import pytest

from longevitynet.modelling.config import CONFIG

from longevitynet.modelling.data import (get_train_image_paths,

get_val_test_image_paths,

get_file_data)

def test_no_overlap_between_train_and_test():

""" Don't test overlap in files, test overlap in people names """

ds_version = CONFIG["DS_VERSION"]

train_image_paths = get_train_image_paths(ds_version)

test_image_paths = get_val_test_image_paths()

people_in_train = [get_file_data(f)['person_name'] for f

in train_image_paths]

people_in_test = [get_file_data(f)['person_name'] for f

in test_image_paths]

overlap = set(people_in_train).intersection(people_in_test)

assert len(overlap) == 0

def test_getting_file_data():

example_paths = get_val_test_image_paths()[:4]

expected_data = [{"person_name": "Carsten Eggers",

"birth": 1957, "death": 2021,

"img_date": 2000, "life_expectancy": 21},

{"person_name": "Béla Éless",

"birth": 1940, "death": 2020,

"img_date": 2014, "life_expectancy": 6},

{"person_name": "Claude Gingras",

"birth": 1931, "death": 2018,

"img_date": 2017, "life_expectancy": 1},

{"person_name": "Franz Muheim",

"birth": 1931, "death": 2020,

"img_date": 1982, "life_expectancy": 32}]

for i, path in enumerate(example_paths):

data = get_file_data(path)

assert data['person_name'] == expected_data[i]['person_name']

assert data['death'] == expected_data[i]['death']

assert data['birth'] == expected_data[i]['birth']