Master branch:

Installation

Using PyPi (recommended):

sudo pip install redisca2

or

wget https://pypi.python.org/packages/source/r/redisca2/redisca2-X.tar.gz
tar xvf redisca2-X.tar.gz
sudo python redisca2-X/setup.py install

Model

from redisca2 import Model
from redisca2 import Email
from redisca2 import DateTime

class User (Model):
        email = Email(
                name='eml',  # Define link with 'eml' hash key.
                index=True,  # Index support.
                unique=True, # Makes sure that field is unique across db.
        )

        created = DateTime(
                name='created',      # Define link with 'created' hash key.
                new=datetime.utcnow, # Value which is used as default in User.new()
        )

        age = Integer(
                name='age',  # Define link with 'age' hash key.
                index=True,  # Enable index.
        )

user = User.new() # Create model with random id and "new" fields values.
user = User.new(model_id='your_id') # Or use custom id if needed.

user.getid() # user id
user.email = 'foo@bar.com'

user.save()   # Saving routines
user.exists() # True

user.delete() # Deletion routines
user.exists() # False

Fields

Field is the way how you should control data in your models. Just define class variables with field-specific options and take classic ORM's advantages:

• data validation;
• native python data types support;
• transparent relations between models;
• indexes support (incl. unique indexes).

Available parameters:

• name - redis hash field name to store value in.
• index - makes field searchable.
• unique - tells that value should be unique across database. Model.save() will raise an Exception if model of same class already exists with given value.
• new - field value which is used as default in Model.new(). Functions, methods and built-in's are acceptable as callback values.
• none - what is returned if field value is None

Built-in fields:

• String - extends IndexField with additional parameters minlen and maxlen.
• Email - extends IndexField field with email validation support.
• Integer - extends RangeIndexField with parameters minval and maxval. Accepts int and numeric strings. Returns int.
• Reference - extends IndexField with cls (reference class) parameter. Accepts and returns instance of cls.
• MD5Pass - extends String field. Acts like string but converts given string to md5 sum.
• DateTime - extends RangeIndexField without additional parameters. Accepts datetime and int(timestamp) values. Returns datetime.

Getting Data

Using id

Here is an example how to get model instance using id (empty model returned if it not exists yet):

user = User('user id')
print(user.email) # 'foo@bar.com'

Raise Exception if model is not exists yet:

user = User('user id', must_exist=True)

You can disable lazy-loading if needed:

user = User('user id', force_load=True)

Each initialized model is saved in registry and returned on each attempt of re-init:

user1 = User('user_id')
user2 = User('user_id')
user1 is user2 # Always is True

user.free()   # Unregister model instance.
User.free_all()  # Cleanup User's registry.
Model.free_all() # Unregister all known models.

Find by Index

users = User.email == 'foo@bar.com'

Subclasses of RangeIndexField has a limited support for ranged queries:

users = User.age >= 10

Dict API

All fields are linked to model dict keys. Use can use model dict API to read and write redis hash data AS IS:

user = User('id')
user['eml'] = 'foo@bar.com'
user['age'] = 10

Note that keys with None values will be removed from model dict:

user['eml'] = None
'eml' in user     # False
print user['eml'] # Raises KeyError

Connecting to Redis

Global database connection setup looks like this:

from redisca2 import conf
from redisca2 import RedisConnector

conf.db = RedisConnector()

Note: redisca2 uses localhost:6379(0) as default database. You can setup inheritable per-model database connection using conf class decorator:

from redisca2 import Model
from redisca2 import conf
from redisca2 import RedisConnector

@conf(db=RedisConnector())
class User (Model):
        pass

Key Format

Model key format is:

model_key_prefix:model_id

Default model_key_prefix is lowercased class name. Use conf class decorator to override it like this:

from redisca2 import Model
from redisca2 import conf

@conf(prefix='usr')
class User (Model):
        pass

print(User.getprefix()) # 'usr'

Tools

ID Generator

from redisca2 import hexid
from redisca2 import intid

print(hexid()) # 59d369790
print(hexid()) # 59d3697bc

print(intid()) # 24116751882
print(intid()) # 24116788848

Flask Support

from redisca2 import FlaskRedisca

app = Flask()

app.config['REDISCA'] = {
        # redis.StrictRedis constructor kwargs dict.
}

FlaskRedisca(app)

Optional autosave constructor parameter tells redisca2 that all known models should be saved at the end of request (if no exception raised). Unchanged and deleted instances are ignored. If you want to skip locally changed instances use free() method during request life.

Requirements

• redis-py 2.7+
• python 2.7/3.2+ or pypy 2.1+

Python 3.x support

Py3k support is still a sort of experiment but I'm looking carefuly into full compability with cutting-edge builds of CPython. There are no known issues with it actually.