#37 Added out-of-core CPU parallel connected components

src/Makefile

@@ -13,7 +13,7 @@ cpu_seq_FLAGS=-Wno-unknown-pragmas -Wno-comment -Wconversion #-Weffc++
 cpu_CXX=$(cpu_seq_CXX)
 cpu_FLAGS=$(cpu_seq_FLAGS) -fopenmp
 
-LIBS=io geometry morphology diffusion histograms
+LIBS=io geometry morphology diffusion histograms connected_components
 TARGETS = $(foreach PLATFORM, $(PLATFORMS), $(foreach LIB, $(LIBS), $(CPP_FOLDER)/$(PLATFORM)/$(LIB)$(PYBIND_SUFFIX)))
 CLEANUP = $(TARGETS) $(foreach PLATFORM, $(PLATFORMS), $(CPP_FOLDER)/$(PLATFORM)/__pycache__)
 PIP_INSTALL = pip_generic

src/lib/cpp/cpu_seq/connected_components.cc

@@ -0,0 +1,10 @@
+#include "connected_components.hh"
+
+namespace cpu_seq {
+
+#pragma GCC diagnostic ignored "-Wunused-parameter"
+int64_t connected_components(const std::string &base_path, std::vector<int64_t> &n_labels, const idx3d &global_shape, const bool verbose) {
+    throw std::runtime_error("Not implemented");
+}
+
+}

src/lib/cpp/gpu/connected_components.cc

@@ -0,0 +1,10 @@
+#include "connected_components.hh"
+
+namespace gpu {
+
+#pragma GCC diagnostic ignored "-Wunused-parameter"
+int64_t connected_components(const std::string &base_path, std::vector<int64_t> &n_labels, const idx3d &global_shape, const bool verbose) {
+    throw std::runtime_error("Not implemented");
+}
+
+}

src/lib/cpp/include/connected_components.hh

@@ -0,0 +1,31 @@
+#ifndef connected_components_h
+#define connected_components_h
+
+#include "datatypes.hh"
+
+namespace NS {
+
+    // External Functions
+    int64_t connected_components(const std::string &base_path, std::vector<int64_t> &n_labels, const idx3d &global_shape, const bool verbose = false);
+
+    // Internal Functions
+    void apply_renaming(std::vector<int64_t> &img, const std::vector<int64_t> &to_rename);
+    void apply_renaming(int64_t *__restrict__ img, const int64_t n, const std::vector<int64_t> &to_rename);
+    void canonical_names_and_size(const std::string &path, int64_t *__restrict__ out, const int64_t n_labels, const idx3d &total_shape, const idx3d &global_shape);
+    std::tuple<mapping_t, mapping_t> get_mappings(const std::vector<int64_t> &a, const int64_t n_labels_a, const std::vector<int64_t> &b, const int64_t n_labels_b, const idx3d &global_shape);
+    std::vector<int64_t> get_sizes(const std::vector<int64_t> &img, const int64_t n_labels);
+    std::vector<std::vector<std::tuple<int64_t, int64_t>>> generate_adjacency_tree(const int64_t chunks);
+    std::vector<idx3d> merge_canonical_names(const std::vector<idx3d> &names_a, const std::vector<idx3d> &names_b);
+    std::vector<int64_t> merge_labels(mapping_t &mapping_a, const mapping_t &mapping_b, const std::vector<int64_t> &to_rename_b);
+    int64_t recount_labels(const mapping_t &mapping_a, mapping_t &mapping_b, std::vector<int64_t> &to_rename_a, std::vector<int64_t> &to_rename_b);
+    std::tuple<std::vector<int64_t>, std::vector<int64_t>, int64_t> relabel(const std::vector<int64_t> &a, const int64_t n_labels_a, const std::vector<int64_t> &b, const int64_t n_labels_b, const idx3d &global_shape, const bool verbose);
+    void rename_mapping(mapping_t &mapping_a, const std::vector<int64_t> &to_rename_other);
+
+    // Debugging functions
+    void print_canonical_names(const std::vector<idx3d> &names_a);
+    void print_mapping(const mapping_t &mapping_);
+    void print_rename(const std::vector<int64_t> &to_rename);
+
+}
+
+#endif // connected_components_h

src/lib/cpp/include/datatypes.hh

@@ -71,6 +71,16 @@ typedef struct {
     int64_t z, y, x;
 } shape_t;
 
+// Struct for a 3d index
+typedef struct {
+    int64_t z, y, x;
+} idx3d;
+
+// Struct for a 3d index range
+struct idx3drange {
+    int64_t z_start, z_end, y_start, y_end, x_start, x_end;
+};
+
 // Struct for accessing the raw bits of a 32-bit float
 typedef struct {
     union {
@@ -81,4 +91,6 @@ typedef struct {
 
 #define GB_VOXEL ((1024 / sizeof(voxel_type)) * 1024 * 1024)
 
+typedef std::vector<std::unordered_set<int64_t>> mapping_t;
+
 #endif

src/pybind/connected_components-pybind.cc

@@ -0,0 +1,22 @@
+#include "connected_components.cc"
+
+namespace python_api {
+
+    void connected_components(const std::string &base_path, np_array<int64_t> &py_n_labels, const std::tuple<int64_t, int64_t, int64_t> &py_global_shape, const bool verbose = false) {
+        auto n_labels_info = py_n_labels.request();
+        int64_t *n_labels = static_cast<int64_t*>(n_labels_info.ptr);
+
+        std::vector<int64_t> n_labels_vec(n_labels, n_labels + n_labels_info.size);
+
+        const idx3d global_shape = {std::get<0>(py_global_shape), std::get<1>(py_global_shape), std::get<2>(py_global_shape)};
+
+        NS::connected_components(base_path, n_labels_vec, global_shape, verbose);
+    }
+
+}
+
+PYBIND11_MODULE(connected_components, m) {
+    m.doc() = "Connected Components"; // optional module docstring
+
+    m.def("connected_components", &python_api::connected_components, py::arg("base_path"), py::arg("np_n_labels"), py::arg("global_shape"), py::arg("verbose") = false);
+}