媒介
近来在用Polymer封装纯数字的输入框，开辟过程当中发明不是坑，也有不少值得研讨的处所。本系列盘算分4篇来叙说这段可歌可泣的踩坑阅历：
《着手写个数字输入框3：痛点——输入法是个魔鬼》
《着手写个数字输入框4：魔鬼在细节——打磨光标位置》
从泉源抓起——阻拦不法字符
从《着手写个数字输入框1：input[type=number]的遗憾》中我们相识到input[type=number]基础不能满足我们的需求，为了简单化我们就直接在input[type=text]上加工出本身的数字输入框吧。
起首很明白的一点是终究数值能够包括以下字符[+-0-9.]，而可输入的功能键为Backspace,Delete,Arrow-Left,Arrow-Right,Arrow-Up,Arrow-Down和Tab。
因而我们能够设置以下划定规矩了
// 断言库
const keyCode = anyPass(prop('keyCode'), prop('which'))
const isBackspace = eq(8)
, isDelete = eq(46)
, isArrowLeft = eq(37)
, isArrowRight = eq(38)
, isArrowUp = eq(39)
, isArrowDown = eq(40)
, isTab = eq(9)
, isMinus = anyPass(eq(109), eq(189))
, isDot = anyPass(eq(110), eq(190))
, isDigit = anyPass(
allPass(lte(49), gte(57))
, allPass(lte(96), gte(105)))
, isPlus = anyPass(
comp(eq(107), keyCode)
, allPass(
prop('shiftKey')
, comp(eq(187), keyCode)))
const isValid = anyPass(
comp(
anyPass(isBackspace, isDelete, isArrowLeft
, isArrowLeft, isArrowUp, isArrowDown
, isTab, isMinus, isDot, isDigit)
, keyCode)
, isPlus)
$('input[type=text]').addEventListener('keydown', e => {
if (!isValid(e)){
e.preventDefault()
}
})扩展不法字符集
还记得min,max,precision吗？
当min大于即是0时，负号应该被归入不法字符；
当max小于0时，正号应该被归入不法字符；
当precision为0时，小数点应该被归入不法字符。因而我们增加以下划定规矩，并修正一下
isValid就好了
// 猎取min,max,precision值
const lensTarget = lens(a => a.target || a.srcElement)
, lensMin = lens(a => Number(a.min) || Number(attr(a, 'min')) || Number.MIN_SAFE_INTEGER)
, lensMax = lens(a => Number(a.max) || Number(attr(a, 'max')) || Number.MAX_SAFE_INTEGER)
, lensPrecision = lens(a => Number(a.precision) || Number(attr(a, 'precision')) || 0)
, lensValue = lens(a => a.value)
const lensTargetMin = lcomp(lensTarget, lensMin)
, lensTargetMax = lcomp(lensTarget, lensMax)
, lensTargetPrecision = lcomp(lensTarget, lensPrecision)
const isValid = anyPass(
comp(
anyPass(isBackspace, isDelete, isArrowLeft
, isArrowLeft, isArrowUp, isArrowDown
, isTab, isDigit)
, keyCode)
, allPass(
comp(gt(0), view(lensTargetMin))
, comp(isMinus, keyCode))
, allPass(
comp(lte(0), view(lensTargetMax))
, isPlus)
, allPass(
comp(lt(0), view(lensTargetPrecision))
, comp(isDot, keyCode)))预推断
到这里为止我们已成功地阻拦了种种不法字符，也就是终究值必需之含[+-0-9.]，但含这些字符跟团体相符数值花样就是两回事了。因而我们要继续补充下面两步，而且因为keydown事宜触发时value值还没被修正，因而我们需要将value值和当前输入值做组合来做预判，进一步扩展不法字符集。
经由过程正则搜检终究值是不是相符花样请求(是不是存在多个小数点也会在这一步处置惩罚掉);
搜检终究值是不是在
min和max局限内。
const isValidStr = precision =>
a => RegExp("^[+-]?[0-9]*"+ (precision ? "(\\.[0-9]{0," + precision + "})?" : "") + "$").test(a)
const lensValue = lens(a => a.value)
, lensTargetValue = lcomp(lensTarget, lensValue)
$('input[type=text]').addEventListener('keydown', e => {
var prevented = true
// 阻拦不法字符
if (isValid(e)){
prevented = false
// 预推断
if (anyPass(comp(anyPass(isMinus, isDigit, isDot), keyCode), isPlus)(e)){
var str = view(lensTargetValue)(e) + prop('key')(e)
// 预推断花样
prevented = !isValidStr(view(lensTargetPrecision)(e))(str)
// 预推断值局限
if (!prevented){
if (str == '-') str = '-0'
if (str == '+') str = '0'
if (str == '.') str = '0'
prevented = !allPass(
gte(view(lensTargetMax)(e))
, lte(view(lensTargetMin)(e)))(Number(str))
}
}
}
if (prevented){
e.preventDefault()
}
})附录：东西函数
// 东西函数，请疏忽我吧:D
const comp =
(...fns) =>
(...args) => {
let len = fns.length
while (len--){
args = [fns[len].apply(null, args)]
}
return args.length > 1 ? args : args[0]
}
const isSome = x => 'undefined' !== typeof x && x !== null
const invokerImpl =
n =>
o =>
m =>
(...args) => {
let args4m = args.splice(0, n)
, times = Number(args[0]) || 1
, ret = []
while (times--){
var tmpRet
try{
tmpRet = o[m].apply(o, args4m)
}
catch(e){
tmpRet = void 0
}
ret.push(tmpRet)
}
return ret.length > 1 ? ret : ret[0]
}
const curry2Partial =
fn =>
(...args) => {
let c = true
, i = 0
, l = args.length
, f = fn
for (;c && i < l; ++i){
c = isSome(args[i])
if (c){
f = f(args[i])
}
}
return f
}
const invoker = curry2Partial(invokerImpl)
const and = (...args) => args.reduce((accu, x) => accu && x, true)
const or = (...args) => args.reduce((accu, x) => accu || x, false)
const allPass = (...fns) => v => fns.reduce((accu, x) => accu && x(v), true)
const anyPass = (...fns) => v => fns.reduce((accu, x) => accu || x(v), false)
const eq = a => b => a === b
const gt = a => b => a > b
const gte = a => anyPass(eq(a), gt(a))
const lt = a => b => a < b
const lte = a => anyPass(eq(a), lt(a))
const prop = k => o => o[k]
const lens = (g, s) => ({getter: g, setter: s})
const lensPath = (...args) => ({ getter: a => args.reduce((accu, x) => accu && accu[x], a) })
const lcomp = (...lenses) => ({ getter: a => lenses.reduce((accu, lens) => accu && lens.getter(accu), a)})
const view = l => a => l.getter(a)
const $ = invoker(1, document, "querySelector")
const attr = (o, a) => invoker(1, o, 'getAttribute')(a)总结
如今能够终究能够紧紧控制住用户输入了，直到用户切换到IME为止:-<当运用IME输入时会发明上述步伐一点用也没有，不必皱眉了，背面我们会一同把IME KO掉！
尊敬原创，转载请说明来自：http://www.cnblogs.com/fsjohn… ^_^肥仔John
