should be it

external/duckdb/test/optimizer/statistics/statistics_setop.test

@@ -0,0 +1,143 @@
+# name: test/optimizer/statistics/statistics_setop.test
+# description: Statistics propagation test with set operations
+# group: [statistics]
+
+statement ok
+CREATE TABLE integers AS SELECT * FROM (VALUES (1), (2), (3)) tbl(i);
+
+statement ok
+CREATE TABLE integers2 AS SELECT * FROM (VALUES (4), (5), (6)) tbl(i);
+
+statement ok
+CREATE TABLE integers3 AS SELECT * FROM (VALUES (4), (5), (NULL)) tbl(i);
+
+statement ok
+PRAGMA explain_output = OPTIMIZED_ONLY;
+
+# union all
+# total min/max after union all is [1, 6]: 7 is out of bounds, so i=7 can be optimized away
+query II
+EXPLAIN SELECT i=7 FROM (SELECT * FROM integers UNION ALL SELECT * FROM integers2) tbl(i);
+----
+logical_opt	<!REGEX>:.*\(i = 7\).*
+
+# 5 is in bounds, so this cannot be optimized away
+query II
+EXPLAIN SELECT i=5 FROM (SELECT * FROM integers UNION ALL SELECT * FROM integers2) tbl(i);
+----
+logical_opt	<REGEX>:.*\(i = 5\).*
+
+# neither integers nor integers2 has null values, so this is false
+query II
+EXPLAIN SELECT i IS NULL FROM (SELECT * FROM integers UNION ALL SELECT * FROM integers2) tbl(i);
+----
+logical_opt	<!REGEX>:.*IS NULL.*
+
+# integers3 has null values, so once we include integers3 we need to execute the IS_NULL predicate
+query II
+EXPLAIN SELECT i IS NULL FROM (SELECT * FROM integers UNION ALL SELECT * FROM integers2 UNION ALL SELECT * FROM integers3) tbl(i);
+----
+logical_opt	<REGEX>:.*IS NULL.*
+
+# except
+# except has the same stats as the LHS (as in the worst case, nothing is filtered out)
+# in this case the LHS stats are [4, 6] without null values
+query II
+EXPLAIN SELECT i=7 FROM (SELECT * FROM integers2 EXCEPT SELECT * FROM integers3) tbl(i);
+----
+logical_opt	<!REGEX>:.*\(i = 7\).*
+
+query II
+EXPLAIN SELECT i=5 FROM (SELECT * FROM integers2 EXCEPT SELECT * FROM integers3) tbl(i);
+----
+logical_opt	<REGEX>:.*\(i = 5\).*
+
+query II
+EXPLAIN SELECT i IS NULL FROM (SELECT * FROM integers2 EXCEPT SELECT * FROM integers3) tbl(i);
+----
+logical_opt	<!REGEX>:.*IS NULL.*
+
+# intersect
+# intersect has as stats the intersection of the LHS with the RHS
+# in this case that would be [4, 5] with no null values
+# however, we don't actually intersect the stats yet, so limit our testing to the stats of the LHS
+query II
+EXPLAIN SELECT i=7 FROM (SELECT * FROM integers2 INTERSECT SELECT * FROM integers3) tbl(i);
+----
+logical_opt	<!REGEX>:.*\(i = 7\).*
+
+query II
+EXPLAIN SELECT i=5 FROM (SELECT * FROM integers2 INTERSECT SELECT * FROM integers3) tbl(i);
+----
+logical_opt	<REGEX>:.*\(i = 5\).*
+
+# now check the results of all these queries
+query I
+SELECT i=7 FROM (SELECT * FROM integers UNION ALL SELECT * FROM integers2) tbl(i);
+----
+0
+0
+0
+0
+0
+0
+
+query I
+SELECT i=5 FROM (SELECT * FROM integers UNION ALL SELECT * FROM integers2) tbl(i);
+----
+0
+0
+0
+0
+1
+0
+
+query I
+SELECT i IS NULL FROM (SELECT * FROM integers UNION ALL SELECT * FROM integers2) tbl(i);
+----
+0
+0
+0
+0
+0
+0
+
+query I
+SELECT i IS NULL FROM (SELECT * FROM integers UNION ALL SELECT * FROM integers2 UNION ALL SELECT * FROM integers3) tbl(i);
+----
+0
+0
+0
+0
+0
+0
+0
+0
+1
+
+query I
+SELECT i=7 FROM (SELECT * FROM integers2 EXCEPT SELECT * FROM integers3) tbl(i);
+----
+0
+
+query I
+SELECT i=5 FROM (SELECT * FROM integers2 EXCEPT SELECT * FROM integers3) tbl(i);
+----
+0
+
+query I
+SELECT i IS NULL FROM (SELECT * FROM integers2 EXCEPT SELECT * FROM integers3) tbl(i);
+----
+0
+
+query I
+SELECT i=7 FROM (SELECT * FROM integers2 INTERSECT SELECT * FROM integers3) tbl(i);
+----
+0
+0
+
+query I
+SELECT i=5 FROM (SELECT * FROM integers2 INTERSECT SELECT * FROM integers3) tbl(i) ORDER BY i;
+----
+0
+1