coverage now over 85%!

R/checkers.R

@@ -43,7 +43,8 @@ warn_if_offsets_present <- function(jags_stuff,
 warn_if_formula_has_lhs <- function(formula,
                                     arg = rlang::caller_arg(formula),
                                     call = rlang::caller_env()) {
-  if (length(formula) > 2) {
+  has_lhs <- rlang::is_formula(formula, lhs = TRUE)
+  if (has_lhs) {
     cli::cli_warn(
       c(
         "Formula has a left hand side",
@@ -53,3 +54,18 @@ warn_if_formula_has_lhs <- function(formula,
     )
   }
 }
+
+check_if_formula <- function(formula,
+                             arg = rlang::caller_arg(formula),
+                             call = rlang::caller_env()){
+  not_formula <- !rlang::is_bare_formula(formula)
+  if (not_formula) {
+    cli::cli_abort(
+      c(
+        "Input must be a formula",
+        "We see that {.code formula} has class, {.cls {class(formula)}}."
+      ),
+      call = call
+    )
+  }
+}

R/jagam2greta.R

@@ -25,9 +25,13 @@ jagam2greta <- function(formula,
 
   # do the jagam call, store the JAGS code gets stored in jags_spec
   jags_spec <- ""
-  jags_stuff <- mgcv::jagam(formula, data,
-    family = stats::gaussian(), knots = knots,
-    file = textConnection("jags_spec",
+  jags_stuff <- mgcv::jagam(
+    formula = formula,
+    data = data,
+    family = stats::gaussian(),
+    knots = knots,
+    file = textConnection(
+      "jags_spec",
       open = "a",
       local = TRUE
     )

R/smooths.R

@@ -77,6 +77,7 @@ smooths <- function(formula,
                     knots = NULL,
                     sp = NULL,
                     tol = 0) {
+  check_if_formula(formula)
   warn_if_formula_has_lhs(formula)
 
   # get all the MGCV objects for Bayesian version, converted to greta arrays

tests/testthat/_snaps/evaluate-smooths.md

@@ -1,4 +1,8 @@
 # evaluate_smooths errors appropriately
 
-    Can only evaluate smooths from greta arrays created with `greta.gam::smooths()`
+    Code
+      evaluate_smooths("thing")
+    Condition
+      Error in `evaluate_smooths()`:
+      ! Can only evaluate smooths from greta arrays created with `greta.gam::smooths()`
 

tests/testthat/_snaps/stop-when-dummy-in-data.md

@@ -1,5 +1,18 @@
 # stop_when_dummy_in_data errors appropriately
 
-    Data cannot already contain column named `dummy`
-    i Rename existing column, perhaps to `dummy1`
+    Code
+      stop_when_dummy_in_data(test_df)
+    Condition
+      Error:
+      ! Data cannot already contain column named `dummy`
+      i Rename existing column, perhaps to `dummy1`
+
+---
+
+    Code
+      smooths(~ s(dummy), data = data.frame(dummy = x))
+    Condition
+      Error in `jagam2greta()`:
+      ! Data cannot already contain column named `dummy`
+      i Rename existing column, perhaps to `dummy1`
 

tests/testthat/test-check-formula.R

@@ -0,0 +1,16 @@
+test_that("check_if_formula works", {
+  expect_snapshot(
+    error = TRUE,
+    check_if_formula("potato")
+  )
+  expect_snapshot(
+    error = TRUE,
+    check_if_formula("y~x")
+  )
+  expect_success(
+    check_if_formula(y~x)
+  )
+  expect_success(
+    check_if_formula(y~s(x))
+  )
+})

tests/testthat/test-evaluate-smooths.R

@@ -1,5 +1,21 @@
+n <- 30
+x <- runif(n, 0, 10)
+f <- function(x) {
+  sin(x * 2) + 1.6 * (x < 3) - 1.4 * (x > 7)
+}
+y <- f(x) + rnorm(n, 0, 0.3)
+x_plot <- seq(0, 10, length.out = 200)
+
+z <- smooths(~ s(x), data = data.frame(x = x))
+
+distribution(y) <- normal(z, 0.3)
+
 test_that("evaluate_smooths errors appropriately", {
-  expect_snapshot_error(
+  expect_snapshot(
+    error = TRUE,
     evaluate_smooths("thing")
   )
+  expect_no_error(
+    z_pred <- evaluate_smooths(z, newdata = data.frame(x = x_plot))
+  )
 })

tests/testthat/test-stop-when-dummy-in-data.R

@@ -3,12 +3,30 @@ test_df <- data.frame(
   thing = LETTERS[1:5]
 )
 
+n <- 30
+x <- runif(n, 0, 10)
+f <- function(x) {
+  sin(x * 2) + 1.6 * (x < 3) - 1.4 * (x > 7)
+}
+y <- f(x) + rnorm(n, 0, 0.3)
+x_plot <- seq(0, 10, length.out = 200)
+
+z <- smooths(~ s(x), data = data.frame(x = x))
+
+
 test_that("stop_when_dummy_in_data errors appropriately", {
-  expect_snapshot_error(
+  expect_snapshot(
+    error = TRUE,
     stop_when_dummy_in_data(test_df)
   )
 
   expect_no_error(
     stop_when_dummy_in_data(mtcars)
   )
+
+  expect_snapshot(
+    error = TRUE,
+    smooths(~ s(dummy), data = data.frame(dummy = x))
+  )
+
 })