A simple, fast, extensible python library for data validation.
Branch: master
Clone or download

README.md

Validr

travis-ci codecov

A simple, fast, extensible python library for data validation.

  • Simple and readable schema
  • 10X faster than jsonschema, 40X faster than schematics
  • Can validate and serialize any object
  • Easy to create custom validators
  • Accurate and friendly error messages

简单,快速,可拓展的数据校验库。

  • 简洁,易读的 Schema
  • jsonschema 快 10 倍,比 schematics 快 40 倍
  • 能够校验&序列化任意类型对象
  • 易于拓展自定义校验器
  • 准确友好的错误提示

Overview

from validr import T, modelclass, asdict

@modelclass
class Model:
    """Base Model"""

class Person(Model):
    name=T.str.maxlen(16).desc('at most 16 chars')
    website=T.url.optional.desc('website is optional')

guyskk = Person(name='guyskk', website='https://github.com/guyskk')
print(asdict(guyskk))

Install

Note: Only support python 3.4+

pip install validr

Document

https://github.com/guyskk/validr/wiki

Performance

benchmark result in Travis-CI:

--------------------------timeits---------------------------
  voluptuous:default             10000 loops cost 0.368s
      schema:default              1000 loops cost 0.318s
        json:loads-dumps        100000 loops cost 1.380s
      validr:default            100000 loops cost 0.719s
      validr:model              100000 loops cost 1.676s
  jsonschema:draft3              10000 loops cost 0.822s
  jsonschema:draft4              10000 loops cost 0.785s
  schematics:default              1000 loops cost 0.792s
---------------------------scores---------------------------
  voluptuous:default               375
      schema:default                43
        json:loads-dumps          1000
      validr:default              1918
      validr:model                 823
  jsonschema:draft3                168
  jsonschema:draft4                176
  schematics:default                17

Develop

Validr is implemented by Cython since v0.14.0, it's 5X faster than original pure python implemented.

setup:

It's better to use virtualenv or similar tools to create isolated Python environment for develop.

After that, install dependencys:

./bootstrap.sh

build, test and benchmark:

inv build
inv test
inv benchmark

License

MIT License