|
| 1 | +#pragma once |
| 2 | + |
| 3 | +#include <algorithm> |
| 4 | +#include <cstddef> |
| 5 | +#include <cstdint> |
| 6 | +#include <stdexcept> |
| 7 | +#include <string> |
| 8 | +#include <unordered_map> |
| 9 | +#include <unordered_set> |
| 10 | +#include <utility> |
| 11 | +#include <vector> |
| 12 | + |
| 13 | +namespace bioimage_cpp::graph { |
| 14 | + |
| 15 | +struct Adjacency { |
| 16 | + std::uint64_t node; |
| 17 | + std::uint64_t edge; |
| 18 | +}; |
| 19 | + |
| 20 | +class UndirectedGraph { |
| 21 | +public: |
| 22 | + using NodeId = std::uint64_t; |
| 23 | + using EdgeId = std::uint64_t; |
| 24 | + using Edge = std::pair<NodeId, NodeId>; |
| 25 | + using AdjacencyList = std::vector<Adjacency>; |
| 26 | + |
| 27 | + explicit UndirectedGraph( |
| 28 | + const NodeId number_of_nodes = 0, |
| 29 | + const EdgeId reserve_number_of_edges = 0 |
| 30 | + ) |
| 31 | + : number_of_nodes_(number_of_nodes), |
| 32 | + adjacency_(static_cast<std::size_t>(number_of_nodes)) { |
| 33 | + edges_.reserve(static_cast<std::size_t>(reserve_number_of_edges)); |
| 34 | + } |
| 35 | + |
| 36 | + virtual ~UndirectedGraph() = default; |
| 37 | + |
| 38 | + void assign( |
| 39 | + const NodeId number_of_nodes = 0, |
| 40 | + const EdgeId reserve_number_of_edges = 0 |
| 41 | + ) { |
| 42 | + number_of_nodes_ = number_of_nodes; |
| 43 | + edges_.clear(); |
| 44 | + edge_lookup_.clear(); |
| 45 | + adjacency_.clear(); |
| 46 | + adjacency_.resize(static_cast<std::size_t>(number_of_nodes)); |
| 47 | + edges_.reserve(static_cast<std::size_t>(reserve_number_of_edges)); |
| 48 | + } |
| 49 | + |
| 50 | + [[nodiscard]] NodeId number_of_nodes() const { |
| 51 | + return number_of_nodes_; |
| 52 | + } |
| 53 | + |
| 54 | + [[nodiscard]] EdgeId number_of_edges() const { |
| 55 | + return static_cast<EdgeId>(edges_.size()); |
| 56 | + } |
| 57 | + |
| 58 | + [[nodiscard]] NodeId node_id_upper_bound() const { |
| 59 | + return number_of_nodes_ == 0 ? 0 : number_of_nodes_ - 1; |
| 60 | + } |
| 61 | + |
| 62 | + [[nodiscard]] EdgeId edge_id_upper_bound() const { |
| 63 | + return edges_.empty() ? 0 : static_cast<EdgeId>(edges_.size() - 1); |
| 64 | + } |
| 65 | + |
| 66 | + [[nodiscard]] std::vector<NodeId> nodes() const { |
| 67 | + std::vector<NodeId> result(static_cast<std::size_t>(number_of_nodes_)); |
| 68 | + for (NodeId node = 0; node < number_of_nodes_; ++node) { |
| 69 | + result[static_cast<std::size_t>(node)] = node; |
| 70 | + } |
| 71 | + return result; |
| 72 | + } |
| 73 | + |
| 74 | + [[nodiscard]] std::vector<EdgeId> edges() const { |
| 75 | + std::vector<EdgeId> result(edges_.size()); |
| 76 | + for (EdgeId edge = 0; edge < edges_.size(); ++edge) { |
| 77 | + result[static_cast<std::size_t>(edge)] = edge; |
| 78 | + } |
| 79 | + return result; |
| 80 | + } |
| 81 | + |
| 82 | + [[nodiscard]] Edge uv(const EdgeId edge) const { |
| 83 | + validate_edge(edge); |
| 84 | + return edges_[static_cast<std::size_t>(edge)]; |
| 85 | + } |
| 86 | + |
| 87 | + [[nodiscard]] NodeId u(const EdgeId edge) const { |
| 88 | + return uv(edge).first; |
| 89 | + } |
| 90 | + |
| 91 | + [[nodiscard]] NodeId v(const EdgeId edge) const { |
| 92 | + return uv(edge).second; |
| 93 | + } |
| 94 | + |
| 95 | + [[nodiscard]] const std::vector<Edge> &uv_ids() const { |
| 96 | + return edges_; |
| 97 | + } |
| 98 | + |
| 99 | + [[nodiscard]] const AdjacencyList &node_adjacency(const NodeId node) const { |
| 100 | + validate_node(node); |
| 101 | + return adjacency_[static_cast<std::size_t>(node)]; |
| 102 | + } |
| 103 | + |
| 104 | + EdgeId insert_edge(const NodeId u, const NodeId v) { |
| 105 | + validate_node(u); |
| 106 | + validate_node(v); |
| 107 | + if (u == v) { |
| 108 | + throw std::invalid_argument("self edges are not supported"); |
| 109 | + } |
| 110 | + |
| 111 | + const auto key = edge_key(u, v); |
| 112 | + const auto found = edge_lookup_.find(key); |
| 113 | + if (found != edge_lookup_.end()) { |
| 114 | + return found->second; |
| 115 | + } |
| 116 | + return insert_new_edge(key.first, key.second); |
| 117 | + } |
| 118 | + |
| 119 | + [[nodiscard]] std::int64_t find_edge(const NodeId u, const NodeId v) const { |
| 120 | + validate_node(u); |
| 121 | + validate_node(v); |
| 122 | + if (u == v) { |
| 123 | + return -1; |
| 124 | + } |
| 125 | + |
| 126 | + const auto key = edge_key(u, v); |
| 127 | + const auto found = edge_lookup_.find(key); |
| 128 | + if (found == edge_lookup_.end()) { |
| 129 | + return -1; |
| 130 | + } |
| 131 | + return static_cast<std::int64_t>(found->second); |
| 132 | + } |
| 133 | + |
| 134 | + [[nodiscard]] std::uint64_t serialization_size() const { |
| 135 | + return 2 + 2 * number_of_edges(); |
| 136 | + } |
| 137 | + |
| 138 | + [[nodiscard]] std::pair<std::vector<EdgeId>, std::vector<EdgeId>> |
| 139 | + extract_subgraph_from_nodes(const std::vector<NodeId> &nodes) const { |
| 140 | + std::unordered_set<NodeId> node_set; |
| 141 | + node_set.reserve(nodes.size()); |
| 142 | + for (const auto node : nodes) { |
| 143 | + validate_node(node); |
| 144 | + node_set.insert(node); |
| 145 | + } |
| 146 | + |
| 147 | + std::vector<EdgeId> inner_edges; |
| 148 | + std::vector<EdgeId> outer_edges; |
| 149 | + for (const auto u : nodes) { |
| 150 | + for (const auto adjacency : node_adjacency(u)) { |
| 151 | + const auto v = adjacency.node; |
| 152 | + const auto edge = adjacency.edge; |
| 153 | + if (node_set.find(v) != node_set.end()) { |
| 154 | + if (u < v) { |
| 155 | + inner_edges.push_back(edge); |
| 156 | + } |
| 157 | + } else { |
| 158 | + outer_edges.push_back(edge); |
| 159 | + } |
| 160 | + } |
| 161 | + } |
| 162 | + return {inner_edges, outer_edges}; |
| 163 | + } |
| 164 | + |
| 165 | +protected: |
| 166 | + EdgeId insert_new_edge(const NodeId u, const NodeId v) { |
| 167 | + const auto edge = static_cast<EdgeId>(edges_.size()); |
| 168 | + edges_.emplace_back(u, v); |
| 169 | + edge_lookup_.emplace(Edge{u, v}, edge); |
| 170 | + adjacency_[static_cast<std::size_t>(u)].push_back(Adjacency{v, edge}); |
| 171 | + adjacency_[static_cast<std::size_t>(v)].push_back(Adjacency{u, edge}); |
| 172 | + return edge; |
| 173 | + } |
| 174 | + |
| 175 | + void validate_node(const NodeId node) const { |
| 176 | + if (node >= number_of_nodes_) { |
| 177 | + throw std::out_of_range( |
| 178 | + "node id must be < number_of_nodes, got node id=" + |
| 179 | + std::to_string(node) + ", number_of_nodes=" + |
| 180 | + std::to_string(number_of_nodes_) |
| 181 | + ); |
| 182 | + } |
| 183 | + } |
| 184 | + |
| 185 | + void validate_edge(const EdgeId edge) const { |
| 186 | + if (edge >= edges_.size()) { |
| 187 | + throw std::out_of_range( |
| 188 | + "edge id must be < number_of_edges, got edge id=" + |
| 189 | + std::to_string(edge) + ", number_of_edges=" + |
| 190 | + std::to_string(edges_.size()) |
| 191 | + ); |
| 192 | + } |
| 193 | + } |
| 194 | + |
| 195 | +private: |
| 196 | + struct EdgeHash { |
| 197 | + std::size_t operator()(const Edge &edge) const { |
| 198 | + const auto first = static_cast<std::size_t>(edge.first); |
| 199 | + const auto second = static_cast<std::size_t>(edge.second); |
| 200 | + return first ^ (second + 0x9e3779b97f4a7c15ULL + (first << 6U) + (first >> 2U)); |
| 201 | + } |
| 202 | + }; |
| 203 | + |
| 204 | + static Edge edge_key(NodeId u, NodeId v) { |
| 205 | + if (v < u) { |
| 206 | + std::swap(u, v); |
| 207 | + } |
| 208 | + return {u, v}; |
| 209 | + } |
| 210 | + |
| 211 | + NodeId number_of_nodes_; |
| 212 | + std::vector<Edge> edges_; |
| 213 | + std::vector<AdjacencyList> adjacency_; |
| 214 | + std::unordered_map<Edge, EdgeId, EdgeHash> edge_lookup_; |
| 215 | +}; |
| 216 | + |
| 217 | +} // namespace bioimage_cpp::graph |
0 commit comments