def test_project_data(db, test_project):
'''
Creates the test project and adds test data to it.
'''
test_data = read_test_data_backend(
file='./core/data/test_files/test_no_labels.csv')
add_data(test_project, test_data)
return test_project
def test_project_all_irr_3_coders_data(db, test_project_all_irr_3_coders):
'''
Creates the test project with 100% irr and adds test data to it.
'''
test_data = read_test_data_backend(
file='./core/data/test_files/test_no_labels.csv')
add_data(test_project_all_irr_3_coders, test_data)
return test_project_all_irr_3_coders
def test_project_all_irr_data(db, test_profile):
'''
Creates the test project with 100% irr and adds test data to it.
'''
project = create_project('test_project', test_profile, 100, 2)
test_data = read_test_data_backend(
file='./core/data/test_files/test_no_labels.csv')
add_data(project, test_data)
return project
def test_project_labeled(test_project):
"""A project that has labeled data."""
for label in SEED_LABELS:
Label.objects.create(name=label, project=test_project)
test_data = read_test_data_backend(
file="./core/data/test_files/test_some_labels.csv"
)
add_data(test_project, test_data)
return test_project
def test_project_gnb_data_tfidf(db, test_profile, tmpdir, settings):
"""This fixture only creates the test project without any data."""
proj = create_project("test_project", test_profile, classifier="gnb")
test_data = read_test_data_backend(file="./core/data/test_files/test_no_labels.csv")
add_data(proj, test_data)
Data.objects.filter(project=proj)
matrix = create_tfidf_matrix(proj.pk)[0]
data_temp = tmpdir.mkdir("data").mkdir("tf_idf")
settings.TF_IDF_PATH = str(data_temp)
save_tfidf_matrix(matrix, proj.pk)
return proj
def test_fill_multiple_projects(db, test_queue, test_profile):
project_data_count = test_queue.project.data_set.count()
test_queue.length = project_data_count + 1
test_queue.save()
test_project2 = create_project("test_project2", test_profile)
project2_data = read_test_data_backend(
file="./core/data/test_files/test_no_labels.csv"
)
add_data(test_project2, project2_data)
fill_queue(test_queue, orderby="random")
# Ensure the queue didn't fill any data from the other project
assert test_queue.data.count() == project_data_count
assert all((d.project == test_queue.project for d in test_queue.data.all()))
def test_project_svm_data_tfidf(db, test_profile, tmpdir, settings):
'''
This fixture only creates the test project without any data.
'''
proj = create_project('test_project', test_profile, classifier="svm")
test_data = read_test_data_backend(
file='./core/data/test_files/test_no_labels.csv')
add_data(proj, test_data)
Data.objects.filter(project=proj)
matrix = create_tfidf_matrix(proj.pk)[0]
data_temp = tmpdir.mkdir('data').mkdir('tf_idf')
settings.TF_IDF_PATH = str(data_temp)
save_tfidf_matrix(matrix, proj.pk)
return proj
def test_add_data_no_labels(db, test_project):
test_data = read_test_data_backend(
file='./core/data/test_files/test_no_labels.csv')
df = add_data(test_project, test_data)
for i, row in df.iterrows():
assert_obj_exists(
Data, {
'upload_id_hash': row['id_hash'],
'hash': row['hash'],
'project': test_project
})
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 test_add_data_no_labels(db, test_project):
test_data = read_test_data_backend(
file="./core/data/test_files/test_no_labels.csv")
df = add_data(test_project, test_data)
for i, row in df.iterrows():
assert_obj_exists(
Data,
{
"upload_id_hash": row["id_hash"],
"hash": row["hash"],
"project": test_project,
},
)
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_add_data_with_labels(db, test_project_labels):
test_data = read_test_data_backend(
file='./core/data/test_files/test_some_labels.csv')
df = add_data(test_project_labels, test_data)
for i, row in df.iterrows():
assert_obj_exists(
Data, {
'upload_id_hash': row['id_hash'],
'hash': row['hash'],
'project': test_project_labels
})
if not pd.isnull(row['Label']):
assert_obj_exists(
DataLabel, {
'data__hash': row['hash'],
'profile': test_project_labels.creator,
'label__name': row['Label']
})
def test_add_data_with_labels(db, test_project_labels):
test_data = read_test_data_backend(
file="./core/data/test_files/test_some_labels.csv")
df = add_data(test_project_labels, test_data)
for i, row in df.iterrows():
assert_obj_exists(
Data,
{
"upload_id_hash": row["id_hash"],
"hash": row["hash"],
"project": test_project_labels,
},
)
if not pd.isnull(row["Label"]):
assert_obj_exists(
DataLabel,
{
"data__hash": row["hash"],
"profile": test_project_labels.creator,
"label__name": row["Label"],
},
)
def test_project_data(db, test_project):
"""Creates the test project and adds test data to it."""
test_data = read_test_data_backend(file="./core/data/test_files/test_no_labels.csv")
add_data(test_project, test_data)
return test_project
def test_project_all_irr_data(db, test_profile):
"""Creates the test project with 100% irr and adds test data to it."""
project = create_project("test_project", test_profile, 100, 2)
test_data = read_test_data_backend(file="./core/data/test_files/test_no_labels.csv")
add_data(project, test_data)
return project
