使用Ramda检测序列中的空位？

Question

In ramda.js, how can I detect if there is a gap in a sequence bigger than some value n?

eg for values that have already been sorted:

[
    {"name":"bob", loginDate:"2017-10-01"},
    {"name":"bob", loginDate:"2017-10-02"},
    {"name":"bob", loginDate:"2017-10-03"},
    {"name":"bob", loginDate:"2017-10-10"},
]

How would I detect if there has been a gap greater than 7 between the loginDates of any two records? But not between the current and the first records - only checking against the previous one. I'm not sure how to compare the current item to the previous item in ramda.

Answer 1

When wanting to process adjacent values in a list it is often useful to create a list that can be traversed like a sliding window of values using R.aperture .

A binary function that performs the comparison between the previous and current items of a list can be wrapped in R.apply so that it will accept an array of two elements to compare.

To illustrate with your example:

const maxGap = 1000 * 60 * 60 * 24 * 7 // 7 days in ms
const gtMaxGap = (prev, curr) =>
  maxGap + Date.parse(prev.loginDate) < Date.parse(curr.loginDate)

const anyGapTooLarge = R.pipe(R.aperture(2), R.any(R.apply(gtMaxGap)))

anyGapTooLarge([
  {"name":"bob", loginDate:"2017-10-01"},
  {"name":"bob", loginDate:"2017-10-02"},
  {"name":"bob", loginDate:"2017-10-03"},
  {"name":"bob", loginDate:"2017-10-10"},
]) // false

Another alternative to R.aperture would be to produce the list of adjacent values by using R.zip to zip the list with its own tail.

Answer 2

As a student of functional programming, I worked on this problem.

I really like the answer by @Scott Christopher; yet I manage work out a different solution before looking at what he did.

That approach follows:

I worked through this a different way. If you want to have the a list of all logins after a given time period and note the difference in time, you could use essentially use reduce then filter for the gap

First, assume the data is contained sorted in a variable called data

example

const logs = [
  {"name":"bob", loginDate:"2017-10-01"},
  {"name":"bob", loginDate:"2017-10-02"},
  {"name":"bob", loginDate:"2017-10-03"},
  {"name":"bob", loginDate:"2017-10-10"},
]

Next, we create a few utility functions and an iterator function

const dateDifference = nextDate => initDate => {
  return ( new Date( nextDate ) - new Date( initDate ) )
}
const daysToMs = R.multiply(1000 * 60 * 60 * 24)

const differenceWithPreviousRecordIterator = dateField => differenceField => (acc,val) => {
  const timeDifference = acc.length  
    ? dateDifference
        (val[dateField])
        (R.last(acc)[dateField])
    : 0

  return R.append(
    R.assoc( differenceField, timeDifference, val ),
    acc
  )
}

Now configure the iterator function

const iteratorFunc = differenceWithPreviousRecordIterator('loginDate')('difference')

Then get the rows and filter for rows where logins occurred after the desired time

const rowsWithDifference = R.reduce(iteratorFunc,[])(logs)
const sevenDaysAndOlder = R.where({
  difference: R.gte( R.__, daysToMs(7) )
})
const filterForLoginsSevenDaysOrGreater = R.filter( sevenDaysAndOlder )

filterForLoginsSevenDaysOrGreater( rowsWithDifference )
// [ { name: 'bob', loginDate: '2017-10-10', difference: 604800000 } ]

---

That said, after looking @Scott's approach, a similar effect can be achieved by slightly modifying his solution. The modifications being namely that we're looking for a time equal to or greater than the gap instead of just greater than. The other is to simply use filter instead of any

const maxGap = 1000 * 60 * 60 * 24 * 7 // 7 days in ms
const gtMaxGap = (prev, curr) =>
  maxGap + Date.parse(prev.loginDate) <= Date.parse(curr.loginDate)

const anyGapTooLarge = R.pipe(R.aperture(2), R.filter( R.apply( gtMaxGap ) ))

anyGapTooLarge([
  {"name":"bob", loginDate:"2017-10-01"},
  {"name":"bob", loginDate:"2017-10-02"},
  {"name":"bob", loginDate:"2017-10-03"},
  {"name":"bob", loginDate:"2017-10-10"},
]) // gives pair where login was greater than or equal to maxGap