Graph Server Project

A high-performance containerized graph computation service in modern C++20, backed by Clang and powered by bulgogi.
Includes Python SDK for easy interaction and automation.

---

🧠 Features

• Implemented in C++20 with LLVM/Clang and -march=native optimizations
• Uses a custom adjacency matrix for SIMD-friendly memory layout
• Graph algorithms supported:
  • Degree, degree statistics
  • Triangle counting
  • Shortest paths (BFS & Dijkstra)
  • Betweenness centrality
  • Isolated node detection
• HTTP interface via bulgogi — a lightweight Boost.Beast wrapper
• Includes Python SDK with automatic cleanup and alias support
• Containerized with Docker for reproducibility and testing

---

🚀 Quick Start

1. Build the Docker image

docker build -t graph-server -f docker/Dockerfile .

2. Run the server

docker run -p 8080:8080 graph-server

3. Use the Python client

Requires Python 3.10+. All Python client files are in the python/ directory.

from client import Graph, AliasGraph

g = Graph(size=100, bi=True, weighed=True)
g.add_edge(0, 1, weight=3)
print(g.stats())

4. Load graph from text file with aliases

g = AliasGraph("graph.txt")
print(g.stats())
print(g.centrality())

---

📑 API

The full HTTP API is documented in docs/api.md. All endpoints support JSON input/output and curl-friendly queries.

---

📦 Dependencies

Development (C++20 build only)

• C++20 compiler (Clang ≥ 13 preferred)
• Boost 1.80+ with JSON module (Docker uses 1.88.0)
• JH-Toolkit 1.3.x-LTS
• CMake ≥ 3.27 (Docker builds from source)
• Docker (for consistent build & deployment)

Runtime (Python client)

• Python 3.10+
• requests library (automatically used by the Python SDK)

---

🛠 Design Highlights

• SIMD-friendly layout: adjacency matrix avoids pointer chains
• ColIter enables efficient column traversal for triangle & centrality calc
• GraphManager supports graph ID registration and locking
• Python-side cleanup uses atexit + __del__ to reduce memory leaks

---

📂 Folder Structure

.
├── docker/           # Dockerfile and build config
├── Application/      # C++ GraphManager and core logic
├── Algo/             # C++ Graph algorithms
├── Web/              # bulgogi-based HTTP server
├── python/           # Python SDK (GraphClient, AliasGraph)
└── docs/             # API documentation

---

📊 Attribution

• Demo graph data sourced from Stanford SNAP
• HTTP framework forked and customized from bulgogi

---

🧑‍💻 Author

Developed and maintained by JH / JeongHan-Bae

• GitHub: @JeongHan-Bae
• Framework: bulgogi-framework