validr

Validr

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
• 比 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 3.5+

pip install validr

When you have c compiler in your system, validr will be c speedup mode. Otherwise validr will fallback to pure python mode.

To force c speedup mode:

VALIDR_SETUP_MODE=c pip install validr

To force pure python mode:

VALIDR_SETUP_MODE=py 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 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

The project is open source under Anti-996 License and GNU GPL License, you can choose one of them.