Refactor do_j_names to reuse logic from setDT (#6702)

[yuzhengcong]

Hi MichaelChirico,

I’ve been working on refactoring the duplicated logic in the setDT function and the [,:=] operation as discussed in issue #6702.

I have extracted common code into new simple_extract() function.
Here is the code of simple_extract();

simple_extract <- function(name, x, checkSingleChar = TRUE, envir = parent.frame()) {
        k = eval(name[[2L]], envir, envir)
        if (is.list(k)) {
          origj = j = if (name[[1L]] == "$") {
            as.character(name[[3L]])
          } else {
            eval(name[[3L]], envir, envir)
          }
          if (is.character(j)) {
            if (checkSingleChar) {
              if (length(j) != 1L) {
                stopf("Cannot assign to an under-allocated recursively indexed list -- L[[i]][,:=] syntax is only valid when i is length 1, but its length is %d", length(j))
              }
              j2 = match(j, names(k))
              if (is.na(j2)) {
                internal_error("item '%s' not found in names of list", origj)
              }
              j = j2
            } else {
              if (length(j) == 1L) {
                j2 = match(j, names(k))
                if (is.na(j2)) {
                  stopf("Item '%s' not found in names of input list", origj)
                }
                j = j2
              }
            }
          }
          .Call(Csetlistelt, k, as.integer(j), x)
        } else if (is.environment(k) && exists(as.character(name[[3L]]), k)) {
          assign(as.character(name[[3L]]), x, k, inherits = FALSE)
        } else if (isS4(k)) {
          .Call(CsetS4elt, k, as.character(name[[3L]]), x)
        }
      }

Best,
Paligi

[MichaelChirico]

try and match the original style of the code, it's much harder to read when the diff is larger than necessary

[MichaelChirico]

in both branches, we do the j2 = match() and if (is.na()) checks -- move them outside checkSingleChar. Try and think through if that applies elsewhere also.

[venom1204]

hi @Paligi
Just checking in—any updates on the testing process? Let me know if you need anything from my side to move things forward.

[MichaelChirico]

Closing in favor of #7064, which is nearly ready for merge and addresses the same call sites.