test_that("mock db: check output format", {
  cdm <- mockIncidencePrevalenceRef() %>%
    generateDenominatorCohortSet(name = "denominator")

  prev <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "years"
  )

  # check estimates tibble
  expect_true(all(c(
    "analysis_id",
    "n_cases",
    "n_population",
    "prevalence",
    "prevalence_95CI_lower",
    "prevalence_95CI_upper",
    "prevalence_start_date",
    "prevalence_end_date",
    "population_obscured",
    "cases_obscured",
    "result_obscured",
    "outcome_cohort_id",
    "outcome_cohort_name",
    "analysis_type",
    "analysis_time_point",
    "analysis_interval",
    "analysis_full_contribution",
    "analysis_complete_database_intervals",
    "analysis_min_cell_count",
    "denominator_cohort_id",
    "denominator_age_group",
    "denominator_sex",
    "denominator_days_prior_observation",
    "denominator_start_date",
    "denominator_end_date",
    "denominator_target_cohort_definition_id",
    "denominator_target_cohort_name",
    "cdm_name"
  ) %in%
    names(prev)))

  expect_true(all(c(
    "analysis_id", "number_records", "number_subjects",
    "reason_id", "reason",
    "excluded_records", "excluded_subjects",
    "outcome_cohort_id",
    "outcome_cohort_name",
    "analysis_type",
    "analysis_time_point",
    "analysis_interval",
    "analysis_full_contribution",
    "analysis_complete_database_intervals",
    "analysis_min_cell_count",
    "denominator_cohort_id",
    "denominator_age_group",
    "denominator_sex",
    "denominator_days_prior_observation",
    "denominator_start_date",
    "denominator_end_date",
    "denominator_target_cohort_definition_id",
    "denominator_target_cohort_name",
    "cdm_name"
  ) %in%
    names(prevalenceAttrition(prev))))

  # by default we don´t return the participants
  expect_true(is.null(participants(prev, analysisId = 1)))
  CDMConnector::cdm_disconnect(cdm)

  cdm <- mockIncidencePrevalenceRef()
  cdm <- generateDenominatorCohortSet(cdm = cdm, name = "denominator")

  prev <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "years",
    temporary = FALSE,
    returnParticipants = TRUE
  )
  # now we do return the participants
  expect_true(is.list(participants(prev, 1))) # list of references to participants
  expect_true(tibble::is_tibble(participants(prev, 1) %>%
    dplyr::collect()))
  expect_true(participants(prev, 1) %>%
    dplyr::collect() %>%
    dplyr::select("subject_id") %>%
    dplyr::pull() == 1)

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: checks on working example", {
  personTable <- tibble::tibble(
    person_id = "1",
    gender_concept_id = "8507",
    year_of_birth = 2000,
    month_of_birth = 01,
    day_of_birth = 01
  )
  observationPeriodTable <- tibble::tibble(
    observation_period_id = "1",
    person_id = "1",
    observation_period_start_date = as.Date("2000-01-01"),
    observation_period_end_date = as.Date("2012-06-01")
  )
  outcomeTable <- tibble::tibble(
    cohort_definition_id = 1,
    subject_id = "1",
    cohort_start_date = c(
      as.Date("2008-02-05"),
      as.Date("2010-02-08"),
      as.Date("2010-02-20")
    ),
    cohort_end_date = c(
      as.Date("2008-02-05"),
      as.Date("2010-02-08"),
      as.Date("2010-02-20")
    )
  )

  cdm <- mockIncidencePrevalenceRef(
    personTable = personTable,
    observationPeriodTable = observationPeriodTable,
    outcomeTable = outcomeTable
  )

  cdm <- generateDenominatorCohortSet(cdm = cdm, name = "denominator")

  prev <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = c("months", "years"),
    minCellCount = 0
  )
  expect_true(nrow(prev) >= 1)

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: working examples 2", {
  skip_on_cran()
  personTable <- tibble::tibble(
    person_id = "1",
    gender_concept_id = "8507",
    year_of_birth = 2000,
    month_of_birth = 01,
    day_of_birth = 01
  )
  observationPeriodTable <- tibble::tibble(
    observation_period_id = "1",
    person_id = "1",
    observation_period_start_date = as.Date("2010-01-01"),
    observation_period_end_date = as.Date("2012-06-01")
  )
  outcomeTable <- tibble::tibble(
    cohort_definition_id = 1,
    subject_id = "1",
    cohort_start_date = c(
      as.Date("2010-02-05"),
      as.Date("2010-02-08"),
      as.Date("2010-02-20")
    ),
    cohort_end_date = c(
      as.Date("2010-02-05"),
      as.Date("2010-02-08"),
      as.Date("2010-02-20")
    )
  )
  cdm <- mockIncidencePrevalenceRef(
    personTable = personTable,
    observationPeriodTable = observationPeriodTable,
    outcomeTable = outcomeTable
  )
  cdm <- generateDenominatorCohortSet(
    cdm = cdm, name = "denominator"
  )

  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "point",
    interval = c("months", "years")
  )
  expect_true(nrow(prev) >= 1)

  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "point",
    interval = c("months", "years")
  )
  expect_true(nrow(prev) >= 1)

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: check minimum counts", {
  skip_on_cran()
  # 20 people
  personTable <- tibble::tibble(
    person_id = as.character(c(1:20)),
    gender_concept_id = rep("8507", 20),
    year_of_birth = rep(2000, 20),
    month_of_birth = rep(01, 20),
    day_of_birth = rep(01, 20)
  )
  observationPeriodTable <- dplyr::bind_rows(
    tibble::tibble(
      observation_period_id = as.character(c(1:17)),
      person_id = as.character(c(1:17)),
      observation_period_start_date = rep(as.Date("2000-01-01"), 17),
      observation_period_end_date = rep(as.Date("2000-01-31"), 17)
    ),
    tibble::tibble(
      observation_period_id = as.character(c(18:20)),
      person_id = as.character(c(18:20)),
      observation_period_start_date = rep(as.Date("2000-01-01"), 3),
      observation_period_end_date = rep(as.Date("2012-06-01"), 3)
    )
  )

  outcomeTable <-
    dplyr::bind_rows(
      # 17 in first period
      tibble::tibble(
        cohort_definition_id = rep(1, 17),
        subject_id = as.character(c(1:17)),
        cohort_start_date = rep(
          as.Date("2000-01-02"), 17
        ),
        cohort_end_date = rep(
          as.Date("2000-01-03"), 17
        )
      ),
      # three in second
      tibble::tibble(
        cohort_definition_id = rep(1, 3),
        subject_id = as.character(c(18:20)),
        cohort_start_date = rep(
          as.Date("2000-02-02"), 3
        ),
        cohort_end_date = rep(
          as.Date("2000-02-03"), 3
        )
      )
    )

  cdm <- mockIncidencePrevalenceRef(
    personTable = personTable,
    observationPeriodTable = observationPeriodTable,
    outcomeTable = outcomeTable
  )

  cdm <- generateDenominatorCohortSet(cdm = cdm, name = "denominator")

  prev <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    minCellCount = 0,
    type = "period",
    interval = "months"
  )
  expect_true(prev$n_cases[1] == 17)
  expect_true(prev$n_cases[2] == 3)
  expect_true(prev$n_cases[3] == 0)
  expect_true(prev$n_population[1] == 20)
  expect_true(prev$n_population[2] == 3)
  expect_true(prev$n_population[3] == 3)
  expect_true(!is.na(prev$prevalence[1]))
  expect_true(!is.na(prev$prevalence[2]))
  expect_true(!is.na(prev$prevalence[3]))
  expect_true(!is.na(prev$prevalence_95CI_lower[1]))
  expect_true(!is.na(prev$prevalence_95CI_upper[1]))

  prev <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    minCellCount = 5,
    type = "period",
    interval = "months"
  )
  expect_true(prev$n_cases[1] == 17)
  expect_true(is.na(prev$n_cases[2]))
  expect_true(!is.na(prev$n_cases[3])) # don't suppress zero
  expect_true(prev$n_population[1] == 20)
  expect_true(is.na(prev$n_population[2]))
  expect_true(is.na(prev$n_population[3]))
  expect_true(!is.na(prev$prevalence[1]))
  expect_true(is.na(prev$prevalence[2]))
  expect_true(is.na(prev$prevalence[3]))
  expect_true(!is.na(prev$prevalence_95CI_lower[1]))
  expect_true(is.na(prev$prevalence_95CI_lower[2]))
  expect_true(is.na(prev$prevalence_95CI_lower[3]))
  expect_true(!is.na(prev$prevalence_95CI_upper[1]))
  expect_true(is.na(prev$prevalence_95CI_upper[2]))
  expect_true(is.na(prev$prevalence_95CI_upper[3]))

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: check study time periods", {
  skip_on_cran()
  personTable <- tibble::tibble(
    person_id = "1",
    gender_concept_id = "8507",
    year_of_birth = 2000,
    month_of_birth = 01,
    day_of_birth = 01
  )
  observationPeriodTable <- tibble::tibble(
    observation_period_id = "1",
    person_id = "1",
    observation_period_start_date = as.Date("2010-01-01"),
    observation_period_end_date = as.Date("2010-12-31")
  )
  outcomeTable <- tibble::tibble(
    cohort_definition_id = 1,
    subject_id = "1",
    cohort_start_date = c(
      as.Date("2010-02-05"),
      as.Date("2010-02-08"),
      as.Date("2010-02-20")
    ),
    cohort_end_date = c(
      as.Date("2010-02-05"),
      as.Date("2010-02-08"),
      as.Date("2010-02-20")
    )
  )

  cdm <- mockIncidencePrevalenceRef(
    personTable = personTable,
    observationPeriodTable = observationPeriodTable,
    outcomeTable = outcomeTable
  )

  cdm <- generateDenominatorCohortSet(cdm = cdm, name = "denominator")

  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "point",
    interval = "months"
  )

  # we expect 12 months of which the last in December
  # the last month should also be included
  # as the person goes up to the last day of the month
  expect_true(nrow(prev) == 12)


  # overall period
  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "period",
    interval = "overall"
  )
  # just one row
  expect_true(nrow(prev) == 1)

  CDMConnector::cdm_disconnect(cdm)

  # should return empty if no study days
  personTable <- tibble::tibble(
    person_id = "1",
    gender_concept_id = "8507",
    year_of_birth = 2000,
    month_of_birth = 01,
    day_of_birth = 01
  )
  observationPeriodTable <- tibble::tibble(
    observation_period_id = "1",
    person_id = "1",
    observation_period_start_date = as.Date("2010-01-01"),
    observation_period_end_date = as.Date("2010-11-15")
  )
  outcomeTable <- tibble::tibble(
    cohort_definition_id = 1,
    subject_id = "1",
    cohort_start_date = c(
      as.Date("2010-02-05"),
      as.Date("2010-02-08"),
      as.Date("2010-02-20")
    ),
    cohort_end_date = c(
      as.Date("2010-02-05"),
      as.Date("2010-02-08"),
      as.Date("2010-02-20")
    )
  )

  cdm <- mockIncidencePrevalenceRef(
    personTable = personTable,
    observationPeriodTable = observationPeriodTable,
    outcomeTable = outcomeTable
  )

  cdm <- generateDenominatorCohortSet(cdm = cdm, name = "denominator")

  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "period",
    fullContribution = TRUE,
    interval = "weeks"
  )
  expect_true(nrow(prev) == 45)

  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "period",
    fullContribution = TRUE,
    interval = "months"
  )
  expect_true(nrow(prev) == 10)

  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "period",
    fullContribution = TRUE,
    interval = "years"
  )
  expect_true(nrow(prev) == 0)

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: check fullContribution requirement", {
  skip_on_cran()
  personTable <- tibble::tibble(
    person_id = c("1", "2", "3"),
    gender_concept_id = "8507",
    year_of_birth = 2000,
    month_of_birth = 01,
    day_of_birth = 01
  )
  observationPeriodTable <- tibble::tibble(
    observation_period_id = c("1", "2", "3"),
    person_id = c("1", "2", "3"),
    observation_period_start_date = c(
      as.Date("2010-01-01"),
      as.Date("2010-01-01"),
      as.Date("2012-04-01")
    ),
    observation_period_end_date = c(
      as.Date("2011-06-01"),
      as.Date("2012-06-01"),
      as.Date("2012-06-01")
    )
  )
  outcomeTable <- tibble::tibble(
    cohort_definition_id = 1,
    subject_id = "1",
    cohort_start_date = c(
      as.Date("2010-02-05"),
      as.Date("2010-02-08"),
      as.Date("2010-02-20")
    ),
    cohort_end_date = c(
      as.Date("2010-02-05"),
      as.Date("2010-02-08"),
      as.Date("2010-02-20")
    )
  )
  cdm <- mockIncidencePrevalenceRef(
    personTable = personTable,
    observationPeriodTable = observationPeriodTable,
    outcomeTable = outcomeTable
  )
  cdm <- generateDenominatorCohortSet(
    cdm = cdm, name = "denominator"
  )

  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "period",
    interval = "years",
    fullContribution = FALSE,
    completeDatabaseIntervals = FALSE,
    minCellCount = 0
  )
  expect_true(all(prev$n_population == 2))

  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "period",
    interval = "years",
    fullContribution = TRUE,
    completeDatabaseIntervals = FALSE,
    minCellCount = 0
  )

  expect_true(all(prev$n_population == c(2, 1, 1)))

  expect_true(prevalenceAttrition(prev) %>%
    dplyr::filter(reason ==
      "Do not satisfy full contribution requirement for an interval") %>%
    dplyr::pull("excluded_subjects") == 1)

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: check periods follow calendar dates", {
  skip_on_cran()
  # check that even if study_start_date is during a period
  # periods still follow calendar dates
  personTable <- tibble::tibble(
    person_id = "1",
    gender_concept_id = "8507",
    year_of_birth = 2000,
    month_of_birth = 01,
    day_of_birth = 01
  )
  observationPeriodTable <- tibble::tibble(
    observation_period_id = "1",
    person_id = "1",
    observation_period_start_date = as.Date("2010-06-05"),
    observation_period_end_date = as.Date("2013-06-15")
  )
  outcomeTable <- tibble::tibble(
    cohort_definition_id = 1,
    subject_id = "1",
    cohort_start_date = c(
      as.Date("2011-01-31"),
      as.Date("2011-02-01"),
      as.Date("2011-03-01")
    ),
    cohort_end_date = c(
      as.Date("2011-01-31"),
      as.Date("2011-02-01"),
      as.Date("2011-03-01")
    )
  )

  cdm <- mockIncidencePrevalenceRef(
    personTable = personTable,
    observationPeriodTable = observationPeriodTable,
    outcomeTable = outcomeTable
  )

  # if completeDatabaseIntervals is TRUE we should go from 2010 to 2013
  # but if FALSE we should go from 2011 to 2012
  # for yearly incidence
  cdm <- generateDenominatorCohortSet(
    cdm = cdm, name = "denominator"
  )
  prev1 <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "period",
    interval = "years",
    minCellCount = 0,
    fullContribution = FALSE,
    completeDatabaseIntervals = FALSE
  )
  expect_true(nrow(prev1) == 4)
  expect_true(all(lubridate::year(prev1$prevalence_start_date) ==
    c("2010", "2011", "2012", "2013")))

  prev2 <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "period",
    interval = "years",
    minCellCount = 0,
    fullContribution = FALSE,
    completeDatabaseIntervals = TRUE
  )
  expect_true(nrow(prev2) == 2)
  expect_true(all(lubridate::year(prev2$prevalence_start_date) ==
    c("2011", "2012")))

  # for months
  cdm <- generateDenominatorCohortSet(
    cdm = cdm, name = "denominator", overwrite = TRUE,
    cohortDateRange = c(as.Date("2011-01-15"), as.Date(NA))
  )

  # where we expect the study to start on 2011-01-15
  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "period",
    interval = "months",
    minCellCount = 0,
    fullContribution = FALSE,
    completeDatabaseIntervals = FALSE
  )
  expect_true(prev$prevalence_start_date[1] ==
    as.Date("2011-01-15"))
  # where we expect the study to start the next month
  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "period",
    interval = "months",
    minCellCount = 0,
    fullContribution = FALSE,
    completeDatabaseIntervals = TRUE
  )
  expect_true(prev$prevalence_start_date[1] ==
    as.Date("2011-02-01"))

  # for overall
  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "period",
    interval = "overall",
    minCellCount = 0,
    fullContribution = FALSE,
    completeDatabaseIntervals = FALSE
  )
  expect_true(prev$prevalence_start_date[1] ==
    as.Date("2011-01-15"))
  expect_true(prev$prevalence_end_date[1] ==
    as.Date("2013-06-15"))

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: check multiple outcome ids", {
  skip_on_cran()
  personTable <- tibble::tibble(
    person_id = c("1", "2"),
    gender_concept_id = "8507",
    year_of_birth = 2000,
    month_of_birth = 01,
    day_of_birth = 01
  )
  observationPeriodTable <- tibble::tibble(
    observation_period_id = c("1", "2"),
    person_id = c("1", "2"),
    observation_period_start_date = as.Date("2011-01-01"),
    observation_period_end_date = as.Date("2012-06-01")
  )
  outcomeTable <- tibble::tibble(
    cohort_definition_id = c(1, 2), # two different outcome ids
    subject_id = c("1", "2"),
    cohort_start_date = c(
      as.Date("2011-02-05")
    ),
    cohort_end_date = c(
      as.Date("2011-02-05")
    )
  )
  cdm <- mockIncidencePrevalenceRef(
    personTable = personTable,
    observationPeriodTable = observationPeriodTable,
    outcomeTable = outcomeTable
  )
  cdm <- generateDenominatorCohortSet(
    cdm = cdm, name = "denominator"
  )

  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "period",
    interval = "years",
    minCellCount = 0
  )
  expect_true(all(prev[["n_cases"]] == 1))

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: some empty result sets", {
  skip_on_cran()
  personTable <- tibble::tibble(
    person_id = c("1", "2"),
    gender_concept_id = "8507",
    year_of_birth = 2000,
    month_of_birth = 01,
    day_of_birth = 01
  )
  observationPeriodTable <- tibble::tibble(
    observation_period_id = c("1", "2"),
    person_id = c("1", "2"),
    observation_period_start_date = as.Date("2012-02-01"),
    observation_period_end_date = as.Date("2012-06-01")
  )
  outcomeTable <- tibble::tibble(
    cohort_definition_id = c(1, 2), # two different outcome ids
    subject_id = c("1", "2"),
    cohort_start_date = c(
      as.Date("2012-02-05")
    ),
    cohort_end_date = c(
      as.Date("2012-02-05")
    )
  )
  cdm <- mockIncidencePrevalenceRef(
    personTable = personTable,
    observationPeriodTable = observationPeriodTable,
    outcomeTable = outcomeTable
  )
  cdm <- generateDenominatorCohortSet(
    cdm = cdm, name = "denominator"
  )

  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "period",
    interval = "years",
    minCellCount = 0
  )
  expect_true(nrow(prev) == 0)

  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "period",
    interval = c("months", "years"),
    minCellCount = 0
  )
  expect_true(nrow(prev) > 0)

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: check expected errors", {
  skip_on_cran()
  personTable <- tibble::tibble(
    person_id = "1",
    gender_concept_id = "8507",
    year_of_birth = 2000,
    month_of_birth = 01,
    day_of_birth = 01
  )
  observationPeriodTable <- tibble::tibble(
    observation_period_id = "1",
    person_id = "1",
    observation_period_start_date = as.Date("2010-01-01"),
    observation_period_end_date = as.Date("2012-06-01")
  )
  outcomeTable <- tibble::tibble(
    cohort_definition_id = 1,
    subject_id = "1",
    cohort_start_date = c(
      as.Date("2010-02-05"),
      as.Date("2010-02-08"),
      as.Date("2010-02-20")
    ),
    cohort_end_date = c(
      as.Date("2010-02-05"),
      as.Date("2010-02-08"),
      as.Date("2010-02-20")
    )
  )
  cdm <- mockIncidencePrevalenceRef(
    personTable = personTable,
    observationPeriodTable = observationPeriodTable,
    outcomeTable = outcomeTable
  )

  cdm <- generateDenominatorCohortSet(cdm = cdm, name = "denominator")

  expect_error(estimatePrevalence(
    cdm = "a",
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    outcomeCohortId = 1,
    denominatorCohortId = 1
  ))

  expect_error(estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    outcomeCohortId = 1,
    denominatorCohortId = 1,
    temporary = TRUE,
    returnParticipants = TRUE
  ))

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: check user point prevalence function", {
  skip_on_cran()
  cdm <- mockIncidencePrevalenceRef()

  cdm <- generateDenominatorCohortSet(cdm = cdm, name = "denominator")

  prev <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome"
  )
  prev_point <- estimatePointPrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome"
  )

  expect_true(all(names(prev) == names(prev_point)))
  expect_true(all(names(prev) ==
    names(prev_point)))

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: check user period prevalence function", {
  skip_on_cran()
  cdm <- mockIncidencePrevalenceRef()

  cdm <- generateDenominatorCohortSet(cdm = cdm, name = "denominator")

  prev <- estimatePrevalence(
    cdm = cdm,
    type = "period",
    denominatorTable = "denominator",
    outcomeTable = "outcome"
  )
  prev_period <- estimatePeriodPrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome"
  )

  expect_true(all(names(prev) == names(prev_period)))
  expect_true(all(names(prev) ==
    names(prev_period)))

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: multiple observation periods", {
  skip_on_cran()
  # create data for hypothetical people to test
  personTable <- tibble::tibble(
    person_id = c("1", "2"),
    gender_concept_id = c("8507", "8507"),
    year_of_birth = c(1998, 1976),
    month_of_birth = c(02, 06),
    day_of_birth = c(12, 01)
  )

  # three observation periods for 1 person
  # and a couple of consecutive events lost to washout
  observationPeriodTable <- tibble::tibble(
    observation_period_id = c("1", "2", "3", "4"),
    person_id = c("1", "1", "1", "2"),
    observation_period_start_date = c(
      as.Date("2005-04-01"),
      as.Date("2009-04-10"),
      as.Date("2010-08-20"),
      as.Date("2012-01-01")
    ),
    observation_period_end_date = c(
      as.Date("2005-11-29"),
      as.Date("2010-01-02"),
      as.Date("2011-12-11"),
      as.Date("2015-06-01")
    )
  )

  conditionX <- tibble::tibble(
    cohort_definition_id = c( 1, 1, 1),
    subject_id = c("1", "1", "2"),
    cohort_start_date = c(
      as.Date("2005-04-01"),
      as.Date("2009-06-10"),
      as.Date("2013-01-01")
    ),
    cohort_end_date = c(
      as.Date("2005-11-29"),
      as.Date("2010-01-02"),
      as.Date("2015-01-01")
    )
  )

  outcomeTable <- tibble::tibble(
    cohort_definition_id = c(1, 1, 1, 1, 1, 1, 1),
    subject_id = c("1", "1", "1", "1", "1", "1", "2"),
    cohort_start_date = c(
      as.Date("2005-08-09"),
      as.Date("2005-08-10"),
      as.Date("2005-08-11"),
      as.Date("2009-11-11"),
      as.Date("2009-11-21"),
      as.Date("2010-12-21"),
      as.Date("2014-04-04")
    ),
    cohort_end_date = c(
      as.Date("2005-08-09"),
      as.Date("2005-08-10"),
      as.Date("2005-08-11"),
      as.Date("2009-11-11"),
      as.Date("2009-11-21"),
      as.Date("2010-12-21"),
      as.Date("2014-04-04")
    )
  )

  cdm <- mockIncidencePrevalenceRef(
    personTable = personTable,
    observationPeriodTable = observationPeriodTable,
    targetCohortTable = conditionX,
    outcomeTable = outcomeTable
  )

  cdm <- generateTargetDenominatorCohortSet(
    cdm = cdm, name = "denominator",
    targetCohortTable = "target",
    targetCohortId = 1
  )
  ppe <- estimatePeriodPrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "years",
    fullContribution = FALSE,
    completeDatabaseIntervals = FALSE,
    minCellCount = 0
  )
  # nobody should appear in 2006
  expect_true(ppe %>%
    dplyr::filter(prevalence_start_date == "2006-01-01") %>%
    dplyr::pull("n_population") == 0)
  expect_true(ppe %>%
    dplyr::filter(prevalence_start_date == "2006-01-01") %>%
    dplyr::pull("n_cases") == 0)

  # one person with an event in 2005
  expect_true(ppe %>%
                dplyr::filter(lubridate::year(prevalence_start_date) == "2005") %>%
                dplyr::pull("n_population") == 1)
  expect_true(ppe %>%
                dplyr::filter(lubridate::year(prevalence_start_date) == "2005") %>%
                dplyr::pull("n_cases") == 1)

  # as for point prevalence, we would expect no positive n_cases at default
  ppo <- estimatePointPrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "months",
    minCellCount = 0
  )
  expect_true(sum(ppo$n_cases) == 0)

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: check confidence intervals", {
  skip_on_cran()
  cdm <- mockIncidencePrevalenceRef(sampleSize = 1000)
  cdm <- generateDenominatorCohortSet(
    cdm = cdm, name = "denominator"
  )
  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "point",
    interval = "years",
    minCellCount = 0
  ) %>%
    dplyr::filter(n_population > 1)

  # compare our wilson CIs with those from Hmisc
  hmisc_ci <- Hmisc::binconf(prev$n_cases, prev$n_population,
    alpha = 0.05,
    method = c("wilson"),
    return.df = TRUE
  )

  expect_equal(prev$prevalence_95CI_lower, hmisc_ci$Lower,
    tolerance = 1e-2
  )
  expect_equal(prev$prevalence_95CI_upper, hmisc_ci$Upper,
    tolerance = 1e-2
  )

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: check attrition", {
  skip_on_cran()
  cdm <- mockIncidencePrevalenceRef(sampleSize = 1000)
  cdm <- generateDenominatorCohortSet(
    cdm = cdm, name = "denominator",
    sex = c("Male", "Female")
  )
  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "point",
    interval = "years",
    minCellCount = 0
  )

  # for female cohort we should have a row for those excluded for not being male
  expect_true(any("Not Female" == prevalenceAttrition(prev) %>%
    dplyr::filter(denominator_sex == "Female") %>%
    dplyr::pull(.data$reason)))
  # for male, the opposite
  expect_true(any("Not Male" == prevalenceAttrition(prev) %>%
    dplyr::filter(denominator_sex == "Male") %>%
    dplyr::pull(.data$reason)))

  # check we can pick out specific analysis attrition
  expect_true(nrow(prevalenceAttrition(result = prev) %>%
    dplyr::filter(analysis_id == 1)) > 1)
  expect_true(nrow(prevalenceAttrition(result = prev) %>%
    dplyr::filter(analysis_id == 2)) > 1)

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: check attrition with complete database intervals", {
  skip_on_cran()
  personTable <- tibble::tibble(
    person_id = c("1", "2", "3"),
    gender_concept_id = "8507",
    year_of_birth = 2000,
    month_of_birth = 01,
    day_of_birth = 01
  )
  observationPeriodTable <- tibble::tibble(
    observation_period_id = c("1", "2", "3"),
    person_id = c("1", "2", "3"),
    observation_period_start_date = c(
      as.Date("2000-06-01"),
      as.Date("2000-06-01"),
      as.Date("2000-06-01")
    ),
    observation_period_end_date = c(
      as.Date("2011-07-01"),
      as.Date("2012-06-01"),
      as.Date("2000-06-15")
    )
  )
  outcomeTable <- tibble::tibble(
    cohort_definition_id = 1,
    subject_id = "1",
    cohort_start_date = c(
      as.Date("2008-02-05"),
      as.Date("2010-02-08"),
      as.Date("2010-02-20")
    ),
    cohort_end_date = c(
      as.Date("2008-02-05"),
      as.Date("2010-02-08"),
      as.Date("2010-02-20")
    )
  )

  cdm <- mockIncidencePrevalenceRef(
    personTable = personTable,
    observationPeriodTable = observationPeriodTable,
    outcomeTable = outcomeTable
  )

  cdm <- generateDenominatorCohortSet(
    cdm = cdm, name = "denominator"
  )
  prev <- estimatePrevalence(cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    type = "point",
    interval = "years",
    completeDatabaseIntervals = TRUE,
    minCellCount = 0
  )

  expect_true(prevalenceAttrition(prev) %>%
    dplyr::filter(reason == "Not observed during the complete database interval") %>%
    dplyr::pull("excluded_subjects") == 1)

  # check min cell suppression
  prev2 <- estimatePrevalence(cdm,
                             denominatorTable = "denominator",
                             outcomeTable = "outcome",
                             type = "point",
                             interval = "years",
                             completeDatabaseIntervals = TRUE,
                             minCellCount = 5
  )

  expect_true(prevalenceAttrition(prev2) %>%
                dplyr::filter(reason == "Not observed during the complete database interval") %>%
                dplyr::pull("excluded_subjects") == "<5")

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: check compute permanent", {
  skip_on_cran()

  # using permanent (no prefix)
  cdm <- mockIncidencePrevalenceRef(sampleSize = 1000)
  attr(attr(cdm, "cdm_source"), "write_schema") <- "main"

  cdm <- generateDenominatorCohortSet(
    cdm = cdm, name = "dpop"
  )
  prev <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "dpop",
    outcomeTable = "outcome",
    interval = "years",
    temporary = FALSE
  )
  # no temp tables created by dbplyr
  expect_false(any(stringr::str_starts(
      CDMConnector::listTables(attr(attr(cdm, "cdm_source"), "dbcon"),
                               schema = attr(attr(cdm, "cdm_source"), "write_schema")
      ),
    "dbplyr_"
  )))

  prev <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "dpop",
    outcomeTable = "outcome",
    interval = "years",
    type = "point",
    temporary = FALSE,
    returnParticipants = TRUE
  )
  expect_true(any(stringr::str_detect(
    CDMConnector::listTables(attr(attr(cdm, "cdm_source"), "dbcon"),
      schema = attr(attr(cdm, "cdm_source"), "write_schema")
    ),
    "point_prev_participants1"
  )))

  prev <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "dpop",
    outcomeTable = "outcome",
    interval = "years",
    type = "period",
    temporary = FALSE,
    returnParticipants = TRUE
  )
  expect_true(any(stringr::str_detect(
    CDMConnector::listTables(attr(attr(cdm, "cdm_source"), "dbcon"),
      schema = attr(attr(cdm, "cdm_source"), "write_schema")
    ),
    "period_prev_participants1"
  )))

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: check participants", {
  skip_on_cran()

  cdm <- mockIncidencePrevalenceRef(sampleSize = 100)

  cdm <- generateDenominatorCohortSet(
    cdm = cdm, name = "dpop",
    sex = c("Male", "Female", "Both"),
    ageGroup = list(
      c(0, 50),
      c(51, 100)
    )
  )

  start_tables <- CDMConnector::listTables(attr(attr(cdm, "cdm_source"), "dbcon"),
                                           schema = attr(attr(cdm, "cdm_source"), "write_schema"))

  prev <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "dpop",
    outcomeTable = "outcome",
    temporary = FALSE,
    returnParticipants = TRUE
  )
  end_tables <- CDMConnector::listTables(attr(attr(cdm, "cdm_source"), "dbcon"),
                                           schema = attr(attr(cdm, "cdm_source"), "write_schema"))

  new_tables <- setdiff(end_tables, start_tables)
  # we should have one new database table with our participants
  # everything else should have been cleaned up along the way
  expect_true(length(new_tables) == 1)

  prev_participants <- participants(prev, 1) %>% dplyr::collect()
  expect_equal(names(prev_participants),
    c("subject_id",
      "cohort_start_date",
      "cohort_end_date",
      "outcome_start_date"))

  expect_true(nrow(prev_participants %>%
    dplyr::filter(is.na(cohort_start_date))) == 0)

  prev2 <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "dpop",
    interval = "months",
    outcomeTable = "outcome",
    temporary = FALSE,
    returnParticipants = TRUE
  )

  end_tables2 <- CDMConnector::listTables(attr(attr(cdm, "cdm_source"), "dbcon"),
                                         schema = attr(attr(cdm, "cdm_source"), "write_schema"))

  new_tables2 <- setdiff(end_tables2, start_tables)
  # we should have another new database table with our participants from the
  # 2nd analysis
  expect_true(length(new_tables2) == 2)

  # with participants from first analysis unchanged
  expect_equal(prev_participants,
               participants(prev, 1) %>%
                 dplyr::collect())

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: overwriting participants", {
  skip_on_cran()

  cdm <- mockIncidencePrevalenceRef(sampleSize = 1000)

  cdm <- generateDenominatorCohortSet(
    cdm = cdm, name = "dpop",
    ageGroup = list(
      c(0, 50),
      c(51, 100)
    )
  )
  prev1 <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "dpop",
    denominatorCohortId = 1,
    outcomeTable = "outcome",
    temporary = FALSE,
    returnParticipants = TRUE
  )
  prev1_count <- nrow(participants(prev1, 1) %>% dplyr::collect())

  prev2 <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "dpop",
    denominatorCohortId = 2,
    outcomeTable = "outcome",
    temporary = FALSE,
    returnParticipants = TRUE
  )
  # participants from prev1 should still be the same
  # (ie no interference from having rerun the function)
  expect_true(nrow(participants(prev1, 1) %>% dplyr::collect()) == prev1_count)

  # we should have two tables with participants
  # one for each function call
  expect_true(length(stringr::str_subset(
    CDMConnector::listTables(attr(attr(cdm, "cdm_source"), "dbcon"),
      schema = attr(attr(cdm, "cdm_source"), "write_schema")
    ),
    "participants"
  )) == 2)

  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: if missing cohort attributes", {
  # missing cohort_set
  cdm <- mockIncidencePrevalenceRef()
  cdm <- generateDenominatorCohortSet(cdm = cdm, name = "denominator")
  attr(cdm$outcome, "cohort_set") <- NULL
  expect_error(estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "years"
  ))
  CDMConnector::cdm_disconnect(cdm)

  # missing cohort_count
  cdm <- mockIncidencePrevalenceRef()
  cdm <- generateDenominatorCohortSet(cdm = cdm, name = "denominator")
  attr(cdm$outcome, "cohort_attrition") <- NULL
  expect_error(estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "years"
  ))
  CDMConnector::cdm_disconnect(cdm)
})

test_that("mock db: test empty outcome table works", {
  skip_on_cran()

  cdm <- mockIncidencePrevalenceRef(sampleSize = 1000)

  cdm[["outcome"]] <- cdm[["outcome"]] %>% dplyr::filter(cohort_definition_id == 33)

  cdm <- generateDenominatorCohortSet(cdm = cdm, name = "denominator")

  expect_no_error(estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "years"
  ))

})

test_that("mock db: prevalence using strata vars", {

  cdm <- mockIncidencePrevalenceRef(sampleSize = 1000,
                                    outPre = 0.7)

  cdm <- generateDenominatorCohortSet(cdm = cdm,
                                      name = "denominator")

  prev_orig <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "years"
  )

  cdm$denominator <- cdm$denominator %>%
    dplyr::mutate(my_strata = dplyr::if_else(year(cohort_start_date) < 1990,
                                             "first", "second")) %>%
    dplyr::compute()

  prev <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "years",
    strata = list(c("my_strata"))
  )
  expect_true(all(c("strata_name", "strata_level") %in%
                    colnames(prev)))
  expect_true(all(c("Overall", "my_strata") %in%
                    unique(prev %>%
                             dplyr::pull("strata_name"))))
  expect_true(all(c("Overall",
                    "first", "second") %in%
                    unique(prev %>%
                             dplyr::pull("strata_level"))))

  # original without strata should be the same as "Overall" strata
  expect_equal(prev_orig,
               prev %>%
                 dplyr::filter(strata_name == "Overall") %>%
                 dplyr::select(!c("strata_name", "strata_level")))


  cdm$denominator <- cdm$denominator %>%
    dplyr::mutate(my_strata2 =  dplyr::if_else(month(cohort_start_date)<7,
                                               "a", "b")) %>%
    dplyr::compute()

  prev2 <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "years",
    strata = list(c("my_strata","my_strata2"))
  )
  expect_true(all(c("strata_name", "strata_level") %in%
                    colnames(prev2)))
  expect_true(all(c("Overall", "my_strata and my_strata2") %in%
                    unique(prev2 %>%
                             dplyr::pull("strata_name"))))
  expect_true(all(c("Overall",
                    "first and a", "first and b",
                    "second and a", "second and b") %in%
                    unique(prev2 %>%
                             dplyr::pull("strata_level"))))

  prev3 <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "years",
    strata = list(c("my_strata"),
                  c("my_strata2"),
                  c("my_strata", "my_strata2")))

  expect_true(all(c("strata_name", "strata_level") %in%
                    colnames(prev3)))
  expect_true(all(c("Overall", "my_strata", "my_strata2",
                    "my_strata and my_strata2") %in%
                    unique(prev3 %>%
                             dplyr::pull("strata_name"))))
  expect_true(all(c("Overall",
                    "first", "second",
                    "a", "b",
                    "first and a", "first and b",
                    "second and a", "second and b") %in%
                    unique(prev3 %>%
                             dplyr::pull("strata_level"))))


  # without overall strata
  prev4 <- estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "years",
    strata = list(c("my_strata"),
                  c("my_strata2"),
                  c("my_strata", "my_strata2")),
    includeOverallStrata = FALSE)
  expect_false("Overall" %in% unique(prev4 %>%
                                       dplyr::pull("strata_name")))
  expect_true(all(c("my_strata", "my_strata2",
                    "my_strata and my_strata2") %in%
                    unique(prev4 %>%
                             dplyr::pull("strata_name"))))
  expect_false("Overall" %in% unique(prev4 %>%
                                       dplyr::pull("strata_level")))
  expect_true(all(c("first", "second",
                    "a", "b",
                    "first and a", "first and b",
                    "second and a", "second and b") %in%
                    unique(prev4 %>%
                             dplyr::pull("strata_level"))))

  expect_error(estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "years",
    strata = list(c("not_a_col"))))

  expect_error(estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "years",
    strata = list(c("my_strata", "not_a_col"))))

  expect_error(estimatePrevalence(
    cdm = cdm,
    denominatorTable = "denominator",
    outcomeTable = "outcome",
    interval = "years",
    strata = list(c("my_strata"), c("not_a_col"))))

  CDMConnector::cdm_disconnect(cdm)
})