Reverting some more auto declarations and adding ctest support

Author: michaelbautin

CMakeLists.txt

@@ -9,19 +9,25 @@ include(FetchContent)
 
 # These example/test targets catch exceptions, so exceptions should always be
 # enabled building these files even if they are disabled in other targets.
+# We check that each target included in this list is a real target.
 set(HNSWLIB_TARGETS_REQUIRING_EXCEPTIONS
     example_mt_filter
     example_mt_replace_deleted
     example_mt_search
     multiThread_replace_test
     test_updates)
 
-
+# Adds an example or test target. The target name parameter is followed by
+# the list of source files. Automatically links with the hnswlib library.
+# Also decides whether to enable exceptions when building the target. 
+# If HNSWLIB_ENABLE_EXCEPTIONS is ON, exceptions are always enabled.
+# If HNSWLIB_ENABLE_EXCEPTIONS is OFF, exceptions are only enabled for the
+# specific targets listed in HNSWLIB_TARGETS_REQUIRING_EXCEPTIONS.
 function(add_example_or_test TARGET_NAME ...)
     add_executable(${ARGV})
-    target_link_libraries("${TARGET_NAME}" ${common_link_libraries})
-    list(FIND HNSWLIB_TARGETS_REQUIRING_EXCEPTIONS "${TARGET_NAME}" _index)
-    if(_index GREATER -1)
+    target_link_libraries(${TARGET_NAME} hnswlib)
+    list(FIND HNSWLIB_TARGETS_REQUIRING_EXCEPTIONS "${TARGET_NAME}" found_at_index)
+    if(found_at_index GREATER -1)
         if(NOT HNSWLIB_ENABLE_EXCEPTIONS)
             message("Enabling exceptions for target ${TARGET_NAME} as a special case")
         endif()
@@ -34,9 +40,20 @@ function(add_example_or_test TARGET_NAME ...)
     else()
         target_compile_options("${TARGET_NAME}" PUBLIC -fno-exceptions)
     endif()
+    if(NOT ${TARGET_NAME} STREQUAL "main")
+        add_test(
+            NAME ${TARGET_NAME}
+            COMMAND ${TARGET_NAME}
+        )
+    endif()
 endfunction()
 
 option(HNSWLIB_ENABLE_EXCEPTIONS "Whether to enable exceptions in hnswlib" ON)
+if(HNSWLIB_ENABLE_EXCEPTIONS)
+    message("Exceptions are enabled using HNSWLIB_ENABLE_EXCEPTIONS=ON (default)")
+else()
+    message("Exceptions are disabled using HNSWLIB_ENABLE_EXCEPTIONS=OFF")
+endif()
 
 set(CMAKE_CXX_STANDARD 11)
 
@@ -62,10 +79,12 @@ install(EXPORT hnswlibTargets
 # Examples and tests
 if(CMAKE_PROJECT_NAME STREQUAL PROJECT_NAME)
     option(HNSWLIB_EXAMPLES "Build examples and tests." ON)
+    message("Building examples and tests")
 else()
     option(HNSWLIB_EXAMPLES "Build examples and tests." OFF)
 endif()
 if(HNSWLIB_EXAMPLES)
+    enable_testing()
     if (CMAKE_CXX_COMPILER_ID MATCHES "Clang")
       SET( CMAKE_CXX_FLAGS  "-Ofast -DHAVE_CXX0X -openmp -fpic -ftree-vectorize" )
       check_cxx_compiler_flag("-march=native" COMPILER_SUPPORT_NATIVE_FLAG)
@@ -85,9 +104,6 @@ if(HNSWLIB_EXAMPLES)
       SET( CMAKE_CXX_FLAGS  "/O2 -DHAVE_CXX0X /W1 /openmp /EHsc" )
     endif()
 
-    set(common_link_libraries
-        hnswlib)
-
     set(EXAMPLE_NAMES
         example_epsilon_search
         example_filter
@@ -103,7 +119,7 @@ if(HNSWLIB_EXAMPLES)
     endforeach()
 
     set(TEST_NAMES
-        multivector_search_test 
+        multivector_search_test
         epsilon_search_test
         searchKnnCloserFirst_test
         searchKnnWithFilter_test
@@ -116,5 +132,15 @@ if(HNSWLIB_EXAMPLES)
     # This test deviates from the above pattern of naming test executables.
     add_example_or_test(test_updates tests/cpp/updates_test.cpp)
 
+    # For historical reasons, the "main" program links with sift_1b.cpp.
     add_example_or_test(main tests/cpp/main.cpp tests/cpp/sift_1b.cpp)
+
+    foreach(target_name IN LISTS HNSWLIB_TARGETS_REQUIRING_EXCEPTIONS)
+        if(NOT TARGET ${target_name})
+            message(FATAL_ERROR
+                    "Target '${target_name}' included in "
+                    "HNSWLIB_TARGETS_REQUIRING_EXCEPTIONS does not exist. "
+                    "Please check if this is a typo.")
+        endif()
+    endforeach()
 endif()

examples/cpp/example_filter.cpp

@@ -45,8 +45,7 @@ int main() {
     // Query the elements for themselves with filter and check returned labels
     int k = 10;
     for (int i = 0; i < max_elements; i++) {
-        std::vector<std::pair<float, hnswlib::labeltype>> result =
-            alg_hnsw->searchKnnCloserFirst(data + i * dim, k, &pickIdsDivisibleByTwo);
+        std::vector<std::pair<float, hnswlib::labeltype>> result = alg_hnsw->searchKnnCloserFirst(data + i * dim, k, &pickIdsDivisibleByTwo);
         for (auto item: result) {
             if (item.second % 2 == 1) std::cout << "Error: found odd label\n";
         }

examples/cpp/example_mt_filter.cpp

@@ -106,7 +106,7 @@ int main() {
     int k = 10;
     std::vector<hnswlib::labeltype> neighbors(max_elements * k);
     ParallelFor(0, max_elements, num_threads, [&](size_t row, size_t threadId) {
-        auto result = alg_hnsw->searchKnn(data + dim * row, k, &pickIdsDivisibleByTwo);
+        std::priority_queue<std::pair<float, hnswlib::labeltype>> result = alg_hnsw->searchKnn(data + dim * row, k, &pickIdsDivisibleByTwo);
         for (int i = 0; i < k; i++) {
             hnswlib::labeltype label = result.top().second;
             result.pop();

examples/cpp/example_mt_search.cpp

@@ -89,7 +89,7 @@ int main() {
     // Query the elements for themselves and measure recall
     std::vector<hnswlib::labeltype> neighbors(max_elements);
     ParallelFor(0, max_elements, num_threads, [&](size_t row, size_t threadId) {
-        auto result = alg_hnsw->searchKnn(data + dim * row, 1);
+        std::priority_queue<std::pair<float, hnswlib::labeltype>> result = alg_hnsw->searchKnn(data + dim * row, 1);
         hnswlib::labeltype label = result.top().second;
         neighbors[row] = label;
     });

examples/cpp/example_search.cpp

@@ -29,7 +29,7 @@ int main() {
     // Query the elements for themselves and measure recall
     float correct = 0;
     for (int i = 0; i < max_elements; i++) {
-        auto result = alg_hnsw->searchKnn(data + i * dim, 1);
+        std::priority_queue<std::pair<float, hnswlib::labeltype>> result = alg_hnsw->searchKnn(data + i * dim, 1);
         hnswlib::labeltype label = result.top().second;
         if (label == i) correct++;
     }
@@ -45,7 +45,7 @@ int main() {
     alg_hnsw = new hnswlib::HierarchicalNSW<float>(&space, hnsw_path);
     correct = 0;
     for (int i = 0; i < max_elements; i++) {
-        auto result = alg_hnsw->searchKnn(data + i * dim, 1);
+        std::priority_queue<std::pair<float, hnswlib::labeltype>> result = alg_hnsw->searchKnn(data + i * dim, 1);
         hnswlib::labeltype label = result.top().second;
         if (label == i) correct++;
     }

hnswlib/bruteforce.h

@@ -74,7 +74,7 @@ class BruteforceSearch : public AlgorithmInterface<dist_t> {
                 idx = search->second;
             } else {
                 if (cur_element_count >= maxelements_) {
-                    HNSWLIB_THROW_RUNTIME_ERROR("The number of elements exceeds the specified limit\n");
+                    HNSWLIB_THROW_RUNTIME_ERROR("The number of elements exceeds the specified limit");
                 }
                 idx = cur_element_count;
                 dict_external_to_internal[label] = idx;

hnswlib/hnswalg.h

@@ -717,6 +717,10 @@ class HierarchicalNSW : public AlgorithmInterface<dist_t> {
 
 
     void loadIndex(const std::string &location, SpaceInterface<dist_t> *s, size_t max_elements_i = 0) {
+        Status status = loadIndexNoExceptions(location, s, max_elements_i);
+        if (!status.ok()) {
+            HNSWLIB_THROW_RUNTIME_ERROR(status.message());
+        }
     }
 
     Status loadIndexNoExceptions(const std::string &location, SpaceInterface<dist_t> *s, size_t max_elements_i = 0) {

hnswlib/hnswlib.h

@@ -124,7 +124,7 @@ static bool AVX512Capable() {
 
 #if defined(__EXCEPTIONS)
 #define HNSWLIB_THROW_RUNTIME_ERROR(message) throw std::runtime_error(message)
-#else 
+#else
 #define HNSWLIB_THROW_RUNTIME_ERROR(message) do { \
     fprintf(stderr, \
         "FATAL: hnswlib compiled without exception support. " \
@@ -142,7 +142,7 @@ class [[nodiscard]] Status {
 public:
     Status() : message_(nullptr) {}
 
-    // Constructor with an error message (can't be nullptr).
+    // Constructor with an error message (nullptr is interpreted as OK status).
     Status(const char* message) : message_(message) {}
 
     // Returns true if the status is OK

tests/cpp/multivector_search_test.cpp

@@ -14,11 +14,11 @@ int main() {
     int num_queries = 100;
     int num_docs = 10;          // Number of documents to search
     int ef_collection = 15;     // Number of candidate documents during the search
-                                // Controlls the recall: higher ef leads to better accuracy, but slower search
+                                // Controls the recall: higher ef leads to better accuracy, but slower search
     docidtype min_doc_id = 0;
     docidtype max_doc_id = 49;
 
-    // Initing index
+    // Initializing index
     hnswlib::MultiVectorL2Space<docidtype> space(dim);
     hnswlib::BruteforceSearch<dist_t>* alg_brute = new hnswlib::BruteforceSearch<dist_t>(&space, max_elements);
     hnswlib::HierarchicalNSW<dist_t>* alg_hnsw = new hnswlib::HierarchicalNSW<dist_t>(&space, max_elements, M, ef_construction);
@@ -67,7 +67,7 @@ int main() {
             *(float*)vec_data = value;
         }
         hnswlib::MultiVectorSearchStopCondition<docidtype, dist_t> stop_condition(space, num_docs, ef_collection);
-        auto hnsw_results =
+        std::vector<std::pair<dist_t, hnswlib::labeltype>> hnsw_results =
             alg_hnsw->searchStopConditionClosest(query_data, stop_condition);
 
         // check number of found documents
@@ -82,7 +82,7 @@ int main() {
         assert(hnsw_docs.size() == num_docs);
 
         // Check overall recall
-        auto gt_results =
+        std::vector<std::pair<dist_t, hnswlib::labeltype>> gt_results =
             alg_brute->searchKnnCloserFirst(query_data, max_elements);
         std::unordered_set<docidtype> gt_docs;
         for (int i = 0; i < gt_results.size(); i++) {
@@ -100,14 +100,16 @@ int main() {
         delete[] query_data;
     }
     float recall = correct / total_num_elements;
-    std::cout << "random elements search recall : " << recall << "\n";
+    std::cout << "random elements search recall : " << recall << std::endl;
+    std::cout << "correct=" << correct << std::endl;
+    std::cout << "total_num_elements=" << total_num_elements << std::endl;
     assert(recall > 0.95);
 
     // Query the elements for themselves and measure recall
     correct = 0;
     for (int i = 0; i < max_elements; i++) {
         hnswlib::MultiVectorSearchStopCondition<docidtype, dist_t> stop_condition(space, num_docs, ef_collection);
-        auto result =
+        std::vector<std::pair<float, hnswlib::labeltype>> result =
             alg_hnsw->searchStopConditionClosest(data + i * data_point_size, stop_condition);
         hnswlib::labeltype label = -1;
         if (!result.empty()) {

tests/cpp/sift_1b.cpp

@@ -179,7 +179,7 @@ test_approx(
     // uncomment to test in parallel mode:
     //#pragma omp parallel for
     for (int i = 0; i < qsize; i++) {
-        auto result = appr_alg.searchKnn(massQ + vecdim * i, k);
+        std::priority_queue<std::pair<int, labeltype >> result = appr_alg.searchKnn(massQ + vecdim * i, k);
         std::priority_queue<std::pair<int, labeltype >> gt(answers[i]);
         unordered_set<labeltype> g;
         total += gt.size();