Add Vamana backwards compatibility data generation and tests (#409)

Author: jparismorgan

apis/python/src/tiledb/vector_search/ivf_pq_index.py

@@ -163,6 +163,11 @@ def create(
     """
     warnings.warn("The IVF PQ index is not yet supported, please use with caution.")
     validate_storage_version(storage_version)
+    # TODO(SC-49166): Support old storage versions with type-erased indexes.
+    if storage_version == "0.1" or storage_version == "0.2":
+        raise ValueError(
+            f"Storage version {storage_version} is not supported for IVFPQIndex. IVFPQIndex requires storage version 0.3 or higher."
+        )
     ctx = vspy.Ctx(config)
     if num_subspaces <= 0:
         raise ValueError(

apis/python/src/tiledb/vector_search/vamana_index.py

@@ -159,6 +159,11 @@ def create(
         If not provided, use the latest stable storage version.
     """
     validate_storage_version(storage_version)
+    # TODO(SC-49166): Support old storage versions with type-erased indexes.
+    if storage_version == "0.1" or storage_version == "0.2":
+        raise ValueError(
+            f"Storage version {storage_version} is not supported for VamanaIndex. VamanaIndex requires storage version 0.3 or higher."
+        )
     ctx = vspy.Ctx(config)
     index = vspy.IndexVamana(
         feature_type=np.dtype(vector_type).name,

apis/python/test/test_backwards_compatibility.py

@@ -3,6 +3,7 @@
 from tiledb.vector_search.flat_index import FlatIndex
 from tiledb.vector_search.ivf_flat_index import IVFFlatIndex
 from tiledb.vector_search.utils import load_fvecs
+from tiledb.vector_search.vamana_index import VamanaIndex
 
 MINIMUM_ACCURACY = 0.85
 
@@ -60,6 +61,8 @@ def test_query_old_indices():
                 index = IVFFlatIndex(uri=index_uri)
             elif "flat" in index_name:
                 index = FlatIndex(uri=index_uri)
+            elif "vamana" in index_name:
+                index = VamanaIndex(uri=index_uri)
             else:
                 assert False, f"Unknown index name: {index_name}"
 

backwards-compatibility-data/generate_data.py

@@ -66,7 +66,7 @@ def generate_indexes(version):
     queries = base[indices]
 
     # Generate each index and query to make sure it works before we write it.
-    index_types = ["FLAT", "IVF_FLAT"]
+    index_types = ["FLAT", "IVF_FLAT", "VAMANA"]
     data_types = ["float32", "uint8"]
     for index_type in index_types:
         for data_type in data_types:

src/include/test/unit_backwards_compatibility.cc

@@ -118,8 +118,26 @@ TEST_CASE("test_query_old_indices", "[backwards_compatibility]") {
         // TODO(paris): Fix flat_l2_index and re-enable. Right now it just tries
         // to load the URI as a tdbMatrix.
       } else if (index_uri.find("vamana") != std::string::npos) {
-        // TODO(paris): Add once we save a vamana index in backwards
-        // compatibility data.
+        // First check that we can query the index.
+        auto index = IndexVamana(ctx, index_uri);
+        auto&& [scores, ids] = index.query(queries_feature_vector_array, 1);
+        auto scores_span =
+            MatrixView<siftsmall_feature_type, stdx::layout_left>{
+                (siftsmall_feature_type*)scores.data(),
+                extents(scores)[0],
+                extents(scores)[1]};
+
+        auto ids_span = MatrixView<siftsmall_ids_type, stdx::layout_left>{
+            (siftsmall_ids_type*)ids.data(), extents(ids)[0], extents(ids)[1]};
+
+        for (size_t i = 0; i < query_indices.size(); ++i) {
+          CHECK(ids_span[0][i] == query_indices[i]);
+          CHECK(scores_span[0][i] == 0);
+        }
+
+        // Next check that we can load the metadata.
+        auto metadata = vamana_index_metadata();
+        metadata.load_metadata(read_group);
       } else {
         REQUIRE(false);
       }