Refactoring

Author: Dmitry Yashunin

hnswlib/hnswalg.h

@@ -66,7 +66,7 @@ class HierarchicalNSW : public AlgorithmInterface<dist_t, filter_func_t> {
     mutable std::atomic<long> metric_distance_computations{0};
     mutable std::atomic<long> metric_hops{0};
 
-    bool replace_deleted = false;
+    bool replace_deleted_ = false;
 
     std::mutex deleted_elements_lock;
     std::unordered_set<tableint> deleted_elements;
@@ -82,7 +82,7 @@ class HierarchicalNSW : public AlgorithmInterface<dist_t, filter_func_t> {
         bool nmslib = false,
         size_t max_elements = 0,
         bool replace_deleted = false)
-        : replace_deleted(replace_deleted) {
+        : replace_deleted_(replace_deleted) {
         loadIndex(location, s, max_elements);
     }
 
@@ -97,7 +97,7 @@ class HierarchicalNSW : public AlgorithmInterface<dist_t, filter_func_t> {
         : link_list_locks_(max_elements),
             link_list_update_locks_(max_update_element_locks),
             element_levels_(max_elements),
-            replace_deleted(replace_deleted) {
+            replace_deleted_(replace_deleted) {
         max_elements_ = max_elements;
         num_deleted_ = 0;
         data_size_ = s->get_data_size();
@@ -703,7 +703,7 @@ class HierarchicalNSW : public AlgorithmInterface<dist_t, filter_func_t> {
         for (size_t i = 0; i < cur_element_count; i++) {
             if (isMarkedDeleted(i)) {
                 num_deleted_ += 1;
-                if (replace_deleted) deleted_elements.insert(i);
+                if (replace_deleted_) deleted_elements.insert(i);
             }
         }
 
@@ -757,7 +757,7 @@ class HierarchicalNSW : public AlgorithmInterface<dist_t, filter_func_t> {
             unsigned char *ll_cur = ((unsigned char *)get_linklist0(internalId))+2;
             *ll_cur |= DELETE_MARK;
             num_deleted_ += 1;
-            if (replace_deleted) deleted_elements.insert(internalId);
+            if (replace_deleted_) deleted_elements.insert(internalId);
         } else {
             throw std::runtime_error("The requested to delete element is already deleted");
         }
@@ -786,7 +786,7 @@ class HierarchicalNSW : public AlgorithmInterface<dist_t, filter_func_t> {
             unsigned char *ll_cur = ((unsigned char *)get_linklist0(internalId)) + 2;
             *ll_cur &= ~DELETE_MARK;
             num_deleted_ -= 1;
-            if (replace_deleted) deleted_elements.erase(internalId);
+            if (replace_deleted_) deleted_elements.erase(internalId);
         } else {
             throw std::runtime_error("The requested to undelete element is not deleted");
         }
@@ -817,9 +817,9 @@ class HierarchicalNSW : public AlgorithmInterface<dist_t, filter_func_t> {
     *
     * If deleted point was replaced returns its label, else returns label of added point
     */
-    labeltype insertPoint(const void* data_point, labeltype label) {
-        if (!replace_deleted) {
-            throw std::runtime_error("Don't use insertPoint when replacement of deleted elements is disabled");
+    labeltype addPointToVacantPlace(const void* data_point, labeltype label) {
+        if (!replace_deleted_) {
+            throw std::runtime_error("Can't use addPointToVacantPlace when replacement of deleted elements is disabled");
         }
 
         std::unique_lock <std::mutex> tmp_del_el_lock(deleted_elements_lock);

python_bindings/bindings.cpp

@@ -230,7 +230,7 @@ class Index {
     }
 
 
-    py::object insertItems_return_numpy(py::object input, py::object ids_ = py::none(), int num_threads = -1) {
+    py::object add_items_to_vacant_place_return_numpy(py::object input, py::object ids_ = py::none(), int num_threads = -1) {
         size_t rows, features;
         hnswlib::labeltype* data_numpy_l = NULL;
 
@@ -261,7 +261,7 @@ class Index {
                     normalize_vector(vector_data, norm_array.data());
                     vector_data = norm_array.data();
                 }
-                hnswlib::labeltype label = appr_alg->insertPoint((void*)vector_data, (size_t)id);
+                hnswlib::labeltype label = appr_alg->addPointToVacantPlace((void*)vector_data, (size_t)id);
                 data_numpy_l[start] = label;
                 start = 1;
                 ep_added = true;
@@ -271,7 +271,7 @@ class Index {
             if (normalize == false) {
                 ParallelFor(start, rows, num_threads, [&](size_t row, size_t threadId) {
                     size_t id = ids.size() ? ids.at(row) : (cur_l + row);
-                    hnswlib::labeltype label = appr_alg->insertPoint((void*)items.data(row), (size_t)id);
+                    hnswlib::labeltype label = appr_alg->addPointToVacantPlace((void*)items.data(row), (size_t)id);
                     data_numpy_l[row] = label;
                     });
             }
@@ -283,7 +283,7 @@ class Index {
                     normalize_vector((float*)items.data(row), (norm_array.data() + start_idx));
 
                     size_t id = ids.size() ? ids.at(row) : (cur_l + row);
-                    hnswlib::labeltype label = appr_alg->insertPoint((void*)(norm_array.data() + start_idx), (size_t)id);
+                    hnswlib::labeltype label = appr_alg->addPointToVacantPlace((void*)(norm_array.data() + start_idx), (size_t)id);
                     data_numpy_l[row] = label;
                     });
             }
@@ -472,7 +472,7 @@ class Index {
             "ef"_a = appr_alg->ef_,
             "has_deletions"_a = (bool)appr_alg->num_deleted_,
             "size_links_per_element"_a = appr_alg->size_links_per_element_,
-            "replace_deleted"_a = appr_alg->replace_deleted,
+            "replace_deleted"_a = appr_alg->replace_deleted_,
 
             "label_lookup_external"_a = py::array_t<hnswlib::labeltype>(
                 { appr_alg->label_lookup_.size() },  // shape
@@ -639,7 +639,7 @@ class Index {
         if (d.contains("replace_deleted")) {
             replace_deleted = d["replace_deleted"].cast<bool>();
         }
-        appr_alg->replace_deleted = replace_deleted;
+        appr_alg->replace_deleted_= replace_deleted;
 
         appr_alg->num_deleted_ = 0;
         bool has_deletions = d["has_deletions"].cast<bool>();
@@ -941,8 +941,8 @@ PYBIND11_PLUGIN(hnswlib) {
             py::arg("data"),
             py::arg("ids") = py::none(),
             py::arg("num_threads") = -1)
-        .def("insert_items",
-            &Index<float>::insertItems_return_numpy,
+        .def("add_items_to_vacant_place",
+            &Index<float>::add_items_to_vacant_place_return_numpy,
             py::arg("data"),
             py::arg("ids") = py::none(),
             py::arg("num_threads") = -1)

python_bindings/tests/bindings_test_replace.py

@@ -72,7 +72,7 @@ def testRandomSelf(self):
         print("Inserting batch 3 by replacing deleted elements")
         # Maximum number of elements is reached therefore we cannot add new items
         # but we can replace the deleted ones
-        labels_replaced = hnsw_index.insert_items(data3, labels3)
+        labels_replaced = hnsw_index.add_items_to_vacant_place(data3, labels3)
         labels2_deleted_list = [l[0] for l in labels2_deleted]
         labels_replaced_list = labels_replaced.tolist()
         labels2_deleted_list.sort()
@@ -114,7 +114,7 @@ def testRandomSelf(self):
 
         # Insert batch 4
         print("Inserting batch 4 by replacing deleted elements")
-        labels_replaced = hnsw_index.insert_items(data4, labels4)
+        labels_replaced = hnsw_index.add_items_to_vacant_place(data4, labels4)
 
         # Check recall
         print("Checking recall")
@@ -133,7 +133,7 @@ def testRandomSelf(self):
         del hnsw_index
         # Insert batch 3
         print("Inserting batch 3 by replacing deleted elements")
-        labels_replaced = hnsw_index_pckl.insert_items(data3, labels3)
+        labels_replaced = hnsw_index_pckl.add_items_to_vacant_place(data3, labels3)
 
         # Check recall
         print("Checking recall")