Create inlabru data mask

Create a data mask for inlabru, which allows for evaluating expressions in the context of the data objects, with support for tidy evaluation pronouns and direct access to the data objects as .name.. This is an internal inlabru method, not intended for general use.

Usage

bru_data_mask(data, pronouns = NULL, objects = NULL)

Arguments

data

list of data objects in priority order. Named elements will be available as whole objects .name. as well as pronouns .name in evaluation.

Can include an object with evaluator functions such as those generated by bru_eval_fun(), names <label>_eval, one for each model component, <label>.

pronouns, objects

If NULL (default), all named elements of data are used as pronouns/objects. If character vector(s), only the matching named elements of data will be available as pronouns/objects.

Value

A data mask environment for use with rlang::eval_tidy().

Examples

m <- bru_data_mask(
  data = list(
    data = data.frame(z = 11:14),
    extra_data = list(something = 1:2)
  )
)
rlang::eval_tidy(
  rlang::quo(z + rep(.extra_data$something, 2)),
  data = m
)
#> [1] 12 14 14 16

mask <- bru_data_mask(list(data = list(x = 1:4)))
rlang::eval_tidy(rlang::quo(x), data = mask)
#> [1] 1 2 3 4
rlang::eval_tidy(rlang::quo(.data$x), data = mask)
#> [1] 1 2 3 4
rlang::eval_tidy(rlang::quo(.data.$x), data = mask)
#> [1] 1 2 3 4

# Using functions that need to call eval_tidy using the data mask:
fun_factory <- function() {
  .get_mask <- function(frame = parent.frame(2L)) {
    rlang::eval_tidy(rlang::parse_expr(".mask."), env = frame)
  }
  .eval_tidy <- function(expr, frame = parent.frame(2L)) {
    mask <- .get_mask(frame)
    rlang::eval_tidy(expr, data = mask, env = frame)
  }
  .addition <- 5L
  fun <- function(nm) {
    val <- .eval_tidy(rlang::expr(.data[[nm]]))
    val + .addition
  }
  list(fun = fun)
}
mask <- bru_data_mask(list(data = list(x = 1:4), fun = fun_factory()))
rlang::eval_tidy(rlang::quo(fun("x")), data = mask)
#> [1] 6 7 8 9
rlang::eval_tidy(rlang::quo(.fun$fun("x")), data = mask)
#> [1] 6 7 8 9
rlang::eval_tidy(rlang::quo(.fun.$fun("x")), data = mask)
#> [1] 6 7 8 9