Database conversion help - db2pickle.py and pickle2db.py

[Skip Montanaro]

I just checked in two scripts to the Python Tools/scripts CVS directory,
db2pickle.py and pickle2db.py.  They should help people upgrade from older
versions of various database modules when Python is updated.  Here's a
simple example of how they are used:

    % python2.2 db2pickle.py some.db some.pickle
    % python2.3 pickle2db.py some.db.new some.pickle

The first walks the bsddb hash database in some.db and writes out pickled
key/value tuples to some.pickle.  The second reads pickled key/value tuples
from some.pickle and writes them out to the bsddb hash database some.db.new.
Something like this will be the most common usage I suspect.  A validator
script isn't really possible, since in the general case, some.db and
some.db.new won't both be readable in the same Python interpreter.

The following five database formats are supported:

    bsddb hash
    bsddb btree
    bsddb recno
    anydbm
    dbm

The default is bsddb hash.  I didn't bother with the dbhash wrapper because
dbhash.open is just bsddb.hashopen.  Dumbdbm didn't seem worth it, either.
In any case, I don't believe its format has changed since it was first
written.

I've tested the scripts to a reasonable degree using the bsddb hash and
bsddb btree formats with Python 2.1.3, 2.2.2 and current CVS (2.3a1+).  I've
also attempted to avoid any gratuitous incompatibilities with earlier
versions, however I would appreciate it if people with 1.5.2 and 2.0 and/or
who use different database formats would test them a bit.  If you don't have
a CVS version of the Python distribution handy, you can grab the scripts
individually from SourceForge:

    http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/python/python/dist/src/Tools/scripts/

Feedback is welcome.

Skip