Fix Issue 1691 - MERGE incorrectly creates multiple vertices

Fixed issue 1691 where MERGE would incorrectly create multiple vertices. This only occurred when MERGE was being driven by a previous clause. NOTE: To be more correct, the issue is with creating duplicate paths. The reason this happened was due to the visibility of tuples that were created during the MERGE instance. It is not possible to add them in to be rescanned. Because of this limitation, it required adding the ability to MERGE to know what path an instance had already created. Added regression tests.
apache · Apr 2, 2024 · 9c17cfd · 9c17cfd
1 parent 8bd45a9
commit 9c17cfd
Show file tree

Hide file tree

Showing 4 changed files with 744 additions and 38 deletions.
diff --git a/regress/expected/cypher_merge.out b/regress/expected/cypher_merge.out
@@ -1263,8 +1263,8 @@ $$) as (a agtype);
 ---
 (0 rows)
 
----
---- Issue 1630 MERGE using array not working in some cases
+--
+-- Issue 1630 - MERGE using array not working in some cases
 --
 SELECT * FROM create_graph('issue_1630');
 NOTICE:  graph "issue_1630" has been created
@@ -1387,7 +1387,151 @@ SELECT * FROM cypher('issue_1630',
  {"id": 844424930131974, "label": "PERSION", "properties": {"last": "snow", "first": "jon"}}::vertex | {"last": "snow", "first": "jon"}
 (1 row)
 
---clean up
+--
+-- Issue 1691 - MERGE incorrectly creates multiple vertices
+--
+SELECT * FROM create_graph('issue_1691');
+NOTICE:  graph "issue_1691" has been created
+ create_graph 
+--------------
+
+(1 row)
+
+SELECT * FROM cypher('issue_1691', $$ MATCH (u) RETURN (u) $$) AS (u agtype);
+ u 
+---
+(0 rows)
+
+-- should only create 2 distinct rows but return 4, the extra 2 being duplicates
+SELECT * FROM cypher('issue_1691', $$ UNWIND ["foo", "bar", "foo", "foo"] as n
+                                      MERGE (u {name: n})-[e:knows]->(v)
+                                      RETURN u, e, v  $$) AS (u agtype, e agtype, v agtype);
+                                      u                                      |                                                             e                                                             |                               v                                
+-----------------------------------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------+----------------------------------------------------------------
+ {"id": 281474976710657, "label": "", "properties": {"name": "foo"}}::vertex | {"id": 844424930131969, "label": "knows", "end_id": 281474976710658, "start_id": 281474976710657, "properties": {}}::edge | {"id": 281474976710658, "label": "", "properties": {}}::vertex
+ {"id": 281474976710659, "label": "", "properties": {"name": "bar"}}::vertex | {"id": 844424930131970, "label": "knows", "end_id": 281474976710660, "start_id": 281474976710659, "properties": {}}::edge | {"id": 281474976710660, "label": "", "properties": {}}::vertex
+ {"id": 281474976710657, "label": "", "properties": {"name": "foo"}}::vertex | {"id": 844424930131969, "label": "knows", "end_id": 281474976710658, "start_id": 281474976710657, "properties": {}}::edge | {"id": 281474976710658, "label": "", "properties": {}}::vertex
+ {"id": 281474976710657, "label": "", "properties": {"name": "foo"}}::vertex | {"id": 844424930131969, "label": "knows", "end_id": 281474976710658, "start_id": 281474976710657, "properties": {}}::edge | {"id": 281474976710658, "label": "", "properties": {}}::vertex
+(4 rows)
+
+-- should only return the same above 4 rows
+SELECT * FROM cypher('issue_1691', $$ UNWIND ["foo", "bar", "foo", "foo"] as n
+                                      MERGE (u {name: n})-[e:knows]->(v)
+                                      RETURN u, e, v  $$) AS (u agtype, e agtype, v agtype);
+                                      u                                      |                                                             e                                                             |                               v                                
+-----------------------------------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------+----------------------------------------------------------------
+ {"id": 281474976710657, "label": "", "properties": {"name": "foo"}}::vertex | {"id": 844424930131969, "label": "knows", "end_id": 281474976710658, "start_id": 281474976710657, "properties": {}}::edge | {"id": 281474976710658, "label": "", "properties": {}}::vertex
+ {"id": 281474976710659, "label": "", "properties": {"name": "bar"}}::vertex | {"id": 844424930131970, "label": "knows", "end_id": 281474976710660, "start_id": 281474976710659, "properties": {}}::edge | {"id": 281474976710660, "label": "", "properties": {}}::vertex
+ {"id": 281474976710657, "label": "", "properties": {"name": "foo"}}::vertex | {"id": 844424930131969, "label": "knows", "end_id": 281474976710658, "start_id": 281474976710657, "properties": {}}::edge | {"id": 281474976710658, "label": "", "properties": {}}::vertex
+ {"id": 281474976710657, "label": "", "properties": {"name": "foo"}}::vertex | {"id": 844424930131969, "label": "knows", "end_id": 281474976710658, "start_id": 281474976710657, "properties": {}}::edge | {"id": 281474976710658, "label": "", "properties": {}}::vertex
+(4 rows)
+
+-- should only return 2 distinct rows from above
+SELECT * FROM cypher('issue_1691', $$ MATCH (u)-[e]->(v)
+                                      RETURN u, e, v  $$) AS (u agtype, e agtype, v agtype);
+                                      u                                      |                                                             e                                                             |                               v                                
+-----------------------------------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------+----------------------------------------------------------------
+ {"id": 281474976710657, "label": "", "properties": {"name": "foo"}}::vertex | {"id": 844424930131969, "label": "knows", "end_id": 281474976710658, "start_id": 281474976710657, "properties": {}}::edge | {"id": 281474976710658, "label": "", "properties": {}}::vertex
+ {"id": 281474976710659, "label": "", "properties": {"name": "bar"}}::vertex | {"id": 844424930131970, "label": "knows", "end_id": 281474976710660, "start_id": 281474976710659, "properties": {}}::edge | {"id": 281474976710660, "label": "", "properties": {}}::vertex
+(2 rows)
+
+SELECT * FROM cypher('issue_1691', $$MATCH ()-[e]->() DELETE e $$) AS (a agtype);
+ a 
+---
+(0 rows)
+
+SELECT * FROM cypher('issue_1691', $$MATCH (u) DELETE u $$) AS (a agtype);
+ a 
+---
+(0 rows)
+
+-- should only create 1 record but return 2, one a dup of the other
+SELECT * FROM cypher('issue_1691', $$ UNWIND ["foo", "foo"] AS each
+                                      MERGE (v:TEST {name: each})
+                                      RETURN v $$) AS (v agtype);
+                                        v                                         
+----------------------------------------------------------------------------------
+ {"id": 1125899906842625, "label": "TEST", "properties": {"name": "foo"}}::vertex
+ {"id": 1125899906842625, "label": "TEST", "properties": {"name": "foo"}}::vertex
+(2 rows)
+
+SELECT * FROM cypher('issue_1691', $$ MATCH (u) RETURN (u) $$) AS (u agtype);
+                                        u                                         
+----------------------------------------------------------------------------------
+ {"id": 1125899906842625, "label": "TEST", "properties": {"name": "foo"}}::vertex
+(1 row)
+
+-- should just return 5 foo records that are all the same one
+SELECT * FROM cypher('issue_1691', $$ UNWIND ["foo", "foo", "bar", "foo", "bar"] AS each
+                                      MERGE (v:TEST {name: "foo"})
+                                      RETURN v $$) AS (v agtype);
+                                        v                                         
+----------------------------------------------------------------------------------
+ {"id": 1125899906842625, "label": "TEST", "properties": {"name": "foo"}}::vertex
+ {"id": 1125899906842625, "label": "TEST", "properties": {"name": "foo"}}::vertex
+ {"id": 1125899906842625, "label": "TEST", "properties": {"name": "foo"}}::vertex
+ {"id": 1125899906842625, "label": "TEST", "properties": {"name": "foo"}}::vertex
+ {"id": 1125899906842625, "label": "TEST", "properties": {"name": "foo"}}::vertex
+(5 rows)
+
+SELECT * FROM cypher('issue_1691', $$ MATCH (u) RETURN (u) $$) AS (u agtype);
+                                        u                                         
+----------------------------------------------------------------------------------
+ {"id": 1125899906842625, "label": "TEST", "properties": {"name": "foo"}}::vertex
+(1 row)
+
+-- should just return 5 bar records that are all the same one
+SELECT * FROM cypher('issue_1691', $$ UNWIND ["foo", "foo", "bar", "foo", "bar"] AS each
+                                      MERGE (v:TEST {name: "bar"})
+                                      RETURN v $$) AS (v agtype);
+                                        v                                         
+----------------------------------------------------------------------------------
+ {"id": 1125899906842626, "label": "TEST", "properties": {"name": "bar"}}::vertex
+ {"id": 1125899906842626, "label": "TEST", "properties": {"name": "bar"}}::vertex
+ {"id": 1125899906842626, "label": "TEST", "properties": {"name": "bar"}}::vertex
+ {"id": 1125899906842626, "label": "TEST", "properties": {"name": "bar"}}::vertex
+ {"id": 1125899906842626, "label": "TEST", "properties": {"name": "bar"}}::vertex
+(5 rows)
+
+SELECT * FROM cypher('issue_1691', $$ MATCH (u) RETURN (u) $$) AS (u agtype);
+                                        u                                         
+----------------------------------------------------------------------------------
+ {"id": 1125899906842625, "label": "TEST", "properties": {"name": "foo"}}::vertex
+ {"id": 1125899906842626, "label": "TEST", "properties": {"name": "bar"}}::vertex
+(2 rows)
+
+SELECT * FROM cypher('issue_1691', $$MATCH (u) DELETE u $$) AS (a agtype);
+ a 
+---
+(0 rows)
+
+-- should create 2 rows foo->bar and bar->bar and the other 3 are just returning dups
+SELECT * FROM cypher('issue_1691', $$ UNWIND ["foo", "bar", "foo", "foo", "bar"] as n
+                                      MERGE (u {name: n})-[e1:knows]->(v {name: "bar"})-[e2:knows]->(w)
+                                      RETURN u, e1, v, e2, w  $$) AS (u agtype, e1 agtype, v agtype, e2 agtype, w agtype);
+                                      u                                      |                                                            e1                                                             |                                      v                                      |                                                            e2                                                             |                               w                                
+-----------------------------------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------+-----------------------------------------------------------------------------+---------------------------------------------------------------------------------------------------------------------------+----------------------------------------------------------------
+ {"id": 281474976710661, "label": "", "properties": {"name": "foo"}}::vertex | {"id": 844424930131972, "label": "knows", "end_id": 281474976710662, "start_id": 281474976710661, "properties": {}}::edge | {"id": 281474976710662, "label": "", "properties": {"name": "bar"}}::vertex | {"id": 844424930131971, "label": "knows", "end_id": 281474976710663, "start_id": 281474976710662, "properties": {}}::edge | {"id": 281474976710663, "label": "", "properties": {}}::vertex
+ {"id": 281474976710664, "label": "", "properties": {"name": "bar"}}::vertex | {"id": 844424930131974, "label": "knows", "end_id": 281474976710665, "start_id": 281474976710664, "properties": {}}::edge | {"id": 281474976710665, "label": "", "properties": {"name": "bar"}}::vertex | {"id": 844424930131973, "label": "knows", "end_id": 281474976710666, "start_id": 281474976710665, "properties": {}}::edge | {"id": 281474976710666, "label": "", "properties": {}}::vertex
+ {"id": 281474976710661, "label": "", "properties": {"name": "foo"}}::vertex | {"id": 844424930131972, "label": "knows", "end_id": 281474976710662, "start_id": 281474976710661, "properties": {}}::edge | {"id": 281474976710662, "label": "", "properties": {"name": "bar"}}::vertex | {"id": 844424930131971, "label": "knows", "end_id": 281474976710663, "start_id": 281474976710662, "properties": {}}::edge | {"id": 281474976710663, "label": "", "properties": {}}::vertex
+ {"id": 281474976710661, "label": "", "properties": {"name": "foo"}}::vertex | {"id": 844424930131972, "label": "knows", "end_id": 281474976710662, "start_id": 281474976710661, "properties": {}}::edge | {"id": 281474976710662, "label": "", "properties": {"name": "bar"}}::vertex | {"id": 844424930131971, "label": "knows", "end_id": 281474976710663, "start_id": 281474976710662, "properties": {}}::edge | {"id": 281474976710663, "label": "", "properties": {}}::vertex
+ {"id": 281474976710664, "label": "", "properties": {"name": "bar"}}::vertex | {"id": 844424930131974, "label": "knows", "end_id": 281474976710665, "start_id": 281474976710664, "properties": {}}::edge | {"id": 281474976710665, "label": "", "properties": {"name": "bar"}}::vertex | {"id": 844424930131973, "label": "knows", "end_id": 281474976710666, "start_id": 281474976710665, "properties": {}}::edge | {"id": 281474976710666, "label": "", "properties": {}}::vertex
+(5 rows)
+
+-- clean up
+SELECT * FROM cypher('issue_1691', $$MATCH ()-[e]->() DELETE e $$) AS (a agtype);
+ a 
+---
+(0 rows)
+
+SELECT * FROM cypher('issue_1691', $$MATCH (u) DELETE u $$) AS (a agtype);
+ a 
+---
+(0 rows)
+
+--
+-- clean up graphs
+--
 SELECT * FROM cypher('cypher_merge', $$MATCH (n) DETACH DELETE n $$) AS (a agtype);
  a 
 ---
@@ -1398,9 +1542,9 @@ SELECT * FROM cypher('issue_1630', $$MATCH (n) DETACH DELETE n $$) AS (a agtype)
 ---
 (0 rows)
 
-/*
- * Clean up graph
- */
+--
+-- delete graphs
+--
 SELECT drop_graph('cypher_merge', true);
 NOTICE:  drop cascades to 19 other objects
 DETAIL:  drop cascades to table cypher_merge._ag_label_vertex
@@ -1439,3 +1583,18 @@ NOTICE:  graph "issue_1630" has been dropped
 
 (1 row)
 
+SELECT drop_graph('issue_1691', true);
+NOTICE:  drop cascades to 4 other objects
+DETAIL:  drop cascades to table issue_1691._ag_label_vertex
+drop cascades to table issue_1691._ag_label_edge
+drop cascades to table issue_1691.knows
+drop cascades to table issue_1691."TEST"
+NOTICE:  graph "issue_1691" has been dropped
+ drop_graph 
+------------
+
+(1 row)
+
+--
+-- End
+--
diff --git a/regress/sql/cypher_merge.sql b/regress/sql/cypher_merge.sql
@@ -606,8 +606,8 @@ SELECT * FROM cypher('cypher_merge', $$
   CREATE (n), (m) WITH n AS r MERGE (m)
 $$) as (a agtype);
 
----
---- Issue 1630 MERGE using array not working in some cases
+--
+-- Issue 1630 - MERGE using array not working in some cases
 --
 SELECT * FROM create_graph('issue_1630');
 SELECT * FROM cypher('issue_1630', $$ MATCH (u) RETURN (u) $$) AS (u agtype);
@@ -670,13 +670,71 @@ SELECT * FROM cypher('issue_1630',
                         MERGE (v:PERSION {first: cols.first, last: cols.last})
                         RETURN v, cols $$) AS (v agtype, cols agtype);
 
+--
+-- Issue 1691 - MERGE incorrectly creates multiple vertices
+--
+SELECT * FROM create_graph('issue_1691');
+SELECT * FROM cypher('issue_1691', $$ MATCH (u) RETURN (u) $$) AS (u agtype);
+
+-- should only create 2 distinct rows but return 4, the extra 2 being duplicates
+SELECT * FROM cypher('issue_1691', $$ UNWIND ["foo", "bar", "foo", "foo"] as n
+                                      MERGE (u {name: n})-[e:knows]->(v)
+                                      RETURN u, e, v  $$) AS (u agtype, e agtype, v agtype);
+-- should only return the same above 4 rows
+SELECT * FROM cypher('issue_1691', $$ UNWIND ["foo", "bar", "foo", "foo"] as n
+                                      MERGE (u {name: n})-[e:knows]->(v)
+                                      RETURN u, e, v  $$) AS (u agtype, e agtype, v agtype);
+-- should only return 2 distinct rows from above
+SELECT * FROM cypher('issue_1691', $$ MATCH (u)-[e]->(v)
+                                      RETURN u, e, v  $$) AS (u agtype, e agtype, v agtype);
+
+SELECT * FROM cypher('issue_1691', $$MATCH ()-[e]->() DELETE e $$) AS (a agtype);
+SELECT * FROM cypher('issue_1691', $$MATCH (u) DELETE u $$) AS (a agtype);
+
+-- should only create 1 record but return 2, one a dup of the other
+SELECT * FROM cypher('issue_1691', $$ UNWIND ["foo", "foo"] AS each
+                                      MERGE (v:TEST {name: each})
+                                      RETURN v $$) AS (v agtype);
+
+SELECT * FROM cypher('issue_1691', $$ MATCH (u) RETURN (u) $$) AS (u agtype);
+
+-- should just return 5 foo records that are all the same one
+SELECT * FROM cypher('issue_1691', $$ UNWIND ["foo", "foo", "bar", "foo", "bar"] AS each
+                                      MERGE (v:TEST {name: "foo"})
+                                      RETURN v $$) AS (v agtype);
+
+SELECT * FROM cypher('issue_1691', $$ MATCH (u) RETURN (u) $$) AS (u agtype);
+
+-- should just return 5 bar records that are all the same one
+SELECT * FROM cypher('issue_1691', $$ UNWIND ["foo", "foo", "bar", "foo", "bar"] AS each
+                                      MERGE (v:TEST {name: "bar"})
+                                      RETURN v $$) AS (v agtype);
+
+SELECT * FROM cypher('issue_1691', $$ MATCH (u) RETURN (u) $$) AS (u agtype);
+SELECT * FROM cypher('issue_1691', $$MATCH (u) DELETE u $$) AS (a agtype);
+
+-- should create 2 rows foo->bar and bar->bar and the other 3 are just returning dups
+SELECT * FROM cypher('issue_1691', $$ UNWIND ["foo", "bar", "foo", "foo", "bar"] as n
+                                      MERGE (u {name: n})-[e1:knows]->(v {name: "bar"})-[e2:knows]->(w)
+                                      RETURN u, e1, v, e2, w  $$) AS (u agtype, e1 agtype, v agtype, e2 agtype, w agtype);
+
+-- clean up
+SELECT * FROM cypher('issue_1691', $$MATCH ()-[e]->() DELETE e $$) AS (a agtype);
+SELECT * FROM cypher('issue_1691', $$MATCH (u) DELETE u $$) AS (a agtype);
 
---clean up
+--
+-- clean up graphs
+--
 SELECT * FROM cypher('cypher_merge', $$MATCH (n) DETACH DELETE n $$) AS (a agtype);
 SELECT * FROM cypher('issue_1630', $$MATCH (n) DETACH DELETE n $$) AS (a agtype);
 
-/*
- * Clean up graph
- */
+--
+-- delete graphs
+--
 SELECT drop_graph('cypher_merge', true);
 SELECT drop_graph('issue_1630', true);
+SELECT drop_graph('issue_1691', true);
+
+--
+-- End
+--