test_that("rob_tools() returns all 14 expected tools", {
  nms <- rob_tools()
  expect_true(all(c(
    "ROB2", "ROB2-Cluster", "ROBINS-I", "ROBINS-E",
    "QUADAS-2", "QUADAS-C", "QUIPS", "ROBIS", "PROBAST",
    "NOS-Cohort", "NOS-CaseControl", "NOS-CrossSectional",
    "GRADE", "Generic"
  ) %in% nms))
})

test_that(".standardise_judgement handles case, whitespace, abbreviations", {
  expect_equal(.standardise_judgement("low"),           "Low")
  expect_equal(.standardise_judgement("HIGH"),          "High")
  expect_equal(.standardise_judgement("some concern"),  "Some concerns")
  expect_equal(.standardise_judgement("ni"),            "No information")
  expect_equal(.standardise_judgement("NI"),            "No information")
  expect_equal(.standardise_judgement("  Low  "),       "Low")
  expect_equal(.standardise_judgement("serious"),       "Serious")
  expect_equal(.standardise_judgement("critical"),      "Critical")
  expect_equal(.standardise_judgement("very low"),      "Very low")
})

test_that(".resolve_colours returns named vector for cochrane palette", {
  cols <- .resolve_colours("cochrane", c("Low", "High", "No information"))
  expect_named(cols)
  expect_true("Low" %in% names(cols))
  expect_true("High" %in% names(cols))
  # NI always gets grey
  expect_true("No information" %in% names(cols))
  expect_match(cols[["No information"]], "^#[0-9A-Fa-f]{6}$")
})

test_that(".resolve_colours works for colourblind palette", {
  cols <- .resolve_colours("colourblind", c("Low", "Some concerns", "High"))
  expect_named(cols)
  expect_equal(length(cols), 3)
})

test_that(".resolve_colours accepts user hex vector", {
  cols <- .resolve_colours(c("#aabbcc", "#ddee00"), c("Low", "High"))
  expect_equal(cols[["Low"]],  "#aabbcc")
  expect_equal(cols[["High"]], "#ddee00")
})

test_that(".resolve_colours errors on mismatched user vector", {
  expect_error(
    .resolve_colours(c("#aabbcc"), c("Low", "High")),
    "Colour vector length mismatch"
  )
})

test_that(".validate_data errors on unknown tool", {
  df <- data.frame(Study = "S1", D1 = "Low")
  expect_error(.validate_data(df, "NOTREAL"), "Unknown tool")
})

test_that(".validate_data errors on empty data", {
  df <- data.frame(Study = character(0))
  expect_error(.validate_data(df, "ROB2"), "non-empty data frame")
})

test_that("ROBINS-I has correct post-Nov-2024 domain order", {
  tmpl <- .rv_tools[["ROBINS-I"]]
  expect_equal(tmpl$domains[2], "D2: Selection of participants")
  expect_equal(tmpl$domains[3], "D3: Classification of interventions")
})

test_that("QUADAS-2 has applicability_domains field", {
  tmpl <- .rv_tools[["QUADAS-2"]]
  expect_true(!is.null(tmpl$applicability_domains))
  expect_equal(length(tmpl$applicability_domains), 3L)
})

test_that("NOS tools are score_based = TRUE", {
  for (t in c("NOS-Cohort", "NOS-CaseControl", "NOS-CrossSectional")) {
    expect_true(.rv_tools[[t]]$score_based, label = paste(t, "score_based"))
    expect_equal(.rv_tools[[t]]$max_total, 9L)
  }
})