docs

ontology_platform/tests/unit/test_dedup_cache.py

@@ -0,0 +1,32 @@
+from __future__ import annotations
+
+from ont_platform.storage.dedup_cache import InMemoryDedupCache
+
+
+def test_dedup_cache_remembers_project_scoped_fingerprint() -> None:
+    cache = InMemoryDedupCache()
+
+    first = cache.check_and_remember(
+        project_id="proj_1",
+        document_id="doc_a",
+        content_hash="hash-a",
+        fingerprint="fingerprint-a",
+    )
+    second = cache.check_and_remember(
+        project_id="proj_1",
+        document_id="doc_b",
+        content_hash="hash-b",
+        fingerprint="fingerprint-a",
+    )
+    other_project = cache.check_and_remember(
+        project_id="proj_2",
+        document_id="doc_c",
+        content_hash="hash-c",
+        fingerprint="fingerprint-a",
+    )
+
+    assert first.is_duplicate is False
+    assert second.is_duplicate is True
+    assert second.existing_document_id == "doc_a"
+    assert other_project.is_duplicate is False
+    assert len(cache) == 2