I’ve been trying to clean up a CSV file which contains some rows with dates and some not – I only want to keep the cells which do have dates so I’ve been trying to work out how to do that.

My first thought was that I’d try and find a function which would convert the contents of the cell into a date if it was in date format and NA if not. I could then filter out the NA values using the is.na function.

I started out with the as.Date function…

> as.Date("2014-01-01")
[1] "2014-01-01"
 
> as.Date("foo")
Error in charToDate(x) : 
  character string is not in a standard unambiguous format

…but that throws an error if we have a non date value so it’s not so useful in this case.

Instead we can make use of the strptime function which does exactly what we want:

> strptime("2014-01-01", "%Y-%m-%d")
[1] "2014-01-01 GMT"
 
> strptime("foo", "%Y-%m-%d")
[1] NA

We can then feed those values into is.na..

> strptime("2014-01-01", "%Y-%m-%d") %>% is.na()
[1] FALSE
 
> strptime("foo", "%Y-%m-%d") %>% is.na()
[1] TRUE

…and we have exactly the behaviour we were looking for.