Added functionality to import/export Graph from/to Binary File (#554)

me-hem · web-flow · commit b0b5466b5b90 · 2025-10-23T15:02:42.000+02:00
diff --git a/include/CXXGraph/Graph/Graph_decl.h b/include/CXXGraph/Graph/Graph_decl.h
@@ -81,6 +81,39 @@ std::ostream &operator<<(std::ostream &o, const AdjacencyList<T> &adj);
 /// Class that implement the Graph. ( This class is not Thread Safe )
 template <typename T>
 class Graph {
+ public:
+  /**
+   * @brief Write the graph to a binary file
+   * @param workingDir The parent directory of the output file
+   * @param fileName The output filename (without extension)
+   * @param writeNodeFeatures Whether to include node features
+   * @param writeEdgeWeights Whether to include edge weights
+   * @return 0 if successful, negative value on error:
+   *         -1: Cannot open file
+   *         -2: Write error
+   */
+  int writeToBinaryFile(const std::string &workingDir,
+                        const std::string &fileName,
+                        bool writeNodeFeatures = false,
+                        bool writeEdgeWeights = true) const;
+
+  /**
+   * @brief Read the graph from a binary file
+   * @param workingDir The parent directory of the input file
+   * @param fileName The input filename (without extension)
+   * @param readNodeFeatures Whether to read node features
+   * @param readEdgeWeights Whether to read edge weights
+   * @return 0 if successful, negative value on error:
+   *         -1: Cannot open file
+   *         -2: Invalid file format
+   *         -3: Unsupported version
+   *         -4: Read error
+   */
+  int readFromBinaryFile(const std::string &workingDir,
+                         const std::string &fileName,
+                         bool readNodeFeatures = false,
+                         bool readEdgeWeights = true);
+
  private:
   T_EdgeSet<T> edgeSet = {};
   T_NodeSet<T> isolatedNodesSet = {};
@@ -104,6 +137,47 @@ class Graph {
   int writeToDot(const std::string &workingDir, const std::string &OFileName,
                  const std::string &graphName) const;
   int readFromDot(const std::string &workingDir, const std::string &fileName);
+
+  // Binary file format constants
+  static constexpr uint32_t BINARY_MAGIC_NUMBER = 0x47525048;  // "GRPH"
+  static constexpr uint32_t BINARY_VERSION = 1;
+  static constexpr uint64_t BINARY_FLAG_HAS_NODE_FEATURES = 0x01;
+  static constexpr uint64_t BINARY_FLAG_HAS_EDGE_WEIGHTS = 0x02;
+
+  // Type trait to check if T is serializable to binary
+  template <typename U, typename = void>
+  struct is_binary_serializable : std::false_type {};
+
+  template <typename U>
+  struct is_binary_serializable<
+      U, std::void_t<decltype(std::declval<std::ofstream &>().write(
+             reinterpret_cast<const char *>(&std::declval<const U &>()),
+             sizeof(U)))>> : std::is_trivially_copyable<U> {};
+
+  // Helper functions for binary I/O
+  void writeBinaryString(std::ofstream &out, const std::string &str) const;
+  std::string readBinaryString(std::ifstream &in) const;
+
+  /**
+   * @brief Write the graph to a binary file
+   * @param filepath The full path to the output file
+   * @param writeNodeFeatures Whether to include node features
+   * @param writeEdgeWeights Whether to include edge weights
+   * @return 0 if successful, negative value on error
+   */
+  int writeToBinary(const std::string &filepath, bool writeNodeFeatures,
+                    bool writeEdgeWeights) const;
+
+  /**
+   * @brief Read the graph from a binary file
+   * @param filepath The full path to the input file
+   * @param readNodeFeatures Whether to read node features
+   * @param readEdgeWeights Whether to read edge weights
+   * @return 0 if successful, negative value on error
+   */
+  int readFromBinary(const std::string &filepath, bool readNodeFeatures,
+                     bool readEdgeWeights);
+
   void recreateGraph(
       std::unordered_map<std::string, std::pair<std::string, std::string>>
           &edgeMap,
diff --git a/include/CXXGraph/Graph/IO/InputOperation_impl.hpp b/include/CXXGraph/Graph/IO/InputOperation_impl.hpp
@@ -248,6 +248,126 @@ int Graph<T>::readFromMTXFile(const std::string &workingDir,
   return 0;
 }
 
+template <typename T>
+int Graph<T>::readFromBinaryFile(const std::string &workingDir,
+                                 const std::string &fileName,
+                                 bool readNodeFeatures, bool readEdgeWeights) {
+  std::string filepath = workingDir + "/" + fileName + ".bin";
+  return readFromBinary(filepath, readNodeFeatures, readEdgeWeights);
+}
+
+template <typename T>
+int Graph<T>::readFromBinary(const std::string &filepath, bool readNodeFeatures,
+                             bool readEdgeWeights) {
+  std::ifstream in(filepath, std::ios::binary);
+  if (!in.is_open()) {
+    return -1;
+  }
+
+  try {
+    // Read and verify header
+    uint32_t magic;
+    in.read(reinterpret_cast<char *>(&magic), sizeof(magic));
+    if (magic != BINARY_MAGIC_NUMBER) {
+      return -2;  // Invalid file format
+    }
+
+    uint32_t version;
+    in.read(reinterpret_cast<char *>(&version), sizeof(version));
+    if (version != BINARY_VERSION) {
+      return -3;  // Unsupported version
+    }
+
+    uint64_t numNodes, numEdges, flags;
+    in.read(reinterpret_cast<char *>(&numNodes), sizeof(numNodes));
+    in.read(reinterpret_cast<char *>(&numEdges), sizeof(numEdges));
+    in.read(reinterpret_cast<char *>(&flags), sizeof(flags));
+
+    bool hasNodeFeatures = (flags & BINARY_FLAG_HAS_NODE_FEATURES) != 0;
+    bool hasEdgeWeights = (flags & BINARY_FLAG_HAS_EDGE_WEIGHTS) != 0;
+
+    // Read nodes
+    std::unordered_map<std::string, shared<Node<T>>> nodeMap;
+    for (uint64_t i = 0; i < numNodes; ++i) {
+      std::string nodeId = readBinaryString(in);
+
+      T nodeData{};
+      if (hasNodeFeatures && readNodeFeatures) {
+        uint32_t dataSize;
+        in.read(reinterpret_cast<char *>(&dataSize), sizeof(dataSize));
+
+        if (dataSize > 0 && is_binary_serializable<T>::value) {
+          in.read(reinterpret_cast<char *>(&nodeData), sizeof(T));
+        }
+      } else if (hasNodeFeatures) {
+        // Skip node data if present but not reading
+        uint32_t dataSize;
+        in.read(reinterpret_cast<char *>(&dataSize), sizeof(dataSize));
+        if (dataSize > 0) {
+          in.seekg(dataSize, std::ios::cur);
+        }
+      } else {
+        uint32_t dataSize;
+        in.read(reinterpret_cast<char *>(&dataSize), sizeof(dataSize));
+      }
+
+      auto node = std::make_shared<Node<T>>(nodeId, std::move(nodeData));
+      nodeMap[nodeId] = node;
+    }
+
+    // Read edges
+    for (uint64_t i = 0; i < numEdges; ++i) {
+      std::string edgeId = readBinaryString(in);
+      std::string node1Id = readBinaryString(in);
+      std::string node2Id = readBinaryString(in);
+
+      uint8_t edgeFlags;
+      in.read(reinterpret_cast<char *>(&edgeFlags), sizeof(edgeFlags));
+
+      bool isDirected = (edgeFlags & 0x01) != 0;
+      bool isWeighted = (edgeFlags & 0x02) != 0;
+
+      double weight = 0.0;
+      if (hasEdgeWeights && isWeighted) {
+        if (readEdgeWeights) {
+          in.read(reinterpret_cast<char *>(&weight), sizeof(weight));
+        } else {
+          in.seekg(sizeof(double), std::ios::cur);
+        }
+      }
+
+      auto node1 = nodeMap[node1Id];
+      auto node2 = nodeMap[node2Id];
+
+      shared<Edge<T>> edge;
+      if (isDirected) {
+        if (isWeighted && readEdgeWeights) {
+          edge = std::make_shared<DirectedWeightedEdge<T>>(edgeId, node1, node2,
+                                                           weight);
+        } else {
+          edge = std::make_shared<DirectedEdge<T>>(edgeId, node1, node2);
+        }
+      } else {
+        if (isWeighted && readEdgeWeights) {
+          edge = std::make_shared<UndirectedWeightedEdge<T>>(edgeId, node1,
+                                                             node2, weight);
+        } else {
+          edge = std::make_shared<UndirectedEdge<T>>(edgeId, node1, node2);
+        }
+      }
+
+      this->addEdge(edge);
+    }
+
+    in.close();
+    return 0;
+
+  } catch (const std::exception &e) {
+    in.close();
+    return -4;
+  }
+}
+
 template <typename T>
 int Graph<T>::readFromDot(const std::string &workingDir,
                           const std::string &fileName) {
@@ -436,5 +556,15 @@ void Graph<T>::recreateGraph(
   }
 }
 
+// Helper function to read string with length prefix
+template <typename T>
+std::string Graph<T>::readBinaryString(std::ifstream &in) const {
+  uint32_t len;
+  in.read(reinterpret_cast<char *>(&len), sizeof(len));
+  std::string str(len, '\0');
+  in.read(&str[0], len);
+  return str;
+}
+
 }  // namespace CXXGraph
-#endif  // __CXXGRAPH_INPUTOPERATION_IMPL_H__
+#endif  // __CXXGRAPH_INPUTOPERATION_IMPL_H__
diff --git a/include/CXXGraph/Graph/IO/OutputOperation_impl.hpp b/include/CXXGraph/Graph/IO/OutputOperation_impl.hpp
@@ -192,6 +192,100 @@ int Graph<T>::writeToMTXFile(const std::string &workingDir,
   return 0;
 }
 
+template <typename T>
+int Graph<T>::writeToBinaryFile(const std::string &workingDir,
+                                const std::string &fileName,
+                                bool writeNodeFeatures,
+                                bool writeEdgeWeights) const {
+  std::string filepath = workingDir + "/" + fileName + ".bin";
+  return writeToBinary(filepath, writeNodeFeatures, writeEdgeWeights);
+}
+
+template <typename T>
+int Graph<T>::writeToBinary(const std::string &filepath, bool writeNodeFeatures,
+                            bool writeEdgeWeights) const {
+  std::ofstream out(filepath, std::ios::binary);
+  if (!out.is_open()) {
+    return -1;
+  }
+
+  try {
+    // Write header
+    out.write(reinterpret_cast<const char *>(&BINARY_MAGIC_NUMBER),
+              sizeof(BINARY_MAGIC_NUMBER));
+    out.write(reinterpret_cast<const char *>(&BINARY_VERSION),
+              sizeof(BINARY_VERSION));
+
+    auto nodeSet = this->getNodeSet();
+    auto edgeSet = this->getEdgeSet();
+
+    uint64_t numNodes = nodeSet.size();
+    uint64_t numEdges = edgeSet.size();
+    uint64_t flags = 0;
+
+    if (writeNodeFeatures) flags |= BINARY_FLAG_HAS_NODE_FEATURES;
+    if (writeEdgeWeights) flags |= BINARY_FLAG_HAS_EDGE_WEIGHTS;
+
+    out.write(reinterpret_cast<const char *>(&numNodes), sizeof(numNodes));
+    out.write(reinterpret_cast<const char *>(&numEdges), sizeof(numEdges));
+    out.write(reinterpret_cast<const char *>(&flags), sizeof(flags));
+
+    // Write nodes
+    for (const auto &node : nodeSet) {
+      writeBinaryString(out, node->getUserId());
+
+      if (writeNodeFeatures) {
+        // For trivially copyable types, write directly
+        if constexpr (is_binary_serializable<T>::value) {
+          uint32_t dataSize = sizeof(T);
+          out.write(reinterpret_cast<const char *>(&dataSize),
+                    sizeof(dataSize));
+          const T &data = node->getData();
+          out.write(reinterpret_cast<const char *>(&data), sizeof(T));
+        } else {
+          // For non-trivially copyable types, write 0 size
+          uint32_t dataSize = 0;
+          out.write(reinterpret_cast<const char *>(&dataSize),
+                    sizeof(dataSize));
+        }
+      } else {
+        uint32_t dataSize = 0;
+        out.write(reinterpret_cast<const char *>(&dataSize), sizeof(dataSize));
+      }
+    }
+
+    // Write edges
+    for (const auto &edge : edgeSet) {
+      writeBinaryString(out, edge->getUserId());
+      writeBinaryString(out, edge->getNodePair().first->getUserId());
+      writeBinaryString(out, edge->getNodePair().second->getUserId());
+
+      uint8_t edgeFlags = 0;
+      if (edge->isDirected().has_value() && edge->isDirected().value()) {
+        edgeFlags |= 0x01;
+      }
+      if (edge->isWeighted().has_value() && edge->isWeighted().value()) {
+        edgeFlags |= 0x02;
+      }
+      out.write(reinterpret_cast<const char *>(&edgeFlags), sizeof(edgeFlags));
+
+      // Write weight if edge is weighted and we're saving weights
+      if (writeEdgeWeights && (edgeFlags & 0x02)) {
+        double weight =
+            std::dynamic_pointer_cast<const Weighted>(edge)->getWeight();
+        out.write(reinterpret_cast<const char *>(&weight), sizeof(weight));
+      }
+    }
+
+    out.close();
+    return 0;
+
+  } catch (const std::exception &e) {
+    out.close();
+    return -2;
+  }
+}
+
 template <typename T>
 int Graph<T>::writeToDot(const std::string &workingDir,
                          const std::string &OFileName,
@@ -284,6 +378,15 @@ void Graph<T>::writeGraphToStream(std::ostream &oGraph, std::ostream &oNodeFeat,
   }
 }
 
+// Helper function to write string with length prefix
+template <typename T>
+void Graph<T>::writeBinaryString(std::ofstream &out,
+                                 const std::string &str) const {
+  uint32_t len = static_cast<uint32_t>(str.length());
+  out.write(reinterpret_cast<const char *>(&len), sizeof(len));
+  out.write(str.c_str(), len);
+}
+
 template <typename T>
 std::ostream &operator<<(std::ostream &os, const Graph<T> &graph) {
   os << "Graph:\n";
@@ -355,4 +458,4 @@ std::ostream &operator<<(std::ostream &os, const AdjacencyList<T> &adj) {
 }
 
 }  // namespace CXXGraph
-#endif  // __CXXGRAPH_OUTPUTOPERATION_IMPL_H__
+#endif  // __CXXGRAPH_OUTPUTOPERATION_IMPL_H__
diff --git a/test/BinaryIOTest.cpp b/test/BinaryIOTest.cpp
