Fix issue where the sf package "sf_column" may not be the last column in an sf object

NEWS.md

@@ -25,6 +25,8 @@ These are all minor breaking changes resulting from enhancements and are not exp
 
 * `get_one_to_one()` no longer errors with near-equal values that become identical factor levels (fix #543, thanks to @olivroy for reporting)
 
+* `clean_names()` for sf objects now works in cases when the sf_column is not the last column name (fix #578, thanks to @ar-puuk for reporting and @billdenney for fixing)
+
 ## Refactoring
 
 * Remove dplyr verbs superseded in dplyr 1.0.0 (#547, @olivroy)

R/clean_names.R

@@ -97,14 +97,14 @@ clean_names.sf <- function(dat, ...) {
   } # nocov end
   # get old names
   sf_names <- names(dat)
-  # identify ending column index to clean
-  n_cols <- length(dat) - 1
+  # Clean the names except for the "sf_column" which is used internally by sf
+  cols_to_rename <- which(!(sf_names %in% attr(dat, "sf_column")))
   # clean all but last column
-  sf_cleaned <- make_clean_names(sf_names[1:n_cols], ...)
+  sf_cleaned <- make_clean_names(sf_names[cols_to_rename], ...)
   # rename original df
-  names(dat)[1:n_cols] <- sf_cleaned
+  names(dat)[cols_to_rename] <- sf_cleaned
 
-  return(dat)
+  dat
 }
 
 #' @rdname clean_names

tests/testthat/test-clean-names.R

@@ -461,6 +461,15 @@ test_that("Names are cleaned appropriately without attaching sf", {
 
   expect_equal(names(clean)[4], "cnty_id")
   expect_s3_class(clean, "sf")
+
+  # Issue #578, sf_column attribute needs to be untouched, it may not be the
+  # last column name
+  issue_578_sf <- readRDS("testdata/issue-578-sf.rds")
+  issue_578_sf_clean <- clean_names(issue_578_sf)
+  expect_error(
+    print(issue_578_sf_clean),
+    NA
+  )
 })
 
 test_that("Names are cleaned appropriately", {