如何在DRF中扁平化嵌套Serializer？

问题背景 #

需要将DRF中嵌套的Serializer输出扁平化，但官方文档里没找到现成实现方法。

当前输出 #

{
    "user_inventory": "UOHvaxFa11R5Z0bPYuihP0RKocn2",
    "quantity": 1,
    "player": {
        "card_id": "c69c0808328fdc3e3f3ee8b9b7d4a7f8",
        "game": "MLB The Show 22",
        "name": "Jesus Tinoco",
        "all_positions": [
            "CP"
        ]
    }
}

期望输出 #

{
    "user_inventory": "UOHvaxFa11R5Z0bPYuihP0RKocn2",
    "quantity": 1,
    "card_id": "c69c0808328fdc3e3f3ee8b9b7d4a7f8",
    "game": "MLB The Show 22",
    "name": "Jesus Tinoco",
    "all_positions": [
        "CP"
    ]
}

现有代码 #

序列化器定义

class PlayerProfileSerializer(serializers.ModelSerializer):
    class Meta:
        model = PlayerProfile
        fields = (
            'card_id',
            'game',
            'name',
            'all_positions',
        )

class UserInventoryItemSerializer(serializers.ModelSerializer):
    player = PlayerProfileSerializer()
    class Meta:
        model = UserInventoryItem
        fields = (
            'user_inventory',
            'quantity',
            'player',
        )

视图代码

class OwnedInventoryView(viewsets.ModelViewSet):
    serializer_class = UserInventoryItemSerializer
    filterset_class = UserInventoryItemFilter

    def get_queryset(self):
        order_by = self.request.query_params.get('order_by', '')

        if order_by:
            order_by_name = order_by.split(' ')[1]
            order_by_sign = order_by.split(' ')[0]
            order_by_sign = '' if order_by_sign == 'asc' else '-'
            return UserInventoryItem.objects.filter(user_inventory=self.kwargs['user_inventory_pk']).order_by(order_by_sign + order_by_name)

        return UserInventoryItem.objects.filter(user_inventory=self.kwargs['user_inventory_pk'])

解决方案 #

方法1：直接用source参数映射字段 #

这是最直接的方式，无需嵌套Serializer，直接通过外键关联映射PlayerProfile的字段：

class UserInventoryItemSerializer(serializers.ModelSerializer):
    card_id = serializers.CharField(source='player.card_id')
    game = serializers.CharField(source='player.game')
    name = serializers.CharField(source='player.name')
    all_positions = serializers.ListField(source='player.all_positions')

    class Meta:
        model = UserInventoryItem
        fields = (
            'user_inventory',
            'quantity',
            'card_id',
            'game',
            'name',
            'all_positions',
        )

注意要根据PlayerProfile中字段的实际类型选择对应Serializer字段，比如all_positions如果是数组类型用ListField，其他类型按需调整即可。

方法2：重写to_representation方法 #

如果PlayerProfile字段较多，手动映射麻烦，可重写序列化器的to_representation方法自动展开嵌套字段：

class UserInventoryItemSerializer(serializers.ModelSerializer):
    class Meta:
        model = UserInventoryItem
        fields = (
            'user_inventory',
            'quantity',
        )

    def to_representation(self, instance):
        # 获取基础序列化数据
        data = super().to_representation(instance)
        # 序列化player对象并合并到根数据
        player_data = PlayerProfileSerializer(instance.player).data
        data.update(player_data)
        return data

这种方式更灵活，自动适配PlayerProfile的字段变化。如果需要支持反序列化（比如创建/更新操作），可以保留player = PlayerProfileSerializer()的定义，或在to_internal_value方法中处理反序列化逻辑。

方法3：使用SerializerMethodField（适合需额外处理的字段） #

如果仅需展开部分字段，且要对字段做自定义处理，可使用SerializerMethodField：

class UserInventoryItemSerializer(serializers.ModelSerializer):
    card_id = serializers.SerializerMethodField()
    game = serializers.SerializerMethodField()
    name = serializers.SerializerMethodField()
    all_positions = serializers.SerializerMethodField()

    class Meta:
        model = UserInventoryItem
        fields = (
            'user_inventory',
            'quantity',
            'card_id',
            'game',
            'name',
            'all_positions',
        )

    def get_card_id(self, obj):
        return obj.player.card_id

    def get_game(self, obj):
        return obj.player.game

    def get_name(self, obj):
        return obj.player.name

    def get_all_positions(self, obj):
        return obj.player.all_positions

每个字段对应一个get_xxx方法，可在方法内添加字段格式化、过滤等额外逻辑。

内容的提问来源于stack exchange，提问作者JeremyE