Django Rest Framework：如何将数据传递给嵌套的Serializer并仅在自定义验证后创建对象

Question

我有两个型号：

class Book(AppModel):
    title = models.CharField(max_length=255)

class Link(AppModel):
    link = models.CharField(max_length=255)

class Page(AppModel):
    book= models.ForeignKey("Book",related_name="pages",on_delete=models.CASCADE)
    link = models.ForeignKey("Link", related_name="pages", on_delete=models.CASCADE)
    page_no = models.IntegerField()
    text = models.TextField()

和serializers

class LinkSerializer(serializers.ModelSerializer):
    class Meta:
       model = Link
       fields = ['link']

class PageSerializer(serializers.ModelSerializer):
    class Meta:
        model = Page
        fields = ('link','text','page_no')

    def validate_text(self, value):
        #some validation is done here.

    def validate_link(self, value):
        #some validation is done here.

class BookSerializer(serializers.ModelSerializer):
    pages = PageSerializer(many=True)
    class Meta:
        model = Book
        fields = ('title','pages')

    @transaction.atomic
    def create(self, validated_data):
        pages_data= validated_data.pop('pages')
        book = self.Meta.model.objects.create(**validated_data)
        for page_data in pages_data:
            Page.objects.create(book=book, **page_data)
        return book

在validate_text有一个PageSerializer方法。 create方法永远不会调用PageSerializer，page_data永远不会被验证。

所以我尝试了另一种方法：

@transaction.atomic
def create(self, validated_data):
    pages_data = validated_data.pop('pages')
    book= self.Meta.model.objects.create(**validated_data)
    for page_data in pages_data:
        page = Page(book=book)
        page_serializer = PageSerializer(page, data = page_data)
        if page_serializer.is_valid():
            page_serializer.save()
        else:
            raise serializers.ValidationError(page_serializer.errors)
    return book

发布数据：

{
"title": "Book Title",
"pages": [{
"link":1,"page_no":52, "text":"sometext"}]
}

但上述方法会引发错误：

{
"link": [
    "Incorrect type. Expected pk value, received Link."
    ]
}

我还发现了为什么会出现这个错误：虽然我发布了pk的1值Link的数据，但是从PageSerializer传递给BookSerializer的数据如下：{"link": "/go_to_link/", "page_no":52, "text": "sometext"}

为什么Link的实例传递给PageSerializer而我发送的是pk的Link？需要知道这里。

Answer 1

当你调用serializer.is_valid(raise_exception=True/False)时，它会自动调用嵌套序列化程序的验证函数。当您调用serializer.save(**kwargs)时，序列化程序会将经过验证的数据传递到序列化程序的create(self, validated_data)或update(self, instance, validated_data)函数中。此外，在验证数据中，ForeignKey字段返回一个对象。

def create(self, validated_data):
    pages_data = validated_data.pop('pages') # [{'link': Linkobject, ...}]
    book= self.Meta.model.objects.create(**validated_data)
    for page_data in pages_data:
        page = Page(book=book)
        page_serializer = PageSerializer(page, data = page_data) # here you are sending object to validation again
        if page_serializer.is_valid():
            page_serializer.save()
        else:
            raise serializers.ValidationError(page_serializer.errors)
    return book

Answer 2

要使用嵌套的序列化程序验证嵌套对象：

@transaction.atomic
def create(self, validated_data):
    pages_data = validated_data.pop('pages') #pages data of a book
    book= self.Meta.model.objects.create(**validated_data)
    for page_data in pages_data:
        page = Page(book=book)
        page_serializer = PageSerializer(page, data = page_data)
        if page_serializer.is_valid(): #PageSerializer does the validation
            page_serializer.save()
        else:
            raise serializers.ValidationError(page_serializer.errors) #throws errors if any
    return book

假设您将数据发送为：

{
    "title": "Book Title",
    "pages": [{
        "link":2,#<= this one here
        "page_no":52, 
        "text":"sometext"}]
}

在上面的数据中，我们发送了id对象的Link。但是在上面定义的create的BookSerializer方法中，我们发送的数据更改为：

{
    "title": "Book Title",
    "pages": [{
        "link":Link Object (2),#<= changed to the Link object with id 2
        "page_no":52, 
        "text":"sometext"}]
}

PageSerializer实际上意味着接受pk的link值，即"link": 2而不是"link":Link Object (2)。因此它会抛出错误：

{
"link": [
    "Incorrect type. Expected pk value, received Link."
    ]
}

因此，解决方法是覆盖嵌套序列化器的to_internal_value方法，将接收到的Link Object (2)对象转换为其pk值。

所以你的PageSerializer课应该是：

class PageSerializer(serializers.ModelSerializer):
    class Meta:
        model = Page
        fields = ('link','text','page_no')

    def to_internal_value(self, data): 
        link_data = data.get("link")
        if isinstance(link_data, Link): #if object is received
            data["link"] = link_data.pk # change to its pk value
        obj = super(PageSerializer, self).to_internal_value(data)
        return obj

    def validate_text(self, value):
        #some validation is done here.

    def validate_link(self, value):
        #some validation is done here.

和父序列化程序：

class BookSerializer(serializers.ModelSerializer):
    pages = PageSerializer(many=True)
    class Meta:
        model = Book
        fields = ('title','pages')

    @transaction.atomic
    def create(self, validated_data):
        pages_data = validated_data.pop('pages')#pages data of a book
        book= self.Meta.model.objects.create(**validated_data)
        for page_data in pages_data:
            page = Page(book=book)
            page_serializer = PageSerializer(page, data = page_data)
            if page_serializer.is_valid(): #PageSerializer does the validation
                page_serializer.save()
            else:
                raise serializers.ValidationError(page_serializer.errors) #throws errors if any
        return book

这应该允许嵌套的序列化程序进行验证，而不是在父序列化器的create方法中编写验证并违反DRY。