@js-bits/process
v1.0.2
Published
Asynchronous multi-step processing
Downloads
9
Maintainers
Readme
Asynchronous multi-step processing
Allows you to organize a complicated processing logic as a set of simpler consecutive steps. Includes Executor capabilities in case you need some metrics.
Installation
Install with npm:
npm install @js-bits/process
Install with yarn:
yarn add @js-bits/process
Import where you need it:
import Process from '@js-bits/process';
or require for CommonJS:
const Process = require('@js-bits/process');
How to use
Let's say you have a long complicated process that consists of some operations (synchronous or asynchronous) which must be performed sequentially, one by one. Process
class can help you to simplify the implementation and make it more readable. Here is an example:
(async () => {
// a synchronous function
const step1 = () => {
console.log('step1');
};
// an asynchronous function
const step2 = async () => {
console.log('step2');
};
// a promise
const step3 = new Promise(resolve => {
setTimeout(() => {
console.log('step3');
resolve();
}, 100);
});
// another process
const step4 = new Process(() => {
console.log('step4');
});
const process = new Process(step1, step2, step3, step4);
await process.start();
// step1
// step2
// step3
// step4
})();
Alternatively you can combine steps into groups like this:
...
const operation1 = [step1, step2];
const operation2 = [step3, step4];
const process = new Process(operation1, operation2);
...
The result will be the same as before, but the code looks more structured and logically organized.
Passing input parameters and returning processing results
You can pass as many input parameters as you need (as an object) into .start()
method of a Process
instance.
Steps of the process may return some results (also as an object) if necessary, but not required to do so.
Each step of the process will receive (as an argument) all input parameters as well as results of all preceding steps,
which means that the step logic can be implemented to behave differently depending on what happened before.
The return value of the whole process will be all step results combined.
(async () => {
const step1 = async args => {
console.log(args); // { inputParam: 1 }
return { step1Result: 'success' };
};
const step2 = async args => {
console.log(args); // { inputParam: 1, step1Result: 'success' }
return { step2Result: 'success' };
};
const step3 = async args => {
console.log(args); // { inputParam: 1, step1Result: 'success', step2Result: 'success' }
return { step3Result: 'success' };
};
const process = new Process(step1, step2, step3);
const result = await process.start({ inputParam: 1 });
console.log(result); // { step1Result: 'success', step2Result: 'success', step3Result: 'success' }
})();
Process.steps() shortcut
There is Process.steps()
method available for your convenience.
It's just a shortcut to new Process(...steps).start(input)
.
...
const result = Process.steps(step1, step2, step3)(inputParams)
You can also use it for better structuring of your source-code.
...
const operation1 = Process.steps(
step1,
step2,
step3,
);
const operation2 = Process.steps(
step4,
step5,
step6,
);
const process = new Process(operation1, operation2);
...
Exit strategy (Process.exit)
Most likely, there will be some exceptional situations when you have to interrupt your processing.
You can account for that using Process.exit
method.
const step1 = async () => {
console.log('step1');
return { step1Result: 'success' };
};
const step2 = async () => {
console.log('step2');
return Process.exit;
};
const step3 = async () => {
// this step won't be performed
console.log('step3');
return { step3Result: 'success' };
};
const process = new Process(step1, step2, step3);
const result = await process.start({ inputParam: 1 });
console.log(result);
// step1
// step2
// { step1Result: 'success', [Symbol(EXIT)]: true }
If you'd like to also return some additional information related to the process interruption,
just use Process.exit
as a function (only accepts objects as an argument).
...
return Process.exit({ exitReason: 'Something bad has happened'});
...
// { step1Result: 'success', [Symbol(EXIT)]: true, exitReason: 'Something bad has happened' }
Conditional processing with Process.switch()
Sometimes business logic requires a processing to go one way or another depending on some condition.
Process.switch()
method is aimed to solve the problem. Here is how it works:
(async () => {
const simulateLatency =
func =>
(...args) =>
new Promise(resolve => {
setTimeout(() => {
resolve(func(...args));
}, Math.ceil(Math.random() * 500));
});
const fetchItem = simulateLatency(() => {
console.log('item loaded');
const item = { id: 'item-id', state: 'draft' };
return { item };
});
const checkNeedsUpdate = ({ newState, item }) => {
if (item.state === newState) return Process.exit;
console.log('item needs update');
};
const publishItem = simulateLatency(() => {
console.log('item published');
});
const notifySubscribers = simulateLatency(() => {
console.log('subscribers notified');
});
const deleteItem = simulateLatency(() => {
console.log('item deleted');
});
const updateItem = simulateLatency(({ item, newState }) => {
console.log('item updated');
return { updateResult: { ...item, state: newState } };
});
const processItem = new Process(
fetchItem,
checkNeedsUpdate,
Process.switch('newState', {
// select next step based on input.newState value
draft: Process.noop,
published: [publishItem, notifySubscribers],
deleted: [deleteItem, Process.exit],
}),
updateItem
);
const input = { id: 'item-id', newState: 'published' };
const result = await processItem.start(input);
console.log(result.updateResult);
// item loaded
// item needs update
// item published
// subscribers notified
// item updated
// { id: 'item-id', state: 'published' }
})();
Process.noop
(abbreviation for "no operation") above is just a convenient shortcut to Promise.resolve()
, which is an equivalent of "do nothing".
Notes
- A process is a one-off operation. You have to create a new instance each time you need to run the same process.