以下是個人在工作中收藏總結的一些關於javascript陣列方法reduce的相關程式碼片段，後續遇到其他使用這個函式的場景，將會陸續新增，這裡作為備忘。

javascript陣列那麼多方法，為什麼我要單挑reduce方法，一個原因是我對這個方法掌握不夠，不能夠用到隨心所欲。另一個方面，我也感覺到了這個方法的龐大魅力，在許多的場景中發揮著神奇的作用。

理解reduce函式

reduce() 方法接收一個函式作為累加器（accumulator），陣列中的每個值（從左到右）開始縮減，最終為一個值。

arr.reduce([callback, initialValue])

關於reduce

的用法，這裡不再做多述，可以去這裡檢視

看如下例子：

let arr = [1, 2, 3, 4, 5];

// 10代表初始值，p代表每一次的累加值，在第一次為10
// 如果不存在初始值，那麼p第一次值為1
// 此時累加的結果為15
let sum = arr.reduce((p, c) => p + c, 10);  // 25

// 轉成es5的寫法即為：
var sum = arr.reduce(function(p, c) {
    console.log(p);
    return p + c;
}, 10);

片段一：字母遊戲

const anagrams = str => {
    if
 
 (str.length <= 2) {
        return str.length === 2 ? [str, str[1] + str[0]] : str;
    }
    return str.split("").reduce((acc, letter, i) => {
        return acc.concat(anagrams(str.slice(0, i) + str.slice(i + 1)).map(val => letter + val));
    }, []);
}

anagrams("abc");    // 結果會是什麼呢？

reduce

負責篩選出每一次執行的首字母，遞迴負責對剩下字母的排列組合。

片段二：累加器

const sum = arr => arr.reduce((acc, val) => acc + val, 0);
sum([1, 2, 3]);

片段三：計數器

const countOccurrences = (arr, value) => arr.reduce((a, v) => v === value ? a + 1 : a + 0, 0);
countOccurrences([1, 2, 3, 2, 2, 5, 1], 1);

迴圈陣列，每遇到一個值與給定值相等，即加1，同時將加上之後的結果作為下次的初始值。

片段四：函式柯里化

const curry = (fn, arity = fn.length, ...args) => 
  arity <= args.length ? fn(...args) : curry.bind(null, fn, arity, ...args);

curry(Math.pow)(2)(10);
curry(Math.min, 3)(10)(50)(2);

通過判斷函式的引數取得當前函式的length(當然也可以自己指定)，如果所傳的引數比當前引數少，則繼續遞迴下面，同時儲存上一次傳遞的引數。

片段五：陣列扁平化

const deepFlatten = arr => 
  arr.reduce((a, v) => a.concat(Array.isArray(v) ? deepFlatten(v) : v), []);
deepFlatten([1, [2, [3, 4, [5, 6]]]]);

片段六：生成菲波列契陣列

const fibonacci = n => Array(n).fill(0).reduce((acc, val, i) => acc.concat(i > 1 ? acc[i - 1] + acc[i - 2] : i), []);
fibonacci(5);

片段七：管道加工器

const pipe = (...funcs) => arg => funcs.reduce((acc, func) => func(acc), arg);
pipe(btoa, x => x.toUpperCase())("Test");

通過對傳遞的引數進行函式加工，之後將加工之後的資料作為下一個函式的引數，這樣層層傳遞下去。

片段八：中介軟體

const dispatch = action => {
    console.log('action', action);
    return action;
}

const middleware1 = dispatch => {
    return action => {
        console.log("middleware1");
        const result = dispatch(action);
        console.log("after middleware1");
        return result;
    }
}

const middleware2 = dispatch => {
    return action => {
        console.log("middleware2");
        const result = dispatch(action);
        console.log("after middleware2");
        return result;
    }
}

const middleware3 = dispatch => {
    return action => {
        console.log("middleware3");
        const result = dispatch(action);
        console.log("after middleware3");
        return result;
    }
}

const compose = middlewares => middlewares.reduce((a, b) => args => a(b(args)))

const middlewares = [middleware1, middleware2, middleware3];
const afterDispatch = compose(middlewares)(dispatch);

const testAction = arg => {
    return { type: "TEST_ACTION", params: arg };
};
afterDispatch(testAction("1111"));

redux中經典的compose函式中運用了這種方式，通過對中介軟體的重重層疊，在真正發起action的時候觸發函式執行。

片段九：redux-actions對state的加工片段

// redux-actions/src/handleAction.js
const handleAction = (type, reducer, defaultState) => {
    const types = type.toString();
    const [nextReducer, throwReducer] = [reducer, reducer];
    return (state = defaultState, action) => {
        const { type: actionType } = action;
        if (!actionType || types.indexOf(actionType.toString()) === -1) {
            return state;
        }

        return (action.error === true ? throwReducer : nextReducer)(state, action);
    }
}

// reduce-reducers/src/index.js
const reduceReducer = (...reducers) => {
    return (previous, current) => {
        reducers.reduce((p, r) => r(p, current), previous);
    }
}

// redux-actions/src/handleActions.js
const handleActions = (handlers, defaultState, { namespace } = {}) => {
    // reducers的扁平化
    const flattenedReducerMap = flattenReducerMap(handles, namespace);
    // 每一種ACTION下對應的reducer處理方式
    const reducers = Reflect.ownkeys(flattenedReducerMap).map(type => handleAction(
        type,
        flattenedReducerMap[type],
        defaultState
    ));
    // 狀態的加工器，用於對reducer的執行
    const reducer = reduceReducers(...reducers);
    // reducer觸發
    return (state = defaultState, action) => reducer(state, action);
}

片段十：資料加工器

const reducers = {
    totalInEuros: (state, item) => {
        return state.euros += item.price * 0.897424392;
    },
    totalInYen: (state, item) => {
        return state.yens += item.price * 113.852;
    }
};

const manageReducers = reducers => {
    return (state, item) => {
        return Object.keys(reducers).reduce((nextState, key) => {
            reducers[key](state, item);
            return state;
        }, {})
    }
}

const bigTotalPriceReducer = manageReducers(reducers);
const initialState = { euros: 0, yens: 0 };
const items = [{ price: 10 }, { price: 120 }, { price: 1000 }];
const totals = items.reduce(bigTotalPriceReducer, initialState);

片段十一：物件空值判斷

let school = {
    name: 'Hope middle school',
    created: '2001',
    classes: [
        {
            name: '三年二班',
            teachers: [
                { name: '張二蛋', age: 26, sex: '男', actor: '班主任' },
                { name: '王小妞', age: 23, sex: '女', actor: '英語老師' }
            ]
        },
        {
            name: '明星班',
            teachers: [
                { name: '歐陽娜娜', age: 29, sex: '女', actor: '班主任' },
                { name: '李易峰', age: 28, sex: '男', actor: '體育老師' },
                { name: '楊冪', age: 111, sex: '女', actor: '藝術老師' }
            ]
        }
    ]
};

// 常規做法
school.classes &&
school.classes[0] &&
school.classes[0].teachers &&
school.classes[0].teachers[0] &&
school.classes[0].teachers[0].name

// reduce方法
const get = (p, o) => p.reduce((xs, x) => (xs && xs[x] ? xs[x] : null), o);
get(['classes', 0, 'teachers', 0, 'name'], school);   // 張二蛋

片段十二：分組

const groupBy = (arr, func) =>
arr.map(typeof func === 'function' ? func : val => val[func]).reduce((acc, val, i) => {
    acc[val] = (acc[val] || []).concat(arr[i]);
    return acc;
}, {});
groupBy([6.1, 4.2, 6.3], Math.floor); 
groupBy(['one', 'two', 'three'], 'length'); 

首先通過map計算出所有的鍵值，然後再根據建值進行歸類

片段十三：物件過濾

const pick = (obj, arr) =>
    arr.reduce((acc, curr) => (curr in obj && (acc[curr] = obj[curr]), acc), {});

pick({ a: 1, b: '2', c: 3 }, ['a', 'c']);

根據給出的鍵值來遍歷，比較物件中是否存在相同鍵值的的值，然後通過逗號表示式把賦值後的物件賦給下一個的初始值

片段十四：陣列中刪除指定位置的值

const remove = (arr, func) =>
    Array.isArray(arr)
    ? arr.filter(func).reduce((acc, val) => {
          arr.splice(arr.indexOf(val), 1);
          return acc.concat(val);
    }, []) : [];

const arr = [1, 2, 3, 4];
remove(arr, n => n % 2 == 0);

首先根據filter函式過濾出陣列中符合條件的值，然後使用reduce在原陣列中刪除符合條件的值，可以得出最後arr的值變成了[1, 3]

片段十五：promise按照順序執行

const runPromisesInSeries = ps => ps.reduce((p, next) => p.then(next), Promise.resolve());

const delay = d => new Promise(r => setTimeout(r, d));
const print = args => new Promise(r => r(args));
runPromisesInSeries([() => delay(1000), () => delay(2000), () => print('hello')]);

片段十六：排序

const orderBy = (arr, props, orders) =>
    [...arr].sort((a, b) =>
        props.reduce((acc, prop, i) => {
            if (acc === 0) {
                const [p1, p2] = orders && orders[i] === 'desc' ? [b[prop], a[prop]] : [a[prop], b[prop]];
                acc = p1 > p2 ? 1 : p1 < p2 ? -1 : 0;
            }
            return acc;
        }, 0)
    );

const users = [{ name: 'fred', age: 48 }, { name: 'barney', age: 36 }, { name: 'fly', age: 26 }];
orderBy(users, ['name', 'age'], ['asc', 'desc']); 
orderBy(users, ['name', 'age']);

片段十七：選擇

const select = (from, selector) =>
    selector.split('.').reduce((prev, cur) => prev && prev[cur], from);

const obj = { selector: { to: { val: 'val to select' } } };
select(obj, 'selector.to.val');

參考地址