feat: Support Cypher function tolower() for strings #106
Description
Feature request
In the graph benchmark, other systems like Neo4j, Kuzu and Ladybug support lowercasing string names in Cypher to allow for case-insensitive matching on query patterns. This will help us write Cypher queries in lance-graph that are more in line with what users coming from other systems may expect.
Another point to note: currently, the behavior isn't ideal because it causes silent failure (see #107) - where the user is unaware that the binder ignored the tolower() function entirely. This could also be a larger issue with binding where not-implemented functions aren't throwing an error that may help inform the user.
Repro
The following query returns unexpected results.
import pyarrow as pa
from lance_graph import CypherQuery, GraphConfig
def main() -> None:
cfg = GraphConfig.builder().with_node_label("Person", "name").build()
datasets = {"Person": pa.table({"name": ["Alice", "BOB", "CaSeY"]})}
query = """
MATCH (p:Person)
RETURN p.name AS name, tolower(p.name) AS lowered
ORDER BY name
"""
print("input:")
print(query.strip())
print("output:")
try:
result = CypherQuery(query).with_config(cfg).execute(datasets)
print([f"{row['name']} -> {row['lowered']}" for row in result.to_pylist()])
except Exception as exc:
print(f"ERROR: {type(exc).__name__}: {exc}")
if __name__ == "__main__":
main()Expected:
Alice -> 'alice', BOB -> 'bob', CaSeY -> 'casey']
Actual
'Alice -> 0', 'BOB -> 0', 'CaSeY -> 0']