Nothing
make_valid_roles <- function() {
r <- propose_roles(iris)
r$role[r$col == "Sepal.Length"] <- "outcome"
r
}
test_that("roles_validate accepts a clean roles table", {
r <- make_valid_roles()
expect_invisible(roles_validate(r))
expect_invisible(roles_validate(r, iris))
})
test_that("roles_validate errors on non-data-frame", {
expect_error(
roles_validate(list(col = "x", role = "outcome", kind = "numeric")),
"must be a data frame"
)
})
test_that("roles_validate errors when required columns are missing", {
r <- make_valid_roles()
expect_error(
roles_validate(r[, c("col", "role")]),
"missing required column.*kind"
)
expect_error(
roles_validate(r[, c("col", "kind")]),
"missing required column.*role"
)
expect_error(
roles_validate(r[, c("role", "kind")]),
"missing required column.*col"
)
})
test_that("roles_validate errors on NA role", {
r <- make_valid_roles()
r$role[1] <- NA
expect_error(roles_validate(r), "NA")
})
test_that("roles_validate errors on unknown role string", {
r <- make_valid_roles()
r$role[1] <- "blocker"
expect_error(roles_validate(r), "Unknown role")
})
test_that("roles_validate accepts a table with no outcome", {
# Two-axis model: no outcome is required; numeric covariates scramble.
r <- propose_roles(iris)
expect_true(all(r$role != "outcome"))
expect_invisible(roles_validate(r))
})
test_that("roles_validate accepts multiple treatment columns", {
# Joint-treatment masking (factorial / split-plot designs) is supported:
# two or more treatment factors must validate cleanly.
r <- make_valid_roles()
r <- set_role(r, c("Sepal.Width", "Petal.Width"), role = "treatment")
expect_invisible(roles_validate(r))
expect_invisible(roles_validate(r, iris))
})
test_that("roles_validate accepts explicit keep actions", {
r <- make_valid_roles()
r <- set_role(r, "Species", action = "keep")
expect_invisible(roles_validate(r))
expect_invisible(roles_validate(r, iris))
})
test_that("roles_validate errors on duplicate col entries", {
r <- make_valid_roles()
r <- rbind(r, r[1, ])
expect_error(roles_validate(r), "Duplicate")
})
test_that("roles_validate errors when unsupported columns are covariates", {
df <- data.frame(
payload = I(list(list(a = 1), list(a = 2), list(a = 3))),
yield = c(1.0, 2.0, 3.0)
)
r <- propose_roles(df, detect = FALSE)
r$role[r$col == "yield"] <- "outcome"
r <- set_role(r, "payload", role = "covariate", action = "scramble")
expect_error(roles_validate(r, df), "unsupported|cannot")
})
test_that("roles_validate cross-checks df columns when df supplied", {
r <- make_valid_roles()
# Drop a row -> df has a column not in roles
expect_error(roles_validate(r[-1, ], iris), "not in.*roles")
# Add a row -> roles has a column not in df
r_extra <- r
r_extra <- rbind(r_extra, data.frame(
col = "ghost", role = "covariate", action = "scramble",
kind = "numeric", freq_or_range = "[0, 1]", pii_suspected = FALSE,
notes = "test", stringsAsFactors = FALSE
))
expect_error(roles_validate(r_extra, iris), "not in.*df")
})
test_that("roles_validate returns the roles table invisibly", {
r <- make_valid_roles()
out <- withVisible(roles_validate(r))
expect_false(out$visible)
expect_identical(out$value, r)
})
Any scripts or data that you put into this service are public.
Add the following code to your website.
For more information on customizing the embed code, read Embedding Snippets.