Incorrect query result for MATCH/OPTIONAL MATCH combination

[meido]

CREATE NODE TABLE V (id STRING, PRIMARY KEY(id)); CREATE NODE TABLE R (id STRING, PRIMARY KEY(id)); CREATE REL TABLE childof (FROM V TO V); CREATE REL TABLE relatedto (FROM R TO V);

CREATE (v1:V {id:"v1"}),(v11:V {id:"v11"}),(v12:V {id:"v12"}),(v111:V {id:"v111"}), (r1:R {id:"r1"}), (r12:R {id:"r12"}), (v111)-[:childof]->(v11),(v11)-[:childof]->(v1),(v12)-[:childof]->(v1), (r1)-[:relatedto]->(v1), (r12)-[:relatedto]->(v12); MATCH (v1:V {id: "v1"}) WITH v1 MATCH (v:V)-[e:childof*]->(v1) OPTIONAL MATCH (r)-[:relatedto]->(v) RETURN r, v; screenshot

What Kuzu returns:

---

| r | v |

| | {_ID: 0:3, _LABEL: V, id: v111} |

| | {_ID: 0:2, _LABEL: V, id: v12} |

| | {_ID: 0:1, _LABEL: V, id: v11} |

(3 tuples) (2 columns) What Neo4j returns (and what is expected here):

╒════════════════╤═════════════════╕ │r │v │ ╞════════════════╪═════════════════╡ │null │(:V {id: "v11"}) │ ├────────────────┼─────────────────┤ │null │(:V {id: "v111"})│ ├────────────────┼─────────────────┤ │(:R {id: "r12"})│(:V {id: "v12"}) │ └────────────────┴─────────────────┘ Why does Kuzu omit the existing optional match (:R {id: "r12"}) for v12 here?

[mayil-ai[bot]]

Summary: Incorrect query result for MATCH/OPTIONAL MATCH combination.

Possible Solution

Based on the provided information, the issue seems to be related to the handling of OPTIONAL MATCH in the query execution engine of Kùzu. The expected behavior is that the query should return all V nodes in the hierarchy of v1 and any related R nodes, with nulls for R when there is no related node. However, Kùzu's result is missing the (:R {id: "r12"}) related to v12.

To address this issue, the following steps should be taken:

• Review the query execution engine code, specifically the part that handles the OPTIONAL MATCH clause.
• Ensure that when an OPTIONAL MATCH does not find a match, it correctly returns null for the optional pattern instead of omitting the row.
• Verify that the engine correctly processes variable length paths, as indicated by childof* in the query.
• Check if the join operation between the MATCH and OPTIONAL MATCH clauses is correctly implemented, as it should not filter out rows when the OPTIONAL MATCH does not find a related node.
• Add or modify test cases to cover scenarios where OPTIONAL MATCH is used in combination with variable length relationships and ensure that the expected behavior matches that of Neo4j.
• Once the issue is identified and fixed, run the entire test suite to ensure that the fix does not introduce regressions in other parts of the query engine.

The provided Java and C++ test code snippets are primarily related to value retrieval and property access, and do not directly address the OPTIONAL MATCH issue. Focus on the parts of the codebase that deal with query parsing, execution, and result construction.

Code snippets to check

• tools/java_api/src/test/java/com/kuzudb/test/ValueTest.java

  The Java API tests for relationship values might be relevant to the issue as they deal with relationship retrieval and properties, which could be related to the MATCH/OPTIONAL MATCH combination issue.

• test/c_api/value_test.cpp

  The C API tests for relationship values could be relevant since they involve relationship retrieval and properties, which might be connected to the MATCH/OPTIONAL MATCH combination issue.