test_that("prepare_panel_data creates correct structure", {
    set.seed(123)
    years <- 2000:2005
    sectors <- c("A", "B", "C")

    direct <- data.frame(
        Year = years,
        A = runif(6, 100, 200),
        B = runif(6, 100, 200),
        C = runif(6, 100, 200)
    )

    production <- data.frame(
        Year = years,
        A = runif(6, 100, 200),
        B = runif(6, 100, 200),
        C = runif(6, 100, 200)
    )

    panel <- prepare_panel_data(direct, production)

    expect_s3_class(panel, "data.frame")
    expect_equal(nrow(panel), length(years) * length(sectors))
    expect_true("year" %in% names(panel))
    expect_true("sector" %in% names(panel))
    expect_true("log_direct" %in% names(panel))
    expect_true("log_production" %in% names(panel))
})

test_that("prepare_panel_data handles missing Year column", {
    direct <- data.frame(A = 1:5, B = 1:5)
    production <- data.frame(A = 1:5, B = 1:5)

    expect_error(
        prepare_panel_data(direct, production),
        "Year"
    )
})

test_that("validate_panel_data accepts valid data", {
    panel <- data.frame(
        year = rep(2000:2002, 3),
        sector = rep(c("A", "B", "C"), each = 3),
        log_direct = rnorm(9),
        log_production = rnorm(9)
    )

    expect_true(validate_panel_data(panel))
})

test_that("validate_panel_data rejects invalid data", {
    panel <- data.frame(
        year = 1:5,
        sector = letters[1:5]
    )

    expect_error(
        validate_panel_data(panel, require_log = TRUE),
        "Missing"
    )
})

test_that("create_mundlak_data adds correct columns", {
    set.seed(123)
    panel <- data.frame(
        year = rep(2000:2002, 3),
        sector = rep(c("A", "B", "C"), each = 3),
        log_direct = rnorm(9, mean = 5),
        log_production = rnorm(9, mean = 5)
    )

    panel_m <- create_mundlak_data(panel)

    expect_true("x_mean_sector" %in% names(panel_m))
    expect_true("x_within" %in% names(panel_m))

    for (s in unique(panel_m$sector)) {
        idx <- panel_m$sector == s
        sector_mean <- mean(panel$log_direct[idx])
        expect_equal(
            unique(panel_m$x_mean_sector[idx]),
            sector_mean,
            tolerance = 1e-10
        )
    }
})

test_that("aggregate_to_timeseries works correctly", {
    set.seed(123)
    panel <- data.frame(
        year = rep(2000:2005, 3),
        sector = rep(c("A", "B", "C"), each = 6),
        log_direct = rnorm(18, mean = 5),
        log_production = rnorm(18, mean = 5)
    )

    ts_agg <- aggregate_to_timeseries(panel)

    expect_equal(nrow(ts_agg), 6)
    expect_true("year" %in% names(ts_agg))
    expect_true("log_direct_mean" %in% names(ts_agg))
    expect_true("log_production_mean" %in% names(ts_agg))
})

test_that("prepare_log_matrices returns correct structure", {
    set.seed(123)
    direct <- data.frame(
        Year = 2000:2005,
        A = runif(6, 100, 200),
        B = runif(6, 100, 200)
    )
    production <- data.frame(
        Year = 2000:2005,
        A = runif(6, 100, 200),
        B = runif(6, 100, 200)
    )

    matrices <- prepare_log_matrices(direct, production)

    expect_type(matrices, "list")
    expect_true("X_log" %in% names(matrices))
    expect_true("Y_log" %in% names(matrices))
    expect_true("X_clean" %in% names(matrices))
    expect_true("Y_clean" %in% names(matrices))
    expect_true(is.matrix(matrices$X_log))
    expect_true(is.matrix(matrices$Y_log))
})

test_that("fit_twoway_fe works with valid data", {
    skip_if_not_installed("plm")

    set.seed(123)
    panel <- data.frame(
        year = rep(2000:2009, 5),
        sector = rep(LETTERS[1:5], each = 10),
        log_direct = rnorm(50, 5, 0.5),
        log_production = rnorm(50, 5, 0.5)
    )
    panel$log_production <- panel$log_direct * 0.95 + rnorm(50, 0, 0.1)

    result <- fit_twoway_fe(panel)

    expect_type(result, "list")
    expect_true("model" %in% names(result))
    expect_true("r2_within" %in% names(result))
    expect_true("metrics" %in% names(result))
    expect_true(result$r2_within > 0.5)
})

test_that("fit_mundlak_cre works with valid data", {
    skip_if_not_installed("plm")

    set.seed(123)
    panel <- data.frame(
        year = rep(2000:2009, 5),
        sector = rep(LETTERS[1:5], each = 10),
        log_direct = rnorm(50, 5, 0.5),
        log_production = rnorm(50, 5, 0.5)
    )
    panel$log_production <- panel$log_direct * 0.95 + rnorm(50, 0, 0.1)

    result <- fit_mundlak_cre(panel)

    expect_type(result, "list")
    expect_true("model" %in% names(result))
    expect_true("variance_components" %in% names(result))
    expect_true("panel_data_augmented" %in% names(result))
})