Don't fail when inserting UDTs with prepared queries with some missing fields #224

[Mokto]

Let's say you have an object:

class Address(object):

    def __init__(self, street, zipcode, **kwargs):
        self.street = street
        self.zipcode = zipcode

cluster.register_user_type('mykeyspace', 'address', Address)

And let's say the type actually contains another field, let's call i raw_address

Then inserting data through a prepared statement will actually fail : the driver will complain raw_address is missing.

This change addresses that, as any field should be optional.

[absurdfarce]

Thanks for the PR @Mokto!

Have you signed the Contributor License Agreement for contributions to DataStax open source projects? If not you can find it at https://cla.datastax.com/. Thanks!

[Mokto]

I just did ;)

[Mokto]

Hi,
Any update on this ?

Thanks.

[fruch]

LGTM

[absurdfarce]

@Mokto Apologies for the delay in getting back to this; we've got a lot of balls in the air at the moment.

A few notes, perhaps mostly to myself. First off this works fine via CQL:

drop keyspace if exists mykeyspace;
create KEYSPACE mykeyspace WITH replication = {'class': 'SimpleStrategy', 'replication_factor': 1};
use mykeyspace;
CREATE TYPE address (street text, zipcode int, raw_address text);
CREATE TABLE users (id int PRIMARY KEY, location frozen<address>);
INSERT INTO users (id, location) VALUES (0, {street: '123 Main St.', zipcode: 78723});
cqlsh:mykeyspace> select * from users;

 id | location
----+-------------------------------------------------------------
  0 | {street: '123 Main St.', zipcode: 78723, raw_address: null}

(1 rows)

Second this change really only affects the prepared statement case. When an "incomplete" (for lack of a better term) instance representing a UDT type is fed to a simple statement the input type is converted to something like the UDT literal above so you get basically the same behaviour. In the prepared statement case we actually know the type that's expected so we can attempt to serialize based on cassandra.cqltypes.UserType, which lands us in the code cited above.

Finally, an explicit repro case:

from cassandra.cluster import Cluster

cluster = Cluster()
session = cluster.connect()
session.execute("drop keyspace if exists mykeyspace")
session.execute("create KEYSPACE mykeyspace WITH replication = {'class': 'SimpleStrategy', 'replication_factor': 1}")
session.set_keyspace('mykeyspace')
session.execute("CREATE TYPE address (street text, zipcode int, raw_address text)")
session.execute("CREATE TABLE users (id int PRIMARY KEY, location frozen<address>)")

# create a class to map to the "address" UDT                                                                                                                                                                         
class Address(object):

    def __init__(self, street, zipcode):
        self.street = street
        self.zipcode = zipcode

#cluster.register_user_type('mykeyspace', 'address', Address)                                                                                                                                                        
insert_statement = session.prepare("INSERT INTO users (id, location) VALUES (?, ?)")
session.execute(insert_statement, [0, Address("123 Main St.", 78723)])

Without this change (and using the repro code above) you get:

$ python foo.py 
Traceback (most recent call last):
  File "cassandra/cqltypes.py", line 1027, in cassandra.cqltypes.UserType.serialize_safe
TypeError: 'Address' object is not subscriptable

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "foo.py", line 28, in <module>
    session.execute(insert_statement, [0, Address("123 Main St.", 78723)])
  File "cassandra/cluster.py", line 2634, in cassandra.cluster.Session.execute
  File "cassandra/cluster.py", line 2677, in cassandra.cluster.Session.execute_async
  File "cassandra/cluster.py", line 2880, in cassandra.cluster.Session._create_response_future
  File "cassandra/query.py", line 506, in cassandra.query.PreparedStatement.bind
  File "cassandra/query.py", line 636, in cassandra.query.BoundStatement.bind
  File "cassandra/cqltypes.py", line 796, in cassandra.cqltypes._ParameterizedType.serialize
  File "cassandra/cqltypes.py", line 1030, in cassandra.cqltypes.UserType.serialize_safe
AttributeError: 'Address' object has no attribute 'raw_address'

With the fix you get no error and a row matching the CQL example above added.

[Mokto]

Thanks!