# tests/testthat/test-data_prep.R

test_that("clean_agri_data removes NA rows", {
  df <- data.frame(yield = c(1, 2, NA, 4), block = c("A","A","B","B"))
  out <- clean_agri_data(df, yield_col = "yield")
  expect_equal(nrow(out), 3L)
  expect_equal(attr(out, "n_removed"), 1L)
})

test_that("clean_agri_data returns agriData class", {
  df  <- data.frame(yield = c(2.1, 3.4, 2.8))
  out <- clean_agri_data(df)
  expect_s3_class(out, "agriData")
})

test_that("clean_agri_data flags IQR outliers", {
  df  <- data.frame(yield = c(3, 3.1, 3.2, 3.1, 100))
  out <- clean_agri_data(df, flag_outliers = TRUE)
  expect_true(".outlier" %in% names(out))
  expect_true(out$.outlier[5])
  expect_false(out$.outlier[1])
})

test_that("clean_agri_data errors on missing column", {
  df <- data.frame(x = 1:5)
  expect_error(clean_agri_data(df, yield_col = "yield"),
               "not found")
})

test_that("yield_normalize zscore produces mean≈0, sd≈1", {
  df  <- data.frame(yield = c(10, 20, 30, 40, 50))
  out <- yield_normalize(df, yield_col = "yield", method = "zscore")
  expect_true("yield_norm" %in% names(out))
  expect_lt(abs(mean(out$yield_norm)), 1e-10)
  expect_lt(abs(sd(out$yield_norm) - 1), 1e-10)
})

test_that("yield_normalize minmax bounds output to [0,1]", {
  df  <- data.frame(yield = c(5, 10, 15, 20))
  out <- yield_normalize(df, method = "minmax")
  expect_equal(min(out$yield_norm), 0)
  expect_equal(max(out$yield_norm), 1)
})

test_that("yield_normalize respects group_by", {
  df <- data.frame(yield = c(1,2,10,20), site = c("A","A","B","B"))
  out <- yield_normalize(df, group_by = "site", method = "minmax")
  grp_a <- out$yield_norm[out$site == "A"]
  grp_b <- out$yield_norm[out$site == "B"]
  expect_equal(min(grp_a), 0); expect_equal(max(grp_a), 1)
  expect_equal(min(grp_b), 0); expect_equal(max(grp_b), 1)
})

test_that("outlier_flag IQR method appends correct column", {
  df  <- data.frame(x = c(1,2,2,2,2,100))
  out <- outlier_flag(df, col = "x", method = "iqr")
  expect_true("x_outlier" %in% names(out))
  expect_true(out$x_outlier[6])
})

test_that("outlier_flag zscore method works", {
  df  <- data.frame(x = c(rep(5,20), 100))
  out <- outlier_flag(df, col = "x", method = "zscore")
  expect_true(out$x_outlier[21])
})

test_that("outlier_flag modz method works", {
  df  <- data.frame(x = c(rep(5,20), 200))
  out <- outlier_flag(df, col = "x", method = "modz")
  expect_true(out$x_outlier[21])
})

test_that("print.agriData outputs without error", {
  df  <- clean_agri_data(data.frame(yield = 1:10))
  expect_output(print(df), "agriData")
})

test_that("summary.agriData outputs without error", {
  df  <- clean_agri_data(data.frame(yield = 1:10))
  expect_output(summary(df), "agriData summary")
})