Data frame into nested list (tree)

You're describing a tree structure: each parent can have multiple children, but each child has exactly one parent. Searching for tree questions, this is similar in principle to Transform a dataframe into a tree structure list of lists, in the sense that we can start from the root and recurse over children. However, as your input and output data are in a different format, we can do something a little more ergonomic. Basically, for each node, create a list with text = node_name, and if there are children, recursively build a subtree for them.

generate_tree <- function(dat, root = dat$Child[is.na(dat$ParentID)]) {
    node <- list(text = root)
    children <- na.omit(dat$Child[dat$ParentID == root])
    if (length(children) > 0) {
        node$children <- lapply(children, \(child) generate_tree(dat, child))
    }
    node
}

If you really want the outer list() you can wrap it in one:

identical(list(generate_tree(roles_data)), desired_output) 
# [1] TRUE

And here is the output as json for printing purposes:

generate_tree(roles_data)  |> jsonlite::toJSON(pretty = TRUE, auto_unbox = TRUE)
{
  "text": "A",
  "children": [
    {
      "text": "B",
      "children": [
        {
          "text": "C"
        },
        {
          "text": "D",
          "children": [
            {
              "text": "E"
            },
            {
              "text": "F"
            }
          ]
        }
      ]
    }
  ]
}