue(test_project, 10)
datum = pop_queue(queue)
assert datum is None
assert not test_redis.exists("queue:" + str(queue.pk))
assert queue.data.count() == 0
def test_pop_nonempty_queue(db, test_project_data, test_redis):
queue_len = 10
queue = add_queue(test_project_data, queue_len)
fill_queue(queue, orderby='random')
datum = pop_queue(queue)
assert isinstance(datum, Data)
assert test_redis.llen('queue:' + str(queue.pk)) == (queue_len - 1)
assert test_redis.scard('set:' + str(queue.pk)) == (queue_len)
assert queue.data.count() == queue_len
def test_pop_first_nonempty_queue_multiple_queues(db, test_project_data,
test_queue, test_redis):
test_queue2 = add_queue(test_project_data, 10)
fill_queue(test_queue2, orderby='random')
queue, data = pop_first_nonempty_queue(test_project_data)
assert isinstance(queue, Queue)
assert queue == test_queue2
fill_queue(test_queue, orderby='random')
queue, data = pop_first_nonempty_queue(test_project_data)
assert isinstance(queue, Queue)
assert queue == test_queue
def test_init_redis_multiple_projects(db, test_project_data, test_redis,
test_profile):
# Try a mix of multiple queues in multiple projects with
# and without data to see if everything initializes as expected.
p1_queue1 = add_queue(test_project_data, 10)
fill_queue(p1_queue1, orderby="random")
add_queue(test_project_data, 10)
project2 = create_project("test_project2", test_profile)
project2_data = read_test_data_backend(
file="./core/data/test_files/test_no_labels.csv")
add_data(project2, project2_data)
p2_queue1 = add_queue(project2, 10)
fill_queue(p2_queue1, orderby="random")
add_queue(project2, 10)
test_redis.flushdb()
init_redis()
assert_redis_matches_db(test_redis)
def done(self, form_list, form_dict, **kwargs):
proj = form_dict['project']
labels = form_dict['labels']
permissions = form_dict['permissions']
advanced = form_dict['advanced']
data = form_dict['data']
codebook_data = form_dict['codebook']
with transaction.atomic():
# Project
proj_obj = proj.save(commit=False)
advanced_data = advanced.cleaned_data
proj_obj.creator = self.request.user.profile
# Advanced Options
proj_obj.save()
proj_pk = proj_obj.pk
# Save the codebook file
cb_data = codebook_data.cleaned_data['data']
if cb_data != "":
cb_filepath = save_codebook_file(cb_data, proj_pk)
else:
cb_filepath = ""
proj_obj.codebook_file = cb_filepath
if advanced_data["batch_size"] == 0:
batch_size = 10 * len([x for x in labels if x.cleaned_data
!= {} and not x.cleaned_data['DELETE']])
else:
batch_size = advanced_data["batch_size"]
proj_obj.batch_size = batch_size
proj_obj.learning_method = advanced_data["learning_method"]
proj_obj.percentage_irr = advanced_data["percentage_irr"]
proj_obj.num_users_irr = advanced_data["num_users_irr"]
proj_obj.classifier = advanced_data["classifier"]
proj_obj.save()
# Training Set
TrainingSet.objects.create(project=proj_obj, set_number=0)
# Labels
labels.instance = proj_obj
labels.save()
# Permissions
permissions.instance = proj_obj
permissions.save()
# Queue
num_coders = len([x for x in permissions if x.cleaned_data
!= {} and not x.cleaned_data['DELETE']]) + 1
q_length = find_queue_length(batch_size, num_coders)
queue = add_queue(project=proj_obj, length=q_length)
# Data
f_data = data.cleaned_data['data']
add_queue(project=proj_obj, length=2000000, type="admin")
irr_queue = add_queue(project=proj_obj, length=2000000, type="irr")
upload_data(f_data, proj_obj, queue, irr_queue, batch_size)
return HttpResponseRedirect(proj_obj.get_absolute_url())
def test_add_queue_no_profile(test_project):
QUEUE_LEN = 10
add_queue(test_project, QUEUE_LEN)
assert_obj_exists(
Queue, {"project": test_project, "length": QUEUE_LEN, "profile": None}
)
def test_add_queue_profile(test_project, test_profile):
QUEUE_LEN = 10
add_queue(test_project, QUEUE_LEN, profile=test_profile)
assert_obj_exists(
Queue, {"project": test_project, "length": QUEUE_LEN, "profile": test_profile}
)
def test_admin_queue(db, test_project_data):
'''
A queue containing data from the test project, with length set to
the global len.
'''
return add_queue(test_project_data, TEST_QUEUE_LEN, type="admin")
def test_queue_labeled(db, test_project_labeled):
'''
A queue containing data from the test project, with length set to
the global len.
'''
return add_queue(test_project_labeled, TEST_QUEUE_LEN, type="normal")
def test_queue(db, test_project_data):
"""A queue containing data from the test project, with length set to the global
len."""
return add_queue(test_project_data, TEST_QUEUE_LEN)
def test_irr_queue(db, test_project_data):
'''
A queue containing data from the test project, with length set to
the global len.
'''
return add_queue(test_project_data, MAX_DATA_LEN, type="irr")
def test_irr_queue_labeled(db, test_project_labeled):
"""A queue containing data from the test project, with length set to the global
len."""
return add_queue(test_project_labeled, MAX_DATA_LEN, type="irr")
def test_admin_queue_labeled(db, test_project_labeled):
"""A queue containing data from the test project, with length set to the global
len."""
return add_queue(test_project_labeled, TEST_QUEUE_LEN, type="admin")
def test_profile_queue(db, test_profile, test_project_data):
"""A queue with test data, associated with the first test profile."""
return add_queue(test_project_data, TEST_QUEUE_LEN, profile=test_profile)
def test_profile_queue2(db, test_profile2, test_project_data):
"""A queue with test data, associated with an additional test profile.
Useful for tests requiring multiple profiles/queues on the same project.
"""
return add_queue(test_project_data, TEST_QUEUE_LEN, profile=test_profile2)
def test_half_irr_all_queues(db, test_project_half_irr):
normal_q = add_queue(test_project_half_irr, TEST_QUEUE_LEN)
admin_q = add_queue(test_project_half_irr, TEST_QUEUE_LEN, type="admin")
irr_q = add_queue(test_project_half_irr, MAX_DATA_LEN, type="irr")
return [normal_q, admin_q, irr_q]
