require(gge)

# ----------------------------------------------------------------------------

# create data for tests

# matrix data
mat1 <- matrix(c(50, 55, 65, 50, 60, 65, 75,
                 67, 71, 76, 80, 82, 89, 95,
                 90, 93, 95, 102, 97, 106, 117,
                 98, 102, 105, 130, 135, 137, 133,
                 120, 129, 134, 138, 151, 153, 155),
               ncol=5, byrow=FALSE)
colnames(mat1) <- c("E1","E2","E3","E4","E5")
rownames(mat1) <- c("G1","G2","G3","G4","G5","G6","G7")

# One missing value in a matrix
mat2 <- mat1 ; mat2[1,1] <- NA


bar <- transform(lattice::barley, env=paste0(site,year))

# ----------------------------------------------------------------------------

test_that("minimum matrix size", {
  x33 <- matrix(c(2.2,3.1,4.0, 2.5,1.9,2.3, 1.5,2.7,2.8), ncol=3)
  m33 <- gge(x33)
  biplot(m33) # ok
  
  x32 <- x33[,1:2] # 2 columns
  m32 <- gge(x32)
  biplot(m32) # ok

  x31 <-  x33[,1,drop=FALSE] # 1 column
  expect_error(gge(x31)) # fails
  
  x23 <- x33[1:2,] # 2 rows
  expect_error(gge(x23)) # fails
})

test_that("errors with gge.matrix", {
  expect_error(gge(mat1,env.group=1:3)) # wrong length
  expect_error(gge(mat1,gen.group=1:3)) # wrong length
  expect_error(gge(mat1, method="NIP")) # unknown method
})


test_that("errors with gge.data.frame", {
  bar <- transform(lattice::barley, env=paste0(site,year))
  gge(bar, yield~variety*site)
  expect_error(gge(yield~variety*site)) # no data
  expect_error(gge(bar, yield~variety*loc)) # no 'loc'
  expect_error(gge(bar, yield~variety*site, gen.group=loc)) # no 'loc'
  expect_error(gge(bar, yield~variety*site, env.group=loc)) # no 'loc'
  bar$junk <- c('A','B','C','D')
  expect_error(m1 <- gge(bar, yield~variety*site, env.group=junk)) # multiple env.group
  expect_error(gge(bar, yield~variety*site, gen.group=junk)) # multiple gen.group
  
})


test_that("nipals",{
  mat2 <- mat1
  mat2[,1] <- 1
  expect_error(gge(mat2, method="nipals")) # constant column
  mat3 <- mat1
  mat3[1:5,1] <- NA
  expect_warning(gge(mat3, method="nipals")) # more than 10 percent missing
})


test_that("Checking arguments of biplot", {
  expect_silent(NULL) # hack to avoid testthat warning
  
  m11 <- gge(mat1)
  plot(m11)

  m21 <- gge(mat2)
  plot(m21)

  biplot(m11)
  biplot(m11, main="Example biplot", subtitle="GGE biplot") # message
  biplot(m11, main="Example biplot", subtitle="GGE biplot")
  
  biplot(m11, subtitle=NULL) # suppress options subtitle
  biplot(m11, xlab=NULL, ylab=NULL)
  biplot(m11, xlab="Axis 1", ylab="Axis 2")
  biplot(m11, main=NULL, subtitle=NULL) # suppress title & subtitle
  biplot(m11, cex.gen=0)
  biplot(m11, cex.gen=2)
  biplot(m11, cex.env=2)
  biplot(m11, col.gen="blue")
  biplot(m11, col.gen=c("blue","red")) # With 1 group, only use first
  biplot(m11, pch.gen=20) # Ignored with 1 group
  biplot(m11, col.env=c("red","blue")) # blue is ignored
  biplot(m11, comps=2:3)
  biplot(m11, lab.env=FALSE)
  
  # flips
  biplot(m11, flip="") # no flipping
  biplot(m11, flip=FALSE)
  biplot(m11, flip=TRUE)
  biplot(m11, flip=c(TRUE,FALSE))
  biplot(m11, flip=c(FALSE,TRUE))
  biplot(m11, flip=c(FALSE,FALSE))
  biplot(m11, flip=c(TRUE,TRUE))
  
  # zooming
  biplot(m11, zoom.gen=.8)
  biplot(m11, zoom.env=.8)
  biplot(m11, zoom.gen=.8, zoom.env=.8)
  
  # princomp methods
  m31 <- gge(mat1, method="svd")
  biplot(m31)
  m32 <- gge(mat1, method="nipals")
  biplot(m32)
  
  m34 <- gge(mat2) # should switch to 'nipals'
  biplot(m34)
  m35 <- gge(mat2, method="svd") # should switch to 'nipals'
  biplot(m35)
  m36 <- gge(mat2, method="nipals")
  biplot(m36)
  
  # matrix data with env.group, gen.group
  m24 <- gge(mat2, env.group=c(1,1,1,2,2))
  biplot(m24)
  m25 <- gge(mat2, gen.group=c(1,1,1,1,2,2,2))
  biplot(m25, col.gen=c('blue','red'), pch.gen=1:2) # group colors, symbols
  
  # Environment groups. Use the lattice::barley data
  require(lattice)
  m32 <- gge(bar, yield~variety*env)
  biplot(m32)
  m33 <- gge(bar, yield~variety*env, env.group=year) # env.group
  plot(m33)
  biplot(m33)
  biplot(m33, lab.env=FALSE) # label locs
  biplot(m33, lab.env=TRUE) # default is to label locs
  # Option to disable residual vectors
  biplot(m33, res.vec=TRUE) # default is to label locs
  biplot(m33, res.vec=FALSE) # default is to label locs
})


test_that("Polygon hull.  Yan 2006 fig 12", {
  require(agridat)
  dat <- yan.winterwheat
  m1 <- gge(dat, yield ~ gen*env, scale=FALSE)
  expect_silent(
    biplot(m1, main="yan.winterwheat - GGE biplot",
           flip=c(1,0), hull=TRUE)
  )
  
  expect_silent(
    biplot(m1, main="yan.winterwheat - GGE biplot",
           flip=c(1,0), origin=0,  hull=TRUE)
  )
})


test_that("AEC biplot option", {
  m11 <- gge(mat1)

  # Basic AEC biplot runs silently
  expect_silent(biplot(m11, AEC = TRUE, cex.env=1))

  # AEC with no environment labels (points instead)
  expect_silent(biplot(m11, AEC = TRUE, lab.env = FALSE))

  # AEC with axis flipping
  expect_silent(biplot(m11, AEC = TRUE, flip = c(TRUE, FALSE)))
  expect_silent(biplot(m11, AEC = TRUE, flip = c(FALSE, TRUE)))
  expect_silent(biplot(m11, AEC = TRUE, flip = c(TRUE, TRUE)))

  # AEC with zoom options
  expect_silent(biplot(m11, AEC = TRUE, zoom.gen = 0.8))
  expect_silent(biplot(m11, AEC = TRUE, zoom.env = 0.8))

  # AEC with alternate components
  expect_silent(biplot(m11, AEC = TRUE, comps = 2:3))

  # AEC with env.group: unit-circle is suppressed (scale=TRUE default)
  m_grp <- gge(mat1, env.group = c(1, 1, 1, 2, 2))
  expect_silent(biplot(m_grp, AEC = TRUE))
  expect_silent(biplot(m_grp, AEC = TRUE, lab.env = FALSE))

  # AEC with unscaled model (no unit circle drawn)
  m_unscaled <- gge(mat1, scale = FALSE)
  expect_silent(biplot(m_unscaled, AEC = TRUE))
})

test_that("rgl works", {
  
  require(rgl) # gives a message, so put this before expect_silent

  expect_silent({
    skip_on_cran()
    require(agridat)
    dat <- yan.winterwheat
    m2 <- gge(dat, yield ~ gen*env, scale=FALSE)
    
    # Tests for 3D
    biplot3d(m2)
    biplot3d(m2, cex.gen=1)
    # biplot3d(m2, cex.gen=0) # omit genotype names
    biplot3d(m2, cex.env=1)
    biplot3d(m2, col.gen="red")
    biplot3d(m2, col.env=c("red","blue")) # should ignore blue
    biplot3d(m2, comps=c(1,2,4))
    biplot3d(m2, lab.env=FALSE)
    biplot3d(m2, res.vec=FALSE)
    biplot3d(m2, zoom.gen=2)
    
    data(crossa.wheat, package="agridat")
    dat2 <- crossa.wheat
    
    #dat2 <- data.frame(
    #  env=c("BH93","EA93","HW93","ID93","KE93","NN93","OA93","RN93","WP93"),
    #  grp=c("G2","G2","G2","G2","G1","G2","G1","G2","G2"))    
  
    dat$eg <- c("G2","G2","G2","G2","G1","G2","G1","G2","G2")[
      match(dat$env, c("BH93","EA93","HW93","ID93","KE93","NN93","OA93","RN93","WP93"))]
  
    m4 <- gge(dat, yield ~ gen*env, env.group=eg, scale=FALSE)
    biplot3d(m4)
    while (rgl.cur() > 0) { close3d() }
  })
  
  expect_error(biplot3d(m2, comps=1:2))

})


# check par() settings are restored
if(FALSE) {
  par()$pty
  # [1] "m"
  B <- matrix(c(50, 67, 90, 98, 120,
                55, 71, 93, 102, 129,
                65, 76, 95, 105, 134,
                50, 80, 102, 130, 138,
                60, 82, 97, 135, 151,
                65, 89, 106, 137, 153,
                75, 95, 117, 133, 155), ncol=5, byrow=TRUE)
  rownames(B) <- c("G1","G2","G3","G4","G5","G6","G7")
  colnames(B) <- c("E1","E2","E3","E4","E5")
      
  m1 <- gge(B)
  plot(m1)
  par()$pty
  # "m"
  biplot(m1, main="Example biplot")
  par()$pty
  # "m"
}

if(0) {
# focus...not yet implemented
dat <- yan.winterwheat

m1 <- gge(dat, yield ~ gen*env, scale=FALSE) # default focus="env"
biplot(m1, hull=TRUE)

m2 <- gge(dat, yield ~ gen*env, scale=FALSE, focus="gen")
biplot(m2, hull=TRUE, flip=c(1,1)) # Yan book p 43

m3 <- gge(dat, yield ~ gen*env, scale=FALSE, focus="env")
biplot(m3, hull=TRUE, flip=c(1,1)) # Yan book p 45

m4 <- gge(dat, yield ~ gen*env, scale=FALSE, focus="symm")  # <-- short lines
biplot(m4, hull=TRUE, flip=c(1,1)) # Yan book p 46

m5 <- gge(dat, yield ~ gen*env, scale=FALSE, focus="dual")
biplot(m5, hull=TRUE, flip=c(1,1)) # OBS! Not Yan book p 48

m2 <- gge(dat, yield ~ gen*env, scale=FALSE, focus="gen")
biplot(m2, aec=TRUE, flip=c(1,1)) # Yan book p 43
#ggbiplot(m2, type="aec")

}

if(FALSE){
  
  # Custom colors for gen/env. Example matrix data from Laffont
  mat6 <- structure(c(120, 140, 131, 144, 120, 129, 131, 135, 107, 132, 123,
                      106, 157, 149, 141, 140, 159, 182, 135, 148, 153, 167,
                      148, 144, 143, 158, 125, 172, 140, 153, 150, 168, 168,
                      151, 148, 152, 171, 139, 157, 178, 157, 134, 169, 178,
                      157, 146, 149, 169, 191, 182, 132, 163, 156, 110, 147,
                      158, 151, 178, 193, 181, 138, 190, 177, 157, 159, 139,
                      154, 157, 145, 151, 185, 151, 197, 187, 195, 173, 166,
                      185, 187, 170, 159, 153, 145, 142, 171, 175, 176, 193,
                      192, 183, 178, 176, 170, 191, 138, 148, 145, 175, 163,
                      136, 178, 131, 179, 171, 161, 148, 196, 182, 159, 114,
                      134, 154, 169, 158, 144, 169, 150, 186, 185, 184, 156,
                      185, 204, 184, 146, 154, 156, 160, 160, 174, 185, 162,
                      160, 179, 210, 168, 193, 179, 167, 181, 167, 172, 187,
                      152, 160, 171, 139, 196, 176, 204, 176, 181, 188, 174,
                      155, 134, 170, 161, 158, 152, 172, 147, 190, 175, 176,
                      189, 186, 193, 204, 173, 178, 175, 165, 188, 191, 169,
                      160, 204, 221, 196, 157, 193, 203, 193, 177, 179, 204,
                      166, 167, 161, 182, 161, 191, 198, 208, 168, 182, 207,
                      148, 173, 195, 192, 206, 202, 193, 207, 167, 195, 186,
                      187, 181, 193, 192, 172, 152, 199, 221, 215, 206, 186,
                      195, 166, 216, 208, 213, 171, 213, 213, 174, 198, 185,
                      207, 219, 181, 195, 202, 169, 207, 190, 199, 168, 199,
                      212, 194, 190, 219, 202, 248, 185, 199, 205, 166, 207,
                      252, 172, 169, 210, 212, 233, 215, 229, 246, 250, 232,
                      242, 242, 192, 231, 222, 226, 189, 224, 249, 209, 205,
                      225, 267, 247, 240, 235, 256, 209, 215, 248, 220, 208,
                      214, 210, 247, 235, 242, 252, 253, 232, 251, 245, 232,
                      260, 237, 242),
                    .Dim = c(15L, 20L),
                    .Dimnames = list(c("G01", "G02", "G03", "G04", "G05",
                                       "G06", "G07", "G08", "G09", "G10",
                                       "G11", "G12", "G13", "G14", "G15"),
                                     c("E01", "E02", "E03", "E04", "E05",
                                       "E06", "E07", "E08", "E09", "E10",
                                       "E11", "E12", "E13", "E14", "E15",
                                       "E16", "E17", "E18", "E19", "E20")))
  
  # specify 'gen.group' and 'env.group' as a vector with matrix data
  m61 <- gge(mat6, scale=TRUE,
             env.group=c(rep("Blk1",3), rep("Blk2",5),rep("Blk3", 5), rep("Blk4", 7)),
             gen.group=rep(letters[1:3], each=5) )
  biplot(m61, flip=c(1,1), col.env=c("red","orange","blue","purple") )
  m61 <- gge(mat6, scale=FALSE,
             env.group=c(rep("Blk1",3), rep("Blk2",5),rep("Blk3", 5), rep("Blk4", 7)),
             gen.group=rep(letters[1:3], each=5), ggb=TRUE)
  biplot(m61, flip=c(1,1),
         col.env=c("red","orange","blue","purple"),
         col.gen=c("gray","darkgreen"))
  
  # Crossa example
  require("reshape2")
  require("agridat")
  # CRAN check doesn't like data() loading into global envt, so keep
  # this commented out.
  # data(crossa.wheat, package="agridat")
  dat1 <- crossa.wheat
  mat7 <- reshape2::acast(dat1, gen~loc, value.var='yield')
  mat7 <- mat7[, c("SR","SG","CA","AK","TB","SE","ES","EB","EG",
                   "KN","NB","PA","BJ","IL","TC","JM","PI","AS","ID",
                   "SC","SS","SJ","MS","MG","MM")]

  egp <- c(rep("Grp2",9), rep("Grp1", 16))
  m7 <- gge(mat7, env.group=egp, ggb=FALSE, lab="Y", scale=FALSE)
  m7 <- gge(mat7, env.group=egp, ggb=TRUE, lab="Y", scale=FALSE)
  biplot(m7, main="CYMMIT wheat")
  plot(m7)
  
  # Specify env.group as column in data frame
  dat2 <- crossa.wheat

  m8 <- gge(dat2, yield~gen*loc, 
            env.group=locgroup, gen.group=gengroup, scale=FALSE)
  biplot(m8)
  
  # No env.group
  m9 <- gge(dat2, yield~gen*loc, scale=FALSE)
  biplot(m9)
  
  # 3D
  biplot3d(m7, cex.gen=1)
  expect_error(biplot3d(m7, comps=1:2))
  
}