hugegraph-toolchain

What is HugeGraph Toolchain?

A comprehensive suite of client SDKs, data tools, and management utilities for Apache HugeGraph graph database. Build applications, load data, and manage graphs with production-ready tools.

Hubble's primary authentication and connection design targets HugeGraph 1.8/master: PD discovery supplies the server address, anonymous mode uses a real unauthenticated client, and account/GraphSpace permissions are reduced to four readable presets. A thin adapter keeps 1.7 usable and limits 1.5 to its standalone core graph workflow; version checks are centralized rather than spread across UI pages.

Hubble Workbench

Hubble brings graph exploration, schema preparation, asynchronous analysis, and distributed cluster operations into one workspace.

Distributed Operations

The cluster overview keeps service topology, node health, source status, and capacity facts in one operational view.

Quick Navigation: Architecture | Quick Start | Modules | Build | Docker | Related Projects

Related Projects

HugeGraph Ecosystem:

1. hugegraph - Core graph database (pd / store / server / commons)
2. hugegraph-computer - Distributed graph computing system
3. hugegraph-ai - Graph AI/LLM/Knowledge Graph integration
4. hugegraph-website - Documentation and website

Architecture Overview

                    ┌─────────────────────────┐
                    │   HugeGraph Server      │
                    │   (Graph Database)      │
                    └───────────┬─────────────┘
                                │ REST API
        ┌ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ┼ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ┐
          Distributed (Optional)│
        │   ┌───────────┐       │       ┌───────────┐   │
            │hugegraph- │◄──────┴──────►│hugegraph- │
        │   │    pd     │               │   store   │   │
            └───────────┘               └───────────┘
        └ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ─ ┘
                                │
          ┌─────────────────────┼─────────────────────┐
          │                     │                     │
          ▼                     ▼                     ▼
 ┌────────────────┐    ┌────────────────┐    ┌────────────────┐
 │ hugegraph-     │    │ Other Client   │    │  Other REST    │
 │ client (Java)  │    │ SDKs (Go/Py)   │    │  Clients       │
 └───────┬────────┘    └────────────────┘    └────────────────┘
         │ depends on
 ┌───────┼───────────┬───────────────────┐
 │       │           │                   │
 ▼       ▼           ▼                   ▼
┌────────┐ ┌────────┐ ┌──────────┐ ┌───────────────────┐
│ loader │ │ hubble │ │  tools   │ │ spark-connector   │
│ (ETL)  │ │ (Web)  │ │  (CLI)   │ │ (Spark I/O)       │
└────────┘ └────────┘ └──────────┘ └───────────────────┘

Quick Start

Prerequisites

For the Java 17 candidate SDK, upgrade procedure and compatibility boundary, see the migration guide.

Choose Your Path

Docker Quick Start

# Hubble Web UI (port 8088)
docker run -d -p 8088:8088 --name hubble hugegraph/hugegraph-hubble

# Loader (batch data import)
docker run --rm hugegraph/hugegraph-loader ./bin/hugegraph-loader.sh -f example.json

Module Overview

Before committing a new source or test file, run the same license-header check used by CI (license-eye from apache/skywalking-eyes is required):

./tools/check-license-header.sh

Do not use shortened Apache headers: the complete header configured in .licenserc.yaml is required.

hugegraph-client

Purpose: Official Java SDK for HugeGraph Server

Key Features:

• Schema management (PropertyKey, VertexLabel, EdgeLabel, IndexLabel)
• Graph operations (CRUD vertices/edges)
• Gremlin query execution
• Built-in traversers (shortest path, k-neighbor, k-out, paths, etc.)
• Multi-graph and authentication support

Entry Point: org.apache.hugegraph.driver.HugeClient

Quick Example:

HugeClient client = HugeClient.builder("http://localhost:8080", "hugegraph").build();

// Schema management
client.schema().propertyKey("name").asText().ifNotExist().create();
client.schema().vertexLabel("person")
    .properties("name")
    .ifNotExist()
    .create();

// Graph operations
Vertex vertex = client.graph().addVertex(T.label, "person", "name", "Alice");

📖 Documentation | 📁 Source

Other Client SDKs

Purpose: Official Go SDK for HugeGraph Server

Key Features:

• RESTful API client for HugeGraph
• Schema and graph operations
• Gremlin query support
• Idiomatic Go interface

Entry Point: github.com/apache/hugegraph-toolchain/hugegraph-client-go

Quick Example:

import "github.com/apache/hugegraph-toolchain/hugegraph-client-go"

client := hugegraph.NewClient("http://localhost:8080", "hugegraph")
// Schema and graph operations

📁 Source

Looking for other languages? See hugegraph-python-client in the hugegraph-ai repository.

hugegraph-loader

Purpose: Batch data import tool from multiple data sources

Key Features:

• Sources: CSV, JSON, HDFS, MySQL, Kafka, existing HugeGraph
• JSON-based mapping configuration
• Parallel loading with configurable threads
• Error handling and retry mechanisms
• Progress tracking and logging

Entry Point: bin/hugegraph-loader.sh

Quick Example:

# Load data from CSV
./bin/hugegraph-loader.sh -f mapping.json -g hugegraph

# Example mapping.json structure
{
  "vertices": [
    {
      "label": "person",
      "input": { "type": "file", "path": "persons.csv" },
      "mapping": { "name": "name", "age": "age" }
    }
  ]
}

📖 Documentation | 📁 Source

hugegraph-hubble

Purpose: Web-based graph management and visualization platform

Key Features:

• Multi-graph workspace & connection management
• Interactive schema management with graphical editor
• Comprehensive data loading dashboard
• Dynamic graph visualization with path and topology canvas
• Built-in Gremlin query console & algorithm explorer
• Fine-grained user authentication & multi-language localization (i18n)

Technology Stack: Spring Boot + React + TypeScript + MobX + Ant Design

Entry Point: bin/start-hubble.sh (default port: 8088)

Quick Start:

cd hugegraph-hubble/apache-hugegraph-hubble-*/bin
./start-hubble.sh      # Background mode
./start-hubble.sh -f   # Foreground mode
./stop-hubble.sh       # Stop server

📖 Documentation | 📁 Source

hugegraph-tools

Purpose: Command-line utilities for graph operations

Key Features:

• Backup and restore graphs
• Graph migration
• Graph cloning
• Metadata management
• Batch operations

Entry Point: bin/hugegraph CLI commands

Quick Example:

# Backup graph
bin/hugegraph backup -t all -d ./backup

# Restore graph
bin/hugegraph restore -t all -d ./backup

📁 Source

hugegraph-spark-connector

Purpose: Spark integration for reading and writing HugeGraph data

Key Features:

• Read HugeGraph vertices/edges as Spark DataFrames
• Write DataFrames to HugeGraph
• Spark SQL support
• Distributed graph processing

Entry Point: Scala API with Spark DataSource v2

Quick Example:

// Read vertices as DataFrame
val vertices = spark.read
  .format("hugegraph")
  .option("host", "localhost:8080")
  .option("graph", "hugegraph")
  .option("type", "vertex")
  .load()

// Write DataFrame to HugeGraph
df.write
  .format("hugegraph")
  .option("host", "localhost:8080")
  .option("graph", "hugegraph")
  .save()

📁 Source

Maven Dependencies

<!-- Note: Use the latest release version in Maven Central -->
<dependency>
    <groupId>org.apache.hugegraph</groupId>
    <artifactId>hugegraph-client</artifactId>
    <version>1.7.0</version>
</dependency>

<dependency>
    <groupId>org.apache.hugegraph</groupId>
    <artifactId>hugegraph-loader</artifactId>
    <version>1.7.0</version>
</dependency>

Check Maven Central for the latest versions.

Build & Development

Full Build

Normal builds use the 1.8.0 SDK and release packaging checks. Select the staging repository in Maven settings and use an isolated local repository:

release_settings=/absolute/path/to/settings-with-selected-staging.xml
release_repository=$(mktemp -d /tmp/hugegraph-release-m2.XXXXXX)
mvn --settings "$release_settings" -Dmaven.repo.local="$release_repository" clean install \
  -DskipTests -Dmaven.javadoc.skip=true -ntp

Distribution packaging requires Python 3.9 or newer. Release mode checks required SDK POM/JARs and packaged library versions and bytes, and rejects candidate manifests. Add -Papache-release for the Apache release profile; see release packaging. To build the pinned same-source 1.8.0 SDK instead, follow the bootstrap instructions and supply -Dmaven.repo.local="$candidate_dir/m2" -Dhugegraph.version=1.8.0 -Dsdk.validation.mode=candidate to every module build.

The aggregate Maven module is org.apache.hugegraph:hugegraph-toolchain-dist:1.8.0, distinct from Server's hugegraph-dist coordinate. Its source directory remains hugegraph-dist; the release archive remains apache-hugegraph-toolchain-1.8.0.tar.gz.

Module-Specific Builds

Until the 1.8.0 SDK is available in Maven Central, apply the same --settings "$release_settings" and -Dmaven.repo.local="$release_repository" options from Full Build to each Maven command below.

Hubble frontend builds require stable Node.js 24 (at least the version in hubble-fe/.node-version) and the exact pnpm version in hubble-fe/package.json. Install Corepack if your Node distribution does not include it, then run corepack enable and pnpm --version from hugegraph-hubble/hubble-fe before the Maven build. Later Node 24 patch/minor releases are supported; Node 25 and prerelease builds require a separate compatibility upgrade.

Running Tests

Code Style

Checkstyle is enforced via tools/checkstyle.xml:

• Max line length: 120 characters
• 4-space indentation (no tabs)
• No star imports
• No System.out.println

Run checkstyle:

mvn checkstyle:check

Docker

Official Docker images are available on Docker Hub:

Examples:

# Hubble
docker run -d -p 8088:8088 --name hubble hugegraph/hugegraph-hubble

# Loader (mount config and data)
docker run --rm \
  -v /path/to/config:/config \
  -v /path/to/data:/data \
  hugegraph/hugegraph-loader \
  ./bin/hugegraph-loader.sh -f /config/mapping.json

Build images from the Toolchain repository root so the context contains the shared SDK bootstrap and verifier:

# Loader
docker build -f hugegraph-loader/Dockerfile \
  -t hugegraph/hugegraph-loader:latest .

# Hubble
docker build -f hugegraph-hubble/Dockerfile \
  -t hugegraph/hugegraph-hubble:latest .

Multi-platform builds use BuildKit's automatic platform arguments. The Maven and Node build stages run on $BUILDPLATFORM, while the final JRE stage uses $TARGETPLATFORM. Java bytecode and frontend assets are architecture-neutral, so they are built once without QEMU. Target-stage package installation still runs for each architecture.

This optimization applies only to architecture-independent build outputs. A component that compiles native code must use a target-platform build stage or separate platform stages. Loader and Hubble packaging is validated on arm64; native dependencies must still be audited. Loader includes arm64 variants for Snappy, LZ4, Commons Crypto, and gRPC tcnative. Some optional legacy HBase and Jansi natives remain x86-only, so their fallback paths require target-runtime validation when those optional features are used.

Documentation

• HugeGraph Toolchain Overview
• hugegraph-client Guide
• hugegraph-loader Guide
• hugegraph-hubble Guide
• API Reference

Contributing

Welcome to contribute to HugeGraph! Please see How to Contribute for more information.

Note: It's recommended to use GitHub Desktop to simplify the PR and commit process.

Thank you to all the people who already contributed to HugeGraph!

Community & Contact

• GitHub Issues - Report bugs and request features
• Email: dev@hugegraph.apache.org (subscriber only)
• Slack: ASF Channel
• WeChat: Apache HugeGraph (scan QR code below)

License

hugegraph-toolchain is licensed under Apache 2.0 License.

See CI behavior for affected-module tests, shared fixtures and retry rules.