library(dplyr)

# function to use inside testthat
add_new_names <- function(.data, x) {
  c(
    tsibble::key_vars(.data),
    tsibble::index_var(.data),
    x,
    tsibble::measured_vars(.data)
  )
}

wages_test <- sample_frac_keys(wages, 0.05)

df_add_l_mean_1 <- wages_test %>%
  features(ln_wages, list(mean = mean)) %>%
  left_join(wages_test, by = "id", multiple = "all")
df_add_n_obs <- add_n_obs(wages_test)
df_add_key_slope <- add_key_slope(wages_test, ln_wages ~ xp)
df_add_key_slope_multi <- add_key_slope(wages_test, ln_wages ~ xp + ged)

updated_dim <- c(nrow(wages_test), ncol(wages_test) + 1)

test_that("add_* funs return a tsibble", {
  expect_s3_class(df_add_n_obs, class = c("tbl_ts"))
  expect_s3_class(df_add_key_slope, class = c("tbl_ts"))
  expect_s3_class(df_add_key_slope_multi, class = c("tbl_ts"))
})


test_that("longnostics returns the right dimensions", {
  expect_equal(dim(df_add_l_mean_1), updated_dim)
  expect_equal(dim(df_add_n_obs), updated_dim)
  expect_equal(dim(df_add_key_slope), c(nrow(wages_test), ncol(wages_test) + 2))
  expect_equal(
    dim(df_add_key_slope_multi),
    c(nrow(wages_test), ncol(wages_test) + 3)
  )
})

test_that("longnostic returns the right names", {
  expect_equal(
    names(df_add_l_mean_1),
    c(
      names(wages_test)[1],
      "mean",
      names(wages_test)[2:length(names(wages_test))]
    )
  )
  expect_equal(names(df_add_n_obs), add_new_names(wages_test, "n_obs"))
  expect_equal(
    names(df_add_key_slope),
    add_new_names(wages_test, c(".intercept", ".slope_xp"))
  )
  expect_equal(
    names(df_add_key_slope_multi),
    add_new_names(wages_test, c(".intercept", ".slope_xp", ".slope_ged"))
  )
})

test_that("longnostic returns a tibble", {
  expect_s3_class(df_add_l_mean_1, class = c("tbl"))
  expect_s3_class(df_add_n_obs, class = c("tbl"))
  expect_s3_class(df_add_key_slope, class = c("tbl"))
  expect_s3_class(df_add_key_slope_multi, class = c("tbl"))
})

classes <- function(x) purrr::map_chr(x, class)

test_that("longnostic returns correct classes", {
  expect_equal(classes(df_add_l_mean_1)[["mean"]], "numeric")
  expect_equal(classes(df_add_n_obs)[["n_obs"]], "integer")
  expect_equal(classes(df_add_key_slope)[[".intercept"]], "numeric")
  expect_equal(classes(df_add_key_slope)[[".slope_xp"]], "numeric")
  expect_equal(classes(df_add_key_slope_multi)[[".intercept"]], "numeric")
  expect_equal(classes(df_add_key_slope_multi)[[".slope_xp"]], "numeric")
  expect_equal(classes(df_add_key_slope_multi)[[".slope_ged"]], "numeric")
})

library(dplyr)
test_that("add-key-slope returns different slopes and intercepts", {
  expect_gte(n_distinct(df_add_key_slope$.intercept), 2)
  expect_gte(n_distinct(df_add_key_slope$.slope_xp), 2)
})