feat(preprocess): publish incremental Evidence corpus

harness/tests/test_corpus_publish.py

@@ -0,0 +1,43 @@
+import pytest
+
+from tht.corpus.models import CorpusManifest
+from tht.corpus.store import CorpusStore, UnsafeCorpusPath
+
+
+def test_publish_switches_active_atomically_and_resolves_materialized_files(tmp_path):
+    store = CorpusStore(tmp_path / "corpus")
+    generation = store.stage(CorpusManifest(), {})
+    seen = []
+    store._replace = lambda source, target: (seen.append(source.read_text()), source.replace(target))
+    published = store.publish(generation)
+    assert published == generation
+    assert store.active_generation() == generation
+    assert seen == [generation + "\n"]
+
+
+def test_active_manifest_is_a_consistent_reader_snapshot(tmp_path):
+    store = CorpusStore(tmp_path / "corpus")
+    first = store.stage(CorpusManifest(metadata={"name": "first"}), {})
+    second = store.stage(CorpusManifest(metadata={"name": "second"}), {})
+    store.publish(first)
+    snapshot = store.active_manifest()
+    store.publish(second)
+    assert snapshot.metadata["name"] == "first"
+    assert store.active_manifest().metadata["name"] == "second"
+
+
+def test_store_rejects_symlinked_generation_root(tmp_path):
+    outside = tmp_path / "outside"
+    outside.mkdir()
+    root = tmp_path / "corpus"
+    root.symlink_to(outside, target_is_directory=True)
+    with pytest.raises(UnsafeCorpusPath):
+        CorpusStore(root)
+
+
+def test_active_pointer_cannot_escape_generation_root(tmp_path):
+    store = CorpusStore(tmp_path / "corpus")
+    store.root.mkdir(parents=True, exist_ok=True)
+    store.active_path.write_text("../outside\n")
+    with pytest.raises(UnsafeCorpusPath):
+        store.active_manifest()