DRF的Serializer组件(源码分析)

1. 数据校验

drf中为我们提供了Serializer，他主要有两大功能：

• 对请求数据校验（底层调用Django的Form和ModelForm）
• 对数据库查询到的对象进行序列化

示例一: 基于Serializer

# models.py
class UserInfo(models.Model):
    username = models.CharField(verbose_name='用户名', max_length=32)
    age = models.CharField(verbose_name='年龄', max_length=32)
    level_choice = ((1, 'VIP'), (2, 'SVIP'), (3, 'PARTNER'))
    level = models.CharField(verbose_name='级别', choices=level_choice, max_length=32)
    email = models.CharField(verbose_name='邮箱', max_length=32)

# views.py 基于Serializer
class UserSerializers(serializers.Serializer):
    username = serializers.CharField(label='用户名', max_length=32)
    age = serializers.CharField(label='年龄', max_length=32)
    level = serializers.ChoiceField(label='级别', choices=models.UserInfo.level_choice)
    email = serializers.CharField(label='用户名', min_length=6, max_length=32, validators=[EmailValidator, ])
    email1 = serializers.CharField(label='用户名', min_length=6, max_length=32)
    email2 = serializers.CharField(label='用户名', min_length=6, max_length=32)

    def validate_email2(self, value):
        """ 钩子函数, 用于验证某个字段 """
        if re.match('^\w+@\w+\.\w+$', value):
            return value
        raise exceptions.ValidationError('邮箱格式错误')
 
class UserView(APIView):
    """ 用户管理 """

    def post(self, request):
        """ 添加用户 """
        ser = UserSerializers(data=request.data)  # 将请求体数据传入, 这个request.data可以解析各种数据
        if not ser.is_valid():
            return Response({'code': 1006, 'data': ser.errors})
        print(ser.validated_data)

        # 将数据保存到数据库
        return Response({'code': 0, 'data': 'xxxx'})

示例二: 基于ModelSerializer

# models.py
from django.db import models


class Role(models.Model):
    """ 角色表 """
    title = models.CharField(verbose_name='名称', max_length=32)


class Department(models.Model):
    """ 部门表 """
    title = models.CharField(verbose_name='名称', max_length=32)


class UserInfo(models.Model):
    username = models.CharField(verbose_name='用户名', max_length=32)
    age = models.CharField(verbose_name='年龄', max_length=32)
    level_choice = ((1, 'VIP'), (2, 'SVIP'), (3, 'PARTNER'))
    level = models.CharField(verbose_name='级别', choices=level_choice, max_length=32)
    email = models.CharField(verbose_name='邮箱', max_length=32)

    # 创建外键
    depart = models.ForeignKey(verbose_name="部门", to="Department", on_delete=models.CASCADE)

    # 多对多
    roles = models.ManyToManyField(verbose_name="角色", to="Role")

# views.py
# 基于ModelSerializer
class UserModelSerializer(serializers.ModelSerializer):
    email1 = serializers.CharField(label='邮箱1', validators=[EmailValidator, ])

    class Meta:
        model = models.UserInfo
        fields = ['username', 'age', 'email', 'email1', 'roles']  # 需要传入的数据, 多对多
        extra_kwargs = {
            'username': {'min_length': 4, 'max_length': 32},
            'age': {'max_length': 3}
        }
       	
        def valicate_email(self, value):
            ....
            return value


class UserView(APIView):
    """ 用户管理 """

    def post(self, request):
        """ 添加用户 """
        ser = UserModelSerializer(data=request.data)  # 将请求体数据传入, 这个request.data可以解析各种数据
        if not ser.is_valid():
            return Response({'code': 1006, 'data': ser.errors})
        print(ser.validated_data)

        # 将数据保存到数据库
        ser.validated_data.pop('email1')  # 删除不需要存入数据库的数据
        ser.save(level=1, depart_id=1)  # 加入初始化数据
        return Response({'code': 0, 'data': '创建成功'})

2. 序列化

示例一: 序列化基本字段

class UserModelSerializer(serializers.ModelSerializer):
    class Meta:
        model = models.UserInfo
        fields = ['username', 'age', 'level', 'email', 'depart', 'roles']  # 序列化基本字段


class UserView(APIView):
    """ 用户管理 """
    def get(self, request):
        """ 序列化数据 """
        queryset = models.UserInfo.objects.all()
        ser = UserModelSerializer(instance=queryset, many=True)
        print(ser.data)
        return Response({'code': 0, 'data': ser.data})

返回值:

HTTP 200 OK
Allow: GET, HEAD, OPTIONS
Content-Type: application/json
Vary: Accept

{
    "code": 0,
    "data": [
        {
            "username": "ifeng",
            "age": "11",
            "level": 1,
            "email": "ifeng190410@gmail.com",
            "depart": 1,
            "roles": []
        },
        {
            "username": "Mcoco",
            "age": "11",
            "level": 1,
            "email": "ifeng190410@gmail.com",
            "depart": 1,
            "roles": [
                1,
                2
            ]
        }
    ]
}

示例二: 自定义字段

from django.forms.models import model_to_dict
from rest_framework import serializers
from rest_framework.response import Response
from rest_framework.views import APIView

from api import models

class UserModelSerializer(serializers.ModelSerializer):
    # 自定义字段
    level_text = serializers.CharField(source="get_level_display")
    depart = serializers.CharField(source='depart.title')

    roles = serializers.SerializerMethodField()
    extra = serializers.SerializerMethodField()

    class Meta:
        model = models.UserInfo
        fields = ['username', 'age', 'level_text', 'email', 'depart', 'roles', 'extra']

    def get_roles(self, obj):
        data_list = obj.roles.all()
        return [model_to_dict(item, ['id', 'title']) for item in data_list]

    def get_extra(self, obj):
        return 666


class UserView(APIView):
    """ 用户管理 """

    def get(self, request):
        """ 序列化数据 """
        queryset = models.UserInfo.objects.all()
        ser = UserModelSerializer(instance=queryset, many=True)
        print(ser.data)
        return Response({'code': 0, 'data': ser.data})

返回值:

{
   "code": 0,
   "data": [
       {
           "username": "ifeng",
           "age": "11",
           "level_text": "SVIP",
           "email": "ifeng190410@gmail.com",
           "depart": "后端",
           "roles": [],
           "extra": 666
       },
       {
           "username": "Mcoco",
           "age": "11",
           "level_text": "VIP",
           "email": "ifeng190410@gmail.com",
           "depart": "销售",
           "roles": [
               {
                   "id": 1,
                   "title": "CEO"
               },
               {
                   "id": 2,
                   "title": "CFO"
               }
           ],
           "extra": 666
       }
   ]
}

示例三: 序列化类的嵌套

嵌套主要是面向外键和多对多表的时候

3. 数据校验&序列化

注意点:

​ 我们在做多对多数据校验的时候, 后面如果需要新增数据, 则需要重写create方法, 如果需要更新数据, 则需要重写update方法

# mdoels.py
from django.db import models


# Create your models here.
class Role(models.Model):
    """ 角色表 """
    title = models.CharField(verbose_name='名称', max_length=32)


class Department(models.Model):
    """ 部门表 """
    title = models.CharField(verbose_name='名称', max_length=32)


class UserInfo(models.Model):
    username = models.CharField(verbose_name='用户名', max_length=32)
    age = models.CharField(verbose_name='年龄', max_length=32)
    level_choice = ((1, 'VIP'), (2, 'SVIP'), (3, 'PARTNER'))
    level = models.SmallIntegerField(verbose_name='级别', choices=level_choice)  # 类型为Int
    email = models.CharField(verbose_name='邮箱', max_length=32)

    # 创建外键
    depart = models.ForeignKey(verbose_name="部门", to="Department", on_delete=models.CASCADE)

    # 多对多
    roles = models.ManyToManyField(verbose_name="角色", to="Role")

# views.py
# 数据校验&序列化
class DepartModelSerializer(serializers.ModelSerializer):
    class Meta:
        model = models.Department
        fields = ['id', "title"]
        extra_kwargs = {
            "id": {"read_only": False},  # 数据验证, 需传入id, 为后续的create做准备
            "title": {"read_only": True}  # 序列化
        }


class RoleModelSerializer(serializers.ModelSerializer):
    class Meta:
        model = models.Role
        fields = ['id', "title"]
        extra_kwargs = {
            "id": {"read_only": False},  # 数据校验, 需传入id, 为后续的create做准备
            "title": {"read_only": True}  # 序列化
        }


class UserModelSerializer(serializers.ModelSerializer):
    level_text = serializers.CharField(source="get_level_display", read_only=True)  # read_only -> 只序列化, 但是不数据校验

    # Serializer嵌套，如果不设置read_only，一定要自定义create和update，自定义新增和更新的逻辑。
    depart = DepartModelSerializer(many=False)
    roles = RoleModelSerializer(many=True)

    extra = serializers.SerializerMethodField(read_only=True)
    email2 = serializers.EmailField(write_only=True)  # write_only -> 只数据校验不序列化

    # 数据校验：username、email、email2、部门、角色信息
    class Meta:
        model = models.UserInfo
        # username, age, email是即read_only也write_only
        fields = [
            "username", "age", "email", "level_text", "depart", "roles", "extra", "email2"
        ]
        # 给字段添加额外参数
        extra_kwargs = {
            "age": {"read_only": True},
            "email": {"validators": [EmailValidator, ]},
        }

    def get_extra(self, obj):
        return 666

    def validate_username(self, value):  # 钩子方法
        return value

    # 新增加数据时, 因为无法解决m2m的储存问题. 所以需要重写create方法
    def create(self, validated_data):
        """
        	如果有嵌套的Serializer，在进行数据校验时，只有两种选择：
              	1. 将嵌套的序列化设置成 read_only
              	2. 自定义create和update方法，自定义新建和更新的逻辑
            注意：用户端提交数据的格式。
        """
        """
        validated_data:
        	OrderedDict([('username', 'xiaoergu'), ('email', 'xiaoergu@gmail.com'), ('depart', OrderedDict([('id', 2)])), ('roles', [OrderedDict([('id', 1)]), OrderedDict([('id', 2)])]), ('email2', 'budianlong@gmail.com')])
        """
        depart_id = validated_data.pop('depart')['id']  # 拿到depart的id

        role_id_list = [ele['id'] for ele in validated_data.pop('roles')]  # 拿到roles的所有id

        # 新增用户表
        validated_data['depart_id'] = depart_id
        user_object = models.UserInfo.objects.create(**validated_data)

        # 在用户表和角色表的关联表中添加对应关系, django-orm知识
        user_object.roles.add(*role_id_list)

        return user_object


class UserView(APIView):
    """ 用户管理 """

    def get(self, request):
        """ 添加用户 """
        queryset = models.UserInfo.objects.all()
        ser = UserModelSerializer(instance=queryset, many=True)
        return Response({"code": 0, 'data': ser.data})

    def post(self, request):
        """ 添加用户 """
        ser = UserModelSerializer(data=request.data)
        if not ser.is_valid():
            return Response({'code': 1006, 'data': ser.errors})

        print(ser.validated_data)
        ser.validated_data.pop('email2')

        instance = ser.save(age=18, level=3)

        # 新增之后的一个对象（内部调用UserModelSerializer进行序列化）
        print(instance)
        # ser = UserModelSerializer(instance=instance, many=False)
        # ser.data

        return Response({'code': 0, 'data': ser.data})

返回值:

4. 源码分析

底层源码实现：

序列化的底层源码实现有别于上述其他的组件，序列化器相关类的定义和执行都是在视图中被调用的，所以源码的分析过程可以分为：定义类、序列化、数据校验。

源码1：序列化过程

源码2：数据校验过程