Django一分钟：DRF模型序列化器处理关联关系的示例与注意事项

DRF的ModelSerializer序列化器与Django的Model模型紧密映射，本文将通过简单的示例介绍几种处理关联关系的方法。

1. 创建模型和初始数据

创建模型

python">from django.db import models

class Product(models.Model):
    product_name = models.CharField(max_length=255)
    quantity = models.IntegerField()

class Component(models.Model):
    product = models.ForeignKey(Product, on_delete=models.CASCADE, related_name='components')
    component_name = models.CharField(max_length=255)
    code = models.CharField(max_length=255)
    quantity = models.IntegerField()

    class Meta:
        ordering = ['product']

    def __str__(self):
        return f'{self.code}-{self.component_name}'

创建数据

python">>>> from api.models import Product, Component
>>> p1 = Product.objects.create(product_name="盖子", quantity=30)
>>> Component.objects.create(product=p1, component_name='纤维板', code='XYB', quantity=5)
>>> Component.objects.create(product=p1, component_name='螺丝', code='LS', quantity=10)
>>> Component.objects.create(product=p1, component_name='密封条', code='MFT', quantity=20)

2. 字符串关系字段

使用StringRelatedField

python">from rest_framework import serializers
from .models import Product

class ProductSerializer(serializers.ModelSerializer):
    components = serializers.StringRelatedField(many=True)

    class Meta:
        model = Product
        fields = ['product_name', 'quantity', 'components']

执行查询和序列化

python">>>> from api.models import Product
>>> p1 = Product.objects.prefetch_related('components').get(pk=1)
>>> from api.serializers import ProductSerializer
>>> ps = ProductSerializer(p1)
>>> ps.data
# {'product_name': '盖子', 'quantity': 30, 'components': ['XYB-纤维板', 'LS-螺丝', 'MFT-密封条']}

效果如下:

python">{
    'product_name': '盖子', 
    'quantity': 30, 
    'components': [
        'XYB-纤维板', 
        'LS-螺丝', 
        'MFT-密封条'
    ]
}

注意观察获取查询集的代码:Product.objects.prefetch_related('components').get(pk=1)不难发现我们使用了prefetch_related来获取查询集，如果一次查询大量的Product，不使用prefetch_related将会导致严重的性能问题（N+1问题）。DRF的序列化器不会为你自动优化。当然我们的示例中影响不大，因为只获取了一个查询集。

3. 主键关系字段

使用PrimaryKeyRelatedField

python">from rest_framework import serializers
from .models import Product

class ProductSerializer(serializers.ModelSerializer):
    components =  serializers.PrimaryKeyRelatedField(many=True, read_only=True)

    class Meta:
        model = Product
        fields = ['product_name', 'quantity', 'components']

效果如下：

python">{
    'product_name': '盖子', 
    'quantity': 30, 
    'components': [
        1, 
        2, 
        3
    ]
}

4. 指定关系字段

使用SlugRelatedField可以指定字段来表示关联关系：

python">from rest_framework import serializers
from .models import Product

class ProductSerializer(serializers.ModelSerializer):
    components =  serializers.SlugRelatedField(
        many=True,
        read_only=True,
        slug_field='code'
     )

    class Meta:
        model = Product
        fields = ['product_name', 'quantity', 'components']

效果如下：

python">{
    'product_name': '盖子', 
    'quantity': 30, 
    'components': [
        'XYB', 
        'LS', 
        'MFT'
    ]
}

5. 嵌套序列化器

python">from rest_framework import serializers
from .models import Product, Component

class ComponentSerializer(serializers.ModelSerializer):
    class Meta:
        model = Component
        fields = ['component_name', 'code', 'quantity']

class ProductSerializer(serializers.ModelSerializer):
    components =  ComponentSerializer(many=True, read_only=True)

    class Meta:
        model = Product
        fields = ['product_name', 'quantity', 'components']

效果如下：

python">{
    "product_name": "盖子",
    "quantity": 30,
    "components": [
        {
            "component_name": "纤维板",
            "code": "XYB",
            "quantity": 5
        },
        {
            "component_name": "螺丝",
            "code": "LS",
            "quantity": 10
        },
        {
            "component_name": "密封条",
            "code": "MFT",
            "quantity": 20
        }
    ]
}

6. 可写嵌套

默认上面创建的嵌套序列化器是只读的，可写嵌套需要实现update、create两者或其一：

python">from rest_framework import serializers
from .models import Product, Component

class ComponentSerializer(serializers.ModelSerializer):
    class Meta:
        model = Component
        fields = ['component_name', 'code', 'quantity']

class ProductSerializer(serializers.ModelSerializer):
    components =  ComponentSerializer(many=True)

    class Meta:
        model = Product
        fields = ['product_name', 'quantity', 'components']

    def create(self, validated_data):
        components_data = validated_data.pop('components')
        product = Product.objects.create(**validated_data)
        for component_data in components_data:
            Component.objects.create(product=product, **component_data)
        return product

可以使用此序列化器创建实例：

python">>>> data = {
    "product_name": "电机",
    "quantity": 1,
    "components": [
        {
            "component_name": "铝型材",
            "code": "LXC",
            "quantity": 5
        },
        {
            "component_name": "螺栓",
            "code": "LSHUN",
            "quantity": 10
        },
    ]
}
>>> from api.serializers import ProductSerializer
>>> s = ProductSerializer(data=data)
>>> s.is_valid()
# True
>>> s.save()
# <Product: Product object (2)>