• [Transducer] Make an Into Helper to Remove Boilerplate and Simplify our Transduce API


    Our transduce function is powerful but requires a lot of boilerplate. It would be nice if we had a way to transduce into arrays and objects without having to specify the inner reducer behaviour, i.e. how to build up values, since a given collection type will almost always have the same inner reducer.

    In this lesson we'll being doing just that with an into() helper function that will know if an array or object collection is passed in and handle each case accordingly.

     The whole point to make 'into' helper is hide the inner reducer logic from user. So 'into' helper will check the target's type, based on the type, will use different inner reducer.
     
    import {isPlainObject, isNumber} from 'lodash';
    import {compose, map, filter, pushReducer} from '../utils';
    
    //current transduce
    const transduce = (xf /** could be composed **/, reducer, seed, collection) => {
        const transformedReducer = xf(reducer);
        let accumulation = seed;
        for (let value of collection) {
            accumulation = transformedReducer(accumulation, value);
        }
    
        return accumulation;
    };
    
    const objectReducer = (obj, value) => Object.assign(obj, value);
    
    const into = (to, xf, collection) => {
        if (Array.isArray(to)) return transduce(xf, pushReducer, to, collection);
        else if (isPlainObject(to)) return transduce(xf, objectReducer, to, collection);
        throw new Error('into only supports arrays and objects as `to`');
    };
    
    into(
      [],
      compose(
        map(x => x/2),
        map(x => x * 10)
      ),
      [1,2,3,4],
    );
    
    into(
      {},
      compose(filter(isNumber), map(val => ({[val]: val}))),
      [1,2,3,4, 'hello', () => 'world'],
    );

    utils:

    export const compose = (...functions) =>
        functions.reduce((accumulation, fn) =>
            (...args) => accumulation(fn(...args)), x => x);
    
    export const map = xf => reducer => {
        return (accumulation, value) => {
            return reducer(accumulation, xf(value));
        };
    };
    
    export const filter = predicate => reducer => {
        return (accumulation, value) => {
            if (predicate(value)) return reducer(accumulation, value);
            return accumulation;
        };
    };
    export const pushReducer = (accumulation, value) => {
        accumulation.push(value);
        return accumulation;
    };
  • 相关阅读:
    矩阵的应用
    PyTorch工具
    python不同开根号速度对比
    GloVe损失函数的理解
    课程回顾-Convolutional Neural Networks
    课程回顾-Structuring Machine Learning Projects
    课程回顾-Improving Deep Neural Networks: Hyperparameter tuning, Regularization and Optimization
    课程回顾-Neural Network & Deep Learning
    使用GOOGLE COLAB训练深度学习模型
    如何使用Keras的Model visualization功能
  • 原文地址:https://www.cnblogs.com/Answer1215/p/8305758.html
Copyright © 2020-2023  润新知