add sorting for results output in expand(), expandAll() and expandN(),

[laminesadiki]

Hey Drew , I hope that you're doing well.
Thank you for your incredible working. I use your library for a small time ago, I found it really usefull and helpful.

Otherwise, for a use case, I need to generate all the possible combinations of letters and numbers. and I see that the results is random and it is not sorted. I need to sort them by my own and that cost memory and time to do this with javascript native sorting function.

For example

const regex = '[a-e][0-5]';
const expander = expand(regex);
console.log(expander);
// output => Expansion { getIterator: [Function: beginIterating], count: 30 }

in this small example I have just 30 examples ,
If I use expandN , it will give me a randomize order

const regex = '[a-e][0-5]';
const expander = expandN(regex,5);
/*
first run    => [ 'b4', 'e3', 'a2', 'b3', 'b1' ]
second run => [ 'b5', 'c3', 'c0', 'b3', 'b0' ]
 */

I tried to use a workaround method using sort() and slice() like this

const regex = '[a-e][0-5]';
const expander = expandAll(regex).sort().slice(0, 5);
console.log(expander);
// output => [ 'a0', 'a1', 'a2', 'a3', 'a4' ]

But this is not practical, imagine I have millions of results, It will cost time and ressource to do this, and causing a memory leak in javascript.

It is better to add sorting for expand, expandAll() and expandN().
I don't know if already exists, please to add this feature.
otherwise, You can give me a clue how I can add it in the library and contribute with you.

My Best.

[wimpyprogrammer]

Hi Ayoub,

Thanks for your feedback! I'm so glad to hear this library has been helpful.

I like your suggestion for supporting output in a non-random order. This was actually the original behavior but I didn't think it was useful, so I implemented the randomness. Hearing that an option for non-random order would be valuable to someone is helpful for prioritizing my effort.

Conceptually I've already solved this problem in the unit tests, where I stub the randomization logic to make the tests deterministic.

regex-to-strings/packages/regex-to-strings/src/pattern.spec.ts

Lines 99 to 112 in f4d5a03

	beforeEach(() => {
	// Keep a stable order for consistent tests.
	sortFn = jest
	.spyOn(randomSort, 'default')
	.mockImplementation((items) => [...items]);
	randomWeightedFn = jest
	.spyOn(chooseRandomWeighted, 'default')
	.mockImplementation(() => 0);
	randomFn = jest
	.spyOn(chooseRandom, 'default')
	.mockImplementation((minValue) => minValue);
	});

I'll need to think about how best to expose this behavior. My first thought is a sub-directory like

// These functions output in random order.
import { expand, expandN, expandAll } from 'regex-to-strings';
// These functions output in sorted order.
import { expand, expandN, expandAll } from 'regex-to-strings/sorted';

similar to how Node.js has fs and fs/promise APIs that mirror each other.

If there's a different API that you think would be more intuitive, please let me know! Otherwise I will try to work on this over the weekend.

[laminesadiki]

Hi Drew.

Thanks for your answer. I found the sub-directory idea is good. but it can be more perfect to add optional argument for sorting or filtering results.
It can be an objet argument or even a callback function to do a specific use case.

something like this :

import { expand, expandN, expandAll } from 'regex-to-strings';
const regex = '[a-e][0-5]';
const expander = expand(regex,{sort: "asc"}); // asc or desc

or if I want to do some filtering, like this

import { expand, expandN, expandAll } from 'regex-to-strings';
const regex = '[a-e][0-5]';
const expander = expand(regex,(str) => str.endsWith("am") ); // or any other filtreing critea.

Looking forward to see and review your next commits.
My Best.