Liquid is a pipeline tool based on pluggy hook technology, which is mainly used for rapid construction of data science applications.
Project description
Liquid-算法应用流程管理工具
介绍
- liquid是一个pipeline工具,主要功能提供算法应用快捷组织功能,主要技术使用基于pluggy的hook技术。
安装
liquid采用Python开发,得益于Python良好的社区环境,安装支持Pythonic风格的各种管理器。
$ pip install liquid-0.1-xxxxxxxxxxxx.whl
快速指南
python-sdk使用
- liquid提供python-sdk的使用方式,以下是liquid主程脚本示例:
from liquid.node import Node
from liquid.pipeline import Pipeline
from liquid.io import DataCatalog
from liquid.hook import HookManager
from liquid.runner import SequentialRunner
### 开始liquid测试
### 设置函数节点对象
def first_func(a,b):
c = a + b
return c
def second_func(c,d):
e = c * d
return e
def third_func(e):
f = e + 1
return f
first_node = Node(func=first_func,inputs=['a','b'],outputs='c',name='firstfunc')
second_node = Node(func=second_func,inputs=['c','d'],outputs='e',name='secondfunc')
third_node = Node(func=third_func,inputs=['e'],outputs=['f'],name='thirdfunc')
### 创建一个pipeline
test_pipeline = Pipeline(nodes=[first_node,second_node,third_node])
### catalog加载初始参数
test_cache_data = DataCatalog()
test_cache_data.save(data_name='a',data_obj=5)
test_cache_data.save(data_name='b',data_obj=6)
test_cache_data.save(data_name='d',data_obj=7)
### hook_manager加载已挂载的前后处理功能函数
hook_manager = HookManager()
sequential_runner = SequentialRunner(pipeline=test_pipeline,catalog=test_cache_data,hook_manager=hook_manager)
# print(sequential_runner.pipeline,sequential_runner.catalog,sequential_runner.hook_manager)
print('--------------------------------------------------------------------------------------------------------------')
sequential_runner.execute(is_release=False)
print(test_cache_data.load(data_name='f'))
print(test_cache_data.cache_data)
### 测试过程(5+6)*7+1=78
设计
- 基于pluggy的hook技术实现灵活扩展性
- 设计关键概念,Node,Pipeline,Runner
- 函数节点化-Node
- 开放挂载点-Pipeline
- 扩展运行方式(顺序运行和并行运行)-Runner
技术列表
- property动态属性
- 基于pluggy的hook
- call
- add
- cache_dict
设计UML图
以下是设计的UML图:
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
shihua-liquid-0.1.tar.gz
(10.1 kB
view hashes)
Built Distribution
Close
Hashes for shihua_liquid-0.1-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 0e067cd2f7e7755e31edb0fefd5f0d1a04e1f80732e431829a2a50a156b2bf2c |
|
MD5 | 4db60ecebb016475191fef8c02d62d6c |
|
BLAKE2b-256 | e63e7300ab99a2278d8c8dc15764c66726d3f230b0c64a8ee792c167531f5b5f |