redis-limpyd-extensions 2.1

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

redis-limpyd-extensions

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

Where to find it:

• Github repository: https://github.com/limpyd/redis-limpyd-extensions

• Pypi package: https://pypi.python.org/pypi/redis-limpyd-extensions

• Documentation: http://documentup.com/limpyd/redis-limpyd-extensions

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

• Dynamic fields

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

• sadd, to set the reverse relation as the fk of the arguments:

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)

• srem works the same way as sadd but for deleting fk:

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

• sadd

• srem

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

M2MSetField

• sadd

The standard:

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

is the same as the new:

somebody.membership.sadd(group2, group3)

• srem works the same way as sadd but for removing relations:

The standard:

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

is the same as the new:

somebody.membership.srem(group2, group3)

M2MListField

• lpush and rpush, that works for M2MListField like sadd for M2MSetField

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)

• lrem works the same way as rpush and lpush but for removing relations:

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

• zadd that works for M2MSortedSetField like sadd for M2MSetField, but managing scores. Arguments can be set the same way as the normal zadd command.

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})

• zrem works the same way as zadd, without the score, but for removing relations:

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:

• use the get_field method of the model:

foo_bar = myinstance.get_field('foo_bar')

• use the get_for method of the field:

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:

You can retrieve them all:

Or only a part:

Filtering

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

• use the classic way, if you now the dynamic part in advance:

MyModel.collection(foo_bar='three')

• use the new dynamic_filter method:

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:

• if you pass a model instance in the get_for method, it will be translated to it’s pk

• the first argument of a “related collection” is the dynamic part (can also be an instance)

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:

• model

  • mixins

    • ModelWithDynamicFieldMixin(object) - A mixin tu use for your model with dynamic fields

• collection

  • mixins

    • CollectionManagerForModelWithDynamicFieldMixin(object) - A mixin to use if you want to add the dynamic_filter method to your own collection manager

  • full classes

    • CollectionManagerForModelWithDynamicField(CollectionManagerForModelWithDynamicFieldMixin, ExtendedCollectionManager) - A simple class inheriting from our mixin and the manager from limpyd.contrib.collection

• field

  • mixins

    • DynamicFieldMixin(object) - A mixin within all the stuff for dynamic fields is done, to use to add dynamic field support to your own fields

  • full classes All fields simply inherits from our mixin and the wanted base field, without anymore addition:

    • DynamicStringField(DynamicFieldMixin, StringField)

    • DynamicInstanceHashField(DynamicFieldMixin, InstanceHashField)

    • DynamicListField(DynamicFieldMixin, ListField)

    • DynamicSetField(DynamicFieldMixin, SetField)

    • DynamicSortedSetField(DynamicFieldMixin, SortedSetField)

    • DynamicHashField(DynamicFieldMixin, HashField)

• related

  • mixins

    • DynamicRelatedFieldMixin(DynamicFieldMixin) - A mixin within all the stuff for dynamic related fields is done, to use to add dynamic field support to your own related fields

  • full classes

    • DynamicFKStringField(DynamicRelatedFieldMixin, FKStringField)

    • DynamicFKInstanceHashField(DynamicRelatedFieldMixin, FKInstanceHashField)

    • DynamicM2MSetField(DynamicRelatedFieldMixin, M2MSetField)

    • DynamicM2MListField(DynamicRelatedFieldMixin, M2MListField)

    • DynamicM2MSortedSetField(DynamicRelatedFieldMixin, M2MSortedSetField)