context("byyr")

library(dplyr)

vft <- readr::read_csv(test_path("data-byyr_toy_vft.csv"))
ba_10 <- basal_area_dbl(10)
ba_100 <- basal_area_dbl(100)

test_that("basal_area_byyr and abundance_byyr fail with informative errors", {
  expect_error(abundance_byyr(1), "data.frame.*is not TRUE")
  expect_error(abundance_byyr(), "is missing")

  msg <- "All expressions.*must refer to `dbh`"

  # Focus on the expected error
  suppressWarnings(
    expect_error(abundance_byyr(vft, exactdate > 0), msg)
  )

  expect_silent(abundance_byyr(vft, DBH > 0))
  expect_silent(abundance_byyr(vft, exactdate > 0, dbh > 0))

  expect_error(abundance_byyr(vft), msg)
  expect_error(abundance_byyr(vft, invalid > 0), msg)

  expect_error(basal_area_byyr(1), "data.frame.*is not TRUE")
  expect_error(basal_area_byyr(), "is missing")

  expect_silent(
    basal_area_byyr(vft, DBH > 0)
  )
  expect_silent(basal_area_byyr(vft, exactdate > 0, dbh > 0))

  expect_error(basal_area_byyr(vft), msg)
  expect_error(basal_area_byyr(vft, exactdate > 0), msg)
  expect_error(basal_area_byyr(vft, invalid > 0), msg)
})

describe("basal_area_byyr and abundance_byyr work with different datasets", {
  # DRY helpers
  byyr <- function(.data, plot, .f) {
    # Fix seed to reproduce random-sampling
    out <- withr::with_seed(123, sample_n(choose_plot(.data, plot), 10))
    .f(out, dbh > 0)
  }
  choose_plot <- function(.data, plot) {
    .data[.data$PlotName %in% plot, , drop = FALSE]
  }

  expect_dataframe <- function(x) expect_is(x, "data.frame")

  expect_named_year <- function(x, year) {
    # Test only 1 year-column to avoid errors as data grows
    expect_named(x[1:3], c("species", "family", year))
  }

  it("works with data from from bci", {
    skip_if_not_installed("bciex")
    # Unexpected name
    vft <- bciex::bci12vft_mini %>% rename(PlotName = Plot)
    plot <- "bci"
    out_basal <- byyr(vft, plot, basal_area_byyr)
    expect_dataframe(out_basal)
    expect_named_year(out_basal, "yr_2005")

    out_abund <- byyr(vft, plot, abundance_byyr)
    expect_dataframe(out_abund)
    expect_named_year(out_abund, "yr_2005")
  })

  it("works with data from from Bukit Timah", {
    skip_if_not_installed("bukittimah")
    bukit <- bukittimah::ViewFullTable_bukit
    plot <- "Bukit Timah Big Trees"

    out_basal <- byyr(bukit, plot, basal_area_byyr)
    expect_dataframe(out_basal)
    expect_named_year(out_basal, "yr_2006")

    out_abund <- byyr(bukit, plot, abundance_byyr)
    expect_dataframe(out_basal)
    expect_named_year(out_basal, "yr_2006")
  })

  it("works with data from from Ngel Niaky", {
    skip_if_not_installed("ngel")

    ngel <- ngel::ViewFullTable_ngel
    plot <- "ngelnyaki"
    out_basal <- byyr(ngel, plot, basal_area_byyr)
    expect_dataframe(out_basal)
    expect_named_year(out_basal, "yr_2015")

    out_abund <- byyr(ngel, plot, abundance_byyr)
    expect_dataframe(out_basal)
    expect_named_year(out_basal, "yr_2015")
  })
})

describe("abundance_byyr and basa_area_byyr return expected output", {
  tiny <- tibble(
    Tag = c("0001", "0001", "0002", "0002", "0003", "0003"),
    TreeID = c("0001", "0001", "0002", "0002", "0003", "0003"),
    PlotName = c("p", "p", "p", "p", "p", "p"),
    Status = c("alive", "alive", "alive", "alive", "alive", "alive"),
    DBH = c(1L),
    ExactDate = c(
      "2000-01-01", "2001-01-01", "2000-01-01", "2001-01-01", "2000-01-01",
      "2001-01-01"
    ),
    PlotCensusNumber = c(1L, 2L, 1L, 2L, 1L, 2L),
    CensusID = c(1L, 2L, 1L, 2L, 1L, 2L),
    Genus = c("A", "A", "B", "B", "B", "B"),
    SpeciesName = c("a", "a", "b", "b", "b", "b"),
    Family = c("f", "f", "f", "f", "f", "f"),
    StemID = 1,
    HOM = 130
  )

  it("outputs basal area multiplied by the abundance", {
    abund <- abundance_byyr(tiny, dbh > 0)
    expect_warning(
      basal <- basal_area_byyr(tiny, dbh > 0), "stemid.*Duplicated values"
    )
    expect_equal(basal$yr_2000, basal_area_dbl(1) * abund$yr_2000)
    expect_equal(basal$yr_2001, basal_area_dbl(1) * abund$yr_2001)

    tiny2 <- mutate(tiny, DBH = 10)
    expect_warning(basal <- basal_area_byyr(
      tiny2, dbh > 0
    ), "stemid.*Duplicated")
    expect_equal(basal$yr_2000, ba_10 * abund$yr_2000)
    expect_equal(basal$yr_2001, ba_10 * abund$yr_2001)
  })

  it("outputs equal to known output", {
    out1 <- suppressWarnings(suppressMessages(abundance_byyr(tiny, dbh > 0)))
    expect_equal(out1$yr_2000, c(1, 2))
    expect_equal(out1$yr_2001, c(1, 2))

    tiny2 <- tiny[1:4, ]
    out2 <- suppressWarnings(suppressMessages(abundance_byyr(tiny2, dbh > 0)))
    expect_equal(out2$yr_2000, c(1, 1))
    expect_equal(out2$yr_2001, c(1, 1))

    tiny3 <- tiny[c(1, 3, 4), ]
    out3 <- suppressWarnings(suppressMessages(abundance_byyr(tiny3, dbh > 0)))
    expect_equal(out3$yr_2000, c(1, 1))
    expect_equal(out3$yr_2001, c(0, 1))
  })
})

describe("abundance_byyr", {
  skip_if_not_installed("readr")

  it("lowercases dbh and only dbh from the expression passed to ...", {
    expect_silent(
      out <- abundance_byyr(vft, dbh >= min(vft$DBH, na.rm = TRUE))
    )
    expect_equal(out, abundance_byyr(vft, dbh > 0))
  })

  it("is sensitive to DBH, so outputs none date-column if dbh is too big ", {
    too_big <- max(vft$DBH, na.rm = TRUE) + 1
    out <- abundance_byyr(vft, dbh > !!too_big)
    expect_named(rlang::set_names(out, tolower), c("species", "family"))
    expect_is(out, "tbl_df")

    # Upper case DBH
    expect_equal(out, abundance_byyr(vft, DBH > !!too_big))
  })

  it("outputs as expected", {
    # All trees are of the same species. There are two trees, each with two
    # stems. In census 1, the count of alive trees should be 2 because both
    # trees are alive, but note that one stem is dead (StemID = 1.2). In census
    # 2 the count of alive trees should be 1:
    #   * One tree is alive (TreeID = 1) although one stem is gone
    #     (StemID = 1.2);
    #   * One tree is dead (TreeID = 2) because both its stems are dead.


    out <- abundance_byyr(vft, dbh > 0)
    expect_is(out, "tbl_df")
    nms <- c("species", "family", "yr_2001", "yr_2002")
    expect_named(set_names(out, tolower), nms)
    expect_equal(out$yr_2001, 2)
    expect_equal(out$yr_2002, 1)
  })

  it("fails if parsed dates are all missing", {
    bad <- mutate(vft[1, ], ExactDate = NA)
    msg <- "Can't parse `exactdates`"
    expect_error(abundance_byyr(bad, dbh > 0), msg)

    # Wrong format: Expecting yyy-mm-dd, so parsing results in NA
    bad <- mutate(vft[1, ], ExactDate = as.character("1/1/2001"))
    expect_error(
      expect_warning(abundance_byyr(bad, dbh > 0)), "Can't parse `exactdates`"
    )
  })

  it("warns if parsed dates are not from 1980 to present", {
    early <- mutate(vft[1, ], ExactDate = "1970-01-01")
    msg <- "Dates should be"
    expect_warning(abundance_byyr(early, dbh > 0), msg)

    late <- mutate(vft[1, ], ExactDate = lubridate::today() + 1)
    msg <- "Dates should be"
    expect_warning(abundance_byyr(late, dbh > 0), msg)

    good <- mutate(vft[1, ], ExactDate = lubridate::today())
    expect_silent(abundance_byyr(good, dbh > 0))
  })
})

describe("*byyr()", {
  it("makes no difference if status is picked before *byyr()", {
    skip_if_not_installed("readr")
    raw <- abundance_byyr(vft, dbh > 0)
    picked <- abundance_byyr(filter(vft, Status == "alive"), dbh > 0)
    expect_equal(picked, raw)
  })
})


describe("basal_area_byyr()", {
  it("sums basal area of all stems which dbh is in the chosen range", {
    vft_simple <- mutate(vft, Status = "alive", DBH = 1)

    # One census; one tree with two stems.
    vft_c1_t1_s2 <- filter(vft_simple, CensusID == 1, TreeID == 1)
    expected <- basal_area_dbl(1) * nrow(vft_c1_t1_s2)
    actual <- basal_area_byyr(vft_c1_t1_s2, dbh >= 1)$yr_2001
    expect_equal(actual, expected)

    # Trees and saplings
    out <- basal_area_byyr(mutate(vft_c1_t1_s2, DBH = c(10, 100)), DBH >= 10)
    expect_equal(ba_10 + ba_100, out$yr_2001)

    # Edge
    out <- basal_area_byyr(mutate(vft_c1_t1_s2, DBH = c(10, 100)), DBH > 10)
    expect_equal(ba_100, out$yr_2001)

    # One census; two trees, each with two stems.
    vft_c1_t2_s4 <- filter(vft_simple, CensusID == 1)
    expected <- basal_area_dbl(1) * nrow(vft_c1_t2_s4)
    actual <- basal_area_byyr(vft_c1_t2_s4, dbh >= 1)$yr_2001
    expect_equal(actual, expected)

    # Trees and samplings
    out <- basal_area_byyr(
      mutate(vft_c1_t2_s4, DBH = rep(c(10, 100), 2)),
      DBH >= 10
    )
    expect_equal(2 * ba_10 + 2 * ba_100, out$yr_2001)
    # Trees
    out <- basal_area_byyr(
      mutate(vft_c1_t2_s4, DBH = rep(c(10, 100), 2)),
      DBH >= 100
    )
    expect_equal(2 * ba_100, out$yr_2001)

    # Two censuses, with one tree in each
    vft_c2_t1_s4 <- filter(vft_simple, TreeID == 1)
    expected_total <- basal_area_dbl(1) * nrow(vft_c2_t1_s4)
    actual_2001 <- basal_area_byyr(vft_c2_t1_s4, dbh >= 1)$yr_2001
    actual_2002 <- basal_area_byyr(vft_c2_t1_s4, dbh >= 1)$yr_2002
    expect_equal(actual_2001 + actual_2002, expected_total)

    # Trees and samplings
    out <- basal_area_byyr(
      mutate(vft_c2_t1_s4, DBH = c(10, 100, 10, 100)),
      DBH >= 10
    )
    both <- ba_10 + ba_100
    expect_equal(both, out$yr_2001)
    expect_equal(both, out$yr_2002)

    # Trees
    out <- basal_area_byyr(
      mutate(vft_c2_t1_s4, DBH = c(10, 100, 10, 100)),
      DBH >= 100
    )
    expect_equal(ba_100, out$yr_2001)
    expect_equal(ba_100, out$yr_2002)
  })

  it("deals with buttresses: Includes only one stem per stemid", {
    vft_hom <- vft %>%
      filter(CensusID == 1) %>%
      slice(1:2) %>%
      select(TreeID, StemID, HOM, DBH, everything()) %>%
      mutate(
        DBH = c(10, 100),
        HOM = c(140, 130),
        Status = "alive",
      )

    # Adapt the data to clearly expose this specific problem
    vft_hom2 <- vft_hom %>% mutate(StemID = c("1.1", "1.1"))
    # TreeID StemID   HOM   DBH
    #  <int>  <dbl>  <dbl> <dbl>
    #      1    1.1   140    10  # Main stem defined by largest HOM
    #      1    1.1   130   100  # This should be removed
    sapl_and_tree <- basal_area_byyr(vft_hom2, dbh >= 10)
    expect_equal(sapl_and_tree$yr_2001, basal_area_dbl(10))
  })
})