Fork me on GitHub

PyPI<br>
Version
Build<br>
Status

redis-limpyd-extensions

Some extensions for
redis-limpyd
(redis orm (sort of) in python)

Where to find it:

Install:

Python versions 2.7, and 3.5 to 3.8 are supported (CPython and PyPy).

Redis-server versions >= 3 are supported.

Redis-py versions >= 3 are supported.

Redis-limpyd versions >= 2 are supported.

You can still use limpyd-extensions versions < 2 if you need something older than the above requirements.

pip install redis-limpyd-extensions

List of available extensions:

Add/remove related on both sides

Say we have the following related models:

class Person(RelatedModel):
    database = main_database
    name = HashableField(indexable=True)

class Group(Relatedmodel):
    database = main_database
    name = HashableField(indexable=True)
    members = M2MSetField(Person, related_name='membership')

And some data:

somebody = Person(name='foobar')
group_1 = Group(name='group 1')
group_2 = Group(name='group 2')
group_3 = Group(name='group 3')

We can add membership the normal way:

group_1.members.sadd(somebody)

And retrieving then this way:

group_1_members = group_1.members()  # somebody !
somebody_membership = somebody.membership()  # group_1

But say that we want to put a person in many groups at ones, we can do:

group_2.members.sadd(somebody)
group_3.members.sadd(somebody)

limpyd_extensions provide a way to add/remove relations via the other
side of the relation:

somebody.membership.sadd(group2, group3)

To use this, simple import the related fields from
limpyd_extensions.related instead of limpyd.contrib.related:

from limpyd_extensions.related import (FKStringField, FKHashableField, 
                                       M2MSetField, M2MListField, 
                                       M2MSortedSetField)

And use them as usual. (Note that for convenience you can also import
the standard RelatedModel from there)

The added methods for the reverse side of each related field are:

FKStringField

Having:

class Group(RelatedModel):
    parent = FKStringField(self, related_name='children')

The standard:

child_group.parent.set(main_group)
other_child_group.parent.set(main_group)

is the same as the new:

main_group.children.sadd(child_group, other_child_group)

The standard:

child_group.parent.delete(main_group)
other_child_group.parent.delete(main_group)

is the same as the new:

main_group.children.srem(child_group, other_child_group)

FKHashableField

Both work the exact same way as for FKStringField, the only difference
is that sadd emulates a hset, not a set.

M2MSetField

The standard:

group_2.members.sadd(somebody)
group_3.members.sadd(somebody)

is the same as the new:

somebody.membership.sadd(group2, group3)

The standard:

group_2.members.srem(somebody)
group_3.members.srem(somebody)

is the same as the new:

somebody.membership.srem(group2, group3)

M2MListField

If in our Person/Group example members is a M2MListField instead of
a M2MSetField,

The standard:

group_2.members.rpush(somebody)
group_3.members.rpush(somebody)

is the same as the new:

somebody.membership.rpush(group2, group3)

The standard:

group_2.members.lrem(0, somebody)  # 0 for "all occurences"
group_3.members.lrem(0, somebody)

is the same as the new:

somebody.membership.lrem(group2, group3)  # the count is forced to 0

M2MSortedSetField

If in our Person/Group example members is a M2MSortedSetField
instead of a M2MSetField, using the score to save the date of
membership

The standard:

group_2.members.zadd({somebody: sometime})  # sometime, a float, can be a call to time.time()
group_3.members.zadd({somebody: another_time})

is the same as the new:

somebody.membership.zadd({group2: sometime, group3: another_time})

The standard:

group_2.members.zrem(somebody)
group_3.members.zrem(somebody)

is the same as the new:

somebody.membership.zrem(group2, group3)

Dynamic fields

Dynamic fields provide a way to add unlimited fields to a model by
defining a (or many) dynamic field, and use it with a dynamic part. ie a
dynamic field name "foo" can be used with as many dynamic parts as you
want to create dynamic variations: "foo_bar" for the dynamic part
"bar", "foo_baz" for the dynamic part "baz", and so on.

A simple API to use them, and filter on them, is provided.

To use a dynamic field, your model must inherit from the following
mixin: ModelWithDynamicFieldMixin, found in
limpyd_extensions.dynamic.model. It's a mixin, you should use it with
another RedisModel class. Fields are available as field classes
(DynamicStringField, DynamicInstanceHashField, DynamicListField,
DynamicSetField, DynamicSortedSetField, DynamicHashField) or as a
mixin (DynamicFieldMixin) if you want to adapt an external field. You
can find them in limpyd_extensions.dynamic.fields

A short example on how to define a dynamic field on a model:

from limpyd.model import RedisModel

from limpyd_extension.dynamic.model import ModelWithDynamicFieldMixin
from limpyd_extension.dynamic.fields import DynamicSetField


class MyModel(ModelWithDynamicFieldMixin, RedisModel):
    foo = DynamicSetField(indexable=True)

As the foo field is dynamic, you cannot run any command on it, but
only on its dynamic variations. How to do it ?

There is two ways:

foo_bar = myinstance.get_field('foo_bar')
foo_bar = myinstance.foo.get_for('bar')

The latter is useful if you have a variable instead of known value:

somebar = 'bar'
foo_bar = myinstance.foo.get_for(somevar)

Note that you can use this shortcut instead of using get_for:

foo_bar = myinstance.foo(somevar)

Knowing this, you can do operations on these fields:

myinstance.foo(somevar).sadd('one', 'two', 'three')
myinstance.foo(othervar).sadd('four', 'five')
myotherinstance.foo(somevar).sadd('three', 'thirty')
print myinstance.foo(somevar).smembers()
print myinstance.foo(othervar).smembers()
print myotherinstance.foo(somevar).smembers()

To know the existing versions in a dynamic_field, you can use
scan_fields.

It takes the same argument as the sscan command of SetField (from
limpyd), because it is applied on the inventory key where all versions
are saved.

So if you have some versions:

myinstance.foo('foo').set('111')
myinstance.foo('bar').set('222')
myinstance.foo('baz').set('333')

You can retrieve them all:

set(myinstance.foo.scan_versions())  # returns {'foo', 'bar', 'baz'}

Or only a part:

set(myinstance.foo.scan_versions('b*'))  # returns {'bar', 'baz'}

Filtering

To filter on indexable dynamic fields, there is two ways too:

MyModel.collection(foo_bar='three')
MyModel.collection().dynamic_filter('foo', 'bar', 'three')

Parameters are: the field name, the dynamic part, the value for the
filter and, not show in the previous example, the index suffix to use.

This suffix is default to ''.

But if what you want to do is

MyModel.collection(foo_bar__eq='three')

You can use dynamic_filter this way:

MyModel.collection().dynamic_filter('foo', 'bar', 'three', 'eq')  # you can use '__eq' too

The collection manager used with ModelWithDynamicFieldMixin depends on
ExtendedCollectionManager, so you can chain filters and dynamic
filters on the resulting collection.

Dynamic related fields

Dynamic fields also work with related fields, exactly the same way.
There is only two additions:

An exemple using dynamic related fields:

from limpyd.fields import PKField
from limpyd_extensions.dynamic.model import ModelWithDynamicFieldMixin
from limpyd_extensions.dynamic.related import DynamicM2MSetField

class Tag(MyBaseModel):
    slug = PKField()

class Person(MyBaseModel):
    name = PKField()

class Movie(ModelWithDynamicFieldMixin, MyBaseModel):
    name = PKField()
    tags = DynamicM2MSetField(Tag, related_name='movies')

somebody = Person(name='Somebody')
matrix = Movie(name='Matrix')
cool = Tag(name='cool')

matrix.tags.get_for(somebody).sadd(cool)
# same as: matrix.tags(somebody).sadd(cool)

cool_movies_for_somebody = cool.movies(somebody)  # the related collection
# ['Matrix']

Provided classes

Here is the list of modules and classes provided with the
limpyd_extensions.dynamic module: