keep filtered aggregates with subsequent select in semi_join

NEWS.md

@@ -1,5 +1,8 @@
 # dbplyr (development version)
 
+* `semi_join()` will no longer inline away an aggregate filter (i.e. `HAVING`
+  clause) that was followed by a `select()` (@ejneer, #1474)
+
 * Refined the `select()` inlining criteria to keep computed columns used to
   `arrange()` subqueries that are eliminated by a subsequent select (@ejneer,
   #1437).

R/lazy-select-query.R

@@ -123,6 +123,9 @@ is_lazy_select_query_simple <- function(x,
   if (!is_empty(x$limit)) {
     return(FALSE)
   }
+  if (!is_empty(x$having)) {
+    return(FALSE)
+  }
 
   TRUE
 }

tests/testthat/_snaps/verb-joins.md

@@ -200,6 +200,24 @@
           (`df_RHS`.`b` = 2)
       )
 
+# filtered aggregates with subsequent select are not inlined away in semi_join (#1474)
+
+    Code
+      out
+    Output
+      <SQL>
+      SELECT `df`.*
+      FROM `df`
+      WHERE EXISTS (
+        SELECT 1 FROM (
+        SELECT `x`
+        FROM `df`
+        GROUP BY `x`
+        HAVING (COUNT(*) = 1.0)
+      ) AS `RHS`
+        WHERE (`df`.`x` = `RHS`.`x`)
+      )
+
 # multiple joins create a single query
 
     Code

tests/testthat/test-verb-joins.R

@@ -616,6 +616,23 @@ test_that("filter() before semi join is not when y has other operations", {
   expect_null(lq$where)
 })
 
+test_that("filtered aggregates with subsequent select are not inlined away in semi_join (#1474)", {
+  lf <- lazy_frame(x = 1, y = 2, z = 3)
+  lf2 <- lazy_frame(x = 1, a = 2, b = 3)
+
+  out <- semi_join(
+    lf,
+    lf2 %>%
+      dplyr::summarize(n = n(), .by = "x") %>%
+      filter(n == 1) %>%
+      select(x)
+  )
+  lq <- out$lazy_query
+
+  expect_equal(lq$y$having, list(quo(n() == 1)), ignore_formula_env = TRUE)
+  expect_snapshot(out)
+})
+
 test_that("multiple joins create a single query", {
   lf <- lazy_frame(x = 1, a = 1, .name = "df1")
   lf2 <- lazy_frame(x = 1, b = 2, .name = "df2")