| """Test basic dataset integrity: format, required fields, property ranges, splits.""" |
| import json |
| from pathlib import Path |
|
|
| DATASET = Path(__file__).parent.parent / "dataset/entries_final_v3.json" |
|
|
|
|
| def load_dataset(): |
| with open(DATASET) as f: |
| return json.load(f) |
|
|
|
|
| _ENTRIES = None |
|
|
| def get_entries(): |
| global _ENTRIES |
| if _ENTRIES is None: |
| _ENTRIES = load_dataset() |
| return _ENTRIES |
|
|
| class TestDatasetIntegrity: |
| def setup_method(self): |
| self.entries = get_entries() |
| self.N = len(self.entries) |
|
|
| def test_dataset_not_empty(self): |
| assert self.N > 0, "Dataset is empty" |
|
|
| def test_all_entries_have_formula(self): |
| missing = [e for e in self.entries if not e.get("formula")] |
| assert len(missing) == 0, f"{len(missing)} entries missing formula" |
|
|
| def test_all_entries_have_structured_formula(self): |
| missing = [e for e in self.entries if not e.get("structured_formula")] |
| assert len(missing) == 0, f"{len(missing)} entries missing structured_formula" |
|
|
| def test_all_entries_have_elements(self): |
| missing = [e for e in self.entries if not e.get("elements")] |
| assert len(missing) == 0, f"{len(missing)} entries missing elements" |
|
|
| def test_all_entries_have_source(self): |
| missing = [e for e in self.entries if not e.get("source")] |
| assert len(missing) == 0, f"{len(missing)} entries missing source" |
|
|
| def test_all_entries_have_quality_score(self): |
| missing = [e for e in self.entries if e.get("quality_score") is None] |
| assert len(missing) == 0, f"{len(missing)} entries missing quality_score" |
|
|
| def test_all_entries_have_tier(self): |
| missing = [e for e in self.entries if not e.get("tier")] |
| assert len(missing) == 0, f"{len(missing)} entries missing tier" |
|
|
| def test_all_entries_have_provenance(self): |
| missing = [e for e in self.entries if not e.get("provenance")] |
| assert len(missing) == 0, f"{len(missing)} entries missing provenance" |
|
|
| def test_quality_score_range(self): |
| scores = [e.get("quality_score", 0) for e in self.entries] |
| assert all(0 <= s <= 100 for s in scores), "Quality score out of [0, 100]" |
|
|
| def test_volume_non_negative(self): |
| volumes = [e.get("volume", -1) for e in self.entries] |
| assert all(v > 0 for v in volumes), f"{sum(1 for v in volumes if v <= 0)} entries with volume <= 0" |
|
|
| def test_density_non_negative(self): |
| densities = [e.get("density", -1) for e in self.entries] |
| assert all(d >= 0 for d in densities), f"{sum(1 for d in densities if d < 0)} entries with density < 0" |
|
|
| def test_known_sources(self): |
| sources = set(e.get("source") for e in self.entries) |
| assert sources <= {"mp", "oqmd", "jarvis"}, f"Unknown sources: {sources - {'mp', 'oqmd', 'jarvis'}}" |
|
|
| def test_known_tiers(self): |
| tiers = set(e.get("tier") for e in self.entries) |
| assert tiers <= {"gold", "validated", "raw"}, f"Unknown tiers: {tiers - {'gold', 'validated', 'raw'}}" |
|
|
| def test_no_stale_missing_spacegroup_flags(self): |
| stale = [e for e in self.entries |
| if e.get("source") == "oqmd" |
| and e.get("space_group") is not None |
| and "missing_spacegroup" in e.get("quality_flags", [])] |
| assert len(stale) == 0, f"{len(stale)} OQMD entries have stale missing_spacegroup flag" |
|
|
| def test_gold_tier_no_extreme_fe(self): |
| extreme = [e for e in self.entries |
| if e.get("tier") == "gold" |
| and e.get("formation_energy_per_atom") is not None |
| and abs(e.get("formation_energy_per_atom")) > 5] |
| assert len(extreme) == 0, f"{len(extreme)} Gold entries have |FE| > 5 eV/atom" |
|
|
| def test_gold_tier_no_extreme_eah(self): |
| extreme = [e for e in self.entries |
| if e.get("tier") == "gold" |
| and e.get("energy_above_hull") is not None |
| and e.get("energy_above_hull") > 5] |
| assert len(extreme) == 0, f"{len(extreme)} Gold entries have EaH > 5 eV/atom" |
|
|
| def test_tier_distribution(self): |
| tiers = {} |
| for e in self.entries: |
| tiers[e.get("tier")] = tiers.get(e.get("tier"), 0) + 1 |
| assert tiers.get("gold", 0) > 0, "No Gold entries" |
| assert tiers.get("validated", 0) > 0, "No Validated entries" |
|
|
| def test_source_distribution(self): |
| for src in ["mp", "oqmd", "jarvis"]: |
| cnt = sum(1 for e in self.entries if e.get("source") == src) |
| assert cnt > 0, f"Source {src} has 0 entries" |
|
|
| def test_splits_distinct(self): |
| import json |
| base = Path(__file__).parent.parent / "dataset/splits" |
| splits = {} |
| for name in ["random_80_10_10", "composition_held_out", "family_held_out", "chemistry_held_out"]: |
| with open(base / f"{name}.json") as f: |
| splits[name] = json.load(f) |
| assert splits["random_80_10_10"]["train"] != splits["composition_held_out"]["train"], \ |
| "composition_held_out is identical to random_80_10_10 — split bug" |
|
|
| def test_splits_no_overlap(self): |
| import json |
| base = Path(__file__).parent.parent / "dataset/splits" |
| for name in ["random_80_10_10", "composition_held_out", "family_held_out", "chemistry_held_out"]: |
| with open(base / f"{name}.json") as f: |
| sp = json.load(f) |
| train_s = set(sp["train"]) |
| val_s = set(sp["val"]) |
| test_s = set(sp["test"]) |
| assert len(train_s & val_s) == 0, f"{name}: train/val overlap" |
| assert len(train_s & test_s) == 0, f"{name}: train/test overlap" |
| assert len(val_s & test_s) == 0, f"{name}: val/test overlap" |
|
|
