add files

tests/test_sample_metadata.py

@@ -0,0 +1,87 @@
+import json
+from pathlib import Path
+
+import fitz
+
+
+ROOT = Path(__file__).resolve().parents[1]
+METADATA_PATH = ROOT / "samples" / "metadata.json"
+
+REQUIRED_TRAITS = {
+    "figure_density",
+    "formula_density",
+    "has_korean_path",
+    "layout_risk",
+    "mixed_scanned_text_pages",
+    "scanned_pages",
+    "table_density",
+    "target_regression_focus",
+    "text_layer_quality",
+}
+
+
+def _load_metadata():
+    with METADATA_PATH.open("r", encoding="utf-8") as handle:
+        return json.load(handle)
+
+
+def _metadata_samples():
+    metadata = _load_metadata()
+    assert isinstance(metadata, dict)
+    samples = metadata.get("samples")
+    assert isinstance(samples, list)
+    return samples
+
+
+def test_metadata_paths_match_current_sample_pdfs_exactly():
+    expected_paths = sorted(
+        path.as_posix()
+        for path in Path("samples").glob("*.pdf")
+    )
+    samples = _metadata_samples()
+    metadata_paths = [sample.get("path") for sample in samples]
+
+    assert sorted(metadata_paths) == expected_paths
+
+
+def test_metadata_paths_are_unique():
+    metadata_paths = [sample.get("path") for sample in _metadata_samples()]
+
+    assert len(metadata_paths) == len(set(metadata_paths))
+
+
+def test_metadata_paths_are_exact_relative_samples_pdf_paths():
+    for sample in _metadata_samples():
+        path = sample.get("path")
+
+        assert isinstance(path, str)
+        assert path.startswith("samples/")
+        assert Path(path).suffix == ".pdf"
+        assert not Path(path).is_absolute()
+        assert (ROOT / path).is_file()
+
+
+def test_required_trait_fields_are_present():
+    for sample in _metadata_samples():
+        traits = sample.get("traits")
+
+        assert isinstance(traits, dict), sample.get("path")
+        assert REQUIRED_TRAITS <= traits.keys(), sample.get("path")
+
+
+def test_page_counts_match_current_sample_pdfs():
+    for sample in _metadata_samples():
+        page_count = sample.get("page_count")
+        path = ROOT / sample["path"]
+
+        assert isinstance(page_count, int), sample["path"]
+        assert page_count > 0, sample["path"]
+        with fitz.open(path) as document:
+            assert page_count == document.page_count, sample["path"]
+
+
+def test_metadata_json_is_deterministic_utf8():
+    raw = METADATA_PATH.read_text(encoding="utf-8")
+    metadata = json.loads(raw)
+
+    assert raw == json.dumps(metadata, ensure_ascii=False, indent=2, sort_keys=True) + "\n"