Refactor MemorySystem to enhance Chroma vector store initialization with conflict handling and backup creation. Added methods for conflict detection and store repair, improving robustness against schema conflicts.

tests/test_memory_chroma_recovery.py

@@ -0,0 +1,78 @@
+from pathlib import Path
+
+import src.ai.memory as memory_module
+
+
+class _FakeJSONStore:
+    def __init__(self, storage_path: Path):
+        self.storage_path = storage_path
+
+
+def test_memory_retries_once_on_chroma_conflict(monkeypatch, tmp_path: Path):
+    class _FlakyChroma:
+        calls = 0
+
+        def __init__(self, persist_directory: Path):
+            type(self).calls += 1
+            self.persist_directory = persist_directory
+            if type(self).calls == 1:
+                raise RuntimeError("table embeddings already exists")
+
+    monkeypatch.setattr(memory_module, "ChromaVectorStore", _FlakyChroma)
+    monkeypatch.setattr(memory_module, "JSONVectorStore", _FakeJSONStore)
+
+    memory = memory_module.MemorySystem(
+        storage_path=tmp_path / "long_term_memory.json",
+        use_vector_db=True,
+    )
+
+    assert isinstance(memory.vector_store, _FlakyChroma)
+    assert _FlakyChroma.calls == 2
+
+
+def test_memory_rebuilds_chroma_directory_when_conflict_persists(monkeypatch, tmp_path: Path):
+    class _RecoverOnThirdTryChroma:
+        calls = 0
+
+        def __init__(self, persist_directory: Path):
+            type(self).calls += 1
+            self.persist_directory = persist_directory
+            if type(self).calls <= 2:
+                raise RuntimeError("table embeddings already exists")
+
+    monkeypatch.setattr(memory_module, "ChromaVectorStore", _RecoverOnThirdTryChroma)
+    monkeypatch.setattr(memory_module, "JSONVectorStore", _FakeJSONStore)
+
+    storage_root = tmp_path / "ai"
+    chroma_dir = storage_root / "chroma_db"
+    chroma_dir.mkdir(parents=True, exist_ok=True)
+    (chroma_dir / "marker.txt").write_text("legacy-data", encoding="utf-8")
+
+    memory = memory_module.MemorySystem(
+        storage_path=storage_root / "long_term_memory.json",
+        use_vector_db=True,
+    )
+
+    assert isinstance(memory.vector_store, _RecoverOnThirdTryChroma)
+    assert _RecoverOnThirdTryChroma.calls == 3
+
+    backups = sorted(storage_root.glob("chroma_db_backup_conflict_*"))
+    assert backups, "expected backup directory after repair"
+    assert (backups[-1] / "marker.txt").exists()
+    assert chroma_dir.exists()
+
+
+def test_memory_falls_back_to_json_if_chroma_repair_fails(monkeypatch, tmp_path: Path):
+    class _AlwaysFailChroma:
+        def __init__(self, persist_directory: Path):
+            raise RuntimeError("table embeddings already exists")
+
+    monkeypatch.setattr(memory_module, "ChromaVectorStore", _AlwaysFailChroma)
+    monkeypatch.setattr(memory_module, "JSONVectorStore", _FakeJSONStore)
+
+    memory = memory_module.MemorySystem(
+        storage_path=tmp_path / "long_term_memory.json",
+        use_vector_db=True,
+    )
+
+    assert isinstance(memory.vector_store, _FakeJSONStore)