enw_formula_as_data_list()s implementation could be improved to better align with DRY principles

[pearsonca]

enw_formula_as_data_list is a bit convoluted and breaks the DRY (don't repeat yourself) principle:

epinowcast/R/model-tools.R

Lines 32 to 71 in c1d9bee

	enw_formula_as_data_list <- function(formula, prefix,
	drop_intercept = FALSE) {
	paste_lab <- function(string, lab = prefix) {
	paste0(lab, "_", string)
	}
	if (!missing(formula)) {
	if (!inherits(formula, "enw_formula")) {
	stop(
	"formula must be an object of class enw_formula as produced using
	enw_formula"
	)
	}
	fintercept <- as.numeric(any(grepl(
	"(Intercept)", colnames(formula$fixed$design), fixed = TRUE
	)))
	data <- list()
	data[[paste_lab("fdesign")]] <- formula$fixed$design
	data[[paste_lab("fintercept")]] <- fintercept
	data[[paste_lab("fnrow")]] <- nrow(formula$fixed$design)
	data[[paste_lab("findex")]] <- formula$fixed$index
	data[[paste_lab("fnindex")]] <- length(formula$fixed$index)
	data[[paste_lab("fncol")]] <-
	ncol(formula$fixed$design) -
	min(as.numeric(drop_intercept), as.numeric(fintercept))
	data[[paste_lab("rdesign")]] <- formula$random$design
	data[[paste_lab("rncol")]] <- ncol(formula$random$design) - 1
	} else {
	data <- list()
	data[[paste_lab("fdesign")]] <- numeric(0)
	data[[paste_lab("fintercept")]] <- 0
	data[[paste_lab("fnrow")]] <- 0
	data[[paste_lab("findex")]] <- numeric(0)
	data[[paste_lab("fnindex")]] <- 0
	data[[paste_lab("fncol")]] <- 0
	data[[paste_lab("rdesign")]] <- numeric(0)
	data[[paste_lab("rncol")]] <- 0
	}
	return(data)
	}

Seems would be clearer as roughly or similar:

# ...initial checking stuff ...
data <- list()
data[c("fintercept", "the other pertinent variables")] <- 0
data[c("fdesign", "these other pertinent variables") <- list(numeric())
if (!missing(formula)) {
  data$fdesign <- formula$fixed$design
   # ... etc
}
names(data) <- sprintf("%s_%s", prefix, names(data))
return(data)

Updating this would make the code easier to maintain and expand in the future.

[seabbs]

Thanks for this @pearsonca

I agree the current implementation can be shortened though and this looks like a workable way of doing it.

Would you like to take this on? If we act fast then we can get in it 0.2.1. Also sharing more widely as this is a good starter issue!

[pearsonca]

Perhaps leave it for another increment, and take advantage of the "good first issue"ness of it.

[seabbs]

Note that @Lnrivas has a draft PR seeking to address this issue here #249. Potentially help still needed supporting this if interested.