test_that("blockMatch returns m and all with expected columns", {
  set.seed(42)
  n <- 40L
  dat <- data.frame(x = rnorm(n), w = rnorm(n), z = rep(0:1, each = n / 2L))
  pr <- runif(n)
  prG <- 2L
  dat$prc <- as.integer(cut(pr, stats::quantile(pr, (0:prG) / prG), include.lowest = TRUE))
  bd <- basicDistance(dat)
  out <- suppressWarnings(
    blockMatch(dat, cost = bd$cost, J = 2L, ratio = 1L, rseed = 42L)
  )
  expect_type(out, "list")
  expect_named(out, c("m", "all"))
  expect_s3_class(out$m, "data.frame")
  expect_s3_class(out$all, "data.frame")
  expect_true("matched" %in% names(out$all))
  expect_true(all(c("mset", "z") %in% names(out$m)))
  J <- 2L
  expect_true(all(table(out$m$mset) == J))
  expect_equal(sum(out$all$matched), nrow(out$m))
  xm <- out$m
  by_mset <- split(seq_len(nrow(xm)), xm$mset)
  diffs <- vapply(by_mset, function(rows) {
    zi <- xm$z[rows]
    xi <- xm$x[rows]
    mean(xi[zi == 1]) - mean(xi[zi == 0])
  }, numeric(1))
  expect_true(all(is.finite(diffs)))
})

test_that("blockMatch requires ratio >= J - 1", {
  dat <- data.frame(x = 1:4, z = c(1, 1, 0, 0), prc = rep(1L, 4))
  bd <- basicDistance(dat)
  expect_error(
    blockMatch(
      dat,
      cost = bd$cost,
      J = 4L,
      ratio = 1L
    )
  )
})

test_that("blockMatch warns when no stratum has both arms", {
  dat <- data.frame(x = rnorm(4), z = c(1, 1, 0, 0), prc = c(1L, 1L, 2L, 2L))
  bd <- basicDistance(dat)
  expect_warning(
    out <- blockMatch(dat, cost = bd$cost, J = 2L, ratio = 1L, rseed = 1L),
    "no matched sets"
  )
  expect_null(out$m)
  expect_false(any(out$all$matched))
})

test_that("blockMatch requires z and prc columns in dat", {
  dat <- data.frame(x = 1:4, prc = rep(1L, 4))
  bd <- basicDistance(data.frame(x = 1:4, z = c(1, 1, 0, 0), prc = rep(1L, 4)))
  expect_error(
    blockMatch(dat, cost = bd$cost, J = 2L, ratio = 1L),
    "z must be a column"
  )
})

test_that("blockMatch runs with multiple strata that each support matching", {
  set.seed(1)
  n <- 24L
  dat <- data.frame(x = rnorm(n), z = c(rep(1L, 6L), rep(0L, 6L), rep(1L, 6L), rep(0L, 6L)),
    prc = c(rep(1L, 12L), rep(2L, 12L)))
  bd <- basicDistance(dat)
  out <- suppressWarnings(
    blockMatch(dat, cost = bd$cost, J = 2L, ratio = 1L, rseed = 1L)
  )
  expect_s3_class(out$m, "data.frame")
  expect_true(length(unique(out$all$prc[out$all$matched])) <= 2L)
})