Keywords
cache-storage, data-science, feature-store-online, python
Licenses
GPL-3.0/GPL-3.0+
Install
pip install data-snack==1.0.5

Documentation

Data Snack

About

Data Snack is a minimalistic framework for storing and accessing structured data.

It uses an Entity objects to define a schema for your data. Snack provides an interface for automatically serializing and storing entities in a cache database of you choice. General interface that allows you to use different backends: redis, memcached.

Features

  • Entity objects are stored in a compress form to reduce memory usage.
  • Snack is using Entity fields to define a unique key to represent an object stored in the db.
  • Snack is supporting batch saving and reading data to achieve high performance.

Core concepts

  • Entity - a class defines a schema of single object stored in db
  • key fields - a list of fields (defined as a list of str values) that will be used to create a key for a given Entity object.
  • key values - a list of values for key fields from given Entity
  • key - a str value created for a given Entity
    • created in a format: <Entity type name>-<key value 1>_<key value 2>...<key value N>

Install

Data Snack can be easily installed using pypi repository.

pip install data_snack

Usage

This examples shows a basic usage of defining an entity and using Snack to save and load it from the cache. More examples can be found in the Examples section.

Example 1 - Creating new entities and saving

1. Define entities

The first thing you need to do is to define an Entity. Entities are used to define a common structure of the objects stores in your database.

We are recommending adding data validation to your entities. The easiest way is using pydantic for type validation of all entity fields.

from pydantic.dataclasses import dataclass
from typing import Text
from data_snack.entities import Entity

@dataclass
class Person(Entity):
    index: Text
    name: Text

2. Connect to Redis

Connect to you a cache database of your choice. In this example we are using Redis, but you could also use Memcached if you want.

import redis
redis_connection = redis.Redis(host='127.0.0.1', port=6379, password='')

3. Create Snack instance

In this step we create a Snack instance and connect it to our Redis database. Notice, that Redis client is wrapped in our RedisConnection class to ensure shared interface. And at least we can register all entities that will be used in our project. For each entity we specify a list of fields that will be used to define keys when saving our data.

from data_snack import Snack
from data_snack.connections.redis import RedisConnection
snack = Snack(connection=RedisConnection(redis_connection))  # create instance
snack.register_entity(Person, key_fields=['index'])  # register your entity

4. Save and load your entities using Snack

You are ready to save and load data using Snack.

snack.set(Person("1", "John"))
# 'Person-1'
entity = snack.get(Person, ["1"])
# Person(index='1', name='John')
snack.set_many([Person("1", "John"), Person("2", "Anna")])
# ['Person-1', 'Person-2']
entities = snack.get_many(Person, [["1"], ["2"]])
# [Person(index='1', name='John'), Person(index='2', name='Anna')]

4. Delete your entities using Snack

After you're done with your data you can delete it using Snack.

snack.delete(Person, ["1"])
# Person(index='1', name='John')
snack.delete_many(Person, [["1"], ["2"]])
# [Person(index='1', name='John'), Person(index='2', name='Anna')]

Documentation

Access documentation

WIP. Documentation will be hosted on github pages.

Setup documentation

Setup documentation directory

mkdir docs
cd docs

Create documentation scaffold. Make sure to select an option with separated directories for source and build.

sphinx-quickstart

Update extensions in docs/source/conf.py.

extensions = ['sphinx.ext.autodoc', 'sphinx.ext.napoleon']

Update apidoc documentation

Before you start make sure to import project src directory at the very top of docs/source/conf.py file.

import sys
import os
sys.path.insert(0, os.path.abspath(os.path.join('..', '..', 'src')))

Since documentation uses additional modules (other than base data-snack), we need to install additional requirements:

pip install -r docs/requirements.txt

Update the scaffold and generate the html docs.

sphinx-apidoc -o ./source ../src/data_snack
make html

Contact

Plugin was created by the Data Science team from Webinterpret.