Implement Louvain as a query module

Reviewers: teon.banek, dsantl

Reviewed By: teon.banek, dsantl

Subscribers: pullbot

Differential Revision: https://phabricator.memgraph.io/D2574

query_modules/louvain/CMakeLists.txt

@@ -7,7 +7,6 @@ include_directories(src)
 
 add_library(louvain-core STATIC ${SOURCES})
 set_target_properties(louvain-core PROPERTIES POSITION_INDEPENDENT_CODE ON)
-target_link_libraries(louvain-core PUBLIC Threads::Threads glog gflags)
 
 add_executable(louvain-main ${MAIN})
 target_link_libraries(louvain-main louvain-core)
@@ -16,6 +15,7 @@ enable_testing()
 add_subdirectory(test)
 
 add_library(louvain SHARED ${MODULE})
+target_link_libraries(louvain louvain-core)
 target_include_directories(louvain PRIVATE ${CMAKE_SOURCE_DIR}/include)
 
 if (NOT MG_COMMUNITY)

query_modules/louvain/src/algorithms/louvain.cpp

@@ -5,8 +5,6 @@
 #include <random>
 #include <unordered_map>
 
-#include <glog/logging.h>
-
 namespace {
 
 void OptimizeLocally(comdata::Graph *graph) {

query_modules/louvain/src/data_structures/graph.cpp

@@ -1,12 +1,11 @@
 #include "data_structures/graph.hpp"
 
+#include <exception>
 #include <numeric>
 #include <unordered_map>
 #include <unordered_set>
 #include <vector>
 
-#include <glog/logging.h>
-
 namespace comdata {
 
 Graph::Graph(uint32_t n_nodes) : n_nodes_(n_nodes), total_w_(0) {
@@ -22,15 +21,9 @@ uint32_t Graph::Size() const {
   return n_nodes_;
 }
 
-uint32_t Graph::Community(uint32_t node) const {
-  CHECK(node < n_nodes_) << "Node index out of range";
-  return community_[node];
-}
+uint32_t Graph::Community(uint32_t node) const { return community_.at(node); }
 
-void Graph::SetCommunity(uint32_t node, uint32_t c) {
-  CHECK(node < n_nodes_) << "Node index out of range";
-  community_[node] = c;
-}
+void Graph::SetCommunity(uint32_t node, uint32_t c) { community_.at(node) = c; }
 
 uint32_t Graph::NormalizeCommunities() {
   std::set<uint32_t> c_id(community_.begin(), community_.end());
@@ -46,10 +39,11 @@ uint32_t Graph::NormalizeCommunities() {
 }
 
 void Graph::AddEdge(uint32_t node1, uint32_t node2, double weight) {
-  CHECK(node1 < n_nodes_) << "Node index out of range";
-  CHECK(node2 < n_nodes_) << "Node index out of range";
-  CHECK(weight > 0) << "Weights must be positive";
-  CHECK(edges_.find({node1, node2}) == edges_.end()) << "Edge already exists";
+  if (node1 >= n_nodes_ || node2 >= n_nodes_)
+    throw std::out_of_range("Node index out of range");
+  if (weight <= 0) throw std::out_of_range("Weights must be positive");
+  if (edges_.find({node1, node2}) != edges_.end())
+    throw std::invalid_argument("Edge already exists");
 
   edges_.emplace(node1, node2);
   edges_.emplace(node2, node1);
@@ -66,13 +60,11 @@ void Graph::AddEdge(uint32_t node1, uint32_t node2, double weight) {
 }
 
 uint32_t Graph::Degree(uint32_t node) const {
-  CHECK(node < n_nodes_) << "Node index out of range";
-  return static_cast<uint32_t>(adj_list_[node].size());
+  return static_cast<uint32_t>(adj_list_.at(node).size());
 }
 
 double Graph::IncidentWeight(uint32_t node) const {
-  CHECK(node < n_nodes_) << "Node index out of range";
-  return inc_w_[node];
+  return inc_w_.at(node);
 }
 
 double Graph::TotalWeight() const {
@@ -98,9 +90,8 @@ double Graph::Modularity() const {
   return ret;
 }
 
-const std::vector<Neighbour>& Graph::Neighbours(uint32_t node) const {
-  CHECK(node < n_nodes_) << "Node index out of range";
-  return adj_list_[node];
+const std::vector<Neighbour> &Graph::Neighbours(uint32_t node) const {
+  return adj_list_.at(node);
 }
 
 } // namespace comdata

query_modules/louvain/src/data_structures/graph.hpp

@@ -40,16 +40,24 @@ public:
   /// @param node1 index of an incident node.
   /// @param node2 index of an incident node.
   /// @param weight real value which represents the weight of the edge.
+  ///
+  /// @throw std::out_of_range
+  /// @throw std::invalid_argument
   void AddEdge(uint32_t node1, uint32_t node2, double weight);
 
   /// @param node index of node.
+  ///
   /// @return community where the node belongs to.
+  ///
+  /// @throw std::out_of_range
   uint32_t Community(uint32_t node) const;
 
   /// Adds a given node to a given community.
   ///
   /// @param node index of node.
   /// @param c community where the given node should go in.
+  ///
+  /// @throw std::out_of_range
   void SetCommunity(uint32_t node, uint32_t c);
 
   /// Normalizes the values of communities. More precisely, after invoking this
@@ -65,14 +73,20 @@ public:
   /// contribute a single edge to the degree.
   ///
   /// @param node index of node.
+  ///
   /// @return degree of given node.
+  ///
+  /// @throw std::out_of_range
   uint32_t Degree(uint32_t node) const;
 
   /// Returns the total weight of incident edges to a given node. Weight
   /// of a self loop contributes once to the total sum.
   ///
   /// @param node index of node.
+  ///
   /// @return total incident weight of a given node.
+  ///
+  /// @throw std::out_of_range
   double IncidentWeight(uint32_t node) const;
 
   /// @return total weight of all edges in a graph.
@@ -91,7 +105,10 @@ public:
   /// Returns nodes adjacent to a given node.
   ///
   /// @param node index of node.
+  ///
   /// @return list of neighbouring nodes.
+  ///
+  /// @throw std::out_of_range
   const std::vector<Neighbour>& Neighbours(uint32_t node) const;
 
 private:

query_modules/louvain/src/louvain_module.cpp

@@ -1,19 +1,128 @@
 #include "mg_procedure.h"
 
+#include <exception>
+#include <unordered_map>
+
+#include "algorithms/algorithms.hpp"
+#include "data_structures/graph.hpp"
+
 static void communities(const mgp_list *args, const mgp_graph *graph,
                         mgp_result *result, mgp_memory *memory) {
-  mgp_result_record *record = mgp_result_new_record(result);
-  mgp_value *hello_world_value =
-      mgp_value_make_string("Louvain, fuck yeah!", memory);
-  mgp_result_record_insert(record, "result", hello_world_value);
-  mgp_value_destroy(hello_world_value);
+  mgp_vertices_iterator *vertices_iterator =
+      mgp_graph_iter_vertices(graph, memory);
+  if (vertices_iterator == nullptr) {
+    mgp_result_set_error_msg(result, "Not enough memory!");
+    return;
+  }
+
+  // Normalize vertex ids
+  std::unordered_map<int64_t, uint32_t> mem_to_louv_id;
+
+  uint32_t louv_id = 0;
+  for (const mgp_vertex *vertex = mgp_vertices_iterator_get(vertices_iterator);
+       vertex != nullptr; vertex = mgp_vertices_iterator_next(vertices_iterator)) {
+    mgp_vertex_id mem_id = mgp_vertex_get_id(vertex);
+    mem_to_louv_id[mem_id.as_int] = louv_id;
+    ++louv_id;
+  }
+
+  mgp_vertices_iterator_destroy(vertices_iterator);
+
+  // Extract the graph structure
+  // TODO(ipaljak): consider filtering nodes and edges by labels.
+  comdata::Graph louvain_graph(louv_id);
+  for (const auto &p : mem_to_louv_id) {
+    mgp_vertex *vertex = mgp_graph_get_vertex_by_id(graph, {p.first}, memory);
+    if (!vertex) {
+      mgp_result_set_error_msg(result, "Not enough memory!");
+      return;
+    }
+
+    // iterate over inbound edges. This is enough because we will eventually
+    // iterate over outbound edges in another direction.
+    mgp_edges_iterator *edges_iterator =
+        mgp_vertex_iter_in_edges(vertex, memory);
+    if (edges_iterator == nullptr) {
+      mgp_vertex_destroy(vertex);
+      mgp_result_set_error_msg(result, "Not enough memory!");
+      return;
+    }
+
+    for (const mgp_edge *edge = mgp_edges_iterator_get(edges_iterator);
+         edge != nullptr; edge = mgp_edges_iterator_next(edges_iterator)) {
+      const mgp_vertex *next_vertex = mgp_edge_get_from(edge);
+      mgp_vertex_id next_mem_id = mgp_vertex_get_id(next_vertex);
+      uint32_t next_louv_id = mem_to_louv_id[next_mem_id.as_int];
+
+      // TODO(ipaljak): retrieve edge weight (default to 1)
+      double weight = 1;
+
+      try {
+        louvain_graph.AddEdge(p.second, next_louv_id, weight);
+      } catch (const std::exception &e) {
+        mgp_vertex_destroy(vertex);
+        mgp_edges_iterator_destroy(edges_iterator);
+        mgp_result_set_error_msg(result, e.what());
+        return;
+      }
+    }
+
+    mgp_vertex_destroy(vertex);
+    mgp_edges_iterator_destroy(edges_iterator);
+  }
+
+  try {
+    algorithms::Louvain(&louvain_graph);
+  } catch (const std::exception &e) {
+    const auto msg = std::string("[Internal error] ") + e.what();
+    mgp_result_set_error_msg(result, msg.c_str());
+    return;
+  }
+
+  // Return node ids and their corresponding communities.
+  for (const auto &p : mem_to_louv_id) {
+    mgp_result_record *record = mgp_result_new_record(result);
+    if (record == nullptr) {
+      mgp_result_set_error_msg(result, "Not enough memory!");
+      return;
+    }
+
+    mgp_value *mem_id_value = mgp_value_make_int(p.first, memory);
+    if (mem_id_value == nullptr) {
+      mgp_result_set_error_msg(result, "Not enough memory!");
+      return;
+    }
+
+    mgp_value *com_value =
+        mgp_value_make_int(louvain_graph.Community(p.second), memory);
+    if (com_value == nullptr) {
+      mgp_value_destroy(mem_id_value);
+      mgp_result_set_error_msg(result, "Not enough memory!");
+      return;
+    }
+
+    int mem_id_inserted =
+        mgp_result_record_insert(record, "id", mem_id_value);
+    int com_inserted =
+        mgp_result_record_insert(record, "community", com_value);
+
+    mgp_value_destroy(mem_id_value);
+    mgp_value_destroy(com_value);
+
+    if (!mem_id_inserted || !com_inserted) {
+      mgp_result_set_error_msg(result, "Not enough memory!");
+      return;
+    }
+  }
 }
 
 extern "C" int mgp_init_module(struct mgp_module *module,
                                struct mgp_memory *memory) {
   struct mgp_proc *proc =
       mgp_module_add_read_procedure(module, "communities", communities);
-  if (!mgp_proc_add_result(proc, "result", mgp_type_string())) return 1;
+  if (!proc) return 1;
+  if (!mgp_proc_add_result(proc, "id", mgp_type_int())) return 1;
+  if (!mgp_proc_add_result(proc, "community", mgp_type_int())) return 1;
   return 0;
 }
 

query_modules/louvain/test/unit/graph.cpp

@@ -70,12 +70,12 @@ TEST(Graph, Communities) {
   for (int i = 0; i < 100; ++i) ASSERT_EQ(graph.Community(i), i % 5);
 
   // Try to set communities on non-existing nodes
-  ASSERT_DEATH({ graph.SetCommunity(100, 2); }, "");
-  ASSERT_DEATH({ graph.SetCommunity(150, 0); }, "");
+  EXPECT_THROW({ graph.SetCommunity(100, 2); }, std::out_of_range);
+  EXPECT_THROW({ graph.SetCommunity(150, 0); }, std::out_of_range);
 
   // Try to get a the community of a non-existing node
-  ASSERT_DEATH({ graph.Community(100); }, "");
-  ASSERT_DEATH({ graph.Community(150); }, "");
+  EXPECT_THROW({ graph.Community(100); }, std::out_of_range);
+  EXPECT_THROW({ graph.Community(150); }, std::out_of_range);
 }
 
 TEST(Graph, CommunityNormalization) {
@@ -108,15 +108,15 @@ TEST(Graph, AddEdge) {
   comdata::Graph graph = GenRandomUnweightedGraph(5, 0);
 
   // Node out of bounds.
-  ASSERT_DEATH({ graph.AddEdge(1, 5, 7); }, "");
+  EXPECT_THROW({ graph.AddEdge(1, 5, 7); }, std::out_of_range);
 
   // Repeated edge
   graph.AddEdge(1, 2, 1);
-  ASSERT_DEATH({ graph.AddEdge(1, 2, 7); }, "");
+  EXPECT_THROW({ graph.AddEdge(1, 2, 7); }, std::invalid_argument);
 
   // Non-positive edge weight
-  ASSERT_DEATH({ graph.AddEdge(2, 3, -7); }, "");
-  ASSERT_DEATH({ graph.AddEdge(3, 4, 0); }, "");
+  EXPECT_THROW({ graph.AddEdge(2, 3, -7); }, std::out_of_range);
+  EXPECT_THROW({ graph.AddEdge(3, 4, 0); }, std::out_of_range);
 }
 
 TEST(Graph, Degrees) {
@@ -183,8 +183,8 @@ TEST(Graph, Degrees) {
   ASSERT_TRUE(DegreeCheck(graph, deg));
 
   // Try to get degree of non-existing nodes
-  ASSERT_DEATH({ graph.Degree(5); }, "");
-  ASSERT_DEATH({ graph.Degree(100); }, "");
+  EXPECT_THROW({ graph.Degree(5); }, std::out_of_range);
+  EXPECT_THROW({ graph.Degree(100); }, std::out_of_range);
 }
 
 TEST(Graph, Weights) {
@@ -256,8 +256,8 @@ TEST(Graph, Weights) {
   EXPECT_NEAR(graph.TotalWeight(), 5.5, 1e-6);
 
   // Try to get incident weight of non-existing node
-  ASSERT_DEATH({ graph.IncidentWeight(5); }, "");
-  ASSERT_DEATH({ graph.IncidentWeight(100); }, "");
+  EXPECT_THROW({ graph.IncidentWeight(5); }, std::out_of_range);
+  EXPECT_THROW({ graph.IncidentWeight(100); }, std::out_of_range);
 }
 
 TEST(Graph, Modularity) {