DRF - 在 M2M 通过模型中创建新实例时出错

Question

我有以下两种型号：

class User(models.Model):
user_id = models.CharField(
max_length=129,
unique=True,
)
user_article = models.ManyToManyField(
Article,
through="UserArticle",
)
occupation = models.CharField(max_length=100, default='null')
def __str__(self):
return self.user_id

和

class Article(models.Model):
uuid = models.UUIDField(editable=False, unique=True)
company = models.ForeignKey(
Company,
on_delete=models.PROTECT,
related_name='article_company_id',
)
articleType = models.ForeignKey(
ArticleType,
on_delete=models.PROTECT,
related_name='type',
)    
date_inserted = models.DateField()    
def __str__(self):
return self.uuid

使用多对多关系建模，通过模型：

class UserArticle(models.Model):    
user = models.ForeignKey(User, to_field='user_id',
on_delete=models.PROTECT,)
article = models.ForeignKey(Article, to_field='uuid',
on_delete=models.PROTECT,)
posted_as = ArrayField(
models.CharField(max_length=100, blank=True),)
post_date = models.DateField()
class Meta:
db_table = "core_user_articles"

以下是我的观点：

class BatchUserArticleList(mixins.ListModelMixin,
mixins.CreateModelMixin,
generics.GenericAPIView):
queryset = UserArticle.objects.all()
serializer_class = BatchUserArticleSerializer
def create(self, request, *args, **kwargs):
serializer = BatchUserArticleSerializer(data=request.data)
if not serializer.is_valid():
return response.Response({'Message': 'POST failed',
'Errors': serializer.errors},
status.HTTP_400_BAD_REQUEST)
self.perform_create(serializer)  # equal to serializer.save()
return response.Response(serializer.data, status.HTTP_201_CREATED)
def post(self, request, *args, **kwargs):
return self.create(request, *args, **kwargs)

我面临的问题是当我想在 M2M 表中使用以下格式的 POST 数据时：

{
"posted_as": ["news"],
"post_date": "2020-05-26",
"user": "jhtpo9jkj4WVQc0000GXk0zkkhv7u",
"article": [
"11111111",
"22222222"
]
}

上面包含许多文章的列表，所以我在serializer中使用了一个custom field来提取每个article，创建一个新的UserArticle对象并使用bulk_create将其插入到我的 M2M 表中。我认为当传入的数据不完全映射到数据库模型时，这就是要走的路，但我可能是错的。因此，如果您发现这种方法有什么不同之处，请发表评论。

下面是序列化程序：

class BatchUserArticleSerializer(serializers.ModelSerializer):
article= ArticleField(source='*') #custom field
class Meta:
model = UserArticle
fields = ('posted_as', 'post_date', 'user', 'article')
def validate(self, data):    
post_date = data['post_date']
if post_date != date.today():
raise serializers.ValidationError(
'post_date: post_date is not valid',
)
return data
def create(self, validated_data):
post_as = list(map(lambda item: item, validated_data['posted_as']))
post_date = validated_data['post_date']
user = validated_data['user']
list_of_articles = validated_data['article']            
user_object = User.objects.get(user_id=user)
articles_objects = list(map(lambda res: Article.objects.get(uuid=res), list_of_articles))    
user_articles_to_insert = list(map(
lambda article: UserArticle(
posted_as=posted_as,
post_date=post_date,
article=article,
user=user_object),
articles_objects))
try:
created_user_articles = UserArticles.objects.bulk_create(user_articles_to_insert)
for res in created_user_articles:
res.save()
return created_user_articles
except Exception as error:
raise Exception('Something went wrong: {0}'.format(error))

和

class ArticleField(serializers.Field):
def to_representation(self, value):
resource_repr = [value.article]
return resource_repr
def to_internal_value(self, data):
internal_repr = {
'article': data
}
return internal_repr

这似乎工作正常，因为我可以看到数据被正确插入UserArticle表中：

id | posted_as | post_date | user | article
1  | news      | 2020-05-26 | jhtpo9jkj4WVQc0000GXk0zkkhv7u | 11111111
2  | news      | 2020-05-26 | jhtpo9jkj4WVQc0000GXk0zkkhv7u | 22222222

当代码到达此行时，问题就来了：

response.Response(serializer.data, status.HTTP_201_CREATED)

更具体地说，我得到的错误是：

AttributeError: Got AttributeError when attempting to get a value for field `posted_as` on serializer `BatchUserArticleSerializer`.
The serializer field might be named incorrectly and not match any attribute or key on the `list` instance. Original exception text was: 'list' object has no attribute 'posted_as'.

原始异常错误在fields.pyDRF 源中def get_attribute(instance, attrs)函数的instance = getattr(instance, attr)行引发。

我在这里错过了什么？

Answer 1

首先，没有理由为每个批量创建的实例调用save方法。

第二个是例外的原因。调用视图集方法create。它调用序列化程序create方法，该方法必须只返回一个实例(创建的对象(。但是您的序列化程序返回列表created_user_articles.列表真的没有字段posted_as.

因此，有两种方法可以修复它。

1. 第一个是覆盖视图中create方法，以更改数据表示的方式。例如，对响应数据使用另一个序列化程序：

   def create(self, request, *args, **kwargs):
   serializer = self.get_serializer(data=request.data)
   serializer.is_valid(raise_exception=True)
   created_user_articles = self.perform_create(serializer)
   # use another way to get representation
   response_data = AnotherUserArticleSerializer(created_user_articles, many=True).data
   return Response(response_data, status=status.HTTP_201_CREATED, headers=headers)
   def perform_create(self, serializer):
   # add return to get created objects
   return serializer.save()
   

2. 第二个是序列化程序create方法中仅返回一个实例。