test_that("`fmls` objects can be fitted", {

	# Reference models
	data("mtcars")
	m0 <- lm(mpg ~ wt + hp, data = mtcars)

	# Original models
	object <- fmls(mpg ~ wt + hp)
	args <- list(model = TRUE)
	data <- mtcars
	m1 <- fit(object, .fn = lm, data = mtcars, raw = TRUE)[[1]]
	expect_equal(class(m0), class(m1))
	expect_equal(m0, m1, ignore_attr = TRUE) # Data argument should be the same name

	# <mdl> subtypes
	m1 <- fit(object, .fn = lm, data = mtcars, raw = FALSE)
	expect_s3_class(m1, "mdl")
	expect_length(m1, 1)

	# Stratified, should have two models
	object <- fmls(mpg ~ wt + hp + .s(am))
	m2 <- fit(object, .fn = lm, data = mtcars, raw = TRUE)
	expect_length(m2, 2)
	expect_s3_class(m2[[1]], "lm")

	# Should also keep strata term information when "tidied" into <mdl> object
	m2 <- fit(object, .fn = lm, data = mtcars, raw = FALSE)
	expect_s3_class(m2, "mdl")
	expect_equal(field(m2, "dataArgs")[[1]]$strataVariable, "am")
	expect_equal(field(m2, "dataArgs")[[1]]$dataName, "mtcars")
	expect_equal(field(m2, "dataArgs")[[1]]$strataLevel, 1)
	expect_equal(field(m2, "dataArgs")[[2]]$strataLevel, 0)


})

test_that("sequential/lengthy formulas can be fitted", {

	object <- fmls(mpg ~ wt + hp + cyl + .s(am), pattern = "sequential")
	m <- fit(object, .fn = lm, data = mtcars, raw = FALSE)
	expect_length(m, 6)

})


test_that("complex terms can be fit", {

	library(survival) # Using lung data
	f <- Surv(time, status) ~ ph.karno + cluster(sex)
	m0 <- coxph(Surv(time, status) ~ ph.karno + cluster(sex), data = lung)
	object <- fmls(f)
	m1 <- fit(object, .fn = coxph, data = lung, raw = TRUE)[[1]]

	# When fitting an object, the data term name should be retained
	expect_equal(m0, m1, ignore_attr = TRUE)

})