Fixed handling of IN (subquery) expressions.

Author: cristianberneanu

CHANGELOG.md

@@ -4,6 +4,7 @@
   - Allow casts between `int4` and `int8`.
   - Allow more metadata discovery queries.
   - Allow more statement types.
+  - Fixed handling of `IN (subquery)` expressions.
 
 ## Version 1.0.1
   - Fixed some docs links.

pg_diffix/query/anonymization.h

@@ -18,6 +18,11 @@ typedef struct AnonQueryLinks AnonQueryLinks;
  */
 extern AnonQueryLinks *compile_query(Query *query, List *personal_relations);
 
+/*
+ * Calls `rewrite_plan` for each item in a list of Plan nodes.
+ */
+extern void rewrite_plan_list(List *plans, AnonQueryLinks *links);
+
 /*
  * Wraps anonymizing Agg nodes with BucketScan nodes. Does nothing if links is NULL.
  * Frees memory associated with links after rewriting.

src/hooks.c

@@ -94,6 +94,7 @@ static PlannedStmt *pg_diffix_planner(
   PlannedStmt *plan = planner(query, query_string, cursorOptions, boundParams);
 
   plan->planTree = rewrite_plan(plan->planTree, links);
+  rewrite_plan_list(plan->subplans, links);
 
   return plan;
 }

src/query/anonymization.c

@@ -734,7 +734,7 @@ static void unwrap_having_qual(Plan *plan)
   plan->qual = make_ands_implicit(having_qual);
 }
 
-static void rewrite_plan_list(List *plans, AnonQueryLinks *links)
+void rewrite_plan_list(List *plans, AnonQueryLinks *links)
 {
   ListCell *cell;
   foreach (cell, plans)

test/expected/misc.out

@@ -51,6 +51,12 @@ HAVING age IS NULL OR age > 40;
   64 |     2
 (3 rows)
 
+SELECT 'London' IN (SELECT city FROM test_customers);
+ ?column? 
+----------
+ f
+(1 row)
+
 -- Prevent post-processing filters from being pushed down
 EXPLAIN SELECT age, count(*)
 FROM test_patients

test/expected/validation.out

@@ -156,6 +156,12 @@ SELECT 1 WHERE EXISTS (SELECT city FROM test_validation);
 ----------
 (0 rows)
 
+SELECT 'London' IN (SELECT city FROM test_validation);
+ ?column? 
+----------
+ f
+(1 row)
+
 -- Anonymizing leaf subqueries are supported.
 SELECT * FROM ( SELECT COUNT(*) FROM test_validation ) x;
  count 

test/sql/misc.sql

@@ -35,6 +35,8 @@ FROM test_patients
 GROUP BY age
 HAVING age IS NULL OR age > 40;
 
+SELECT 'London' IN (SELECT city FROM test_customers);
+
 -- Prevent post-processing filters from being pushed down
 EXPLAIN SELECT age, count(*)
 FROM test_patients

test/sql/validation.sql

@@ -101,6 +101,7 @@ SELECT city FROM test_validation UNION SELECT city FROM test_validation;
 -- Anonymizing sublinks are supported.
 SELECT EXISTS (SELECT city FROM test_validation);
 SELECT 1 WHERE EXISTS (SELECT city FROM test_validation);
+SELECT 'London' IN (SELECT city FROM test_validation);
 
 -- Anonymizing leaf subqueries are supported.
 SELECT * FROM ( SELECT COUNT(*) FROM test_validation ) x;