library(testthat)
library(dplyr)

convertToScriDataWrapper <- function(cases,
                                     eras,
                                     exposureId = NULL,
                                     covariateSettings = NULL,
                                     controlIntervalSettings = NULL,
                                     naivePeriod = 0,
                                     firstOutcomeOnly = FALSE,
                                     excludeeraIds = NULL,
                                     minAge = NULL,
                                     maxAge = NULL) {
  if (is.null(covariateSettings)) {
    covariateSettings <- createEraCovariateSettings(
      includeEraIds = exposureId,
      start = 0,
      end = 0,
      endAnchor = "era end"
    )
  }

  if (is.null(controlIntervalSettings)) {
    controlIntervalSettings <- createControlIntervalSettings(
      includeEraIds = exposureId,
      start = -14,
      end = -7,
      endAnchor = "era start"
    )
  }

  covariateIds <- c()
  if (is.list(covariateSettings) && class(covariateSettings) != "EraCovariateSettings") {
    for (i in 1:length(covariateSettings)) {
      covariateIds <- c(covariateIds, covariateSettings[[i]]$includeEraIds)
    }
  } else {
    covariateIds <- covariateSettings$includeEraIds
  }
  eraRef <- eras |>
    distinct(.data$eraId, .data$eraType) |>
    mutate(eraName = "")

  data <- Andromeda::andromeda(
    cases = cases |>
      mutate(observationPeriodStartDate = observationPeriodStartDate),
    eras = eras,
    eraRef = eraRef
  )
  attr(data, "metaData") <- list(
    outcomeIds = 10,
    attrition = tibble(outcomeId = 10),
    prevalences = tibble(outcomeId = 10)
  )
  class(data) <- "SccsData"
  attr(class(data), "package") <- "SelfControlledCaseSeries"

  studyPop <- createStudyPopulation(
    sccsData = data,
    outcomeId = 10,
    createStudyPopulationArgs = createCreateStudyPopulationArgs(
      firstOutcomeOnly = firstOutcomeOnly,
      naivePeriod = naivePeriod,
      minAge = minAge,
      maxAge = maxAge
    )
  )

  result <- createScriIntervalData(
    studyPopulation = studyPop,
    sccsData = data,
    createScriIntervalDataArgs = createCreateScriIntervalDataArgs(
      eraCovariateSettings = covariateSettings,
      controlIntervalSettings = controlIntervalSettings
    )
  )
  return(list(outcomes = collect(result$outcomes), covariates = collect(result$covariates)))
}

test_that("Simple SCRI era construction", {
  cases <- tibble(
    observationPeriodId = "1000",
    caseId = 1,
    personId = "1",
    ageAtObsStart = 0,
    observationPeriodStartDate = as.Date("2000-5-1"),
    startDay = 0,
    endDay = 100,
    noninformativeEndCensor = 0
  )
  eras <- tibble(
    eraType = c("hoi", "rx"),
    caseId = c(1, 1),
    eraId = c(10, 11),
    eraValue = c(1, 1),
    eraStartDay = c(50, 25),
    eraEndDay = c(50, 75)
  )
  result <- convertToScriDataWrapper(cases, eras, exposureId = 11)
  expect_equal(result$outcomes$rowId, c(0, 1))
  expect_equal(result$outcomes$stratumId, c(1, 1))
  expect_equal(result$outcomes$time, c(8, 51))
  expect_equal(result$outcomes$y, c(0, 1))
  expect_equal(result$covariates$rowId, c(1))
  expect_equal(result$covariates$stratumId, c(1))
  expect_equal(result$covariates$covariateId, c(1000))
})


test_that("Outcome on boundary or control interval", {
  cases <- tibble(
    observationPeriodId = "1000",
    caseId = 1,
    personId = "1",
    ageAtObsStart = 0,
    observationPeriodStartDate = as.Date("2000-5-1"),
    startDay = 0,
    endDay = 100,
    noninformativeEndCensor = 0
  )
  eras <- tibble(
    eraType = c("hoi", "rx"),
    caseId = c(1, 1),
    eraId = c(10, 11),
    eraValue = c(1, 1),
    eraStartDay = c(25 - 7, 25),
    eraEndDay = c(25 - 7, 75)
  )
  result <- convertToScriDataWrapper(cases, eras, exposureId = c(11))
  expect_equal(result$outcomes$rowId, c(0, 1))
  expect_equal(result$outcomes$stratumId, c(1, 1))
  expect_equal(result$outcomes$time, c(8, 51))
  expect_equal(result$outcomes$y, c(1, 0))
  expect_equal(result$covariates$rowId, c(1))
  expect_equal(result$covariates$stratumId, c(1))
  expect_equal(result$covariates$covariateId, c(1000))
})

test_that("Merging overlapping control intervals", {
  cases <- tibble(
    observationPeriodId = "1000",
    caseId = 1,
    personId = "1",
    ageAtObsStart = 0,
    observationPeriodStartDate = as.Date("2000-5-1"),
    startDay = 0,
    endDay = 100,
    noninformativeEndCensor = 0
  )
  eras <- tibble(
    eraType = c("hoi", "rx", "rx"),
    caseId = c(1, 1, 1),
    eraId = c(10, 11, 11),
    eraValue = c(1, 1, 1),
    eraStartDay = c(25 - 7, 25, 28),
    eraEndDay = c(25 - 7, 26, 29)
  )
  result <- convertToScriDataWrapper(cases, eras, exposureId = c(11))
  expect_equal(result$outcomes$rowId, c(0, 1))
  expect_equal(result$outcomes$stratumId, c(1, 1))
  expect_equal(result$outcomes$time, c(11, 4))
  expect_equal(result$outcomes$y, c(1, 0))
  expect_equal(result$covariates$rowId, c(1))
  expect_equal(result$covariates$stratumId, c(1))
  expect_equal(result$covariates$covariateId, c(1000))
})

test_that("Control intervals overlapping with a risk window", {
  cases <- tibble(
    observationPeriodId = "1000",
    caseId = 1,
    personId = "1",
    ageAtObsStart = 0,
    observationPeriodStartDate = as.Date("2000-5-1"),
    startDay = 0,
    endDay = 100,
    noninformativeEndCensor = 0
  )
  eras <- tibble(
    eraType = c("hoi", "rx", "rx"),
    caseId = c(1, 1, 1),
    eraId = c(10, 11, 11),
    eraValue = c(1, 1, 1),
    eraStartDay = c(32, 30, 50),
    eraEndDay = c(32, 40, 60)
  )
  result <- convertToScriDataWrapper(cases, eras, exposureId = c(11))
  expect_equal(result$outcomes$rowId, c(0, 1))
  expect_equal(result$outcomes$stratumId, c(1, 1))
  expect_equal(result$outcomes$time, c(11, 22))
  expect_equal(result$outcomes$y, c(0, 1))
  expect_equal(result$covariates$rowId, c(1))
  expect_equal(result$covariates$stratumId, c(1))
  expect_equal(result$covariates$covariateId, c(1000))
})