Conveyor Belt
Last updated
Last updated
The util class Conveyor Belt is used to implement the underlying design pattern that composes all the pipelines implemented in Stellar Plus' core. It aims to encapsulate core processes into specific contexts while keeping a great deal of flexibility and personalization.
Each conveyor belt represents a process that is implemented into its internal method 'process'. Using Typescript's generics, one can constrict the belt into only handling a specific Input and Output, also associating its pipeline type.
When initialized, Conveyor belts and their sub-classes can, along with their custom arguments, provide an array of plugins that are compatible with this belt type. These plugins are core to personalizing the workflow of a pipeline by acting at predefined moments of the execution.
The execute method serves as the heart of the ConveyorBelt class, orchestrating the entire processing pipeline from pre-processing to post-processing, including error management. This method demonstrates the class's ability to transform an input item into an output item, leveraging the flexibility and customizability provided by the plugins. Here's an in-depth look at each step of the execution process:
1. Invocation
The execution process begins when the execute method is called with an item of the specified input type. An optional existingItemId can also be passed to identify the item throughout the process, especially useful for tracking and logging purposes as well as leveraging the plugins' capabilities to act on an item through different pipelines.
2. Pre-Processing
Before the main processing phase, the item undergoes pre-processing. This step allows for initial modifications or setup actions to be performed on the input. Each plugin with a preProcess method will be applied in sequence to the item. The preProcess functions are called with the item and metadata (including itemId, beltId, and beltType) allowing plugins to perform context-aware operations.
3. Main Processing
After pre-processing, the item is passed to the main process method. This method is intended to be overridden in subclasses to implement the core transformation or action of the conveyor belt. The process method is where the input item is transformed into the output item.
4. Error Handling
If an error occurs during the main processing phase, the error is caught and passed to the processError method. This method allows for standardized error handling and potentially converting errors into a manageable format. Plugins with a processError method will be invoked in sequence, providing an opportunity to log errors, retry operations, or modify the error before it is re-thrown.
5. Post-Processing
Following successful processing (or error handling if an error occurs), the item is subjected to post-processing. Similar to pre-processing, this step involves invoking the postProcess method of each plugin, allowing for final adjustments or clean-up operations on the processed item or the handling of any residual effects from the main processing or error-handling phases.
6. Completion
Once the post-processing is complete, the final processed item is returned. If an error is caught and not resolved, the modified error is thrown, allowing the caller to handle it as needed.
The ConveyorBelt utilizes plugins to provide extensibility and customization at various stages of the execution process. Plugins can define preProcess, processError, and postProcess methods to interact with the item at different phases. This design allows for a highly modular and flexible system where specific behaviors can be injected or modified without altering the core processing logic.
The error-handling mechanism is designed to ensure robust processing pipelines. By encapsulating error handling within the execution flow, ConveyorBelt offers a systematic approach to managing exceptions, allowing for error logging, transformation, or recovery strategies to be implemented seamlessly within the pipeline.
