Mostly Invisible Database

A database access layer that acts like python in-memory objects.

Disclaimer: Only minimal attempts have been made at speed. And mainly intended for personal projects. Deeply nested data structures will likely be slow.

Example:

>>> import midb
>>> root = midb.get_root('db_file.db')
>>> root['test'] = 1
>>> root['test2'] = {'test3': 3}
>>> exit()

later ...

>>> import midb
>>> root = midb.get_root('db_file.db')
>>> root['test']
1
>>> root['test2']['test3']
3
>>> root['test2']
PDict({'test3': 3}, _backend=SQLiteBackend(filename="db_file.db"), _id=1, _temp=None)
>>> root['test2'].in_memory()
{'test3': 3}
>>> exit()

What's currently supported?

Currently, the types that are supported are:

• Simple types:

  • str
  • None
  • bool
  • int
  • float
  • complex
  • datetime.datetime
  • datetime.date
  • datetime.time

• container types (must not contain unsupported types)

  • dict (silently converted to PDict)
  • tuple (silently convert to PTuple)
  • list (silently convert to PList)
  • custom objects (see below for example)

Custom objects

Custom objects are still not fully tested, but the basics have been including property descriptors. Also, custom objects that are hashable may be used as keys in dictionaries. The following has been tested and works.

import midb

# define or import an class
#   Note: subclassing midb.MemoryObject is not necessary but useful 
#         because it defines __repr__ and __eq__.
class MyObj(midb.MemoryObject):
    def __init__(self, a_value, b_value):
        self.a = a_value
        self.b = b_value

    def some_other_method(self, input):
        self.methods_may_set = input
        return f"{self.a}, {self.b}"

# define a persistent object with an _in_memory_class attribute set to the object you 
# want to have it emulate and decorate it with the @midb.setup_pobject decorator.
@midb.setup_pobject
class PMyObj(midb.PObject):
    _in_memory_class = MyObj

With those objects imported or defined:

# first run
root = midb.get_root("filename.db")

root['my object'] = MyObj(1, 2)
print(root['my object'].a)
print(root['my object'].b)
print(root['my object'].some_other_method(3))
print(root['my object'].methods_may_set)

Later (also with the above classes defined or imported) ...

root = midb.get_root("filename.db")

print(root['my object'].a)
print(root['my object'].b)
print(root['my object'].some_other_method(4))
print(root['my object'].methods_may_set)

If you encounter any problems with the classes you have defined, please file an issue at https://gitlab.com/gossrock/midb/-/issues.

Future:

• Support for and/or documentation of additional or custom serializers.
• More examples and documentation
• Add support for other standard library classes that represent storable data.
  • decimal.Decimal and fractions.Fraction
  • set and frozenset
  • collections.namedtuple()/typing.NamedTuple and dataclasses.dataclass
  • colletions.OrderdDict
  • enum.Enum and friends
  • pathlib.Path
• Add generic type annotation to collection types.

Bugs

Please report bugs at https://gitlab.com/gossrock/midb/-/issues