Add tests for percent

Some of these fail but I think they should pass.

I guess we don't necessarily want to depend on `{janitor}` but is there any reason not to copy the code for use here? Then we'd be confident that the functionality matched exactly.

https://github.com/sfirke/janitor/blob/main/R/round_half_up.R

Either way, I think round and signif half-up should be in their own script (and probably split the tests, too) as they're not necessarily related to percent, and we have an option to export them in the future if we want.

tests/testthat/test-percent_james.R

@@ -0,0 +1,117 @@
+test_that("as_percent works correctly", {
+  expect_s3_class(as_percent(0.5), "percent")
+  expect_type(as_percent(0.5), "double")
+
+  expect_equal(as.character(as_percent(50)), "5,000%")
+  expect_equal(as.character(as_percent(5)), "500%")
+  expect_equal(as.character(as_percent(0.5)), "50%")
+  expect_equal(as.character(as_percent(0.05)), "5%")
+  expect_equal(as.character(as_percent(0.005)), "0.5%")
+  expect_equal(as.character(as_percent(0.0005)), "0.05%")
+  expect_equal(as.character(as_percent(0.00005)), "0.01%")
+})
+
+test_that("as_percent handles non-numeric input", {
+  expect_error(as_percent("not a number"), regexp = "must be a <numeric> vector, not a <character> vector")
+  expect_error(as_percent(TRUE), regexp = "must be a <numeric> vector, not a <logical> vector")
+  expect_error(as_percent(list(1, 2, 3)), regexp = "must be a <numeric> vector, not a <list> vector")
+})
+
+test_that("round_half_up handles non-numeric input", {
+  expect_error(round_half_up("not a number"), regexp = "must be a <numeric> vector, not a <character> vector")
+  expect_error(round_half_up(TRUE), regexp = "must be a <numeric> vector, not a <logical> vector")
+  expect_error(round_half_up(list(1, 2, 3)), regexp = "must be a <numeric> vector, not a <list> vector")
+})
+
+
+test_that("signif_half_up handles non-numeric input", {
+  expect_error(signif_half_up("not a number"), regexp = "must be a <numeric> vector, not a <character> vector")
+  expect_error(signif_half_up(TRUE), regexp = "must be a <numeric> vector, not a <logical> vector")
+  expect_error(signif_half_up(list(1, 2, 3)), regexp = "must be a <numeric> vector, not a <list> vector")
+})
+
+
+test_that("signif_half_up results are as expected", {
+  # scalars
+  expect_equal(signif_half_up(x = 12.5, digits = 2),
+               13)
+  expect_equal(signif_half_up(x = 0),
+               0)
+  expect_equal(signif_half_up(x = -2.5, digits = 1),-3)
+  expect_equal(signif_half_up(x = 123.45, digits = 4),
+               123.5)
+  expect_equal(signif_half_up(x = -123.45, digits = 4),-123.5)
+  # vectors
+  expect_equal(signif_half_up(
+    x = c(12.5, 0,-2.5, 123.45,-123.45),
+    digits = 2
+  ),
+  c(13, 0,-2.5, 120,-120))
+})
+
+test_that("signif_half_up works correctly", {
+  expect_equal(signif_half_up(0.555, 2), 0.56)
+  expect_equal(signif_half_up(0.555, 1), 0.6)
+  expect_equal(signif_half_up(12345, 3), 12300)
+})
+
+test_that("as.character.percent works correctly", {
+  p <- as_percent(0.1234)
+  expect_equal(as.character(p), "12.34%")
+  expect_equal(as.character(p, digits = 1), "12.3%")
+})
+
+test_that("format.percent works correctly", {
+  p <- as_percent(0.1234)
+  expect_equal(format(p), "12.34%")
+  expect_equal(format(p, digits = 1), "12.3%")
+  expect_equal(format(p, symbol = " (%)"), "12.34 (%)")
+})
+
+test_that("print.percent works correctly", {
+  expect_output(print(as_percent(0.1234)), "12.34%")
+  expect_output(print(as_percent(0.567890)), "56.79%")
+})
+
+test_that("subsetting percent objects works correctly", {
+  p <- as_percent(c(0.1, 0.2, 0.3))
+  expect_s3_class(p[1:2], "percent")
+  expect_equal(p[1:2], as_percent(c(0.1, 0.2)))
+})
+
+test_that("unique.percent works correctly", {
+  p <- as_percent(c(0.1, 0.2, 0.1, 0.3))
+  expect_s3_class(unique(p), "percent")
+  expect_equal(unique(p), as_percent(c(0.1, 0.2, 0.3)))
+})
+
+test_that("rep.percent works correctly", {
+  p <- as_percent(0.1)
+  expect_s3_class(rep(p, 3), "percent")
+  expect_equal(rep(p, 3), as_percent(c(0.1, 0.1, 0.1)))
+})
+
+test_that("Math.percent works correctly", {
+  p <- as_percent(c(0.1234, 0.5678))
+  expect_equal(floor(p), as_percent(c(0.12, 0.56)))
+  expect_equal(ceiling(p), as_percent(c(0.13, 0.57)))
+  expect_equal(trunc(p), as_percent(c(0.12, 0.56)))
+  expect_equal(round(p), as_percent(c(0.12, 0.57)))
+  expect_equal(signif(p, 2), as_percent(c(0.12, 0.57)))
+})
+
+test_that("Summary.percent works correctly", {
+  p <- as_percent(c(0.1, 0.2, 0.3))
+  expect_s3_class(sum(p), "percent")
+  expect_equal(sum(p), as_percent(0.6))
+  expect_equal(prod(p), as_percent(0.006))
+  expect_equal(min(p), as_percent(0.1))
+  expect_equal(max(p), as_percent(0.3))
+  expect_equal(range(p), as_percent(c(0.1, 0.3)))
+})
+
+test_that("mean.percent works correctly", {
+  p <- as_percent(c(0.1, 0.2, 0.3))
+  expect_s3_class(mean(p), "percent")
+  expect_equal(mean(p), as_percent(0.2))
+})