Python 参数校验的进化

摘要：事情的起因是感觉目前项目中的参数校验方法写的太简单了很多时候需要在层再处理于是就动手准备写一个好用一点的可以自定义校验参数规则的参数校验器考虑到要可以灵活的配置就萌生了大概的印象使用参数参数等等对参数进行规则绑定使用装饰器可扩展可以自定

事情的起因是感觉目前项目中的参数校验方法写的太简单了,很多时候需要在server层再if else处理,于是就动手准备写一个好用一点的,可以自定义校验参数规则的参数校验器,考虑到要可以灵活的配置就萌生了大概的印象:

使用map - 参数A:ruleA,参数B-ruleB..等等,对参数进行规则绑定

使用装饰器

可扩展,可以自定义校验规则

于是第一个版本实现如下:

# -*- coding:utf-8 -*-
__author__ = "aleimu"
__date__ = "2018-12-6"
__doc__ = "一个实用的入参校验装饰器--针对目前,前端 url?&a=1&b=2或-d"a=1&b=2c=qwe"形式的非json(所有参数都是str类型)" 
          "入参的校验"

import copy
import traceback
from collections import OrderedDict
from functools import wraps
from flask import Flask, json, jsonify, request

app = Flask(__name__)


def verify_args(need=None, length=None, check=None, strip=True, default=(False, None), diy_func=None, release=False):
    """
    约束:
    1. 简化了传参校验,使用位置传参或者关键词传参(一个参数对应一个参数),不允许使用one to list等python高级传参特性
    2. 所有的参数都是str/unicode类型的,前端没有使用json带参数类型的入参方式
    :param need: 必须参数,且不能为None或者""
    :param length: 参数长度范围
    :param check:  str的常用类方法/属性如下:
        isalnum 判断字符串中只能由字母和数字的组合，不能有特殊符号
        isalpha 字符串里面都是字母，并且至少是一个字母，结果就为真，（汉字也可以）其他情况为假
        isdigit 函数判断是否全为数字
    :param strip:对字段进行前后过滤空格
    :param default:将"" 装换成None
    :param diy_func:自定义的对某一参数的校验函数格式: {key:func},类似check, diy_func={"a": lambda x: x + "aa"})
    :param release:发生参数校验异常后是否依然让参数进入主流程函数
    :return:
    """

    def wraps_1(f):
        @wraps(f)
        def wraps_2(*args, **kwargs):
            if release:
                args_bak = args[:]
                kwargs_bak = copy.deepcopy(kwargs)  # 下面流程异常时,是否直接使用 原参数传入f todo
            print ("in", args, kwargs)
            args_template = f.func_code.co_varnames
            print("args_template:", args_template)
            args_dict = OrderedDict()
            req_args_need_list = []
            req_args_types_list = []
            try:
                for i, x in enumerate(args):
                    args_dict[args_template[i]] = x
                sorted_kwargs = sort_by_co_varnames(args_template, kwargs)
                args_dict.update(sorted_kwargs)
                print("args_dict:", args_dict)
                # need
                if need:
                    for k in need:
                        if k not in args_dict:
                            req_args_need_list.append(k)
                        else:
                            if args_dict[k] == None or args_dict[k] == "":
                                req_args_need_list.append(k)
                    if req_args_need_list:
                        return False, "%s is in need" % req_args_need_list
                # strip
                if strip:
                    for k in args_dict:
                        if args_dict[k]:
                            args_dict[k] = args_dict[k].strip()
                # length
                if length:
                    for k in args_dict:
                        if k in length:
                            if not (len(args_dict[k]) >= length[k][0] and len(args_dict[k]) <= length[k][1]):
                                return False, "%s length err" % k
                # default:
                if default[0]:
                    for x in args_dict:
                        if args_dict[x] == "":
                            args_dict[x] = default[1]
                # check
                if check:
                    for k in check:
                        check_func = getattr(type(args_dict[k]), check[k], None)
                        if not (k in args_dict and check_func and check_func(args_dict[k])):
                            req_args_types_list.append(k)
                    if req_args_types_list:
                        return False, "%s type err" % req_args_types_list
                # diy_func
                if diy_func:
                    for k in args_dict:
                        if k in diy_func:
                            args_dict[k] = diy_func[k](args_dict[k])
            except Exception as e:
                print("verify_args catch err: ", traceback.format_exc())
                if release:
                    return f(*args_bak, **kwargs_bak)
                else:
                    return False, str(e)
            return f(*args_dict.values())

        return wraps_2

    return wraps_1


def sort_by_co_varnames(all_args, kwargs):
    new_ordered = OrderedDict()
    for x in all_args:
        if x in kwargs:
            new_ordered[x] = kwargs[x]
    return new_ordered


@app.route("/", methods=["GET", "POST", "PUT"])
def index():
    a = request.values.get("a")
    b = request.values.get("b")
    c = request.values.get("c")
    d = request.values.get("d")
    e = request.values.get("e")
    f = request.values.get("f")
    g = request.values.get("g")
    status, data = todo(a, b, c, d, e=e, f=f, g=g)
    if status:
        return jsonify({"code": 200, "data": data, "err": None})
    else:
        return jsonify({"code": 500, "data": None, "err": data})


@verify_args(need=["a", "b", "c"], length={"a": (6, 50)}, strip=True,
             check={"b": "isdigit", "c": "isalnum"},
             default=(True, None),
             diy_func={"a": lambda x: x + "aa"})
def todo(a, b, c, d, e="  1  ", f="2    ", g=""):
    return True, {"a": a, "b": b, "c": c, "d": d, "e": e, "f": f, "g": g}


if __name__ == "__main__":
    app.run(host="0.0.0.0", port=6000, debug=True)

"""
# curl "http://127.0.0.1:6000/" -d "pwd=123&a=1111111&b=2&c=3&d=d&e=eeeeee&f=12345&g="
{
  "code": 200,
  "data": {
    "a": "1111111aa",
    "b": "2",
    "c": "3",
    "d": "d",
    "e": "eeeeee",
    "f": "12345",
    "g": null
  },
  "err": null
}

# curl "http://127.0.0.1:6000/" -d "pwd=123&a=1111111&b=2&c=3346()*&d=d&e=eeeeee&f=12345&g="
{
  "code": 500,
  "data": null,
  "err": "["c"] type err"
}

# curl "http://127.0.0.1:6000/" -d "pwd=123&a=1111111&b=2&c=&d=d&e=eeeeee&f=12345&g="    
{                                                                                        
  "code": 500,                                                                           
  "data": null,                                                                          
  "err": "["c"] is in need"                                                              
}   

# curl "http://127.0.0.1:6000/" -d "pwd=123&a=1111111&b=2&c=  1  &d=d&e=eeeeee&f=12345&g="  
{                                                                                           
  "code": 200,                                                                              
  "data": {                                                                                 
    "a": "1111111aa",                                                                       
    "b": "2",                                                                               
    "c": "1",                                                                               
    "d": "d",                                                                               
    "e": "eeeeee",                                                                          
    "f": "12345",                                                                           
    "g": null                                                                               
  },                                                                                        
  "err": null                                                                               
}                                                                                                                                                                                
"""

第一个版本切合了当前项目中经常遇到的校验问题,实现起来较简单,基本满足要求.
想要更通用点,更多校验规则一些,就需要每次为verify_args添加参数写if else了,嗯.....有点不优雅啊,于是去看github上有啥好的实现.
找到了如下几个项目:

https://github.com/keleshev/s... 嗯,1.6K的star,思路一致,实现的优雅,但是不好扩展啊....

https://github.com/kvesteri/v... 额,Python Data Validation for Humans™. not for me....

https://github.com/mansam/val... 嗯,思路一致,实现也简单,挺好扩展的,就用它了!

这里说说validator.py ,给个例子

from validator import Required, Not, Truthy, Blank, Range, Equals, In, validate

# let"s say that my dictionary needs to meet the following rules...
rules = {
    "foo": [Required, Equals(123)],
    "bar": [Required, Truthy()],
    "baz": [In(["spam", "eggs", "bacon"])],
    "qux": [Not(Range(1, 100))] # by default, Range is inclusive
}

# then this following dict would pass:
passes = {
    "foo": 123,
    "bar": True, # or a non-empty string, or a non-zero int, etc...
    "baz": "spam",
    "qux": 101
}
print validate(rules, passes)
# (True, {})

# but this one would fail
fails = {
    "foo": 321,
    "bar": False, # or 0, or [], or an empty string, etc...
    "baz": "barf",
    "qux": 99
}
print validate(rules, fails)
# (False,
#  {
#  "foo": ["must be equal to "123""],
#  "bar": ["must be True-equivalent value"],
#  "baz": ["must be one of ["spam", "eggs", "bacon"]"],
#  "qux": ["must not fall between 1 and 100"]
#  })

嗯,使用第一个版本封装一下validator.py就好了!考虑到需要写个dome来试试,就选了flask,嗯,对了,先去github 上搜一下 flask validator 没准已经有现成的呢,实现思路基本一致,但是......前几个star多的都不令人满意,还是自己造轮子吧.
先实现常见的在route上加装饰器版本,这样的话,就可以直接接收request收到的参数,然后直接校验了,有问题就直接返回错误给调用者,于是有了版本2

rules_example = {
    "a": [Required, Equals("123")],  # foo must be exactly equal to 123
    "b": [Required, Truthy()],  # bar must be equivalent to True
    "c": [In(["spam", "eggs", "bacon"])],  # baz must be one of these options
    "d": [Not(Range(1, 100))],  # qux must not be a number between 1 and 100 inclusive
    "e": [Length(0, maximum=5)],
    "f": [Required, InstanceOf(str)],
    "g": [Required, Not(In(["spam", "eggs", "bacon"]))],
    "h": [Required, Pattern("dd\%")],
    "i": [Required, GreaterThan(1, reverse=True, auto=True)],  # auto 自动转换成float类型来做比较
    "j": [lambda x: x == "bar"],
    "k": [Required, Isalnum()],  # 判断字符串中只能由字母和数字的组合，不能有特殊符号
    "l": [Required, Isalpha()],  # 字符串里面都是字母，并且至少是一个字母，结果就为真，（汉字也可以）其他情况为假
    "m": [Required, Isdigit()],  # 判断字符串是否全为数字
}


def validator_wrap(rules, strip=True, diy_func=None):
    """装饰器版 - 只能检测是否符合规则,不能修改参数
    :param rules:参数的校验规则,map
    :param strip:对字段进行前后空格检测
    :param diy_func:自定义的对某一参数的校验函数格式: {key:func},类似check, diy_func={"a": lambda x: x=="aa"})
    """

    def decorator(f):
        @wraps(f)
        def decorated_func(*args, **kwargs):
            try:
                args_dict = OrderedDict()
                if request.values:
                    args_dict.update(request.values)
                if request.json:
                    args_dict.update(request.json)
                # strip
                if strip:
                    for k in args_dict:
                        if args_dict[k] and isstr(args_dict[k]):
                            if args_dict[k][0] == " " or args_dict[k][-1] == " ":
                                return jsonify({"code": 500, "data": None, "err": "%s should not contain spaces" % k})
                # diy_func
                if diy_func:
                    for k in args_dict:
                        if k in diy_func:
                            args_dict[k] = diy_func[k](args_dict[k])
                # rules
                if rules:
                    result, err = validate(rules, args_dict)
                    if not result:
                        return jsonify(
                            {"code": 500, "data": None, "err": err})
            except Exception as e:
                print("verify_args catch err: ", traceback.format_exc())
                return jsonify({"code": 500, "data": None, "err": str(e)})
            return f(*args, **kwargs)

        return decorated_func

    return decorator
    
@app.route("/wrap", methods=["GET", "POST", "PUT"])
@validator_wrap(rules=rules_example, strip=True)  # 姿势 1:只能检测是否符合规则,不能修改参数,不符合就会直接返回json给调用者
def wrap_example():
    a = request.values.get("a")
    b = request.values.get("b")
    c = request.values.get("c")
    d = request.values.get("d")
    e = request.values.get("e")
    f = request.values.get("f")
    g = request.values.get("g")
    h = request.values.get("h")
    i = request.values.get("i")
    j = request.values.get("j")
    k = request.values.get("k")
    l = request.values.get("l")
    m = request.values.get("m")
    status, data = todo(a=a, b=b, c=c, d=d, e=e, f=f, g=g, h=h, i=i, j=j, k=k, l=l, m=m)
    if status:
        return jsonify({"code": 200, "data": data, "err": None})
    else:
        return jsonify({"code": 500, "data": None, "err": data})

好像挺好的,基本满足要求了,但是再route上加装饰器,那就改变不了参数的值了,虽然有些参数不一定符合要求,但是简单修补一下还是可以用的,还得继续寻找能够改变入参的方式,第一反应是在装饰器中修改request.values或者request.json的值,让进入到主函数后获取更新后的值,上下求索未得门径,request.value.update方法是被禁用的,继续看源码,后面的实现使用了dict的复杂封装,不好改啊,这样太绕了,还是直接调用函数吧,不玩装饰器了.于是又了版本3

def validator_func(rules, strip=True, default=(False, None), diy_func=None, release=False):
    """函数版-返回dict,代替request.values/request.json
    :param rules:参数的校验规则,map
    :param strip:对字段进行前后过滤空格
    :param default:将"" 装换成None
    :param diy_func:自定义的对某一参数的校验函数格式: {key:func},类似check, diy_func={"a": lambda x: x + "aa"})
    :param release:发生参数校验异常后是否依然让参数进入主流程函数
    """
    args_dict = OrderedDict()
    try:
        if request.values:
            args_dict.update(request.values)
        if request.json:
            args_dict.update(request.json)
        if release:
            args_dict_copy = copy.deepcopy(args_dict)  # 下面流程异常时,是否直接使用 原参数传入f # fixme
        # strip
        if strip:
            for k in args_dict:
                if isstr(args_dict[k]):
                    args_dict[k] = args_dict[k].strip()
        # default
        if default[0]:
            for x in args_dict:
                if args_dict[x] == "":
                    args_dict[x] = default[1]
        # diy_func
        if diy_func:
            for k in args_dict:
                if k in diy_func:
                    args_dict[k] = diy_func[k](args_dict[k])
        # rules
        if rules:
            result, err = validate(rules, args_dict)
            if not result:
                return False, err
    except Exception as e:
        print("verify_args catch err: ", traceback.format_exc())  # TODO
        if release:
            return True, args_dict_copy
        else:
            return False, str(e)
    return True, args_dict

@app.route("/func", methods=["GET", "POST", "PUT"])
def func_example():
    result, request_args = validator_func(rules=rules_example, strip=True)  # 姿势 2
    if not result:
        return jsonify({"code": 500, "data": None, "err": request_args})
    a = request_args.get("a")
    b = request_args.get("b")
    c = request_args.get("c")
    d = request_args.get("d")
    e = request_args.get("e")
    f = request_args.get("f")
    g = request_args.get("g")
    h = request_args.get("h")
    i = request_args.get("i")
    j = request_args.get("j")
    k = request_args.get("k")
    l = request_args.get("l")
    m = request_args.get("m")
    status, data = todo(a=a, b=b, c=c, d=d, e=e, f=f, g=g, h=h, i=i, j=j, k=k, l=l, m=m)
    if status:
        return jsonify({"code": 200, "data": data, "err": None})
    else:
        return jsonify({"code": 500, "data": None, "err": data})

嗯,还行吧,就是不怎么优雅,还是有点喜欢装饰器版本,但是苦于能力有限,不想看ImmutableMultiDict,MultiDict的实现,还是将第一个版本融合一下吧,装饰route不行,装饰todo还不行吗.于是有了版本4

def validator_args(rules, strip=True, default=(False, None), diy_func=None, release=False):
    """针对普通函数的参数校验的装饰器
    :param rules:参数的校验规则,map
    :param strip:对字段进行前后过滤空格
    :param default:将"" 装换成None
    :param diy_func:自定义的对某一参数的校验函数格式: {key:func},类似check, diy_func={"a": lambda x: x + "aa"})
    :param release:发生参数校验异常后是否依然让参数进入主流程函数
    """

    def decorator(f):
        @wraps(f)
        def decorated_func(*args, **kwargs):
            if release:
                args_bak = args[:]
                kwargs_bak = copy.deepcopy(kwargs)  # 下面流程异常时,是否直接使用 原参数传入f # fixme
            try:
                args_template = f.func_code.co_varnames
            except:
                args_template = f.__code__.co_varnames
            args_dict = OrderedDict()
            try:
                for i, x in enumerate(args):
                    args_dict[args_template[i]] = x
                sorted_kwargs = sort_by_co_varnames(args_template, kwargs)
                args_dict.update(sorted_kwargs)
                # strip
                if strip:
                    for k in args_dict:
                        if isstr(args_dict[k]):
                            args_dict[k] = args_dict[k].strip()
                # default
                if default[0]:
                    for x in args_dict:
                        if args_dict[x] == "":
                            args_dict[x] = default[1]
                # diy_func
                if diy_func:
                    for k in args_dict:
                        if k in diy_func:
                            args_dict[k] = diy_func[k](args_dict[k])
                # rules
                if rules:
                    result, err = validate(rules, args_dict)
                    if not result:
                        return False, err
            except Exception as e:
                print("verify_args catch err: ", traceback.format_exc())
                if release:
                    return f(*args_bak, **kwargs_bak)
                else:
                    return False, str(e)
            return f(*args_dict.values())

        return decorated_func

    return decorator
    
    
@validator_args(rules=rules_example, strip=True)  # 姿势 3
def todo(a, b, c, d, e, f, g, h, i, j, k, l, m):
    return True, {"a": a, "b": b, "c": c, "d": d, "e": e, "f": f, "g": g, "h": h, "i": i, "j": j, "k": k, "l": l,
                  "m": m}
                  

哎,就这样吧,打包一下,随便选吧,爱用哪个用哪个,反正我都写出来了.简单说就是:

validator_func 针对flask的request.json/requests.values的参数校验以及修改,修改的方式有限,可以自己控制

validator_wrap 是针对flask route的装饰器,针对request.json/requests.values的参数校验,只是校验,当然校验的方式可以自己写扩展

validator_args 针对普通函数的参数校验以及修改,注意不要使用python传参的高级特性(一个参数对应多个值),这个方法可以脱离flask使用,所以如果需要就直接copy过去吧.

嗯,最后还是分享一下到git上吧, https://github.com/aleimu/flask-validator 喜欢的点个star.