| import unittest |
|
|
| import disposable_email_domains |
|
|
| from app.ml.build_dataset import NORMAL_DOMAIN_SEEDS, NORMAL_LOCAL_SEEDS |
|
|
|
|
| class NormalSeedIntegrityTests(unittest.TestCase): |
| """The NORMAL class teaches the model what legitimate looks like. A single |
| disposable domain in the seeds would poison the decision boundary - the |
| exact false-positive that flagged stripe.com as DISPOSABLE (0.977).""" |
|
|
| def test_no_normal_domain_is_in_the_disposable_blocklist(self): |
| poisoned = sorted(d for d in NORMAL_DOMAIN_SEEDS |
| if d in disposable_email_domains.blocklist) |
| self.assertEqual(poisoned, [], f"disposable domains in NORMAL seeds: {poisoned}") |
|
|
| def test_seed_diversity_is_sufficient_to_generalize(self): |
| |
| |
| |
| self.assertGreaterEqual(len(NORMAL_DOMAIN_SEEDS), 200, |
| f"only {len(NORMAL_DOMAIN_SEEDS)} NORMAL domains") |
|
|
| def test_seeds_span_many_distinct_tlds(self): |
| tlds = {d.rsplit(".", 1)[-1] for d in NORMAL_DOMAIN_SEEDS} |
| self.assertGreaterEqual(len(tlds), 12, |
| f"NORMAL seeds concentrated in {sorted(tlds)}") |
|
|
| def test_all_seeds_are_wellformed_domains(self): |
| bad = sorted(d for d in NORMAL_DOMAIN_SEEDS |
| if "." not in d or " " in d or d != d.strip().lower()) |
| self.assertEqual(bad, [], f"malformed seeds: {bad}") |
|
|
| def test_no_normal_seed_is_a_known_free_provider(self): |
| |
| |
| from app.validator.disposable import FREE_PROVIDERS |
| overlap = sorted(d for d in NORMAL_DOMAIN_SEEDS if d in FREE_PROVIDERS) |
| self.assertEqual(overlap, [], f"free providers in NORMAL seeds: {overlap}") |
|
|
| def test_local_seeds_are_name_like(self): |
| digits_only = [s for s in NORMAL_LOCAL_SEEDS |
| if s.replace(".", "").replace("_", "").isdigit()] |
| self.assertEqual(digits_only, []) |
| self.assertGreaterEqual(len(NORMAL_LOCAL_SEEDS), 20) |
|
|
|
|
| if __name__ == "__main__": |
| unittest.main() |
