我将从完成的解决方案开始......
这个答案的最后有一个很大的解释。让我们先尝试着想一下大局。
readdirp('.')
.fmap(filter(match(/\.json$/)))
.fmap(map(readfilep))
.fmap(map(fmap(JSON.parse)))
.fmap(concatp)
.fmap(flatten)
.fmap(reduce(createMap)({}))
.fmap(data=> JSON.stringify(data, null, '\t'))
.fmap(writefilep(resolve(__dirname, 'result.json')))
.then(filename=> console.log('wrote results to %s', filename), err=>console.error(err));
控制台输出
wrote results to /path/to/result.json
result.json（我添加了一个c.json一些数据表明这适用于 2 个以上的文件）
{
"addEmoticon1": "Hello, {name}!",
"addPhoto1": "How are you??",
"close1": "Close!",
"somethingelse": "Something!"
}
执行
I made Promise基于接口readdir and readFile and writeFile
import {readdir, readFile, writeFile} from 'fs';
const readdirp = dir=>
new Promise((pass,fail)=>
readdir(dir, (err, filenames) =>
err ? fail(err) : pass(mapResolve (dir) (filenames))));
const readfilep = path=>
new Promise((pass,fail)=>
readFile(path, 'utf8', (err,data)=>
err ? fail(err) : pass(data)));
const writefilep = path=> data=>
new Promise((pass,fail)=>
writeFile(path, data, err=>
err ? fail(err) : pass(path)));
为了将函数映射到我们的 Promise，我们需要一个fmap公用事业。注意我们如何注意冒泡错误。
Promise.prototype.fmap = function fmap(f) {
return new Promise((pass,fail) =>
this.then(x=> pass(f(x)), fail));
};
这是其余的实用程序
const fmap = f=> x=> x.fmap(f);
const mapResolve = dir=> map(x=>resolve(dir,x));
const map = f=> xs=> xs.map(x=> f(x));
const filter = f=> xs=> xs.filter(x=> f(x));
const match = re=> s=> re.test(s);
const concatp = xs=> Promise.all(xs);
const reduce = f=> y=> xs=> xs.reduce((y,x)=> f(y)(x), y);
const flatten = reduce(y=> x=> y.concat(Array.isArray(x) ? flatten (x) : x)) ([]);
最后，一个可以完成您工作的自定义函数
const createMap = map=> ({id, defaultMessage})=>
Object.assign(map, {[id]: defaultMessage});
这是c.json
[
{
"id": "somethingelse",
"description": "something",
"defaultMessage": "Something!"
}
]
“为什么有这么多小功能？”
不管你怎么想，你都有一个很大的问题。大问题是通过组合几个小解决方案来解决的。该代码最突出的优点是每个函数都有非常独特的目的，并且对于相同的输入它总是会产生相同的结果。这意味着每个函数都可以在程序中的其他地方使用。另一个优点是较小的函数更容易阅读、推理和调试。
将所有这些与此处给出的其他答案进行比较； @BlazeSahlen 特别是。这有 60 多行代码，基本上只能用于解决这一特定问题。它甚至不会过滤掉非 JSON 文件。因此，下次您需要创建一系列读/写文件的操作时，您每次都必须重写这 60 行中的大部分。由于样板文件耗尽，它会产生大量重复的代码和难以发现的错误。以及所有手动错误处理...哇，现在就杀了我吧。他/她认为回调地狱很糟糕？哈哈，他/她刚刚又自己创造了另一个地狱圈。
所有代码放在一起...
函数（大致）按其使用顺序出现
import {readdir, readFile, writeFile} from 'fs';
import {resolve} from 'path';
// logp: Promise<Value> -> Void
const logp = p=> p.then(x=> console.log(x), x=> console.err(x));
// fmap : Promise<a> -> (a->b) -> Promise<b>
Promise.prototype.fmap = function fmap(f) {
return new Promise((pass,fail) =>
this.then(x=> pass(f(x)), fail));
};
// fmap : (a->b) -> F<a> -> F<b>
const fmap = f=> x=> x.fmap(f);
// readdirp : String -> Promise<Array<String>>
const readdirp = dir=>
new Promise((pass,fail)=>
readdir(dir, (err, filenames) =>
err ? fail(err) : pass(mapResolve (dir) (filenames))));
// mapResolve : String -> Array<String> -> Array<String>
const mapResolve = dir=> map(x=>resolve(dir,x));
// map : (a->b) -> Array<a> -> Array<b>
const map = f=> xs=> xs.map(x=> f(x));
// filter : (Value -> Boolean) -> Array<Value> -> Array<Value>
const filter = f=> xs=> xs.filter(x=> f(x));
// match : RegExp -> String -> Boolean
const match = re=> s=> re.test(s);
// readfilep : String -> Promise<String>
const readfilep = path=>
new Promise((pass,fail)=>
readFile(path, 'utf8', (err,data)=>
err ? fail(err) : pass(data)));
// concatp : Array<Promise<Value>> -> Array<Value>
const concatp = xs=> Promise.all(xs);
// reduce : (b->a->b) -> b -> Array<a> -> b
const reduce = f=> y=> xs=> xs.reduce((y,x)=> f(y)(x), y);
// flatten : Array<Array<Value>> -> Array<Value>
const flatten = reduce(y=> x=> y.concat(Array.isArray(x) ? flatten (x) : x)) ([]);
// writefilep : String -> Value -> Promise<String>
const writefilep = path=> data=>
new Promise((pass,fail)=>
writeFile(path, data, err=>
err ? fail(err) : pass(path)));
// -----------------------------------------------------------------------------
// createMap : Object -> Object -> Object
const createMap = map=> ({id, defaultMessage})=>
Object.assign(map, {[id]: defaultMessage});
// do it !
readdirp('.')
.fmap(filter(match(/\.json$/)))
.fmap(map(readfilep))
.fmap(map(fmap(JSON.parse)))
.fmap(concatp)
.fmap(flatten)
.fmap(reduce(createMap)({}))
.fmap(data=> JSON.stringify(data, null, '\t'))
.fmap(writefilep(resolve(__dirname, 'result.json')))
.then(filename=> console.log('wrote results to %s', filename), err=>console.error(err));
仍然遇到困难吗？
一开始要了解这些东西是如何工作的并不容易。这是一个特别奇怪的问题，因为数据嵌套得非常快。值得庆幸的是，这并不意味着我们的代码必须是一大堆嵌套混乱才能解决问题！请注意，即使我们正在处理诸如 JSON Promises 数组的 Promise 之类的事情，代码仍然保持良好且平坦的状态...
// Here we are reading directory '.'
// We will get a Promise<Array<String>>
// Let's say the files are 'a.json', 'b.json', 'c.json', and 'run.js'
// Promise will look like this:
// Promise<['a.json', 'b.json', 'c.json', 'run.js']>
readdirp('.')
// Now we're going to strip out any non-JSON files
// Promise<['a.json', 'b.json', 'c.json']>
.fmap(filter(match(/\.json$/)))
// call `readfilep` on each of the files
// We will get <Promise<Array<Promise<JSON>>>>
// Don't freak out, it's not that bad!
// Promise<[Promise<JSON>, Promise<JSON>. Promise<JSON>]>
.fmap(map(readfilep))
// for each file's Promise, we want to parse the data as JSON
// JSON.parse returns an object, so the structure will be the same
// except JSON will be an object!
// Promise<[Promise<Object>, Promise<Object>, Promise<Object>]>
.fmap(map(fmap(JSON.parse)))
// Now we can start collapsing some of the structure
// `concatp` will convert Array<Promise<Value>> to Array<Value>
// We will get
// Promise<[Object, Object, Object]>
// Remember, we have 3 Objects; one for each parsed JSON file
.fmap(concatp)
// Your particular JSON structures are Arrays, which are also Objects
// so that means `concatp` will actually return Promise<[Array, Array, Array]
// but we'd like to flatten that
// that way each parsed JSON file gets mushed into a single data set
// after flatten, we will have
// Promise<Array<Object>>
.fmap(flatten)
// Here's where it all comes together
// now that we have a single Promise of an Array containing all of your objects ...
// We can simply reduce the array and create the mapping of key:values that you wish
// `createMap` is custom tailored for the mapping you need
// we initialize the `reduce` with an empty object, {}
// after it runs, we will have Promise<Object>
// where Object is your result
.fmap(reduce(createMap)({}))
// It's all downhill from here
// We currently have Promise<Object>
// but before we write that to a file, we need to convert it to JSON
// JSON.stringify(data, null, '\t') will pretty print the JSON using tab to indent
// After this, we will have Promise<JSON>
.fmap(data=> JSON.stringify(data, null, '\t'))
// Now that we have a JSON, we can easily write this to a file
// We'll use `writefilep` to write the result to `result.json` in the current working directory
// I wrote `writefilep` to pass the filename on success
// so when this finishes, we will have
// Promise<Path>
// You could have it return Promise<Void> like writeFile sends void to the callback. up to you.
.fmap(writefilep(resolve(__dirname, 'result.json')))
// the grand finale
// alert the user that everything is done (or if an error occurred)
// Remember `.then` is like a fork in the road:
// the code will go to the left function on success, and the right on failure
// Here, we're using a generic function to say we wrote the file out
// If a failure happens, we write that to console.error
.then(filename=> console.log('wrote results to %s', filename), err=>console.error(err));
全做完了 ！
