Understanding LiteFlow: A Lightweight Component‑Based Flow Engine for Decoupling Complex Business Logic

Many enterprise systems contain tangled business logic that becomes hard to maintain as projects evolve; LiteFlow is presented as a lightweight, fast component‑based flow engine designed to split such logic into independent nodes and orchestrate them through a configurable rule file, enabling hot‑reloading and flexible process changes.

The framework follows a "workbench" model: workers (components) pick up resources from a shared workbench (context) and produce parts that are later assembled, allowing each worker to operate without knowledge of others, which brings decoupling, stability, reusability, and real‑time reconfiguration.

In a non‑Spring environment you add the core dependency:

<dependency>
    <groupId>com.yomahub</groupId>
    <artifactId>liteflow-core</artifactId>
    <version>2.6.13</version>
</dependency>

Then you implement custom nodes by extending NodeComponent and overriding process , define an XML flow file under resources , build a LiteflowConfig with the XML path, create a FlowExecutor , and invoke execute2Resp("chain1") to run the defined chain.

When using Spring Boot you add the starter dependency:

<dependency>
    <groupId>com.yomahub</groupId>
    <artifactId>liteflow-spring-boot-starter</artifactId>
    <version>2.6.13</version>
</dependency>

Custom nodes are annotated with @LiteflowComponent , the XML file path is set in application.properties , and the framework automatically registers the nodes, so no explicit XML <component> tags are required.

The core components include:

Parser : parses rule files (XML/JSON/YAML) into Java objects.

FlowBus : stores parsed Node and Chain objects.

FlowExecutor : drives execution by retrieving a Chain from FlowBus and invoking its conditions.

Slot : shared context for a flow execution.

DataBus : manages Slot instances.

During FlowExecutor construction the framework loads rule files, creates Document objects, and through FlowParser builds Node , Condition (Pre, Then, When, Finally) and Chain objects, registering them in FlowBus .

Execution proceeds in the order PreCondition → ThenCondition (serial) → WhenCondition (parallel) → FinallyCondition. Each Condition contains Executable objects, which are either Node or nested Chain . Nodes are executed via a NodeExecutor that supports retries and invokes the node’s process method.

Overall, LiteFlow provides a clear separation between business code and orchestration, allowing developers to modify process flows by editing rule files without changing Java code, thus reducing coupling and improving maintainability.