Short intro to transducers

Transducers are a generic and composable way to operate on a collection of values, producing a new value or new collection of new values. The word ‘transducer’ itself can be split into two parts that reflect this definition: ‘transform’ — to produce some value from another — and ‘reducer’ — to combine the values of a data structure to produce a new one.

Transducers are a composable way to build algorithmic transformations. We kind of do this already with algorithmic transformations as map, filter, etc. combined with the usage of compose.

But there are issues.

Reducing functions

Array (or other collection) operations like map, filter and reverse can be defined in terms of a reduce.

[1,2,3,4].map(function (input) {
    return input + 1;
}) // => [2,3,4,5]

[1,2,3,4].reduce(function (result, input) {
    return result.concat([input + 1]);
}, []) // => [2,3,4,5]

A reducing function is any function that can be passed to reduce. They have the form: (something, input) -> something. They’re the inner-most function in the map and filter examples.

..

A map and filter function extracted out to be a reducing function:

const mapper = xf => {
    return (acc, value) => {
        return concat(acc, xf(value));
    }
}

const filterer = predicate => reducer => {
    return (acc, value) => {
        if (predicate(value)) concat(acc, value)
        return acc
    }
}

function greaterThanTwo(x) {
    return (x > 2);
}

function inc(x) {
    return x + 1;
}

[1,2,3,4].reduce(mapper(inc), []) // => [2,3,4,5]
[1,2,3,4].reduce(filterer(greaterThanTwo), []) // => [3,4]

Impl

Lookup