如何編寫自定義序列化程序？

Question

我想在使用DRF發回JSON響應之前執行一些數據操作。

形勢

我的模式是：

class ThirdParty(models.Model): 
    label = models.CharField(verbose_name=_("Third party label"), null=False, blank=False, default=DEFAUT_LABEL, max_length=255) 

class CashFlow(TimeStampedModel): 
    date = models.DateField(verbose_name=_("Due date"), null=True, blank=True) 
    forecasted_value = models.DecimalField(verbose_name=_("Forecasted value"), null=True, blank=True, max_digits=11, decimal_places=2) 
    third_party = models.ForeignKey(ThirdParty, null=False, blank=False, related_name='cashflows') 

目前我有兩個串行：

class CashFlowSerializer(serializers.ModelSerializer): 
    third_party = serializers.PrimaryKeyRelatedField(many=False, read_only=True, allow_null=True) 
    class Meta: 
     model = CashFlow 
     fields = ('id', 'date', 'forecasted_value', 'edited_value', 'third_party') 

class ThirdPartyReadSerializer(serializers.ModelSerializer): 
    cashflows = CashFlowSerializer(many=True, read_only=True) 
    class Meta: 
     model = ThirdParty 
     fields = ('id', 'label', 'category', 'cashflows',) 

我的第三方觀點是正確返回一個不錯的JSON爲：

{ 
     "id": 15, 
     "label": "Adeo", 
     "category": 7, 
     "cashflows": [ 
      { 
       "id": 1, 
       "date": "2016-11-01", 
       "forecasted_value": "2000.00", 
       "edited_value": null, 
       "third_party": 15 
      }, 
      { 
       "id": 2, 
       "date": "2017-01-17", 
       "forecasted_value": "3000.00", 
       "edited_value": null, 
       "third_party": 15 
      }, 
      { 
       "id": 3, 
       "date": "2017-01-31", 
       "forecasted_value": "1000.00", 
       "edited_value": null, 
       "third_party": 15 
      } 
     ] 
    } 

問題

我想按月對現金流進行分組並添加它們的值。 問題是：什麼是最好的辦法呢？

預期的結果是：

{ 
     "id": 15, 
     "label": "Adeo", 
     "category": 7, 
     "cashflows": [ 
      { 
       "date": "2016-11-01", 
       "forecasted_value": "2000.00", 
       "edited_value": null, 
       "third_party": 15 
      }, 
      { 
       "date": "2017-01-01", 
       "forecasted_value": "4000.00", 
       "third_party": 15 
      } 
     ] 
    } 

這將是一個只讀序列化。

Answer 1

使用串行的to_representation：

def to_representation(self, obj): 
    data = super().to_representation(obj) 
    # manipulate data['cashflows'] to group by month 
    return data