Cleanup IVF PQ C++ index code (#421)

Author: jparismorgan

src/include/index/ivf_flat_index.h

@@ -524,20 +524,6 @@ class ivf_flat_index {
     return true;
   }
 
-  auto write_index_arrays(
-      const tiledb::Context& ctx,
-      const std::string& centroids_uri,
-      const std::string& parts_uri,
-      const std::string& ids_uri,
-      const std::string& indices_uri) const {
-    write_matrix(ctx, centroids_, centroids_uri, 0, true);
-    write_matrix(ctx, *partitioned_vectors_, parts_uri, 0, true);
-    write_vector(ctx, partitioned_vectors_->ids(), ids_uri, 0, true);
-    write_vector(ctx, partitioned_vectors_->indices(), indices_uri, 0, true);
-
-    return true;
-  }
-
   /*****************************************************************************
    *
    * Queries, infinite and finite.
@@ -1033,23 +1019,14 @@ class ivf_flat_index {
     return indices;
   }
 
-  void dump_group(const std::string& msg) {
+  void dump(const std::string& msg) {
     if (!group_) {
       throw std::runtime_error(
-          "[ivf_flat_index@dump_group] Cannot dump group because there is no "
+          "[ivf_flat_index@dump] Cannot dump group because there is no "
           "group");
     }
     group_->dump(msg);
   }
-
-  void dump_metadata(const std::string& msg) {
-    if (!group_) {
-      throw std::runtime_error(
-          "[ivf_flat_index@dump_metadata] Cannot dump metadata because there "
-          "is no group");
-    }
-    group_->metadata.dump(msg);
-  }
 };
 
 #endif  // TILEDB_ivf_flat_index_H

src/include/index/ivf_pq_group.h

@@ -51,8 +51,7 @@
          {"flat_ivf_centroids_array_name", "uncompressed_centroids"},
          {"pq_ivf_centroids_array_name", "partition_centroids"},
 
-         {"ivf_index_array_name", "partition_indexes"},
-
+         {"pq_ivf_indices_array_name", "partition_indexes"},
          {"pq_ivf_vectors_array_name", "shuffled_vectors"},
 
          {"distance_tables_array_name", "pq_symmetric_distance_tables"},
@@ -113,12 +112,6 @@ class ivf_pq_group : public base_index_group<index_type> {
           array_name_to_uri_[array_name] =
               array_name_to_uri(group_uri_, array_name);
         }
-      } else {
-        valid_array_keys_.insert(array_key);
-        valid_array_names_.insert(array_name);
-        array_key_to_array_name_[array_key] = array_name;
-        array_name_to_uri_[array_name] =
-            array_name_to_uri(group_uri_, array_name);
       }
       valid_array_keys_.insert(array_key);
       valid_array_names_.insert(array_name);
@@ -135,7 +128,8 @@ class ivf_pq_group : public base_index_group<index_type> {
         cached_ctx_, flat_ivf_centroids_uri(), 0, timestamp);
     tiledb::Array::delete_fragments(
         cached_ctx_, pq_ivf_centroids_uri(), 0, timestamp);
-    tiledb::Array::delete_fragments(cached_ctx_, ivf_index_uri(), 0, timestamp);
+    tiledb::Array::delete_fragments(
+        cached_ctx_, pq_ivf_indices_uri(), 0, timestamp);
     tiledb::Array::delete_fragments(cached_ctx_, this->ids_uri(), 0, timestamp);
     tiledb::Array::delete_fragments(
         cached_ctx_, pq_ivf_vectors_uri(), 0, timestamp);
@@ -181,8 +175,8 @@ class ivf_pq_group : public base_index_group<index_type> {
   [[nodiscard]] auto pq_ivf_centroids_uri() const {
     return this->array_key_to_uri("pq_ivf_centroids_array_name");
   }
-  [[nodiscard]] auto ivf_index_uri() const {
-    return this->array_key_to_uri("ivf_index_array_name");
+  [[nodiscard]] auto pq_ivf_indices_uri() const {
+    return this->array_key_to_uri("pq_ivf_indices_array_name");
   }
   [[nodiscard]] auto pq_ivf_vectors_uri() const {
     return this->array_key_to_uri("pq_ivf_vectors_array_name");
@@ -200,8 +194,8 @@ class ivf_pq_group : public base_index_group<index_type> {
   [[nodiscard]] auto pq_ivf_centroids_array_name() const {
     return this->array_key_to_array_name("pq_ivf_centroids_array_name");
   }
-  [[nodiscard]] auto ivf_index_array_name() const {
-    return this->array_key_to_array_name("ivf_index_array_name");
+  [[nodiscard]] auto pq_ivf_indices_array_name() const {
+    return this->array_key_to_array_name("pq_ivf_indices_array_name");
   }
   [[nodiscard]] auto pq_ivf_vectors_array_name() const {
     return this->array_key_to_array_name("pq_ivf_vectors_array_name");
@@ -331,12 +325,12 @@ class ivf_pq_group : public base_index_group<index_type> {
 
     create_empty_for_vector<typename index_type::indices_type>(
         cached_ctx_,
-        ivf_index_uri(),
+        pq_ivf_indices_uri(),
         default_domain,
         default_tile_extent,
         default_compression);
     tiledb_helpers::add_to_group(
-        write_group, ivf_index_uri(), ivf_index_array_name());
+        write_group, pq_ivf_indices_uri(), pq_ivf_indices_array_name());
 
     create_empty_for_vector<typename index_type::id_type>(
         cached_ctx_,

src/include/index/ivf_pq_index.h

@@ -873,9 +873,6 @@ class ivf_pq_index {
    * This will read the centroids, indices, partitioned_ids, and
    * and the complete set of partitioned_pq_vectors, along with metadata
    * from a group_uri.
-   *
-   * @param group_uri
-   * @return bool indicating success or failure of read
    */
   auto read_index_infinite() {
     if (!group_) {
@@ -899,7 +896,7 @@ class ivf_pq_index {
     partitioned_pq_vectors_ = std::make_unique<tdb_pq_storage_type>(
         group_->cached_ctx(),
         group_->pq_ivf_vectors_uri(),
-        group_->ivf_index_uri(),
+        group_->pq_ivf_indices_uri(),
         group_->get_num_partitions() + 1,
         group_->ids_uri(),
         infinite_parts,
@@ -951,7 +948,7 @@ class ivf_pq_index {
     partitioned_pq_vectors_ = std::make_unique<tdb_pq_storage_type>(
         group_->cached_ctx(),
         group_->pq_ivf_vectors_uri(),
-        group_->ivf_index_uri(),
+        group_->pq_ivf_indices_uri(),
         group_->get_num_partitions() + 1,
         group_->ids_uri(),
         active_partitions,
@@ -1073,7 +1070,7 @@ class ivf_pq_index {
     write_vector(
         ctx,
         partitioned_pq_vectors_->indices(),
-        write_group.ivf_index_uri(),
+        write_group.pq_ivf_indices_uri(),
         0,
         false,
         temporal_policy_);
@@ -1104,23 +1101,6 @@ class ivf_pq_index {
     return true;
   }
 
-  auto write_index_arrays(
-      const tiledb::Context& ctx,
-      const std::string& centroids_uri,
-      const std::string& parts_uri,
-      const std::string& ids_uri,
-      const std::string& indices_uri,
-      bool overwrite) const {
-    tiledb::VFS vfs(ctx);
-
-    write_matrix(ctx, flat_ivf_centroids_, centroids_uri, 0, true);
-    write_matrix(ctx, *partitioned_pq_vectors_, parts_uri, 0, true);
-    write_vector(ctx, partitioned_pq_vectors_->ids(), ids_uri, 0, true);
-    write_vector(ctx, partitioned_pq_vectors_->indices(), indices_uri, 0, true);
-
-    return true;
-  }
-
   static void clear_history(
       const tiledb::Context& ctx,
       const std::string& group_uri,
@@ -1786,23 +1766,14 @@ class ivf_pq_index {
     return indices;
   }
 
-  void dump_group(const std::string& msg) const {
+  void dump(const std::string& msg) const {
     if (!group_) {
       throw std::runtime_error(
-          "[ivf_flat_index@dump_group] Cannot dump group because there is no "
+          "[ivf_flat_index@dump] Cannot dump group because there is no "
           "group");
     }
     group_->dump(msg);
   }
-
-  void dump_metadata(const std::string& msg) const {
-    if (!group_) {
-      throw std::runtime_error(
-          "[ivf_flat_index@dump_metadata] Cannot dump metadata because there "
-          "is no group");
-    }
-    group_->metadata.dump(msg);
-  }
 };
 
 #endif  // TILEDB_PQ_INDEX_H