check_*() error on empty input when allow_empty = FALSE (#30)

[Melkiades]

What changes are proposed in this pull request?

• check_*() functions now error on empty input when allow_empty = FALSE (previously empty but correctly-typed input silently passed class/type checks). (allow_empty updates #30, @Melkiades)

Empty but correctly-typed input (e.g. character(0), numeric(0), 0-row/0-column data frames) previously fell through the empty-handling guard and silently passed the downstream class/type checks even under the default allow_empty = FALSE. The guard is now explicit: it returns invisibly only when allow_empty = TRUE, otherwise it errors. This was applied uniformly across all affected guard sites (check_class(), check_length(), check_range(), check_binary(), check_formula_list_selector(), check_integerish(), check_scalar_integerish(), check_numeric(), and the delegating check_data_frame() / check_logical()).

Downstream consumers (cards, cardx, crane, pager) were scanned: no call site passes a legitimately-empty argument without allow_empty = TRUE, and no consumer test relies on the old silent-pass behavior, so this change is safe.

Reference GitHub issue associated with pull request. closes #30

---

Pre-review Checklist (if item does not apply, mark is as complete)

• All GitHub Action workflows pass with a ✅
• PR branch has pulled the most recent updates from master branch: usethis::pr_merge_main()
• If a bug was fixed, a unit test was added.
• If a standalone script was updated, a comment is added to the script header (changelog) AND the last-updated field has been updated.
• Code coverage is suitable for any new functions/features (generally, 100% coverage for new code): devtools::test_coverage()
• Request a reviewer

Reviewer Checklist (if item does not apply, mark is as complete)

• If a bug was fixed, a unit test was added.
• If a standalone script was updated, a comment is added to the script header (changelog) AND the last-updated field has been updated.
• Run pkgdown::build_site(). Check the R console for errors, and review the rendered website.
• Code coverage is suitable for any new functions/features: devtools::test_coverage()

When the branch is ready to be merged:

• All GitHub Action workflows pass with a ✅
• Approve Pull Request
• Merge the PR. Please use "Squash and merge" or "Rebase and merge".
• Create an issue in any repositories using {standalone} to update the standalone scripts.

[github-actions]

✅ All contributors have signed the CLA
_{Posted by the CLA Assistant Lite bot.}

[Melkiades]

I have read the CLA Document and I hereby sign the CLA

[Melkiades]

recheck

[Copilot]

Pull request overview

Updates {standalone} input validation so check_*() helpers consistently error on empty inputs when allow_empty = FALSE, aligning behavior with the documented/default expectation and Issue #30.

Changes:

• Make empty-input handling explicit across multiple check_*() functions (error by default; return invisibly when allow_empty = TRUE).
• Update/add test coverage to assert the new empty-input behavior.
• Document the behavior change in the standalone script changelog and NEWS.md.

Reviewed changes

Copilot reviewed 3 out of 3 changed files in this pull request and generated 4 comments.

File	Description
R/standalone-checks.R	Tightens guard logic so empty inputs no longer silently pass type/class checks when allow_empty = FALSE.
tests/testthat/test-standalone-checks.R	Updates existing expectations and adds a regression test block for empty-input behavior.
NEWS.md	Notes the user-visible behavior change for check_*() functions.

Suppressed comments (3)

R/standalone-checks.R:626

• With the new empty-input erroring behavior, the empty-branch abort currently reuses message ("must be numeric."), which is confusing for numeric(0) because it is numeric. Use a dedicated message for the empty-input branch so the error explains the actual failure condition.

    cli::cli_abort(message, class = c(class, "standalone-checks"), call = call, .envir = envir)

R/standalone-checks.R:508

• Similar to check_numeric(): in the empty-input branch this abort reuses the general message, which is confusing for integer(0) because it satisfies the "integer vector" predicate (rlang::is_integerish()) but is rejected due to emptiness. Using an explicit empty-input message here will make the failure reason clear.

    cli::cli_abort(message, class = c(class, "standalone-checks"), call = call, .envir = envir)

R/standalone-checks.R:540

• The default error message text is ungrammatical ("must an scalar integer"). This will surface in snapshots/consumer logs and is easy to correct while touching this function.

  # if empty: return invisibly when allowed, otherwise error
  if (rlang::is_empty(x)) {
    if (isTRUE(allow_empty)) {
      return(invisible(x))
    }
    cli::cli_abort(message, class = c(class, "standalone-checks"), call = call, .envir = envir)

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[shajoezhu]

lgtm