Export query result to arrow: NodeID, Node, Rel, LIST data types

[ray6080]

No description provided.

[mewim]

Overall LGTM.

Two general comments:

1. In the Python class for query result (

   kuzu/tools/python_api/src_py/query_result.py

   Line 44 in 88a989c

   def get_as_arrow(self, chunk_size):

   ), we should add import pyarrow before calling the CPP function, so that if pyarrow is not installed, the user gets a ModuleNotFoundError exception early without having to deal with the confusing exceptions in CPP code.
2. The function seems to throw error for un-processable data types. Should we consider not throwing error and do a best-effort with warning if we add a new type but the arrow conversion has not been implemented?

[ray6080]

Overall LGTM.

Two general comments:

1. In the Python class for query result (

   kuzu/tools/python_api/src_py/query_result.py

   Line 44 in 88a989c

   def get_as_arrow(self, chunk_size):

   
   ), we should add import pyarrow before calling the CPP function, so that if pyarrow is not installed, the user gets a ModuleNotFoundError exception early without having to deal with the confusing exceptions in CPP code.
2. The function seems to throw error for un-processable data types. Should we consider not throwing error and do a best-effort with warning if we add a new type but the arrow conversion has not been implemented?

Thanks! Added import check for get_as_df and get_as_arrow. For exception, we've covered data types we have right now. for new data types, I think it's better to throw exception than trying to convert unknown type by ourself.