R custom data.table function with multiple variable inputs

Question

I am writing a custom aggregation function with data.table (v 1.9.6) and struggle to pass function arguments to it. there have been similar questions on this but none deals

Answer 1

Looks like a question to me :)
I prefer computing on the language over get/mget.

fn_agg = function(dt, var_list, var_name_list, by_var_list, order_var_list) {
    j_call = as.call(c(
        as.name("."),
        sapply(setNames(var_list, var_name_list), function(var) as.call(list(as.name("sum"), as.name(var), na.rm=TRUE)), simplify=FALSE)
    ))
    order_call = as.call(c(
        as.name("order"),
        lapply(order_var_list, as.name)
    ))
    j2_call = as.call(c(
        as.name(":="),
        c(
            sapply(setNames(var_name_list, paste0(var_name_list,"_del")), function(var) {
                substitute(.var - shift(x = .var, n = 1, type = "lag"), list(.var=as.name(var)))
            }, simplify=FALSE),
            sapply(setNames(var_name_list, paste0(var_name_list,"_del_rel")), function(var) {
                substitute((.var - shift(x = .var, n = 1, type = "lag")) / (shift(x = .var, n = 1, type = "lag")), list(.var=as.name(var)))
            }, simplify=FALSE)
        )
    ))
    dt[eval(order_call), eval(j_call), by=by_var_list
       ][, eval(j2_call)
         ][]
}

ans = fn_agg(dt, var_list=c("x","y"), var_name_list=c("x_sum","y_sum"), by_var_list=c("a","b"), order_var_list=c("a","b"))
all.equal(temp2, ans)
#[1] TRUE

Some extra notes:

1. make strict input validation as debugging issues is more difficuilt against meta programming.
2. optimization of step2 is possible as shift is computed multiple times, easy way is just to compute _del in step2 and _del_rel in step3.
3. if order variables is always the same as by variables you can put them into keyby argument.