[R] rle with NA values?

[Justin Haynes]

Happy Friday!

Using this function:

fixSeq <- function(df) {
  shift1 <- function(x) c(1, x[-length(x)])
  df$state_shift<-df$state
  df.rle<-rle(df$state_shift)
  repeat {
    shifted.sf<-shift1(df.rle$values)
    change <- df.rle$values >= 4 & shifted.sf >= 4 & shifted.sf != df.rle$values
    if(any(change))
        df.rle$values[change] <- shifted.sf[change] else break
    }
    gc()
  df$state_shift<-inverse.rle(df.rle)
  return(df)
}

I would like to separate runs where the removed NAs will separate runs
into two separate runs.
to illustrate with a short example:

> dat<-data.frame(id=1,state=c(1,2,4,4,5,NA,5,5,1))
>
> fixSeq(dat)
Error in df.rle$values[change] <- shifted.sf[change] :
  NAs are not allowed in subscripted assignments
>
> fixSeq(na.omit(dat))
  id state state_shift
1  1     1           1
2  1     2           2
3  1     4           4
4  1     4           4
5  1     5           4
7  1     5           4
8  1     5           4
9  1     1           1
>

rather than the true output of 1 2 4 4 4 5 5 1.  The NA makes the
second pair of 5s a unique state rather than a continuation of the
previous state 4.  Is this best accomplished by assigning NA to a
value like -99?  or do I have other options?