#' @srrstats {G5.0}
#' @srrstats {G5.1}
#' @srrstats {G5.2}
#' @srrstats {G5.2a}
#' @srrstats {G5.2b}
#' @srrstats {G2.12}

test_that("multiplication works", {
  # test 1
  amp_acro(
    time_col = time,
    n_components = 2,
    group = "X",
    period = c(24, 12),
    .data = vitamind,
    .formula = vit_d ~ X + amp_acro("time",
      n_components = 2,
      group = "X",
      period = c(24, 12)
    )
  )

  f <- Y ~ X + amp_acro(time, n_components = 2, group = c(NA, "X"))

  cglmm(
    vit_d ~ X + amp_acro(time,
      n_components = 2,
      group = c(NA, "X"),
      period = c(12, 24)
    ),
    data = vitamind
  )
  expect(
    ok = TRUE,
    "unsuccessful multiplication of group to match n_components"
  )
})

test_that("bad inputs return useful errors", {
  # test 2
  f <- function() {
    amp_acro(
      time_col = time, n_components = 1.1, group = "X", period = 12,
      .data = vitamind,
      .formula = vit_d ~ X + amp_acro(time,
        n_components = 1.1,
        group = "X",
        period = 12
      )
    )
  }
  expect_error( # non-count n_components
    f(),
    regexp = "assertthat::is.count(n_components) is not TRUE", fixed = TRUE
  )

  # test 3
  f <- function() {
    amp_acro(
      time_col = time, n_components = 1, group = "X", period = -12,
      .data = vitamind,
      .formula = vit_d ~ X + amp_acro(time,
        n_components = 1,
        group = "X",
        period = -12
      )
    )
  }
  expect_error(f(), regexp = "period > 0") # negative period

  # test 4
  data(vitamind)
  vitamind_mod <- vitamind
  vitamind_mod$time <- as.factor(rbinom(length(vitamind$time), 1, 0.5))
  f <- function() {
    amp_acro(
      time_col = time, n_components = 1, group = "X", period = 12,
      .data = vitamind_mod,
      .formula = vit_d ~ X + amp_acro(time,
        n_components = 1,
        group = "X",
        period = 12
      )
    )
  }
  expect_error(f(), regexp = "ttt is not a numeric or integer vector")

  # test 5
  data(vitamind)
  vitamind_2 <- vitamind
  vitamind_2$time <- cbind(vitamind_2$time, vitamind_2$time) # two time columns
  f <- function() {
    amp_acro(
      time_col = time, n_components = 1, group = "X", period = 12,
      .data = vitamind_2,
      .formula = vit_d ~ X + amp_acro(time,
        n_components = 1,
        group = "X",
        period = 12
      )
    )
  }
  expect_error(f(), regexp = "time_col must be univariate")

  # test 7
  data(vitamind)
  f <- function() {
    amp_acro(
      time_col = time_values, n_components = 1, group = "X", period = 12,
      .data = vitamind,
      .formula = vit_d ~ X + amp_acro(time,
        n_components = 1,
        group = "X",
        period = 12
      )
    )
  }
  expect_error(f(),
    regexp = "time_col must be the name of a column in dataframe"
  )

  # test 8
  data(vitamind)
  vitamind_mod <- as.list(vitamind)
  f <- function() {
    amp_acro(
      time_col = time, n_components = 1, group = "X", period = 12,
      .data = vitamind_mod,
      .formula = vit_d ~ X + amp_acro(time,
        n_components = 1,
        group = "X",
        period = 12
      )
    )
  }
  expect_error(f(),
    regexp = "'data' must be of class 'data.frame', 'matrix', or 'tibble'"
  )

  # test 9
  data(vitamind)
  vitamind_mod <- vitamind
  colnames(vitamind_mod)[1] <- "rrr2"
  f <- function() {
    amp_acro(
      time_col = time, n_components = 1, group = "rrr2", period = 12,
      .data = vitamind_mod,
      .formula = vit_d ~ X + amp_acro(time,
        n_components = 1,
        group = "rrr2",
        period = 12
      )
    )
  }
  expect_error(f(),
    regexp = "Group variable names cannot contain 'rrr' or 'sss'"
  )

  # test 10
  data(vitamind)
  vitamind_two_groups <- vitamind
  vitamind_two_groups["X2"] <- vitamind_two_groups["X"]
  f <- function() {
    amp_acro(
      time_col = time, n_components = 1, group = c("X", "X2"), period = 12,
      .data = vitamind_two_groups,
      .formula = vit_d ~ X + amp_acro(time,
        n_components = 1,
        group = c("X", "X2"),
        period = 12
      )
    )
  }
  expect_error(f(),
    regexp = paste(
      "Grouping variable in amp_acro() must be of",
      "length 1 or the same as n_components"
    ),
    fixed = TRUE
  )

  # test 11
  data(vitamind)
  f <- function() {
    amp_acro(
      time_col = time, n_components = 1, group = "X", period = c(8, 12),
      .data = vitamind,
      .formula = vit_d ~ X + amp_acro(time,
        n_components = 1,
        group = "X",
        period = c(8, 12)
      )
    )
  }
  expect_error(f(),
    regexp = paste(
      "period value(s) in amp_acro() must be of",
      "length 1 or the same as n_components"
    ),
    fixed = TRUE
  )

  # test 12
  data(vitamind)
  f <- function() {
    amp_acro(
      time_col = time, n_components = 1, group = "Z", period = c(8, 12),
      .data = vitamind,
      .formula = vit_d ~ X + amp_acro(time,
        n_components = 1,
        group = "Z",
        period = c(8, 12)
      )
    )
  }
  expect_error(f(),
    regexp = "Grouping variable(s) not found in input data:",
    fixed = TRUE
  )
})

test_that("matrix, or tibble inputs are converted to dataframe ", {
  # test 13
  data(vitamind)
  vitamind_mod <- dplyr::as_tibble(vitamind)

  f <- function() {
    amp_acro(
      time_col = time, n_components = 1, group = "X", period = 12,
      .data = vitamind_mod,
      .formula = vit_d ~ X + amp_acro(time,
        n_components = 1,
        group = "X",
        period = 12
      ),
      .quietly = FALSE
    )
  }
  suppressMessages(expect_message(f(),
    regexp = "main_rrr1 and main_sss1 have been added to dataframe"
  ))
  # test 14
  suppressMessages(expect_message(f(),
    regexp = "Data has been reformatted as dataframe"
  ))


  # test 15
  vitamind_mod <- as.matrix(vitamind)
  f <- function() {
    amp_acro(
      time_col = time, n_components = 1, group = "X", period = 12,
      .data = vitamind_mod,
      .formula = vit_d ~ X + amp_acro(time,
        n_components = 1,
        group = "X",
        period = 12
      )
    )
  }
  expect(f(), ok = TRUE, "unsuccessful conversion of matrix to dataframe")

  library(dplyr)

  withr::with_seed(
    50,
    {
      f_sample_id <- function(id_num,
                              n = 30,
                              mesor,
                              amp,
                              acro,
                              family = "gaussian",
                              sd = 0.2,
                              period,
                              n_components,
                              beta.group = TRUE) {
        data <- simulate_cosinor(
          n = n,
          mesor = mesor,
          amp = amp,
          acro = acro,
          family = family,
          sd = sd,
          period = period,
          n_components = n_components
        )
        data$subject <- id_num
        data
      }

      dat_mixed <- do.call(
        "rbind",
        lapply(1:30, function(x) {
          f_sample_id(
            id_num = x,
            mesor = rnorm(1, mean = 0, sd = 1),
            amp = rnorm(1, mean = 3, sd = 0.5),
            acro = rnorm(1, mean = 1.5, sd = 0.2),
            period = 24,
            n_components = 1
          )
        })
      )

      dat_mixed$subject <- as.factor(dat_mixed$subject)

      mixed_mod <- cglmm(
        Y ~ amp_acro(times,
          n_components = 1,
          period = 24
        ) + (1 + amp_acro1 | subject),
        data = dat_mixed
      )
      f_round <- function(x) {
        unname(round(x, digits = 4))
      }


      # Testing mixed model specification
      testthat::expect_no_error(
        update_formula_and_data(
          formula = Y ~ amp_acro(times,
            n_components = 1,
            period = 24
          ) +
            (1 + amp_acro1 * treatment * hospital:patient | subject),
          data = dat_mixed
        )
      )

      # Testing mixed model specification
      testthat::expect_no_error(
        update_formula_and_data(
          formula = Y ~ amp_acro(times,
            n_components = 1,
            period = 24
          ) +
            (amp_acro1 * treatment * hospital:patient + 1 | subject),
          data = dat_mixed
        )
      )

      testthat::expect_no_error(
        update_formula_and_data(
          formula = Y ~ amp_acro(times,
            n_components = 2,
            period = c(12, 6)
          ) +
            (amp_acro1 * treatment * hospital:patient + 1 +
              amp_acro2 * treatment | subject),
          data = dat_mixed
        )
      )

      testthat::expect_true(all.equal(
        f_round(mixed_mod$coefficients),
        c(-0.0932, 2.8613, 1.5077)
      ))
    }
  )
})