context("Tests for lets.attrcells")

test_that("lets.attrcells returns a data.frame with expected columns", {
  set.seed(1)
  df <- data.frame(
    Species = paste0("sp", 1:50),
    trait_a = rnorm(50),
    trait_b = rnorm(50)
  )
  
  x <- lets.attrpam(df, n_bins = 10)
  y <- lets.attrcells(x)
  
  expect_s3_class(y, "data.frame")
  expect_gt(nrow(y), 0)
  
  expected_cols <- c(
    "Cell_attr",
    "Weighted Mean Distance to midpoint",
    "Mean Distance to midpoint",
    "Minimum Zero Distance",
    "Minimum 10% Zero Distance",
    "Distance to MCP border",
    "Frequency Weighted Distance"
  )
  
  expect_true(all(expected_cols %in% colnames(y)))
})



test_that("lets.attrcells output has one row per attribute cell", {
  set.seed(2)
  df <- data.frame(
    Species = paste0("sp", 1:40),
    trait_a = rnorm(40),
    trait_b = rnorm(40)
  )
  
  x <- lets.attrpam(df, n_bins = 8)
  y <- lets.attrcells(x)
  
  expect_equal(
    nrow(y),
    terra::ncell(x$Attr_Richness_Raster)
  )
})


test_that("lets.attrcells handles case with no zero-richness cells", {
  set.seed(4)
  df <- data.frame(
    Species = paste0("sp", 1:20),
    trait_a = rnorm(20),
    trait_b = rnorm(20)
  )
  
  x <- lets.attrpam(df, n_bins = 5)
  terra::values(x$Attr_Richness_Raster) <- 1
  
  y <- lets.attrcells(x)
  
  expect_true(all(!is.na(y$`Weighted Mean Distance to midpoint`)))
  expect_true(all(!is.na(y$`Mean Distance to midpoint`)))
  expect_true(all(is.na(y$`Minimum Zero Distance`)))
  expect_true(all(is.na(y$`Minimum 10% Zero Distance`)))
  
  expect_true(all(!is.na(y$`Distance to MCP border`)))
})


test_that("MCP distance requires at least 3 occupied cells", {
  set.seed(5)
  df <- data.frame(
    Species = paste0("sp", 1:5),
    trait_a = rnorm(5),
    trait_b = rnorm(5)
  )
  
  x <- lets.attrpam(df, n_bins = 3)
  
  rich <- terra::values(x$Attr_Richness_Raster)[,1]
  idx <- which(!is.na(rich))
  rich[idx] <- c(1, 1, rep(0, length(idx) - 2))
  terra::values(x$Attr_Richness_Raster) <- rich
  
  y <- lets.attrcells(x)
  
  expect_true(all(is.na(y$`Distance to MCP border`)))
})



test_that("lets.attrcells respects perc argument", {
  set.seed(6)
  df <- data.frame(
    Species = paste0("sp", 1:60),
    trait_a = rnorm(60),
    trait_b = rnorm(60)
  )
  
  x <- lets.attrpam(df, n_bins = 10)
  
  rich <- terra::values(x$Attr_Richness_Raster)[,1]
  zero_idx <- sample(seq_along(rich), size = 10)
  rich[zero_idx] <- 0
  terra::values(x$Attr_Richness_Raster) <- rich
  
  y1 <- lets.attrcells(x, perc = 0.1)
  y2 <- lets.attrcells(x, perc = 0.5)
  
  expect_false(isTRUE(all.equal(
    y1$`Minimum 10% Zero Distance`,
    y2$`Minimum 10% Zero Distance`
  )))
})



test_that("midpoint distances are negated (centrality direction)", {
  set.seed(7)
  df <- data.frame(
    Species = paste0("sp", 1:30),
    trait_a = rnorm(30),
    trait_b = rnorm(30)
  )
  
  x <- lets.attrpam(df, n_bins = 8)
  y <- lets.attrcells(x)
  
  expect_true(all(y$`Weighted Mean Distance to midpoint` <= 0, na.rm = TRUE))
  expect_true(all(y$`Mean Distance to midpoint` <= 0, na.rm = TRUE))
})


test_that("lets.plot.attrcells works without plotting (plot_ras = FALSE)", {

  set.seed(1)
  df <- data.frame(
    Species = paste0("sp", 1:50),
    trait_a = rnorm(50),
    trait_b = rnorm(50)
  )
  
  x <- lets.attrpam(df, n_bins = 10)
  y <- lets.attrcells(x)
  
  expect_silent(
    lets.plot.attrcells(x, y, plot_ras = FALSE)
  )
})


test_that("lets.plot.attrcells returns rasters when ras = TRUE", {
  
  set.seed(2)
  df <- data.frame(
    Species = paste0("sp", 1:40),
    trait_a = rnorm(40),
    trait_b = rnorm(40)
  )
  
  x <- lets.attrpam(df, n_bins = 8)
  y <- lets.attrcells(x)
  
  ras_list <- lets.plot.attrcells(x, y, ras = TRUE, plot_ras = FALSE)
  
  expect_type(ras_list, "list")
  expect_gt(length(ras_list), 0)
  expect_true(
    all(vapply(ras_list, inherits, logical(1), "SpatRaster"))
  )
  
})


test_that("lets.attrcells works correctly when y is provided", {
  
  skip_on_cran()
  
  data("PAM")
  
  # Simulate attributes
  n <- length(PAM$Species_name)
  Species <- PAM$Species_name
  trait_a <- rnorm(n)
  trait_b <- trait_a * 0.2 + rnorm(n)
  df <- data.frame(Species, trait_a, trait_b)
  
  # Build AttrPAM
  x <- lets.attrpam(df, n_bins = 4)
  
  # Run function with geographic PAM
  d <- lets.attrcells(x, y = PAM)
  
  # --- Basic structure ---
  expect_true(is.data.frame(d))
  expect_equal(nrow(d), terra::ncell(x$Attr_Richness_Raster))
  
  # --- Required columns ---
  expected_cols <- c(
    "Cell_attr",
    "Frequency",
    "Isolation (Min.)",
    "Isolation (1st Qu.)",
    "Isolation (Median)",
    "Isolation (Mean)",
    "Isolation (3rd Qu.)",
    "Isolation (Max.)",
    "Weighted Mean Distance to midpoint",
    "Mean Distance to midpoint",
    "Minimum Zero Distance",
    "Minimum 10% Zero Distance",
    "Distance to MCP border",
    "Frequency Weighted Distance"
  )
  
  expect_true(all(expected_cols %in% colnames(d)))
  
  # --- Frequency properties ---
  expect_true(all(d$Frequency >= 0))
  expect_true(is.numeric(d$Frequency))
  
  # --- Isolation metrics ---
  iso_cols <- grep("Isolation", colnames(d), value = TRUE)
  expect_true(all(sapply(d[iso_cols], is.numeric)))
  
  # --- Midpoint distances (should be <= 0 because negated) ---
  expect_true(all(d$`Weighted Mean Distance to midpoint` <= 0, na.rm = TRUE))
  expect_true(all(d$`Mean Distance to midpoint` <= 0, na.rm = TRUE))
  
  # --- Border metrics ---
  expect_true(all(sapply(d[, c(
    "Minimum Zero Distance",
    "Minimum 10% Zero Distance",
    "Distance to MCP border"
  )], function(x) is.numeric(x))))
  
  # --- Weighted distance ---
  expect_true(is.numeric(d$`Frequency Weighted Distance`))
  
  # --- No unexpected NA inflation (except expected edge cases) ---
  expect_true(sum(is.na(d$Frequency)) == 0)
  
})