ls
我的建议是不要把所有的鸡蛋都放在一个篮子里。ls
我们可以使用 Node 的fs.Dirent对象编写一个超快速的函数,并绕过对每个文件进行慢速fs.stat
调用的需要——
// fsext.js
import { readdir } from "fs/promises"
import { join } from "path"
async function* ls (path = ".")
{ yield { dir: path }
for (const dirent of await readdir(path, { withFileTypes: true }))
if (dirent.isDirectory())
yield *ls(join(path, dirent.name))
else
yield { file: join(path, dirent.name) }
}
async function toArray (iter)
{ const r = []
for await (const v of iter)
r.push(v)
return r
}
export { ls, toArray }
// main.js
import { ls, toArray } from "./fsext.js"
toArray(ls("./node_modules")).then(console.log, console.error)
为了测试它,让我们添加一些流行的npm
包,以便我们有一个大的层次结构来测试我们的程序。我们将安装很多并计算目录和文件的数量 -
$ npm install async chalk commander debug express immutable lodash moment prop-types react react-dom request webpack
$ find ./node_modules | wc -l
5453
现在让我们运行我们的程序,time
它 -
$ time node main.js
[
{ dir: './node_modules' },
{ dir: 'node_modules/.bin' },
{ file: 'node_modules/.bin/acorn' },
{ file: 'node_modules/.bin/browserslist' },
{ file: 'node_modules/.bin/loose-envify' },
{ file: 'node_modules/.bin/mime' },
{ file: 'node_modules/.bin/sshpk-conv' },
{ file: 'node_modules/.bin/sshpk-sign' },
{ file: 'node_modules/.bin/sshpk-verify' },
{ file: 'node_modules/.bin/terser' },
{ file: 'node_modules/.bin/uuid' },
{ file: 'node_modules/.bin/webpack' },
{ file: 'node_modules/.package-lock.json' },
{ dir: 'node_modules/@types' },
{ dir: 'node_modules/@types/eslint' },
{ file: 'node_modules/@types/eslint/LICENSE' },
{ file: 'node_modules/@types/eslint/README.md' },
{ file: 'node_modules/@types/eslint/helpers.d.ts' },
{ file: 'node_modules/@types/eslint/index.d.ts' },
{ dir: 'node_modules/@types/eslint/lib' },
... 5433 more items
]
node main.js 0.09s user 0.02s system 116% cpu 0.099 total
目录
如果我们只想要目录,我们可以写成dirs
我们泛型的简单特化ls
——
// fsext.js (continued)
async function* dirs (path)
{ for await (const f of ls(path))
if (f.dir)
yield f.dir
}
$ find ./node_modules -type d | wc -l
457
现在将它与我们的程序进行比较
// main.js
import { dirs, toArray } from "./fsext.js"
toArray(dirs("./node_modules")).then(console.log, console.error)
$ time node.main.js
[
'./node_modules',
'node_modules/.bin',
'node_modules/@types',
'node_modules/@types/eslint',
'node_modules/@types/eslint/lib',
'node_modules/@types/eslint/lib/rules',
'node_modules/@types/eslint/rules',
'node_modules/@types/eslint-scope',
'node_modules/@types/estree',
'node_modules/@types/json-schema',
'node_modules/@types/node',
'node_modules/@types/node/assert',
'node_modules/@types/node/dns',
'node_modules/@types/node/fs',
'node_modules/@types/node/stream',
'node_modules/@types/node/timers',
'node_modules/@types/node/ts3.6',
'node_modules/@webassemblyjs',
'node_modules/@webassemblyjs/ast',
'node_modules/@webassemblyjs/ast/esm',
... 437 more items
]
node main2.js 0.09s user 0.02s system 108% cpu 0.099 total
排除
如果我们想要exclude
某些目录或文件,我们也可以通用地编写它 -
// fsext.js (continued)
async function* exclude (iter, test)
{ for await (const v of iter)
if (Boolean(test(v)))
continue
else
yield v
}
// main.js
import { dirs, exclude, toArray } from "./fsext.js"
toArray(exclude(dirs("./node_modules"), v => /@/.test(v)))
.then(console.log, console.error)
$ time node main.js
[
'./node_modules',
'node_modules/.bin',
'node_modules/accepts',
'node_modules/acorn',
'node_modules/acorn/bin',
'node_modules/acorn/dist',
'node_modules/ajv',
'node_modules/ajv/dist',
'node_modules/ajv/lib',
'node_modules/ajv/lib/compile',
'node_modules/ajv/lib/dot',
'node_modules/ajv/lib/dotjs',
'node_modules/ajv/lib/refs',
'node_modules/ajv/scripts',
'node_modules/ajv-keywords',
'node_modules/ajv-keywords/keywords',
'node_modules/ajv-keywords/keywords/dot',
'node_modules/ajv-keywords/keywords/dotjs',
'node_modules/ansi-styles',
'node_modules/array-flatten',
... 351 more items
]
node main.js 0.09s user 0.02s system 105% cpu 0.104 total
改组
在我们的文件系统扩展模块中fsext
,我们编写了两个可用于任何可迭代对象的函数,而不仅仅是ls
or dirs
。我建议将这些分解成他们自己的iter
模块。这种类型的重组有助于在整个程序中解耦关注点并最大限度地重用代码 -
// iter.js
async function* empty () {}
async function* exclude (iter = empty(), test = Boolean)
{ for await (const v of iter)
if (Boolean(test(v)))
continue
else
yield v
}
async function toArray (iter = empty())
{ const r = []
for await (const v of iter)
r.push(v)
return r
}
export { empty, exclude, toArray }
// fsext.js
import { readdir } from "fs/promises"
import { join } from "path"
async function* ls (path = ".")
{ yield { dir: path }
for (const dirent of await readdir(path, { withFileTypes: true }))
if (dirent.isDirectory())
yield *ls(join(path, dirent.name))
else
yield { file: join(path, dirent.name) }
}
async function* dirs (path)
{ for await (const f of ls(path))
if (f.dir)
yield f.dir
}
async function* files (path)
{ for await (const f of ls(path))
if (f.file)
yield f.file
}
export { ls, dirs, files }
// main.js
import { dirs } from "./fsext.js"
import { exclude, toArray } from "./iter.js"
const somePath = "..."
const someTest = v => ...
toArray(exclude(dirs(somePath), someTest))
.then(console.log, console.error)