Fix WITH clause variable passing in aggregation queries

[DecisionNerd]

Problem

Queries using WITH clause for aggregation fail with KeyError when trying to access variables in subsequent clauses. This prevents common aggregation patterns from working.

Example Query

MATCH (p:Person)-[:ACTED_IN]->(m:Movie)
WITH p, count(m) as movie_count
RETURN p.name as actor, movie_count
ORDER BY movie_count DESC

Error

KeyError: 'p'

The variable p from the WITH clause is not accessible in the RETURN clause.

Impact

• Cannot perform common aggregation patterns
• Prevents queries like "find top N by count"
• Blocks basic graph analytics workflows
• Discovered during dataset integration testing

Expected Behavior

Variables listed in WITH clause should be passed through to subsequent clauses. In the example above:

• p should be available as a node reference
• movie_count should be available as the aggregation result

Test Case

from graphforge import GraphForge

gf = GraphForge()
gf.execute("CREATE (p1:Person {name: 'Alice'})")
gf.execute("CREATE (p2:Person {name: 'Bob'})")
gf.execute("CREATE (m1:Movie {title: 'Movie1'})")
gf.execute("CREATE (m2:Movie {title: 'Movie2'})")
gf.execute("MATCH (p:Person {name: 'Alice'}), (m) CREATE (p)-[:ACTED_IN]->(m)")
gf.execute("MATCH (p:Person {name: 'Bob'}), (m:Movie {title: 'Movie1'}) CREATE (p)-[:ACTED_IN]->(m)")

# This should work but fails
results = gf.execute("""
    MATCH (p:Person)-[:ACTED_IN]->(m:Movie)
    WITH p, count(m) as movie_count
    RETURN p.name as actor, movie_count
    ORDER BY movie_count DESC
""")
# KeyError: 'p'

Root Cause

The executor's handling of WITH clause doesn't properly preserve variable bindings when transitioning between aggregation and projection phases.

Priority

HIGH - This is a core query feature that should work. Blocks common use cases.

Related

• Discovered during dataset integration testing (v0.2.1)
• See tests/integration/datasets/test_working_datasets.py
• Affects real-world queries on dataset graphs

Labels

• bug
• executor
• high-priority