fix: grepl(x = NA) gives correct results

R/relational-duckdb.R

@@ -34,7 +34,7 @@ duckplyr_macros <- c(
   # https://github.com/duckdb/duckdb/discussions/8599
   # "as.Date" = '(x) AS strptime(x, \'%Y-%m-%d\')',
 
-  "grepl" = "(pattern, x) AS regexp_matches(x, pattern)",
+  "grepl" = "(pattern, x) AS (CASE WHEN x IS NULL THEN FALSE ELSE regexp_matches(x, pattern) END)",
   "as.integer" = "(x) AS CAST(x AS int32)",
   "if_else" = "(test, yes, no) AS (CASE WHEN test THEN yes ELSE no END)",
   "|" = "(x, y) AS (x OR y)",

tests/testthat/test-as_duckplyr_df.R

@@ -1345,6 +1345,19 @@ test_that("as_duckplyr_df() and mutate(c = 0, d = -1, e = log(c), f = log10(d))"
   expect_equal(pre, post)
 })
 
+
+test_that("as_duckplyr_df() and mutate(c = NA_character_, d = grepl('.', c))", {
+  # Data
+  test_df <- data.frame(a = 1:6 + 0, b = 2, g = rep(1:3, 1:3))
+
+  # Run
+  pre <- test_df %>% as_duckplyr_df() %>% mutate(c = NA_character_, d = grepl('.', c))
+  post <- test_df %>% mutate(c = NA_character_, d = grepl('.', c)) %>% as_duckplyr_df()
+
+  # Compare
+  expect_equal(pre, post)
+})
+
 test_that("as_duckplyr_df() and nest_by()", {
   withr::local_envvar(DUCKPLYR_FORCE = "FALSE")
 

tools/00-funs.R

@@ -608,6 +608,9 @@ test_extra_arg_map <- list(
     "c = 0, d = -1, e = log(c), f = log(d)",
     "c = 0, d = -1, e = log(c), f = log10(d)",
 
+    # grepl with NA
+    "c = NA_character_, d = grepl('.', c)",
+
     NULL
   ),
   nest_join = "join_by(a)",