def test_project_unlabeled_and_tfidf(test_project_data, test_tfidf_matrix,
tmpdir, settings):
data_temp = tmpdir.mkdir('data').mkdir('tf_idf')
settings.TF_IDF_PATH = str(data_temp)
save_tfidf_matrix(test_tfidf_matrix, test_project_data.pk)
return test_project_data
def test_project_labeled_and_tfidf(test_project_labeled,
test_tfidf_matrix_labeled,
test_tfidf_vectorizer_labeled, tmpdir,
settings):
data_temp = tmpdir.mkdir('data').mkdir('tf_idf')
settings.TF_IDF_PATH = str(data_temp)
save_tfidf_matrix(test_tfidf_matrix_labeled, test_project_labeled.pk)
save_tfidf_vectorizer(test_tfidf_vectorizer_labeled,
test_project_labeled.pk)
return test_project_labeled
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_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 send_tfidf_creation_task(project_pk):
"""Create and Save tfidf"""
from core.utils.utils_model import create_tfidf_matrix, save_tfidf_matrix, save_tfidf_vectorizer
tf_idf, vectorizer = create_tfidf_matrix(project_pk)
file = save_tfidf_matrix(tf_idf, project_pk)
save_tfidf_vectorizer(vectorizer, project_pk)
return file
def test_save_tfidf_matrix(test_project_data, test_tfidf_matrix, tmpdir,
settings):
data_temp = tmpdir.mkdir('data').mkdir('tf_idf')
settings.TF_IDF_PATH = str(data_temp)
file = save_tfidf_matrix(test_tfidf_matrix, test_project_data.pk)
assert os.path.isfile(file)
assert file == os.path.join(
settings.TF_IDF_PATH,
'project_' + str(test_project_data.pk) + '_tfidf_matrix.pkl')
def test_save_tfidf_matrix(test_project_data, test_tfidf_matrix, tmpdir, settings):
data_temp = tmpdir.mkdir("data").mkdir("tf_idf")
settings.TF_IDF_PATH = str(data_temp)
file = save_tfidf_matrix(test_tfidf_matrix, test_project_data.pk)
assert os.path.isfile(file)
assert file == os.path.join(
settings.TF_IDF_PATH,
"project_" + str(test_project_data.pk) + "_tfidf_matrix.pkl",
)