Re: [Configobj-develop] Making ConfigObj pickle-able

[Fuzzyman <fuz...@vo...>]

Hello Jack,

Sorry for the delay in getting to this. Work on Movable Python (etc) has
taken longer than expected.

Thanks for the code. I'll try and get it integrated into an official
release as soon as possible.

All the best,


Michael Foord
http://www.voidspace.org.uk/python/index.shtml


Jack Kuan wrote:

> Hi all,
>
> sorry, I couldn't wait for ConfigObj to be pickle-able (for pickle
> protocol 2). So, after some investigation here are some changes
> that I've found necessary:
>
> def __newobj__(cls, *args):
>     """Required by Section.__reduce__ for using Pickle protocol 2 with
> newstyle class. """
>     return cls.__new__(cls, *args)
>
>
> class Section(dict):
>
>     def __reduce__(self):
>         d = {}
>         d.update(self.__dict__)
>         d['parent'] = d['main'] = None      # remove recursive references
>         d['sections'] = d['scalars'] = []   # must be empty since we
> are going to Section.__setitem__
>         return __newobj__, (self.__class__, d), self.__dict__, None,
> self.iteritems()
>        
>     def __new__(cls, *args, **kws):
>         """Partially initialize the object so that it's safe to call
> Section.__setitem__ before
>         the full states are restored during unpickling.
>         """
>         inst = super(Section, cls).__new__(cls)
>
>         # Check the args to make sure it's called by __newobj__ before
> updating
>         # there's probably a better way to verify this....
>         if len(args) == 1 and isinstance(args[0], dict):
>             d = args[0]
>             try:
>                 if d['parent'] == d['main'] == None:
>                     inst.__dict__.update(d)
>             except:
>                 pass
>         return inst
>
> then we also need to change just one line in Section.__setitem__:
> the line where it says "if not self.main.stringify:", change it to "if
> self.main and not self.main.stringify:" since if we are unpickling
> self.main will be None, plus there's no need to check the value during
> unpickling anyway.
>
> I did some tests on a my configuration, it seems to work. But anyway,
> the basic idea I think is that we need to have some states
> restored so that during unpickling Section.__setitem__ can rebuild the
> section tree structure, and then we complete the unpickling by restoring
> the the original states.
>
> Thanks,
> Jack
>
>------------------------------------------------------------------------
>
>-------------------------------------------------------------------------
>Using Tomcat but need to do more? Need to support web services, security?
>Get stuff done quickly with pre-integrated technology to make your job easier
>Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo
>http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642
>
>------------------------------------------------------------------------
>
>_______________________________________________
>Configobj-develop mailing list
>Con...@li...
>https://lists.sourceforge.net/lists/listinfo/configobj-develop
>  
>