## non-CRAN use
# if (!require("survHEhmc")) remotes::install_github('giabaio/survHEhmc')

if (require("survHEhmc")) {

  library(survHE)
  library(survHEhmc)

  # GitHub Actions only allows 2 cores on Windows
  options("mc.cores" = 1)
  # options(cores = 1)

  data("TA174_FCR", package = "blendR")


  test_that("different distributions in survHE hmc", {

    data_sim <- ext_surv_sim(t_info = 144,
                             S_info = 0.05,
                             T_max = 180)

    obs_Surv2 <- fit.models(formula = Surv(death_t, death) ~ 1,
                            data = dat_FCR,
                            distr = "exponential",
                            method = "hmc")

    blend_interv <- list(min = 48, max = 150)
    beta_params <- list(alpha = 3, beta = 3)

    # exponential

    ext_Surv2 <- fit.models(formula = Surv(time, event) ~ 1,
                            data = data_sim,
                            distr = "exponential",
                            method = "hmc")

    expect_type(blendsurv(obs_Surv2, ext_Surv2, blend_interv, beta_params), "list")

    # weibull

    ext_Surv2 <- fit.models(formula = Surv(time, event) ~ 1,
                            data = data_sim,
                            distr = "weibull",
                            method = "hmc")

    suppressWarnings(
      expect_type(blendsurv(obs_Surv2, ext_Surv2, blend_interv, beta_params), "list")
    )

    # gompertz

    ext_Surv2 <- fit.models(formula = Surv(time, event) ~ 1,
                            data = data_sim,
                            distr = "gompertz",
                            method = "hmc")

    suppressWarnings(
      expect_type(blendsurv(obs_Surv2, ext_Surv2, blend_interv, beta_params), "list")
    )

    # log normal

    ext_Surv2 <- fit.models(formula = Surv(time, event) ~ 1,
                            data = data_sim,
                            distr = "lognormal",
                            method = "hmc")

    suppressWarnings(
      expect_type(blendsurv(obs_Surv2, ext_Surv2, blend_interv, beta_params), "list")
    )

    # gamma

    ext_Surv2 <- fit.models(formula = Surv(time, event) ~ 1,
                            data = data_sim,
                            distr = "gamma",
                            method = "hmc")

    suppressWarnings(
      expect_type(blendsurv(obs_Surv2, ext_Surv2, blend_interv, beta_params), "list")
    )

    # log logistic

    ext_Surv2 <- fit.models(formula = Surv(time, event) ~ 1,
                            data = data_sim,
                            distr = "loglogistic",
                            method = "hmc")

    suppressWarnings(
      expect_type(blendsurv(obs_Surv2, ext_Surv2, blend_interv, beta_params), "list")
    )
  })


  test_that("user-supplied time points for survival distribution", {

    data_sim <- ext_surv_sim(t_info = 144,
                             S_info = 0.05,
                             T_max = 180)

    obs_Surv2 <- fit.models(formula = Surv(death_t, death) ~ 1,
                            data = dat_FCR,
                            distr = "exponential",
                            method = "hmc")

    blend_interv <- list(min = 48, max = 150)
    beta_params <- list(alpha = 3, beta = 3)

    ext_Surv2 <- fit.models(formula = Surv(time, event) ~ 1,
                            data = data_sim,
                            distr = "exponential",
                            method = "hmc")

    expect_type(blendsurv(obs_Surv2, ext_Surv2, blend_interv, beta_params, times = 0:100), "list")
    expect_type(blendsurv(obs_Surv2, ext_Surv2, blend_interv, beta_params, times = -100:100), "list")
    expect_type(blendsurv(obs_Surv2, ext_Surv2, blend_interv, beta_params, times = seq(0, 100, by = 0.5)), "list")
    expect_type(blendsurv(obs_Surv2, ext_Surv2, blend_interv, beta_params, times = 0:300), "list")
    expect_type(blendsurv(obs_Surv2, ext_Surv2, blend_interv, beta_params, times = seq(0, 300, by = 0.5)), "list")

    ## INLA

    obs_Surv_inla <- fit_inla_pw(data = dat_FCR,
                                 cutpoints = seq(0, 180, by = 5),
                                 num.threads = 2)

    expect_type(blendsurv(obs_Surv_inla, ext_Surv2, blend_interv, beta_params, times = 0:100), "list")
    expect_type(blendsurv(obs_Surv_inla, ext_Surv2, blend_interv, beta_params, times = seq(0, 100, by = 0.5)), "list")
    expect_type(blendsurv(obs_Surv_inla, ext_Surv2, blend_interv, beta_params, times = 0:300), "list")
    expect_type(blendsurv(obs_Surv_inla, ext_Surv2, blend_interv, beta_params, times = seq(0, 300, by = 0.5)), "list")

    # # error
    # xx <- blendsurv(obs_Surv_inla, ext_Surv2, blend_interv, beta_params, times = -100:100)


    # flexsurv

    ext_Surv_flex <- flexsurv::flexsurvreg(formula = Surv(time, event) ~ 1,
                                           data = data_sim,
                                           dist = "gompertz")

    expect_type(blendsurv(obs_Surv2, ext_Surv_flex, blend_interv, beta_params, times = 0:100), "list")
    expect_type(blendsurv(obs_Surv2, ext_Surv_flex, blend_interv, beta_params, times = -100:100), "list")
    expect_type(blendsurv(obs_Surv2, ext_Surv_flex, blend_interv, beta_params, times = seq(0, 100, by = 0.5)), "list")
    expect_type(blendsurv(obs_Surv2, ext_Surv_flex, blend_interv, beta_params, times = 0:300), "list")
    expect_type(blendsurv(obs_Surv2, ext_Surv_flex, blend_interv, beta_params, times = seq(0, 300, by = 0.5)), "list")

  })
}