Reiter NPM | npm.io

reiter

Various utilities for working with Iterators and Iterables.

Building on the iterator and iterable protocols, this library aims to provide various utilities for working with iterators.

To install, run either yarn add reiter or npm install reiter. Including it is simple:

// Node.js
const reiter = require("reiter");

// Environments with EcmaScript modules such as Babel and Webpack.
import * as reiter from "reiter";

Motivation

Normal functions on collections such as map and filter will construct entire arrays for each intermediate result. For small arrays of maybe 20 or so elements this will likely be very fast, possibly faster than using iterators depending on run-time optimizations. For larger collections, intermediate representations can be avoided, only applying methods to each element as it is needed.

Arrays and other finite collections also have the disadvantage of not being able to deal with endless sequences, due to their inherently lazy evaluation.

Curried

All methods with multiple arguments are curried to allow for functional programming patterns. The curry function is not particularly optimized, so builds using Lodash's curry are available. The package exports curry and placeholder. _ is an alias of placeholder. A compose function is included to aid in composing methods together.

Different Builds:

By default the UMD bundle (without Lodash's curry) will be loaded by require.

Each build will have a UMD bundle (index.js), a CommonJS bundle (index.common.js), and an EcmaScript module bundle (index.esm.js). Builds with the regular built-in curry function are in reiter/dist/, while builds referencing the Lodash curry function are in reiter/dist/lodash/.

All Lodash builds will reference the curry function from the lodash package on npm. If you are planning to use this build, be sure to have already installed Lodash.

Examples

Iterate over values and indices cleanly without a traditional for loop:

for (const [index, value] of reiter.enumerate([1, 2, 3])) {
  console.log(`[${index}]: value`);
}
// [0]: 1
// [1]: 2
// [2]: 3

Take the first three odds and then take the first three evens from a single iterator created by the count method. Only 6 values are ever created. Note that this could be done more simply with odds = reiter.count(1, 2) and evens = reiter.count(2, 2). This is just a simple proof of concept.

const naturals = reiter.count(1, 1);
const [naturals1, naturals2] = reiter.tee(2, naturals);
// partial applications
const filterOdds = reiter.filter(n => n % 2 === 1);
const filterEvens = reiter.filter(n => n % 2 === 0);
const odds = filterOdds(naturals1);
const evens = filterEvens(naturals2);
console.log(...reiter.take(3, odds));
console.log(...reiter.take(3, evens));
// 1, 3, 5
// 2, 4, 6

Avoid deeply nested for loops with cartesianAll and other combinatorial functions.

const domain = range(-10, 11, 1); // stop value is excluded
for (const [x, y, z] of reiter.cartesianAll(reiter.tee(3, domain)) {
  // do something with each coordinate pair ...
}

Caveats

For better or worse, iterators are unidirectional and cannot be used more than once. Once an iterator is passed to a method, one should consider it exhausted and no longer attempt to use it. If you must iterate an iterator multiple times, consider using tee or reiter, e.g.:

// Iterator copies of iterator.
const [copy1, copy2, copy3] = reiter.tee(3, iterator);

// A reiterable collection created from iterator.
const reiterable = reiter.reiter(iterator);

When passing iterable collections such as Arrays, Sets, and Maps to methods of reiter, one should not mutate the collections until after all iterators are not in-use. Mutating the collections could result in undefined behavior.