SQLAlchemy committing pickle types

Question

I'm having an issue committing changes to pickle types (lists) in sqlalchemy. It will act as if nothing happened after the committal.

Here's my my function where I try to commit:

def commit_move(game_id, player, move):
    game = game_query(game_id)
    if player == 'human':
        game.human_spaces.append(move)
    if player == 'ai':
        game.ai_spaces.append(move)
    game.available_spaces.remove(move)
    print game.human_spaces
    print game.ai_spaces
    print game.available_spaces
    print "----"
    session.add(game)
    session.commit()

here's how the table is setup:

class Game(Base):
    __tablename__ = 'game'
    id = Column(Integer, primary_key=True)
    human_spaces = Column(PickleType)
    ai_spaces = Column(PickleType)
    available_spaces = Column(PickleType)

here's the code I'm using to test it:

game_id = create_game()
print game_id
print get_available_spaces(game_id)
print get_human_spaces(game_id)
print get_ai_spaces(game_id)
print "---------"
commit_move(game_id, 'human', 7)
print get_available_spaces(game_id)
print get_human_spaces(game_id)
print get_ai_spaces(game_id)

and here's what the good ol' terminal is telling me:

1
[1, 2, 3, 4, 5, 6, 7, 8, 9]
[]
[]
---------
[7]
[]
[1, 2, 3, 4, 5, 6, 8, 9]
----
[1, 2, 3, 4, 5, 6, 7, 8, 9]
[]
[]

I'm sure it's something simple I'm missing here, but any help would be greatly appreciated!

Answer 1

The problem is that the ORM is not alerted to changes inside a mutable type, like a list. SQLAlchemy therefore offers mutation tracking with the sqlalchemy.ext.mutable extension.

From the examples in the documentation , in particular with reference to the sqlalchemy.ext.mutable.MutableList class, it looks like the column declaration should go (eg):

human_spaces = Column(MutableList.as_mutable(PickleType))

I quote from the documentation on the as_mutable method: "This establishes listeners that will detect ORM mappings against the given type, adding mutation event trackers to those mappings."

Answer 2

I've written a package to help make this easy. You can choose different encodings, including pickle , and easily dump and store objects to a database. It can connect to any database that sqlalchemy understands. There is a dictionary interface to a SQL table, and you can store any type that dill can serialize:

>>> import klepto
>>> db = klepto.archives.sqltable_archive('playgame')
>>> db['human'] = [1,2,3,4]
>>> db['ai'] = [1,2]
>>> db
sqltable_archive('sqlite:///:memory:?table=playgame', {'ai': [1, 2], 'human': [1, 2, 3, 4]}, cached=True)
>>> db.dump()
>>>