Redis storage backend for schematics.
Provides Redis persistence to Schematics models with cutomizable abstraction levels.
pip install redis_schematics
Understanding Persistence layers
There are several ways to implement complex objects persitence on a key-value-set database such as redis. The best way to do it depends on your application constraints. We think that providing a good storage model for your application is to allow you to choose which abstraction you want to use. Below you can find a comparison on different provided abstraction layers.
Currently we only support a SimpleRedisMixin and SimpleRedisModel, but you can use BaseRedisMixin to build your own persistance layers.
Add Redis persistance to an object using a simple approach. Each object correnspond to a single key on redis prefixed with the object namespace, which correnponds to a serialized object. To use this mixin you just need to declare a primary key as on the example below.
You may use this Mixin when you have frequent matches on primary key and set operations, unique expires, hard memory contraints or just wants a 1-1 object-key approach. You may not use this Mixin if you need performance on filter, all and get on non primary key operations.
Add Redis persistance to an object using a single hash approach. Each type correnspond to a single key on redis containing a hash set with every instance as an entry on the set which contains a serialized object.
You may use this Mixin when you have frequent matches on primary key, set and all operations, hard memory contraints or wants a single key approach. You may not use this Mixin if you need performance on filter and get on non primary key operations.
Creating models with persistence
Note: you should include a pk, but don’t bother setting it’s value manually. We can infer it from an id field or by setting a tuple of field names using __unique_together__.
from datetime import datetime, timedelta from redis import StrictRedis from redis_schematics import SimpleRedisMixin from schematics import models, types class IceCreamModel(models.Model, SimpleRedisMixin): pk = types.StringType() # Just include a pk id = types.StringType() flavour = types.StringType() amount_kg = types.IntType() best_before = types.DateTimeType()
Setting on Redis
Saving is simple as set().
vanilla = IceCreamModel(dict( id='vanilla', flavour='Sweet Vanilla', amount_kg=42, best_before=datetime.now() + timedelta(days=2), )) chocolate = IceCreamModel(dict( id='chocolate', flavour='Delicious Chocolate', amount_kg=12, best_before=datetime.now() + timedelta(days=3), )) vanilla.set() chocolate.set()
Getting from Redis
There are two basic ways to get an element from Redis: by pk or by value. You can use the classmethods match_for_pk(pk) or match_for_values(**Kwargs) or just simply match(**kwargs) to let us choose which one. Notice that the performance from both methods is a lot different, so you may avoid matching for values on high performance environments. You may also use refresh to reload an object from storage if it has been modified.
IceCreamModel.match_for_pk('vanilla') IceCreamModel.match_for_values(amount__gte=30) IceCreamModel.match(id='vanilla') # match on pk IceCreamModel.match(best_before__gte=datetime.now()) # match on values vanilla.refresh()
Fetching all and filtering
You can also use all() to deserialize all and filters. Notice that this invlolves deserializing all stored objects.
Deleting and expiring
To remove objects, you can set __expire__ or use the delete() method. Notice that expires work differently on single key and multiple keys approaches.
class MyVolatileModel(models.Model, SimpleRedisMixin): __expire__ = 3600 # model expire (in seconds) pk = types.StringType() vanilla.delete()
If you want json serialization, you have at least two options:
- Patch the default serializer.
- Write a custom JSONEncoder.
We’ve implemented a handy patch funtion, you need to add this code to somewhere at the top of everything to automagically add json serialization capabilities:
from redis_schematics.patches import patch_json patch_json()
Eventually __json__ will be added to the stdlib, see https://bugs.python.org/issue27362
We are still 0.x, but we are very close to a stable API. Check our roadmap for a glance of what’s missing.
|Filename, size||File type||Python version||Upload date||Hashes|
|Filename, size redis_schematics-0.2.2.tar.gz (9.2 kB)||File type Source||Python version None||Upload date||Hashes View hashes|