Address Copilot review: NULL semantics, iterator validation, single() perf, tests

- Rewrite predicate functions from EXISTS_SUBLINK to EXPR_SUBLINK with
  aggregate-based CASE expressions (bool_or + IS TRUE/FALSE/NULL) to
  preserve three-valued Cypher NULL semantics
- Add list_length check in extract_iter_variable_name() to reject
  qualified names like x.y as iterator variables
- Add copy/read support for cypher_predicate_function ExtensibleNode
  to prevent query rewriter crashes
- Use IS TRUE filtering in single() count (LIMIT 2 optimization
  breaks correlated variable refs in graph contexts -- documented)
- Add 13 NULL regression tests: null list input, null elements,
  null predicates for all four functions

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Author: gregfelice

regress/expected/predicate_functions.out

@@ -154,6 +154,132 @@ $$) AS (result agtype);
  false
 (1 row)
 
+--
+-- NULL list input: all/any/none return null, single returns false
+-- (unnest of NULL produces zero rows; aggregates return NULL over
+-- empty input, but count(*) returns 0)
+--
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN all(x IN null WHERE x > 0)
+$$) AS (result agtype);
+ result 
+--------
+ 
+(1 row)
+
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN any(x IN null WHERE x > 0)
+$$) AS (result agtype);
+ result 
+--------
+ 
+(1 row)
+
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN none(x IN null WHERE x > 0)
+$$) AS (result agtype);
+ result 
+--------
+ 
+(1 row)
+
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN single(x IN null WHERE x > 0)
+$$) AS (result agtype);
+ result 
+--------
+ false
+(1 row)
+
+--
+-- NULL predicate results: three-valued logic
+--
+-- Note: In AGE's agtype, null is a first-class value.  The comparison
+-- agtype_null > agtype_integer evaluates to true (not SQL NULL).
+-- Three-valued logic only applies when the predicate itself is a
+-- literal null constant, which becomes SQL NULL after coercion.
+-- agtype null in list: null > 0 = true in AGE, so any() = true
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN any(x IN [null] WHERE x > 0)
+$$) AS (result agtype);
+ result 
+--------
+ true
+(1 row)
+
+-- agtype null + real values: all comparisons are true
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN any(x IN [null, 1, 2] WHERE x > 0)
+$$) AS (result agtype);
+ result 
+--------
+ true
+(1 row)
+
+-- literal null predicate: pred = SQL NULL -> three-valued logic
+-- all([1] WHERE null) = null (unknown)
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN all(x IN [1] WHERE null)
+$$) AS (result agtype);
+ result 
+--------
+ 
+(1 row)
+
+-- agtype null in list: null > 0 = true in AGE, so all() = true
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN all(x IN [1, null, 2] WHERE x > 0)
+$$) AS (result agtype);
+ result 
+--------
+ true
+(1 row)
+
+-- -1 > 0 = false, so all() = false
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN all(x IN [1, null, -1] WHERE x > 0)
+$$) AS (result agtype);
+ result 
+--------
+ false
+(1 row)
+
+-- agtype null > 0 = true in AGE, so none() = false
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN none(x IN [null] WHERE x > 0)
+$$) AS (result agtype);
+ result 
+--------
+ false
+(1 row)
+
+-- 5 > 0 = true, so none() = false
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN none(x IN [null, 5] WHERE x > 0)
+$$) AS (result agtype);
+ result 
+--------
+ false
+(1 row)
+
+-- agtype null > 0 = true AND 5 > 0 = true: 2 matches, single = false
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN single(x IN [null, 5] WHERE x > 0)
+$$) AS (result agtype);
+ result 
+--------
+ false
+(1 row)
+
+-- single() with null list: count(*) = 0, 0 = 1 -> false
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN single(x IN null WHERE x > 0)
+$$) AS (result agtype);
+ result 
+--------
+ false
+(1 row)
+
 --
 -- Integration with graph data
 --

regress/sql/predicate_functions.sql

@@ -99,6 +99,81 @@ SELECT * FROM cypher('predicate_functions', $$
     RETURN single(x IN [] WHERE x > 0)
 $$) AS (result agtype);
 
+--
+-- NULL list input: all/any/none return null, single returns false
+-- (unnest of NULL produces zero rows; aggregates return NULL over
+-- empty input, but count(*) returns 0)
+--
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN all(x IN null WHERE x > 0)
+$$) AS (result agtype);
+
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN any(x IN null WHERE x > 0)
+$$) AS (result agtype);
+
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN none(x IN null WHERE x > 0)
+$$) AS (result agtype);
+
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN single(x IN null WHERE x > 0)
+$$) AS (result agtype);
+
+--
+-- NULL predicate results: three-valued logic
+--
+-- Note: In AGE's agtype, null is a first-class value.  The comparison
+-- agtype_null > agtype_integer evaluates to true (not SQL NULL).
+-- Three-valued logic only applies when the predicate itself is a
+-- literal null constant, which becomes SQL NULL after coercion.
+
+-- agtype null in list: null > 0 = true in AGE, so any() = true
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN any(x IN [null] WHERE x > 0)
+$$) AS (result agtype);
+
+-- agtype null + real values: all comparisons are true
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN any(x IN [null, 1, 2] WHERE x > 0)
+$$) AS (result agtype);
+
+-- literal null predicate: pred = SQL NULL -> three-valued logic
+-- all([1] WHERE null) = null (unknown)
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN all(x IN [1] WHERE null)
+$$) AS (result agtype);
+
+-- agtype null in list: null > 0 = true in AGE, so all() = true
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN all(x IN [1, null, 2] WHERE x > 0)
+$$) AS (result agtype);
+
+-- -1 > 0 = false, so all() = false
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN all(x IN [1, null, -1] WHERE x > 0)
+$$) AS (result agtype);
+
+-- agtype null > 0 = true in AGE, so none() = false
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN none(x IN [null] WHERE x > 0)
+$$) AS (result agtype);
+
+-- 5 > 0 = true, so none() = false
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN none(x IN [null, 5] WHERE x > 0)
+$$) AS (result agtype);
+
+-- agtype null > 0 = true AND 5 > 0 = true: 2 matches, single = false
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN single(x IN [null, 5] WHERE x > 0)
+$$) AS (result agtype);
+
+-- single() with null list: count(*) = 0, 0 = 1 -> false
+SELECT * FROM cypher('predicate_functions', $$
+    RETURN single(x IN null WHERE x > 0)
+$$) AS (result agtype);
+
 --
 -- Integration with graph data
 --

src/backend/nodes/ag_nodes.c

@@ -134,7 +134,7 @@ const ExtensibleNodeMethods node_methods[] = {
     DEFINE_NODE_METHODS_EXTENDED(cypher_delete_information),
     DEFINE_NODE_METHODS_EXTENDED(cypher_delete_item),
     DEFINE_NODE_METHODS_EXTENDED(cypher_merge_information),
-    DEFINE_NODE_METHODS(cypher_predicate_function)
+    DEFINE_NODE_METHODS_EXTENDED(cypher_predicate_function)
 };
 
 static bool equal_ag_node(const ExtensibleNode *a, const ExtensibleNode *b)

src/backend/nodes/cypher_copyfuncs.c

@@ -169,3 +169,15 @@ void copy_cypher_merge_information(ExtensibleNode *newnode, const ExtensibleNode
     COPY_SCALAR_FIELD(merge_function_attr);
     COPY_NODE_FIELD(path);
 }
+
+/* copy function for cypher_predicate_function */
+void copy_cypher_predicate_function(ExtensibleNode *newnode,
+                                    const ExtensibleNode *from)
+{
+    COPY_LOCALS(cypher_predicate_function);
+
+    COPY_SCALAR_FIELD(kind);
+    COPY_STRING_FIELD(varname);
+    COPY_NODE_FIELD(expr);
+    COPY_NODE_FIELD(where);
+}

src/backend/nodes/cypher_readfuncs.c

@@ -311,3 +311,17 @@ void read_cypher_merge_information(struct ExtensibleNode *node)
     READ_INT_FIELD(merge_function_attr);
     READ_NODE_FIELD(path);
 }
+
+/*
+ * Deserialize a string representing the cypher_predicate_function
+ * data structure.
+ */
+void read_cypher_predicate_function(struct ExtensibleNode *node)
+{
+    READ_LOCALS(cypher_predicate_function);
+
+    READ_ENUM_FIELD(kind, cypher_predicate_function_kind);
+    READ_STRING_FIELD(varname);
+    READ_NODE_FIELD(expr);
+    READ_NODE_FIELD(where);
+}