Remove unused code previously used to define arrays for testing (#346)

Author: jparismorgan

apis/python/test/array_paths.py

@@ -9,14 +9,16 @@
 this_file_path = os.path.join(os.path.dirname(__file__))
 vector_search_root = this_file_path + "/../../../"
 
-# vector_search_root = "../../"
-
 test_data_root = vector_search_root + "external/test_data/"
 test_array_root = test_data_root + "arrays/"
 test_file_root = test_data_root + "files/"
 
-siftsmall_root = test_array_root + "siftsmall/"
 siftsmall_file_root = test_file_root + "siftsmall/"
+siftsmall_inputs_file = siftsmall_file_root + "input_vectors.fvecs"
+siftsmall_query_file = siftsmall_file_root + "queries.fvecs"
+siftsmall_groundtruth_file = siftsmall_file_root + "groundtruth.ivecs"
+
+siftsmall_root = test_array_root + "siftsmall/"
 siftsmall_group_uri = siftsmall_root + "group"
 siftsmall_inputs_uri = siftsmall_root + "input_vectors"
 siftsmall_centroids_uri = siftsmall_root + "partition_centroids"
@@ -26,36 +28,17 @@
 siftsmall_query_uri = siftsmall_root + "queries"
 siftsmall_groundtruth_uri = siftsmall_root + "groundtruth"
 
-siftsmall_inputs_file = siftsmall_file_root + "input_vectors.fvecs"
-siftsmall_query_file = siftsmall_file_root + "queries.fvecs"
-siftsmall_groundtruth_file = siftsmall_file_root + "groundtruth.ivecs"
-
 siftsmall_uint8_root = test_array_root + "siftsmall_uint8/"
-siftsmall_uint8_group_uri = siftsmall_uint8_root + "group"
 siftsmall_uint8_inputs_uri = siftsmall_uint8_root + "input_vectors"
-siftsmall_uint8_centroids_uri = siftsmall_uint8_root + "partition_centroids"
-siftsmall_uint8_index_uri = siftsmall_uint8_root + "partition_indexes"
 siftsmall_uint8_ids_uri = siftsmall_uint8_root + "shuffled_vector_ids"
-siftsmall_uint8_parts_uri = siftsmall_uint8_root + "shuffled_vectors"
-siftsmall_uint8_query_uri = siftsmall_uint8_root + "queries"
 siftsmall_uint8_groundtruth_uri = siftsmall_uint8_root + "groundtruth"
 
 bigann10k_root = test_array_root + "bigann10k/"
-bigann10k_group_uri = bigann10k_root + "group"
 bigann10k_inputs_uri = bigann10k_root + "input_vectors"
-bigann10k_centroids_uri = bigann10k_root + "partition_centroids"
-bigann10k_index_uri = bigann10k_root + "partition_indexes"
 bigann10k_ids_uri = bigann10k_root + "shuffled_vector_ids"
-bigann10k_parts_uri = bigann10k_root + "shuffled_vectors"
-bigann10k_query_uri = bigann10k_root + "queries"
 bigann10k_groundtruth_uri = bigann10k_root + "groundtruth"
 
 fmnistsmall_root = test_array_root + "fmnistsmall/"
-fmnistsmall_group_uri = fmnistsmall_root + "group"
 fmnistsmall_inputs_uri = fmnistsmall_root + "input_vectors"
-fmnistsmall_centroids_uri = fmnistsmall_root + "partition_centroids"
-fmnistsmall_index_uri = fmnistsmall_root + "partition_indexes"
 fmnistsmall_ids_uri = fmnistsmall_root + "shuffled_vector_ids"
-fmnistsmall_parts_uri = fmnistsmall_root + "shuffled_vectors"
-fmnistsmall_query_uri = fmnistsmall_root + "queries"
 fmnistsmall_groundtruth_uri = fmnistsmall_root + "groundtruth"

apis/python/test/data/siftsmall/siftsmall_base.fvecs

@@ -17,12 +17,11 @@ def create_sift_micro():
     base_uri = os.path.join(
         base_dir,
         "..",
-        "apis",
-        "python",
-        "test",
-        "data",
+        "external",
+        "test_data",
+        "files",
         "siftsmall",
-        "siftsmall_base.fvecs",
+        "input_vectors.fvecs",
     )
     write_fvecs(
         os.path.join(base_dir, "siftmicro_base.fvecs"), load_fvecs(base_uri)[:100]

apis/python/test/data/siftsmall/siftsmall_groundtruth.ivecs

@@ -145,29 +145,6 @@ static std::string siftsmall_parts_uri{siftsmall_root / "shuffled_vectors"};
 static std::string siftsmall_query_uri{siftsmall_root / "queries"};
 static std::string siftsmall_groundtruth_uri{siftsmall_root / "groundtruth"};
 
-using siftsmall_uint8_feature_type = uint8_t;
-using siftsmall_uint8_groundtruth_type = uint64_t;
-using siftsmall_uint8_centroids_type = float;
-using siftsmall_uint8_ids_type = uint64_t;
-using siftsmall_uint8_indices_type = uint64_t;
-constexpr size_t num_siftsmall_uint8_vectors = 10'000;
-constexpr size_t siftsmall_uint8_dimension = 128;
-static std::string siftsmall_uint8_root{test_array_root / "siftsmall_uint8"};
-static std::string siftsmall_uint8_group_uri{siftsmall_uint8_root / "group"};
-static std::string siftsmall_uint8_inputs_uri{
-    siftsmall_uint8_root / "input_vectors"};
-static std::string siftsmall_uint8_centroids_uri{
-    siftsmall_uint8_root / "partition_centroids"};
-static std::string siftsmall_uint8_index_uri{
-    siftsmall_uint8_root / "partition_indexes"};
-static std::string siftsmall_uint8_ids_uri{
-    siftsmall_uint8_root / "shuffled_vector_ids"};
-static std::string siftsmall_uint8_parts_uri{
-    siftsmall_uint8_root / "shuffled_vectors"};
-static std::string siftsmall_uint8_query_uri{siftsmall_uint8_root / "queries"};
-static std::string siftsmall_uint8_groundtruth_uri{
-    siftsmall_uint8_root / "groundtruth"};
-
 using bigann1M_feature_type = uint8_t;
 using bigann1M_groundtruth_type = uint64_t;
 using bigann1M_centroids_type = float;
@@ -298,33 +275,6 @@ static std::string diskann_mem_index{
 static std::string diskann_truth_index_data =
     diskann_root / "truth_index_siftsmall_learn_256pts_R4_L50_A1.2.data";
 
-/*
- * Definitions for the "nano" reference group for vamana index
- */
-using vamana_nano_feature_type = float;
-using vamana_nano_groundtruth_type = uint64_t;
-using vamana_nano_ids_type = uint64_t;
-using vamana_nano_indices_type = uint64_t;
-constexpr size_t num_vamana_nano_vectors = 231;
-constexpr size_t vamana_nano_dimension = 128;
-static std::string vamana_nano_root{nano_root / "vamana"};
-static std::string vamana_nano_group_uri{
-    vamana_nano_root / "vamana_test_index"};
-
-static std::string vamana_nano_inputs_uri{vamana_nano_root / "feature_vectors"};
-static std::string vamana_nano_feature_vectors_uri{
-    vamana_nano_root / "feature_vectors"};
-static std::string vamana_nano_adjacency_scores_uri{
-    vamana_nano_root / "adjacency_scores"};
-static std::string vamana_nano_adjacency_ids_uri{
-    vamana_nano_root / "adjacency_ids"};
-static std::string vamana_nano_adjacency_row_index_uri{
-    vamana_nano_root / "adjacency_row_index"};
-
-static std::string vamana_nano_query_uri{vamana_nano_root / "queries"};
-static std::string vamana_nano_groundtruth_uri{
-    vamana_nano_root / "groundtruth"};
-
 #define TEMP_LEGACY_URIS
 #ifdef TEMP_LEGACY_URIS
 using db_type = siftsmall_feature_type;

apis/python/test/data/siftsmall/siftsmall_learn.fvecs

@@ -85,18 +85,6 @@ std::vector<std::string> siftsmall_array_uris{
     siftsmall_groundtruth_uri,
 };
 
-// Note that we don't have a canonical IVF index for siftsmall_uint8 yet, so
-// some of these URIs are placeholders
-std::vector<std::string> siftsmall_uint8_array_uris{
-    siftsmall_uint8_inputs_uri,
-    siftsmall_uint8_centroids_uri,
-    siftsmall_uint8_index_uri,
-    siftsmall_uint8_ids_uri,
-    siftsmall_uint8_parts_uri,
-    siftsmall_uint8_query_uri,
-    siftsmall_uint8_groundtruth_uri,
-};
-
 // Note that we don't have a canonical IVF index for bigann10k yet, so some
 // of these URIs are placeholders
 std::vector<std::string> bigann10k_array_uris{