使用Django Rest Framework 3.2.2的可写嵌套序列化器和现有对象

Question

考虑具有关联类别的Listing模型.我想通过对数据执行POST来为现有类别创建新的列表: {"title": "myapp", "category": {"name": "Business"}},其中title是应创建的列表的标题,并且Business是用于此新列表的现有类别的名称.

当我尝试发出这样的请求并ListingSerializer为此实例化时,我收到一个错误,指示类别名称必须是唯一的 - 我不想创建新的类别,而是使用现有的类别.我已尝试在类别字段上设置验证器[],但这并没有改变行为.

我可以使用a SlugRelatedField,但这会强制我的请求数据看起来更像{"title": "myapp", "category": "Business"},这不是我想要的.我尝试使用source参数SlugRelatedField来指定嵌套关系,但这也不起作用:

category = serializers.SlugRelatedField(
        slug_field='category.name',
        queryset=models.Category.objects.all()
    )

收益率:

  "category": [
    "Object with name={'name': 'Business'} does not exist."
  ]

models.py:

import django.contrib.auth
from django.db import models
from django.conf import settings

class Profile(models.Model):
    display_name = models.CharField(max_length=255)
    user = models.OneToOneField(settings.AUTH_USER_MODEL)

class Category(models.Model):
    name = models.CharField(max_length=50, unique=True)
    description = models.CharField(max_length=200)

class Listing(models.Model):
    title = models.CharField(max_length=50, unique=True)
    category = models.ForeignKey(Category, related_name='listings', null=True)
    owners = models.ManyToManyField(
        Profile,
        related_name='owned_listings',
        db_table='profile_listing',
        blank=True
    )

serializers.py:

import logging
import django.contrib.auth
from rest_framework import serializers
import myapp.models as models

logger = logging.getLogger('mylogger')

class ShortUserSerializer(serializers.ModelSerializer):
    class Meta:
        model = django.contrib.auth.models.User
        fields = ('username', 'email')

class ProfileSerializer(serializers.ModelSerializer):
    user = ShortUserSerializer()
    class Meta:
        model = models.Profile
        fields = ('user', 'display_name')
        read_only = ('display_name',)

class CategorySerializer(serializers.ModelSerializer):
    class Meta:
        model = models.Category
        fields = ('name', 'description')
        read_only = ('description',)


class ListingSerializer(serializers.ModelSerializer):
    owners = ProfileSerializer(required=False, many=True)
    # TODO: how to indicate that this should look for an existing category?
    category = CategorySerializer(required=False, validators=[])

    class Meta:
        model = models.Listing
        depth = 2

    def validate(self, data):
        logger.info('inside ListingSerializer validate')
        return data

    def create(self, validated_data):
        logger.info('inside ListingSerializer.create')
        # not even getting this far...

views.py:

import logging

from django.http import HttpResponse
from django.shortcuts import get_object_or_404

import django.contrib.auth

from rest_framework import viewsets
from rest_framework.response import Response

import myapp.serializers as serializers
import myapp.models as models


# Get an instance of a logger
logger = logging.getLogger('mylogger')

class CategoryViewSet(viewsets.ModelViewSet):
    queryset = models.Category.objects.all()
    serializer_class = serializers.CategorySerializer

class UserViewSet(viewsets.ModelViewSet):
    queryset = django.contrib.auth.models.User.objects.all()
    serializer_class = serializers.ShortUserSerializer

class ProfileViewSet(viewsets.ModelViewSet):
    queryset = models.Profile.objects.all()
    serializer_class = serializers.ProfileSerializer

class ListingViewSet(viewsets.ModelViewSet):
    logger.info('inside ListingSerializerViewSet')
    queryset = models.Listing.objects.all()
    serializer_class = serializers.ListingSerializer

完整示例:https://github.com/arw180/drf-example

Answer 1

这并不理想，但我确实找到了解决我的问题的解决方案（我正在等待接受它作为答案，希望其他人可以做得更好）。有两个部分：

首先，partial=True在初始化ListingSerializer( http://www.django-rest-framework.org/api-guide/serializers/#partial-updates )时使用参数。然后使用序列化器的validate方法获取输入数据对应的实际模型实例。

其次，明确移除验证的name领域CategorySerializer。这尤其糟糕，因为它影响的不仅仅是ListingSerializer.

忽略任一部分将导致在序列化程序实例化时抛出验证错误。

对 views.py 的修改：

class ListingViewSet(viewsets.ModelViewSet):
    queryset = models.Listing.objects.all()
    serializer_class = serializers.ListingSerializer

    def create(self, request):
        serializer = serializers.ListingSerializer(data=request.data,
            context={'request': request}, partial=True)
        if not serializer.is_valid():
            logger.error('%s' % serializer.errors)
            return Response(serializer.errors,
                  status=status.HTTP_400_BAD_REQUEST)

        serializer.save()
        return Response(serializer.data, status=status.HTTP_201_CREATED)

对 serializers.py 的修改：

class CategorySerializer(serializers.ModelSerializer):
    class Meta:
        model = models.Category
        fields = ('name', 'description')
        read_only = ('description',)
        # also need to explicitly remove validators for `name` field
        extra_kwargs = {
            'name': {
                'validators': []
            }
        }

class ListingSerializer(serializers.ModelSerializer):
    owners = ProfileSerializer(required=False, many=True)
    category = CategorySerializer(required=False)
    class Meta:
        model = models.Listing
        depth = 2

    def validate(self, data):
        # manually get the Category instance from the input data
        data['category'] = models.Category.objects.get(name=data['category']['name'])
        return data

    def create(self, validated_data):
        title = validated_data['title']

        listing = models.Listing(title=validated_data['title'],
            category=validated_data['category'])

        listing.save()

        if 'owners' in validated_data:
            logger.debug('owners: %s' % validated_data['owners'])
            for owner in validated_data['owners']:
                print ('adding owner: %s' % owner)
                listing.owners.add(owner)
        return listing

如果有人可以提出更好的解决方案（例如如何source使用 a使参数正常工作SlugRelatedField），我将稍等一下接受这个作为答案- 我在https://github 上有一个使用上述解决方案的工作示例如果您想进行实验，请访问 .com/arw180/drf-example。我也很想听听关于为什么这些extra_kwargs东西是必要的评论CategorySerializer- 为什么不像这样实例化它：category = CategorySerializer(required=False, validators=[])足够（在ListingSerializer）？更新：我认为这不起作用，因为唯一的验证器是从数据库约束中自动添加的，并且无论此处设置的任何显式验证器如何都会运行，如本答案所述：http : //iswwwup.com/t/3bf20dfabe1f/python- order-of-serializer-validation-in-django-rest-framework.html

Answer 2

转变CategorySerializer.create成一个update_or_create方法name

class CategorySerializer(serializers.ModelSerializer):

    ...

    # update_or_create on `name`
    def create(self, validated_data):
        try:
            self.instance = Category.objects.get(name=validated_data['name'])
            self.instance = self.update(self.instance, validated_data)
            assert self.instance is not None, (
                '`update()` did not return an object instance.'
            )
            return self.instance
        except Category.DoesNotExist:
            return super(CategorySerializer, self).create(validated_data)

    ...

我建议您在需要创建自定义功能时查看DRF源代码。

相关问题由创建者回答DRF：django-rest-framework 3.0 create or update in Nested Serializer

编辑

所以我仍然处于 DRF 2 思维模式中，其中嵌套的可写字段是自动处理的。您可以在这里阅读该主题：http ://www.django-rest-framework.org/topics/3.0-announcement/

我已经测试了以下代码并且它有效：

class CategorySerializer(serializers.ModelSerializer):
    class Meta:
        ...
        extra_kwargs = {
            'name': {'validators': []},
            'description': {'required': False},
        }


class ListingSerializer(serializers.ModelSerializer):

    ...

    def update_or_create_category(self, validated_data):
        data = validated_data.pop('category', None)
        if not data:
            return None

        category, created = models.Category.objects.update_or_create(
            name=data.pop('name'), defaults=data)

        validated_data['category'] = category

    def create(self, validated_data):
        self.update_or_create_category(validated_data)
        return super(ListingSerializer, self).create(validated_data)

    def update(self, instance, validated_data):
        self.update_or_create_category(validated_data)
        return super(ListingSerializer, self).update(instance, validated_data)

编辑

正确的使用方法SlugRelatedField是这样的，如果你想知道：

class ListingSerializer(serializers.ModelSerializer):

    ...

    # slug_field should be 'name', i.e. the name of the field on the related model
    category = serializers.SlugRelatedField(slug_field='name',
        queryset=models.Category.objects.all())

    ...