数组可以代替字符串来处理嵌套对象和数组，例如:["my_field"， "another_field"， 0， "last_field"， 10]

下面是一个基于该数组表示方式更改字段的示例。我在react.js中使用类似的东西来控制输入字段，改变嵌套结构的状态。

let state = {
        test: "test_value",
        nested: {
            level1: "level1 value"
        },
        arr: [1, 2, 3],
        nested_arr: {
            arr: ["buh", "bah", "foo"]
        }
    }

function handleChange(value, fields) {
    let update_field = state;
    for(var i = 0; i < fields.length - 1; i++){
        update_field = update_field[fields[i]];
    }
    update_field[fields[fields.length-1]] = value;
}

handleChange("update", ["test"]);
handleChange("update_nested", ["nested","level1"]);
handleChange(100, ["arr",0]);
handleChange('changed_foo', ["nested_arr", "arr", 3]);
console.log(state);

在这里我提供了更多的方法，这些方法在很多方面看起来都更快:

选项1:Split string on。Or [Or] Or ' Or "，颠倒过来，跳过空项。

function getValue(path, origin) {
    if (origin === void 0 || origin === null) origin = self ? self : this;
    if (typeof path !== 'string') path = '' + path;
    var parts = path.split(/\[|\]|\.|'|"/g).reverse(), name; // (why reverse? because it's usually faster to pop off the end of an array)
    while (parts.length) { name=parts.pop(); if (name) origin=origin[name]; }
    return origin;
}

选项2(最快的，除了eval):低级字符扫描(没有regex/split/等等，只是一个快速字符扫描)。 注意:该命令不支持索引引用。

function getValue(path, origin) {
    if (origin === void 0 || origin === null) origin = self ? self : this;
    if (typeof path !== 'string') path = '' + path;
    var c = '', pc, i = 0, n = path.length, name = '';
    if (n) while (i<=n) ((c = path[i++]) == '.' || c == '[' || c == ']' || c == void 0) ? (name?(origin = origin[name], name = ''):(pc=='.'||pc=='['||pc==']'&&c==']'?i=n+2:void 0),pc=c) : name += c;
    if (i==n+2) throw "Invalid path: "+path;
    return origin;
} // (around 1,000,000+/- ops/sec)

选项3:(新:选项2扩展到支持引号-有点慢，但仍然很快)

function getValue(path, origin) {
    if (origin === void 0 || origin === null) origin = self ? self : this;
    if (typeof path !== 'string') path = '' + path;
    var c, pc, i = 0, n = path.length, name = '', q;
    while (i<=n)
        ((c = path[i++]) == '.' || c == '[' || c == ']' || c == "'" || c == '"' || c == void 0) ? (c==q&&path[i]==']'?q='':q?name+=c:name?(origin?origin=origin[name]:i=n+2,name='') : (pc=='['&&(c=='"'||c=="'")?q=c:pc=='.'||pc=='['||pc==']'&&c==']'||pc=='"'||pc=="'"?i=n+2:void 0), pc=c) : name += c;
    if (i==n+2 || name) throw "Invalid path: "+path;
    return origin;
}

JSPerf： http://jsperf.com/ways-to-dereference-a-delimited-property-string/3

"eval(...)" is still king though (performance wise that is). If you have property paths directly under your control, there shouldn't be any issues with using 'eval' (especially if speed is desired). If pulling property paths "over the wire" (on the line!? lol :P), then yes, use something else to be safe. Only an idiot would say to never use "eval" at all, as there ARE good reasons when to use it. Also, "It is used in Doug Crockford's JSON parser." If the input is safe, then no problems at all. Use the right tool for the right job, that's it.

请注意，以下选项并不适用于所有有效的unicode属性名(但据我所知，其他选项也不适用)。

const PATTERN = /[\^|\[|\.]([$|\w]+)/gu function propValue(o, s) { const names = [] for(let [, name] of [...s.matchAll(PATTERN)]) names.push(name) return names.reduce((p, propName) => { if(!p.hasOwnProperty(propName)) throw 'invalid property name' return p[propName] }, o) } let path = 'myObject.1._property2[0][0].$property3' let o = { 1: { _property2: [ [{ $property3: 'Hello World' }] ] } } console.log(propValue(o, path)) // 'Hello World'

使用eval:

var part1name = eval("someObject.part1.name");

换行以在错误时返回undefined

function path(obj, path) {
    try {
        return eval("obj." + path);
    } catch(e) {
        return undefined;
    }
}

http://jsfiddle.net/shanimal/b3xTw/

在使用评估的权力时，请使用常识和谨慎。它有点像光剑，如果你打开它，有90%的几率你会切断一个肢体。并不是每个人都适合。

我只是基于一些类似的代码，我已经有了，它似乎工作:

Object.byString = function(o, s) {
    s = s.replace(/\[(\w+)\]/g, '.$1'); // convert indexes to properties
    s = s.replace(/^\./, '');           // strip a leading dot
    var a = s.split('.');
    for (var i = 0, n = a.length; i < n; ++i) {
        var k = a[i];
        if (k in o) {
            o = o[k];
        } else {
            return;
        }
    }
    return o;
}

用法::

Object.byString(someObj, 'part3[0].name');

在http://jsfiddle.net/alnitak/hEsys/上可以看到一个工作演示

有些人注意到，如果传递的字符串中最左边的索引与对象中正确嵌套的条目不对应，这段代码将抛出错误。这是一个有效的问题，但我认为最好在调用时使用try / catch块来解决，而不是让这个函数对于无效的索引无声地返回undefined。

你必须自己解析字符串:

function getProperty(obj, prop) {
    var parts = prop.split('.');

    if (Array.isArray(parts)) {
        var last = parts.pop(),
        l = parts.length,
        i = 1,
        current = parts[0];

        while((obj = obj[current]) && i < l) {
            current = parts[i];
            i++;
        }

        if(obj) {
            return obj[last];
        }
    } else {
        throw 'parts is not valid array';
    }
}

这要求你也用点表示法定义数组下标:

var part3name1 = "part3.0.name";

它使解析更容易。

DEMO