r - Aggregating a data.frame with NAs using data.table -

i have large data.frame character column , several numeric columns contain na's.

here few example rows:

df <- data.frame(id=rep("a",3),v1=c(na,1,na),v2=c(2,5,2),v3=c(na,na,na),v4=c(0,0,0),stringsasfactors=f) 

since df$id repeats want aggregate df df$id, , apply sum other columns.

i did this:

require(data.table) setdt(df)[,lapply(.sd, function(x) sum(x,na.rm=t)),by=.(id)] 

and getting this:

   id v1 v2 v3 v4 1:   1  9  0  0 

so column v3 na's in df , hence gets value of 0, poses problem me since in such case i'd keep na value in other cases (where aggregation on mix of numerics , na's, i'd remove na's otherwise sum na). example shows (df$v4) have columns 0, therefore can't replace 0's na's aggregated data.frame.

in other words desired outcome:

   id v1 v2 v3 v4 1:   1  9  na 0 

any idea how data.table's .sd aggregation achieve this?

df[,lapply(.sd, function(x) ifelse(all(is.na(x)),na,sum(x,na.rm=t))),by=.(id)] id v1 v2 v3 v4 1:   1  9 na  0