printInc <- FALSE

test_that("GaussSuppressionFromData works", {
  expect_equal(which(GaussSuppressionFromData(SSBtoolsData("z1"), 1:2, 3, printInc = printInc)$suppressed), c(12, 13, 22, 23, 42, 43))
})


# Sample with seed inside test_that do not work
z3 <- SSBtoolsData("z3")
upper <- z3$region %in% LETTERS
z3$region[upper] <- paste0(z3$region[upper], 2)
z3$region[!upper] <- paste0(toupper(z3$region[!upper]), 1)

mm <- SSBtools::ModelMatrix(z3[, 1:6], crossTable = TRUE, sparse = FALSE)
x <- mm$modelMatrix  
k <- 1:20000
set.seed(123)
sample_k <- sample(k)
x[k] <- x[sample_k]


test_that("Advanced with integer overflow", {
  #skip("Strange behaviour. Test works, but not when run inside Check package")
  skip_on_cran()  # The above problem was caused by different character sorting in different systems
  
  a <- GaussSuppressionFromData(z3, c(1:6), 7, x = mm$modelMatrix , crossTable = mm$crossTable, maxN = 5, singletonMethod = "anySumOld", printInc = printInc)
  expect_identical(sum(which(a$suppressed)), 599685L)
  
  # This test involves integer overflow in AnyProportionalGaussInt  
  a <- GaussSuppressionFromData(z3, c(1:6), 7, x = x, crossTable = mm$crossTable, singletonMethod = "anySumOld", printInc = printInc)
  expect_identical(sum(which(a$suppressed)), 525957L)
  
  # This test involves integer overflow in AnyProportionalGaussInt  
  a <- GaussSuppressionFromData(z3, c(1:6), 7, x = x, crossTable = mm$crossTable, protectZeros = FALSE, secondaryZeros = TRUE, singletonMethod = "anySumNOTprimaryOld", printInc = printInc)
  expect_identical(sum(which(a$suppressed)), 411693L)
  
  # This test involves all ways of updating A$r[[i]], A$x[[i]], B$r[[i]], B$x[[i]]  (Including integer overflow)
  a <- GaussSuppressionFromData(z3, c(1:6), 7, x = x, crossTable = mm$crossTable, protectZeros = FALSE, secondaryZeros = TRUE, testMaxInt = 10, singletonMethod = "anySumNOTprimaryOld", printInc = printInc)
  expect_identical(sum(which(a$suppressed)), 411693L)
  
  a <- GaussSuppressionFromData(z3, c(1:6), 7, x = x, crossTable = mm$crossTable, protectZeros = FALSE, secondaryZeros = TRUE, allNumeric = TRUE, singletonMethod = "anySumNOTprimaryOld", printInc = printInc)
  expect_identical(sum(which(a$suppressed)), 411693L)
  
  # This test involves TRUE return in AnyProportionalGaussInt after ReduceGreatestDivisor (identical length 3 vectors)
  x[, 201:300] <- round(0.6 * x[, 201:300] + 0.6 * x[, 301:400])
  a <- GaussSuppressionFromData(z3, c(1:6), 7, x = x, crossTable = mm$crossTable, singletonMethod = "anySumOld", printInc = printInc)
  expect_identical(sum(which(a$suppressed)), 576555L)
  
})


test_that("structuralEmpty and removeEmpty", {
  expect_warning(a1 <- GaussSuppressionFromData(z3[100:300, ], 1:6, 7, printInc = printInc))
  a2 <- GaussSuppressionFromData(z3[100:300, ], 1:6, 7, printInc = printInc, structuralEmpty = TRUE)
  a3 <- GaussSuppressionFromData(z3[100:300, ], 1:6, 7, printInc = printInc, removeEmpty = TRUE)
  k <- a1$suppressed != a2$suppressed
  expect_equal(a1[!k, ], a3, ignore_attr = TRUE)
  expect_equal(a2[!k, ], a3, ignore_attr = TRUE)
  expect_equal(unique(a1[k, "ant"]), 0)
})



test_that("extend0 and various hierarchy input", {
  z2 <- SSBtoolsData("z2")
  dimLists <- SSBtools::FindDimLists(z2[, -5])
  hi <- list(c("region", "fylke", "kostragr"), hovedint = dimLists$hovedint)
  
  a1 <- GaussSuppressionFromData(z2, 1:4, 5, printInc = printInc)
  a2 <- GaussSuppressionFromData(z2, freqVar = "ant", hierarchies = dimLists, printInc = printInc)
  a3 <- GaussSuppressionFromData(z2, freqVar = "ant", hierarchies = hi, printInc = printInc)
  
  expect_identical(a1, a2)
  expect_identical(a3, a2)
  
  z2_ <- z2[z2$ant != 0, ]
  
  a1 <- GaussSuppressionFromData(z2_, 1:4, 5, extend0 = TRUE, output = "publish_inner", printInc = printInc)
  
  expect_identical(a1$publish, a2)
  
  a2 <- GaussSuppressionFromData(z2_, freqVar = "ant", hierarchies = dimLists, extend0 = TRUE, output = "publish_inner", printInc = printInc)
  a3 <- GaussSuppressionFromData(z2_, freqVar = "ant", hierarchies = hi, extend0 = TRUE, output = "publish_inner", printInc = printInc)
  
  if (FALSE) { # Include code that shows differences 
    tail(a1$inner)
    tail(a2$inner)
    tail(a3$inner)
  }
  
  expect_identical(a1$publish, a2$publish)
  expect_identical(a3$publish, a2$publish)
  
  expect_equal(a1$inner[names(a2$inner)], a2$inner, ignore_attr = TRUE)
  expect_equal(a3$inner[names(a1$inner)], a1$inner, ignore_attr = TRUE)
  
  a1_ <- GaussSuppressionFromData(z2_, 1:4, 5, extend0 = "all", output = "publish_inner", printInc = printInc)
  a2_ <- GaussSuppressionFromData(z2_, freqVar = "ant", hierarchies = dimLists, extend0 = "all", output = "publish_inner", printInc = printInc)
  a3_ <- GaussSuppressionFromData(z2_, freqVar = "ant", hierarchies = hi, extend0 = "all", output = "publish_inner", printInc = printInc)
  
  expect_identical(a1, a1_)
  expect_identical(a2, a2_)
  expect_identical(a3, a3_)
  
  z2__ <- z2_[z2_$hovedint != "trygd", ]
  
  a2 <- GaussSuppressionFromData(z2__, freqVar = "ant", hierarchies = dimLists, extend0 = "all", output = "publish_inner", printInc = printInc)
  a3 <- GaussSuppressionFromData(z2__, freqVar = "ant", hierarchies = hi, extend0 = "all", output = "publish_inner", printInc = printInc)
  
  expect_identical(a3$publish, a2$publish)
  expect_equal(a3$inner[names(a2$inner)], a2$inner, ignore_attr = TRUE)
  
  expect_identical(lapply(c(a2, a3), dim), lapply(c(a2_, a3_), dim))
  
  z2___ <- z2__[z2__$fylke != 10, ]
  
  a2_ <- GaussSuppressionFromData(z2___, freqVar = "ant", hierarchies = dimLists, extend0 = "all", output = "publish_inner", printInc = printInc)
  a3_ <- GaussSuppressionFromData(z2___, freqVar = "ant", hierarchies = hi, extend0 = "all", output = "publish_inner", printInc = printInc)
  
  expect_identical(lapply(a2, dim), lapply(a2_, dim))
  
  expect_true(nrow(a3_$inner) < nrow(a3$inner))
  expect_true(nrow(a3_$publish) < nrow(a3$publish))
})



test_that("DominanceRule and NcontributorsRule + CandidatesNum + singleton + forced/unsafe", {
  set.seed(123)
  z <- SSBtools::MakeMicro(SSBtoolsData("z2"), "ant")
  z$char <- sample(paste0("char", 1:10), nrow(z), replace = TRUE)
  z$value <- rnorm(nrow(z))^2
  
  a <- GaussSuppressionFromData(z, dimVar = c("region", "fylke", "kostragr", "hovedint"), numVar = "value", charVar = "char", 
                                candidates = CandidatesNum, primary = DominanceRule, singletonMethod = "sub2Sum",
                                n = c(1, 2), k = c(65, 85), printInc = printInc)
  
  
  b <- GaussSuppressionFromData(z, dimVar = c("region", "fylke", "kostragr", "hovedint"), numVar = "value", charVar = "char", 
                                candidates = CandidatesNum, primary = NcontributorsRule, singletonMethod = "none",
                                removeCodes = paste0("char", 1:2), printInc = printInc)
  
  expect_identical(as.numeric(which(a$primary)), c(8, 17, 18, 23, 52, 53, 58, 63, 73, 77, 78, 80, 83, 87, 90, 92, 97, 98))
  expect_identical(as.numeric(which(b$primary)), c(8, 18, 23, 53, 63, 78, 83, 87, 90, 97, 98))
  
  
  z$seq2 <- (1:nrow(z))^2 
  
  aseq2 <- GaussSuppressionFromData(z, dimVar = c("region", "fylke", "kostragr", "hovedint"), 
                                    numVar = c("seq2", "value"), 
                                    candidatesVar = "value",
                                    dominanceVar = "value",
                                    charVar = "char", candidates = CandidatesNum, 
                                    primary = DominanceRule, singletonMethod = "sub2Sum",
                                    n = c(1, 2), k = c(65, 85), printInc = printInc)
  
  expect_identical(a[names(a)], aseq2[names(a)])
  
  
  z$char <- paste0("char", 1:nrow(z))
  d1 <- GaussSuppressionFromData(z, dimVar = c("region", "fylke", "kostragr", "hovedint"), numVar = "value", charVar = "char", 
                                candidates = CandidatesNum, primary = NcontributorsRule, singletonMethod = "none",
                                removeCodes = paste0("char", 1:20), printInc = printInc, 
                                freqVar = "ant", preAggregate = FALSE, maxN = 10,
                                whenEmptyUnsuppressed = "stop")
  
  d2 <- GaussSuppressionFromData(z, dimVar = c("region", "fylke", "kostragr", "hovedint"), numVar = "value", 
                                 candidates = CandidatesNum, primary = NContributorsRule, singletonMethod = "none",
                                 removeCodes = 1:20, printInc = printInc, 
                                 preAggregate = FALSE, maxN = 10, # Empty freq in CandidatesNum
                                 whenEmptyUnsuppressed = "stop") 
  
  expect_equal(d1[names(d1) != "ant"], d2, ignore_attr = TRUE)
  
  
  if(TRUE){   
    set.seed(123)
    z$value <- rnorm(nrow(z))^2  # Need to generate again ... not same as above 
    set.seed(1986) # Seed is not randomly chosen
    z$char <- sample(paste0("char", c(1, 1, 1, 1, 1, 2, 2, 2, 3, 4)), nrow(z), replace = TRUE)
    b0 <- GaussSuppressionFromData(z, dimVar = c("region", "fylke", "kostragr", "hovedint"), numVar = "value", charVar = "char", 
                                   maxN = 2, candidates = CandidatesNum, primary = NcontributorsRule, printInc = printInc, 
                                   singleton = SingletonUniqueContributor, 
                                   singletonMethod = "none") 
    b1 <- GaussSuppressionFromData(z, dimVar = c("region", "fylke", "kostragr", "hovedint"), numVar = "value", charVar = "char", 
                                   maxN = 2, candidates = CandidatesNum, primary = NcontributorsRule, printInc = printInc, 
                                   singleton = SingletonUniqueContributor, 
                                   singletonMethod = "sub2Sum")
    b2 <- GaussSuppressionFromData(z, dimVar = c("region", "fylke", "kostragr", "hovedint"), numVar = "value", charVar = "char", 
                                   maxN = 2, candidates = CandidatesNum, primary = NcontributorsRule, printInc = printInc, 
                                   singleton = SingletonUniqueContributor, 
                                   singletonMethod = "numFTT") 
    suppressWarnings({b3 <- GaussSuppressionFromData(z, dimVar = c("region", "fylke", "kostragr", "hovedint"), numVar = "value", charVar = "char", 
                                   maxN = 2, candidates = CandidatesNum, 
                                   primary = c(63, 73, 77),   # primary = c(8, 18, 23, 53, 63, 73, 77, 78, 90, 97, 98, 100), 
                                   forced = c(11, 13, 18, 20, 40),
                                   printInc = printInc, 
                                   singleton = SingletonUniqueContributor, 
                                   singletonMethod = "numFTT")}) 
    suppressWarnings({b4 <- GaussSuppressionFromData(z, dimVar = c("region", "fylke", "kostragr", "hovedint"), numVar = "value", charVar = "char", 
                                   maxN = 2, candidates = CandidatesNum, 
                                   primary = c(8, 18, 23, 53, 63, 73, 77, 78, 90, 97, 98, 100), 
                                   forced = c(11, 13, 18, 20, 40),
                                   printInc = printInc, 
                                   singleton = SingletonUniqueContributor, 
                                   singletonMethod = "numFTT")}) 
    
    suppressWarnings({b5 <- GaussSuppressionFromData(z, dimVar = c("region", "fylke", "kostragr", "hovedint"), numVar = "value", charVar = "char", 
                                                     maxN = 2, candidates = CandidatesNum, 
                                                     primary = c(8, 18, 23, 53, 63, 73, 77, 78, 90, 97, 98, 100), 
                                                     forced =  c(11, 13, 18, 20, 40),
                                                     printInc = printInc,
                                                     protectZeros = TRUE)})
    
    
    suppressWarnings({b6 <- GaussSuppressionFromData(z, dimVar = c("region", "fylke", "kostragr", "hovedint"), numVar = "value", charVar = "char", 
                                                     maxN = 2, candidates = CandidatesNum, 
                                                     primary = c(8, 18, 23, 53, 63, 73, 77, 78, 90, 97, 98, 100), 
                                                     forced = 1:30,
                                                     printInc = printInc,
                                                     protectZeros = FALSE)})
    
    
    expect_equal(sum(b0$suppressed), 32)
    expect_equal(sum(b1$suppressed), 33)
    expect_equal(sum(b2$suppressed), 35)
    expect_equal(sum(b3$suppressed), 12)
    expect_equal(sum(b4$suppressed), 32)
    expect_equal(sum(b5$suppressed), 27)
    expect_equal(sum(b6$suppressed), 19)
    expect_equal(sum(b3$unsafe), 0)
    expect_equal(sum(b4$unsafe), 1)
    expect_equal(sum(b5$unsafe), 1)
    expect_equal(sum(b6$unsafe), 3)
    
    skip_on_cran()
    
    # Code to see differences:
    #"sub2Sum" solves G-problem 
    #"numFTT" needed to solve K-problem. 
    if (FALSE) for (myChar in c("G", "K")) {
      kp <- b0[b0$region == myChar & b0$primary, ]
      k0 <- b0[b0$region == myChar & b0$suppressed, ]
      k1 <- b1[b2$region == myChar & b1$suppressed, ]
      k2 <- b2[b2$region == myChar & b2$suppressed, ]
      cat("===============", myChar, "=============== \n")
      for (kk in c("kp", "k0", "k1", "k2")) {
        cat("   -----", kk, "-----\n")
        ma <- Match(z[c("region", "hovedint")], get(kk)[c("region", "hovedint")])
        print(z[!is.na(ma), ])
      }
    }
    sn <- c(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 2, 0, 0, 0, 0, 1, 0, 1, 
           0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0)
    sf <- c(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 1, 
      0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0)
    sum_suppressed <- integer(0)
    for (m1 in c("none", "anySumNOTprimary")) 
      for (m2 in c("none", "sub2Sum", "numFTT")) {
        b <- GaussSuppressionFromData(z, 
                                    dimVar = c("region", "fylke", "kostragr", "hovedint"), 
                                    numVar = "value", charVar = "char", maxN = 2, 
                                    candidates = CandidatesNum, 
                                    primary = NcontributorsRule, 
                                    printInc = printInc, 
            singleton = list(freq = as.logical(sf), num = as.integer(sn)), 
            singletonMethod = c(freq = m1, num = m2))
            sum_suppressed <- c(sum_suppressed, sum(b$suppressed))
      }
    expect_equal(sum_suppressed, c(32, 33, 35, 35, 38, 40))
    
    
    set.seed(1138)
    sum_suppressed <- integer(0)
    zz = z[sample.int(nrow(z), 100, replace = TRUE), ]
    for (c2 in c("F", "T")) 
      for (c3 in c("F", "T", "H"))
       for (c4 in c("F", "T")) {
        b <- GaussSuppressionFromData(zz, 
                                      dimVar = c("region", "fylke", "kostragr", "hovedint"), 
                                      numVar = "value", charVar = "char", 
                                      maxN = 2, printInc = printInc, 
                                      candidates = CandidatesNum, 
                                      primary = NcontributorsRule,  
                                      singleton = SingletonUniqueContributor, 
                                      singletonMethod = paste0("numF", c2, c3, c4))
        sum_suppressed <- c(sum_suppressed, sum(b$suppressed))
      }
    expect_equal(sum_suppressed, c(49, 55, 51, 55, 53, 55, 49, 57, 52, 57, 55, 57))
    
    # Why extra primary needed for 5:Total when "numFTH"
    # can be seen by looking at 
    # b[b$region == 5, ]
    # zz[zz$fylke == 5 & zz$hovedint == "annet", ]
    # zz[zz$fylke == 5 & zz$hovedint == "arbeid", ]
    # zz[zz$fylke == 5 & zz$hovedint == "soshjelp", ]  
    
    sum_suppressed <- integer(0)
    for (singletonMethod  in c("numFFF", "numtFF","numTFF", "numtTT", "numtTH", "numtTFT", "numtTHT")) {
        b <- GaussSuppressionFromData(zz, 
                                      dimVar = c("region", "fylke", "kostragr", "hovedint"), 
                                      numVar = "value", charVar = "char", 
                                      maxN = 2, printInc = printInc, 
                                      candidates = CandidatesNum, 
                                      primary = NcontributorsRule,  
                                      singleton = SingletonUniqueContributor, 
                                      singletonMethod = singletonMethod,
          inputInOutput = c(FALSE, TRUE)) # singleton not in publish and therefore not primary suppressed  
        sum_suppressed <- c(sum_suppressed, sum(b$suppressed))
      }
    expect_equal(sum_suppressed, c(17, 18, 18, 19, 19, 23, 23))
    
    
    # To make non-suppressed singletons
    SUC <- function(..., removeCodes, primary) SingletonUniqueContributor(..., removeCodes = character(0), primary = integer(0))
    sum_suppressed <- integer(0)
    for (singletonMethod  in c("numFFF", "numtFF","numTFF")) {
      b <- GaussSuppressionFromData(zz, 
                                  dimVar = c("region", "fylke", "kostragr", "hovedint"), 
                                  numVar = "value", charVar = "char", 
                                  maxN = 2, printInc = printInc, 
                                  candidates = CandidatesNum, 
                                  primary = NcontributorsRule,  
                                  removeCodes = "char1",
                                  singleton = SUC, 
                                  singletonMethod = singletonMethod,
                                  whenEmptyUnsuppressed = NULL)
      sum_suppressed <- c(sum_suppressed, c(59, 59, 67))
    }
    
    zz$char[1:15] <- "char5"
    expect_warning({b <- GaussSuppressionFromData(zz, 
                                  dimVar = c("region", "fylke", "kostragr", "hovedint"), 
                                  numVar = "value", charVar = "char", 
                                  maxN = 2, printInc = printInc, 
                                  candidates = CandidatesNum, 
                                  primary = NcontributorsRule,  
                                  singleton = SingletonUniqueContributor, 
                                  singletonMethod = "numFTFW")})
    expect_equal(sum(b$suppressed), 51)  # Here "if (s_unique == primarySingletonNum[i])" in SSBtools::GaussSuppression matters. 
    
    
    set.seed(193)
    zz$A <- sample(paste0("A", c(1, 1, 1, 1, 1, 2, 2, 2, 3, 4)), nrow(zz), replace = TRUE)
    zz$B <- sample(paste0("B", c(1, 1, 1, 1, 1, 2, 2, 2, 3, 4)), nrow(zz), replace = TRUE)
    rcd <- data.frame(char = "char2", A = c("A1", "A2"), B = "B1")
    removeCodes <- list(NULL, rcd, as.list(rcd))
    k <- integer(0)
    for (specialMultiple in c(FALSE, TRUE)) for (i in 1:3) {
      b <- GaussSuppressionFromData(zz, 
                                    dimVar = c("region", "fylke", "kostragr", "hovedint"), 
                                    numVar = "value", charVar = c("char","A","B"), 
                                    maxN = 2, printInc = printInc, 
                                    candidates = CandidatesNum, 
                                    primary = NcontributorsRule,  
                                    singleton = SingletonUniqueContributor, 
                                    singletonMethod = "numTTTTT", output = "inputGaussSuppression",
                                    specialMultiple = specialMultiple,
                                    removeCodes = removeCodes[[i]])
      k <- c(k, 0L, as.vector(table(b$singleton)[as.character(unique(b$singleton))]))
    }
    expect_equal(sort(k), sort(c(0, 1, 1, 1, 1, 1, 2, 19, 1, 1, 1, 1, 0, 1, 1, 1, 1, 1, 
                      2, 20, 1, 1, 1, 0, 1, 29, 0, 2, 6, 3, 9, 9, 1, 0, 2, 
                      5, 3, 9, 10, 1, 0, 2, 5, 1, 1, 2, 17, 2)))
  }
})



test_that("Interpret primary output correctly", {
  x <- SSBtoolsData("sprt_emp_withEU")[, c(1, 2, 5, 3, 4)]
  
  p1 <- function(num, ...) round(10 * num[, 1])%%10 == 3
  p2 <- function(num, ...) round(10 * num)%%10 == 3
  p3 <- function(num, ...) as.data.frame(round(10 * num)%%10 == 3)
  p4 <- function(num, ...) list(primary = as.data.frame(round(10 * num)%%10 == 3), 
                                numExtra = data.frame(numExtra = round(10 * num[, 1])%%10))
  
  p12 <- function(...) {
    p <- p2(...)
    p[] <- as.integer(p)
    p
  }
  
  G <- function(primary, formula = ~eu * year + age:geo) {
    which(GaussSuppressionFromData(data = x, formula = formula, numVar = "ths_per", 
                                   primary = primary, singleton = NULL, 
                                   output = "inputGaussSuppression", 
                                   printInc = printInc)$primary)
  }
  
  # Case when x is square
  gp1 <- G(p1)
  expect_identical(G(p2), gp1)
  expect_identical(G(p3), gp1)
  expect_identical(G(p4), gp1)
  expect_identical(length(G(p12)), 0L)  # since interpret as xExtraPrimary

  # Case when x is not square
  gp1_ <- G(p1, formula = ~age * geo)
  expect_identical(G(p2, formula = ~age * geo), gp1_)
  expect_identical(G(p3, formula = ~age * geo), gp1_)
  expect_identical(G(p4, formula = ~age * geo), gp1_)
  expect_error(G(p12, formula = ~age * geo)) #  Error 0 index found in primary output (change to logical?)
  
  
  # Single column xExtraPrimary, Matrix and matrix 
  
  x$freq <- round(sqrt(x$ths_per) + as.integer(x$year) - 2014 + 0.2 * (-7:10))
  z <- x[x$year == "2014", -(4:5)]
  
  
  K <- function(primary) {
    GaussSuppressionFromData(data = z, formula = ~geo + age, freqVar = "freq", coalition=7, 
                             primary = primary, 
                             mc_hierarchies = NULL, upper_bound = Inf, 
                             protectZeros = FALSE, secondaryZeros = TRUE, 
                             output ="outputGaussSuppression_x", 
                             printInc = printInc)$xExtraPrimary
  }
  
  e1 <- K(KDisclosurePrimary)
  e2 <- K(function (...) as.matrix(KDisclosurePrimary(...)))
    
  expect_equal(max(abs(e2 - e1)), 0)
  expect_warning({e3 <- K(function (...) round(1 + 0.1*as.matrix(KDisclosurePrimary(...))))}) # Warning message: Primary output interpreted as xExtraPrimary (rare case of doubt)
  expect_true(all(dim(e3) == c(6, 1)))
  
})


test_that("More NumSingleton", {
  
  sum_suppressed <- integer(0)
  for (seed in c(116162, 643426)) {
    set.seed(seed)
    z <- SSBtoolsData("magnitude1")
    set.seed(seed)
    z$company <- z$company[sample.int(20)]
    z$value <- z$value[sample.int(20)]
    dataset <- SSBtools::SortRows(aggregate(z["value"], z[1:5], sum))
    for (c3 in c("F", "T", "H")) for (c4 in c("F", "t", "T")) for (c5 in c("F", "t", "T")) {
      if (!(c4 == "F" & c5 != "F")) {
        singletonMethod <- paste0("numTt", c3, c4, c5)
        output <- SuppressDominantCells(data = dataset, numVar = "value", dimVar = c("sector4", "geo"), contributorVar = "company", n = 1, k = 80, singletonMethod = singletonMethod,
                                        printInc = FALSE)
        sum_suppressed <- c(sum_suppressed, sum(output$suppressed))
      }
    }
    
  }
  
  expect_equal(sum_suppressed, c(8, 11, 13, 13, 11, 13, 13, 10, 11, 13, 13, 11, 13, 13, 10, 
                                 11, 13, 13, 11, 13, 13, 7, 9, 10, 12, 10, 11, 12, 8, 10, 10, 
                                 12, 11, 11, 12, 8, 10, 10, 12, 11, 11, 12))  
  
})



test_that("data.table and NA", {
  if (!requireNamespace("data.table", quietly = TRUE)) {
    skip()
  }
  
  z3 <- SSBtoolsData("z3")
  
  set.seed(123)
  z <- z3[sample.int(nrow(z3), 100), ]
  z <- z3[sample.int(nrow(z), 300, replace = TRUE), ]
  
  z$char <- sample(paste0("char", 1:10), nrow(z), replace = TRUE)
  z$value <- rnorm(nrow(z))^2
  z$pop <- c("1", "2")
  
  z[sample.int(nrow(z), 5), 1:3] <- NA
  z[sample.int(nrow(z), 5), 4] <- NA
  z[sample.int(nrow(z), 5), "pop"] <- NA
  
  
  f <- ~pop:(region + (fylke + kostragr) * hovedint) - 1
  
  a <- vector("list", 8)
  
  i <- 0
  for (NAomit in c(FALSE, TRUE)) 
    for (aggregateNA in c(FALSE, TRUE)) 
      for (aggregatePackage in c("base", "data.table")) {
        i <- i + 1
        a[[i]] <- SuppressDominantCells(data = z, 
                                        numVar = "value", 
                                        formula = f, 
                                        contributorVar = "char", 
                                        k = c(80, 90),
                                        NAomit = NAomit, 
                                        aggregateNA = aggregateNA, 
                                        aggregatePackage = aggregatePackage, 
                                        printInc = printInc)
        names(a)[i] <- paste(substr(as.character(NAomit), 1, 1), 
                             substr(as.character(aggregateNA), 1, 1), 
                             aggregatePackage,
                             sep = "_")
        
      }
  
  
  expect_equal(sum(a[["F_F_base"]]$freq), 1716)
  expect_equal(sum(a[["F_T_base"]]$freq), 1800)
  expect_equal(sum(a[["T_T_base"]]$freq), 1733)
  
  expect_equal(a[["T_F_base"]], a[["F_F_base"]])
  
  expect_equal(a[["F_F_base"]], a[["F_F_data.table"]])
  expect_equal(a[["F_T_base"]], a[["F_T_data.table"]])
  expect_equal(a[["T_F_base"]], a[["T_F_data.table"]])
  expect_equal(a[["T_T_base"]], a[["T_T_data.table"]])
  
})