context("trelliscope plotting functions")

test_that("trelliPlots check the correct inputs", {
  testthat::skip_on_cran()
  # Quick trelliscope function check for future changes
  # file_coverage(c("./R/as.trelliData.R", "./R/summary_trelliData.R", "./R/trelliPlots.R"), 
  #               c("./tests/testthat/test_as_trelliData_edata.R", "./tests/testthat/test_as_trelliData.R", "./tests/testthat/test_as_trelliData_summary.R", "./tests/testthat/test_trelliPlots.R")) %>% 
  #     report()
  
  # Load: lipid expression data-------------------------------------------------

  load(system.file('testdata',
    'metaboliteData.RData',
    package = 'pmartR'
  ))

  # Generate: trelliData objects for testing------------------------------------

  # Add another condition to fdata
  fdata$Condition <- c(rep("Mock", 3), rep("InfectionA", 4), rep("InfectionB", 5))

  # Create the omics object
  metabData <- as.metabData(
    e_data = edata,
    edata_cname = "Metabolite",
    f_data = fdata,
    fdata_cname = "SampleID",
    e_meta = emeta,
    emeta_cname = "MClass"
  )

  # Log transform and set main effects
  metabData <- edata_transform(metabData, "log2")
  metabData <- group_designation(metabData, "Condition")

  # Filter for imd_anova and normalize
  suppressMessages(metabData <- applyFilt(imdanova_filter(metabData), metabData, min_nonmiss_anova = 2))
  metabData <- normalize_global(metabData, "all", "median", apply_norm = TRUE, backtransform = TRUE)

  # Now run statistics
  metabStat <- imd_anova(metabData, test_method = "combined")

  metabStat_gtest <- imd_anova(metabData, test_method = "gtest")

  # Create the four trelliData object testers
  mtrelliData1 <- as.trelliData.edata(e_data = edata, edata_cname = "Metabolite", omics_type = "metabData")
  mtrelliData2 <- as.trelliData(omicsData = metabData)
  mtrelliData3 <- as.trelliData(statRes = metabStat)
  mtrelliData4 <- as.trelliData(omicsData = metabData, statRes = metabStat)

  mtrelliData5 <- as.trelliData(omicsData = metabData, statRes = metabStat_gtest)
  
  ds_test <- data.frame(
    p_value_anova = 0.7,
    fold_change = NaN
  )
  expect_null(pmartR:::determine_significance(ds_test, 0.05))

  # Test: trelli plotting functions---------------------------------------------

  ## trelli_abundance_boxplot

  # The object must be a trelliData object
  expect_error(
    trelli_abundance_boxplot(metabData),
    "trelliData must be of the class trelliData."
  )

  # The object must be paneled with trelli_panel_by
  expect_error(
    trelli_abundance_boxplot(mtrelliData1),
    "trelliData must be paneled with trelli_panel_by."
  )

  # The object must have omicsData
  expect_error(
    mtrelliData3 %>% trelli_panel_by("Metabolite") %>% trelli_abundance_boxplot(),
    "trelliData must have omicsData for this plotting function."
  )

  # Only acceptable cognostics are allowed
  expect_error(
    mtrelliData1 %>% trelli_panel_by("Metabolite") %>% trelli_abundance_boxplot(cognostics = "test"),
    "Unacceptable cognostic option included. Acceptable options are:  count, mean abundance, median abundance, cv abundance, anova p-value, fold change"
  )

  # ggplot params should be strings
  expect_error(
    mtrelliData1 %>% trelli_panel_by("Metabolite") %>% trelli_abundance_boxplot(ggplot_params = 1),
    "ggplot_params must be a string, vector of strings, or NULL."
  )

  # Interactive parameter should be a true or false
  expect_error(
    mtrelliData1 %>% trelli_panel_by("Metabolite") %>% trelli_abundance_boxplot(interactive = "Yes"),
    "interactive must be a TRUE or FALSE."
  )

  # Include points should be a true or false
  expect_error(
    mtrelliData1 %>% trelli_panel_by("Metabolite") %>% trelli_abundance_boxplot(include_points = "Yes"),
    "include_points must be a TRUE or FALSE."
  )

  # Test mode should be a true or false
  expect_error(
    mtrelliData1 %>% trelli_panel_by("Metabolite") %>% trelli_abundance_boxplot(test_mode = "test"),
    "test_mode must be a TRUE or FALSE"
  )

  # Test example should be a non-zero integer
  expect_error(
    mtrelliData1 %>% trelli_panel_by("Metabolite") %>% trelli_abundance_boxplot(test_mode = TRUE, test_example = 0),
    "test_example should be a non-zero integer."
  )

  # Test example also can't be more than the number of panels
  expect_error(
    mtrelliData1 %>% trelli_panel_by("Metabolite") %>% trelli_abundance_boxplot(test_mode = TRUE, test_example = -2000.3),
    "test_example must be in the range of possibilities, of 1 to 80"
  )

  # Single plot must be true or false
  expect_error(
    mtrelliData1 %>% trelli_panel_by("Metabolite") %>% trelli_abundance_boxplot(single_plot = "test"),
    "single_plot must be a TRUE or FALSE."
  )

  # Blank biomolecules should be removed, and if no data, an error produced
  blankExample <- mtrelliData1 %>% trelli_panel_by("Metabolite")
  blankExample$trelliData.omics <- blankExample$trelliData.omics[1, ]
  blankExample$trelliData.omics$Metabolite <- ""

  expect_error(
    suppressMessages(blankExample %>% trelli_abundance_boxplot()),
    "No data to build trelliscope with."
  )

  # Test mode for windows should pass
  expect_equal(
    .getDownloadsFolder(.test_mode = TRUE),
    file.path(dirname("~"), "Downloads")
  )

  # Expect a single plot object to be made
  abun_boxplot <- mtrelliData1 %>%
    trelli_panel_by("Metabolite") %>%
    trelli_abundance_boxplot(single_plot = TRUE)
  expect_true(inherits(abun_boxplot, "ggplot"))

  # Generate a tests folder
  testFolder <- file.path(tempdir(), "/Trelli_Tests")

  # If the folder exists, remove it
  if (file.exists(testFolder)) {
    unlink(testFolder, recursive = TRUE, force = TRUE)
  }

  # Build a trelliscope that tests multiple functions
  suppressWarnings(mtrelliData4 %>% trelli_panel_by("Metabolite") %>%
    trelli_abundance_boxplot(
      path = file.path(testFolder, "BoxAbundanceTest1"),
      test_mode = TRUE,
      test_example = 2,
      ggplot_params = "xlab('')",
      interactive = TRUE
    ))
  expect_true(file.exists(file.path(testFolder, "BoxAbundanceTest1")))

  # Build a trelliscope that tries to call stats cognostics without stats data and
  # has one plot
  singlePlot <- mtrelliData1 %>% trelli_panel_by("Metabolite") 
  singlePlot$trelliData.omics <- singlePlot$trelliData.omics[1, ]
  suppressWarnings(singlePlot %>% 
   trelli_abundance_boxplot(path = file.path(testFolder, "BoxAbundanceTest2"), 
                            cognostics = "anova p-value")
  )
  expect_true(file.exists(file.path(testFolder, "BoxAbundanceTest2")))

  # Build a trelliscope that tries to call stats cognostics without cognostics data
  suppressWarnings(mtrelliData1 %>% trelli_panel_by("Metabolite") %>%
    trelli_abundance_boxplot(
      path = file.path(testFolder, "BoxAbundanceTest3"),
      test_mode = TRUE,
      test_example = 2,
      cognostics = NULL
    ))
  expect_true(file.exists(file.path(testFolder, "BoxAbundanceTest3")))

  ## trelli_abundance_histogram

  # trelliData must be grouped by edata_cname
  expect_error(
    mtrelliData2 %>% trelli_panel_by("SampleID") %>% trelli_abundance_histogram(),
    "trelliData must be grouped by edata_cname."
  )

  # Expect a single plot object to be made
  abun_histplot <- mtrelliData1 %>%
    trelli_panel_by("Metabolite") %>%
    trelli_abundance_histogram(single_plot = TRUE)
  expect_true(inherits(abun_histplot, "ggplot"))
  
  
  # Test that trelliscope builds when passed cognostic doesn't exist  
  suppressWarnings(mtrelliData1 %>% trelli_panel_by("Metabolite") %>% 
    trelli_abundance_histogram(path = file.path(testFolder, "HistAbundanceTest1"),
                               test_mode = TRUE, 
                               test_example = 2,
                               ggplot_params = "xlab('')",
                               cognostics = c("sample count", "cv abundance"),
                               interactive = TRUE)
  )
  expect_true(file.exists(file.path(testFolder, "HistAbundanceTest1")))

  # Test that trelliscope builds even with missing cognostics and a single omic
  suppressWarnings(singlePlot %>%
    trelli_abundance_histogram(
      path = file.path(testFolder, "HistAbundanceTest2"),
      cognostics = NULL
    ))
  expect_true(file.exists(file.path(testFolder, "HistAbundanceTest2")))

  # Test that trelliscope builds with all cognostics
  suppressWarnings(mtrelliData4 %>% trelli_panel_by("Metabolite") %>%
    trelli_abundance_histogram(
      path = file.path(testFolder, "HistAbundanceTest3"),
      test_mode = TRUE,
      test_example = 3
    ))
  expect_true(file.exists(file.path(testFolder, "HistAbundanceTest3")))

  ## trelli_abundance_heatmap

  # Test that the data has been grouped by an emeta column
  expect_error(
    mtrelliData1 %>% trelli_panel_by("Metabolite") %>% trelli_abundance_heatmap(),
    "trelliData must be paneled_by an e_meta column."
  )

  # Return a single interactive plot
  abun_hmplot <- mtrelliData4 %>%
    trelli_panel_by("MClass") %>%
    trelli_abundance_heatmap(single_plot = TRUE, ggplot_params = "xlab('')", interactive = TRUE)
  expect_true(inherits(abun_hmplot, "plotly"))

  # Test that trelliscope builds with all cognostics
  suppressWarnings(mtrelliData4 %>% trelli_panel_by("MClass") %>%
    trelli_abundance_heatmap(
      path = file.path(testFolder, "HmAbundanceTest1"),
      test_mode = TRUE,
      test_example = 2
    ))
  expect_true(file.exists(file.path(testFolder, "HmAbundanceTest1")))

  # Test that the trelliscope still builds without group data and only one plot
  nogroup <- mtrelliData4 %>% trelli_panel_by("MClass")
  attributes(nogroup$omicsData)$group_DF <- NULL
  nogroup$trelliData.omics <- nogroup$trelliData.omics[1, ]
  suppressWarnings(nogroup %>%
    trelli_abundance_heatmap(path = file.path(testFolder, "HmAbundanceTest2")))
  expect_true(file.exists(file.path(testFolder, "HmAbundanceTest2")))

  ## trelli_missingness_bar

  # Trigger proportion check warning with something outrageous
  expect_error(
    mtrelliData3 %>% trelli_panel_by("Metabolite") %>% trelli_missingness_bar(proportion = "cantelope"),
    "proportion must be a TRUE or FALSE."
  )

  # Test trelliscope builds with all cognostics
  suppressWarnings(mtrelliData1 %>% trelli_panel_by("Metabolite") %>% 
    trelli_missingness_bar(path = file.path(testFolder, "MissingTest1"),
                           ggplot_params = "ylab('')",
                           test_mode = TRUE, 
                           test_example = 2,
                           cognostics = "observed proportion",
                           interactive = TRUE)
  )
  expect_true(file.exists(file.path(testFolder, "MissingTest1")))
  
  # Test trelliscope with just a statRes object 
  suppressWarnings(mtrelliData3 %>% trelli_panel_by("Metabolite") %>% 
                     trelli_missingness_bar(path = file.path(testFolder, "MissingTest2"),
                                            test_mode = TRUE, 
                                            test_example = 2,
                                            proportion = FALSE,
                                            cognostics = "total count")
  )
  expect_true(file.exists(file.path(testFolder, "MissingTest2")))

  # Build a single plot
  miss_plot <- singlePlot %>% trelli_missingness_bar(single_plot = TRUE)
  expect_true(inherits(miss_plot, "ggplot"))

  ## trelli_foldchange_bar

  # statRes data is required for this function
  expect_error(
    mtrelliData2 %>% trelli_panel_by("Metabolite") %>% trelli_foldchange_bar(),
    "trelliData must have statRes for this plotting function."
  )

  # Requested test example must not be out of the range of possibilities
  expect_error(
    mtrelliData3 %>% trelli_panel_by("Metabolite") %>% trelli_foldchange_bar(test_mode = TRUE, test_example = -2000.3),
    "test_example must be in the range of possibilities, of 1 to 65"
  )

  # trelliData must be paneled_by edata_cname
  expect_error(
    mtrelliData4 %>% trelli_panel_by("SampleID") %>% trelli_foldchange_bar(),
    "trelliData must be grouped by edata_cname."
  )

  # Test that the p-value threshold is a numeric
  expect_error(
    mtrelliData4 %>% trelli_panel_by("Metabolite") %>% trelli_foldchange_bar(p_value_test = "anova", p_value_thresh = "test"),
    "p_value_thresh must be a numeric."
  )

  # Test that the p-value threshold is a number between 0 and 1
  expect_error(
    mtrelliData4 %>% trelli_panel_by("Metabolite") %>% trelli_foldchange_bar(p_value_test = "anova", p_value_thresh = -0.05),
    "p_value_thresh must be between 0 and 1."
  )
  expect_error(
    mtrelliData4 %>% trelli_panel_by("Metabolite") %>% trelli_foldchange_bar(p_value_test = "anova", p_value_thresh = 5),
    "p_value_thresh must be between 0 and 1."
  )
  
  # Generate a trelliscope without proportions and a single plot
  singleStatPlot <- mtrelliData4 %>% trelli_panel_by("Metabolite")
  singleStatPlot$trelliData.omics <- singleStatPlot$trelliData.omics[1, ]
  singleStatPlot$trelliData.stat <- singleStatPlot$trelliData.stat[singleStatPlot$trelliData.stat$Metabolite == singleStatPlot$trelliData.omics$Metabolite, ]
  suppressWarnings(singleStatPlot %>%
     trelli_foldchange_bar(path = file.path(testFolder, "barFoldChangeTest1"),
                           ggplot_params = "ylab('')",
                           cognostics = "anova p-value",
                           interactive = TRUE)
  )
  expect_true(file.exists(file.path(testFolder, "barFoldChangeTest1")))
  
  # Generate a trelliscope with proportions and a gtest 
  suppressWarnings(mtrelliData3 %>% trelli_panel_by("Metabolite") %>% 
                     trelli_foldchange_bar(path = file.path(testFolder, "barFoldChangeTest2"),
                                           ggplot_params = "ylab('')",
                                           test_mode = TRUE, 
                                           test_example = 2,
                                           cognostics = "fold change",
                                           p_value_thresh = 0,
                                           interactive = TRUE)
  )
  expect_true(file.exists(file.path(testFolder, "barFoldChangeTest2")))
  
  # Generate a trelliscope with proportions 
  suppressWarnings(mtrelliData3 %>% trelli_panel_by("Metabolite") %>% 
                     trelli_foldchange_bar(path = file.path(testFolder, "barFoldChangeTest3"),
                                           test_mode = TRUE, 
                                           test_example = 2)
  )
  expect_true(file.exists(file.path(testFolder, "barFoldChangeTest3")))

  # Generate a trelliscope with proportions and no p-values
  fc_barplot <- mtrelliData3 %>% trelli_panel_by("Metabolite") %>% 
                     trelli_foldchange_bar(test_mode = TRUE, 
                                           test_example = 2,
                                           p_value_test = "gtest",
                                           p_value_thresh = 0,
                                           single_plot = TRUE)
  expect_true(inherits(fc_barplot, "ggplot"))

  ## trelli_foldchange_boxplot

  # Data must be grouped by an e_meta column
  expect_error(
    mtrelliData4 %>% trelli_panel_by("Metabolite") %>% trelli_foldchange_boxplot(),
    "trelliData must be paneled_by an e_meta column."
  )

  # Include points should be a true or false
  expect_error(
    mtrelliData4 %>% trelli_panel_by("MClass") %>% trelli_foldchange_boxplot(include_points = "FALSE"),
    "include_points must be a TRUE or FALSE."
  )
  
  # Run a standard trelli foldchange boxplot with change ggplot parameters 
  suppressWarnings(mtrelliData4 %>% trelli_panel_by("MClass") %>% 
   trelli_foldchange_boxplot(path = file.path(testFolder, "boxFoldChangeTest1"),
                             ggplot_params = "xlab('')",
                             test_mode = TRUE, 
                             test_example = 2,
                             interactive = TRUE)
  )
  expect_true(file.exists(file.path(testFolder, "boxFoldChangeTest1")))

  # Create a second trelliscope from the single data with no include_points and p_value in the cognostics
  singleEmetaPlot <- mtrelliData4 %>% trelli_panel_by("MClass")
  singleEmetaPlot$trelliData.omics <- singleEmetaPlot$trelliData.omics[singleEmetaPlot$trelliData.omics$MClass == "MClass5", ]
  singleEmetaPlot$trelliData.stat <- singleEmetaPlot$trelliData.stat[singleEmetaPlot$trelliData.stat$MClass == "MClass5", ]

  suppressWarnings(singleEmetaPlot %>%
                     trelli_foldchange_boxplot(path = file.path(testFolder, "boxFoldChangeTest2"),
                                               p_value_thresh = 0,
                                               include_points = TRUE)
  )
  expect_true(file.exists(file.path(testFolder, "boxFoldChangeTest2")))

  # Generate a single plot
  fc_boxplot <- singleEmetaPlot %>% trelli_foldchange_boxplot(single_plot = TRUE)
  expect_true(inherits(fc_boxplot, "ggplot"))

  ## trelli_foldchange_volcano
  
  # Data must be grouped by an e_meta column
  expect_error(
    mtrelliData4 %>% trelli_panel_by("MClass") %>% trelli_foldchange_volcano(comparison = "Tigers"),
    "is not an acceptable comparison"
  )

  # Data must be grouped by an e_meta column
  expect_error(
    mtrelliData4 %>% trelli_panel_by("Metabolite") %>% trelli_foldchange_volcano(comparison = "Mock_vs_InfectionA"),
    "trelliData must be paneled_by an e_meta column."
  )

  # Generate a volcano trelliscope with a modified ggplot parameter
  suppressWarnings(mtrelliData4 %>% trelli_panel_by("MClass") %>% 
   trelli_foldchange_volcano(path = file.path(testFolder, "volFoldChangeTest1"),
                             comparison = "Mock_vs_InfectionA",
                             ggplot_params = "xlab('')",
                             test_mode = TRUE, 
                             test_example = 2,
                             interactive = TRUE)
  )
  expect_true(file.exists(file.path(testFolder, "volFoldChangeTest1")))

  # Generate a single plot volcano trelliscope with no p_value
  suppressWarnings(singleEmetaPlot %>%
                     trelli_foldchange_volcano(path = file.path(testFolder, "volFoldChangeTest2"), comparison = "Mock_vs_InfectionA",
                                               p_value_thresh = 0)
  )
  expect_true(file.exists(file.path(testFolder, "volFoldChangeTest2")))

  # Test the creation of a single plot
  fc_volcano <- singleEmetaPlot %>% trelli_foldchange_volcano(single_plot = TRUE, p_value_test = NULL, comparison = "Mock_vs_InfectionA")
  expect_true(inherits(fc_volcano, "ggplot"))

  ## trelli_foldchange_heatmap

  # Data must be grouped by an e_meta column
  expect_error(
    mtrelliData4 %>% trelli_panel_by("Metabolite") %>% trelli_foldchange_heatmap(),
    "trelliData must be paneled_by an e_meta column."
  )

  # Generate the fold change heatmap trelliscope with a modified ggplot parameter
  suppressWarnings(mtrelliData4 %>% trelli_panel_by("MClass") %>%
    trelli_foldchange_heatmap(
      path = file.path(testFolder, "hmFoldChangeTest1"),
      ggplot_params = "xlab('')",
      test_mode = TRUE,
      test_example = 2,
      interactive = TRUE
    ))
  expect_true(file.exists(file.path(testFolder, "hmFoldChangeTest1")))

  # Generate the fold change heatmap trelliscope with smaller dataset
  suppressWarnings(singleEmetaPlot %>%
    trelli_foldchange_heatmap(path = file.path(testFolder, "hmFoldChangeTest2"), ))
  expect_true(file.exists(file.path(testFolder, "hmFoldChangeTest2")))

  # Generate a single plot
  fc_heatmap <- singleEmetaPlot %>% trelli_foldchange_heatmap(single_plot = TRUE)
  expect_true(inherits(fc_heatmap, "ggplot"))
  
})