如何在Django Rest Framework的SlugRelatedField中按其他字段过滤？

我有一个包含多个ForeignKey关联的模型，想要构建一个POST API，允许提交关联对象的可读名称而非id。这似乎适合使用SlugRelatedField，但该字段无法根据JSON请求中的其他字段过滤查询集。

具体需求：

• subdivision的查询集需根据organization字段过滤
• prediction_model的查询集需同时根据organization和subdivision字段过滤

组织名称唯一，SlugField可正常使用；但分部名称不唯一，必须按组织过滤；预测模型需同时按组织和分部过滤。我未找到在Django Rest Framework中便捷实现链式过滤的方法，且扩展SlugRelatedField类无法获取完整数据结构，请问有可行方案吗？

代码示例：

class GenericReportSerializer(serializers.ModelSerializer):
    """
    GenericReportSerializer for POST.
    """
    organization = serializers.SlugRelatedField(slug_field='name', queryset=models.Organization.objects.all())
    subdivision = serializers.SlugRelatedField(slug_field='name', queryset=models.Subdivision.objects.all())  # queryset needs to be filtered by organization
    prediction_model = serializers.SlugRelatedField(slug_field='name', queryset=models.PredictionModel.objects.all())  # queryset needs to be filtered by both organization and subdivision

    class Meta:
        model = models.GenericReport
        fields = '__all__'  # use all fields

下面提供两种实用方案，均能实现根据请求中其他字段动态过滤SlugRelatedField的查询集。

方案一：自定义SlugRelatedField子类（通用型） #

通过重写get_queryset方法，从请求上下文获取提交的字段值，针对性过滤查询集，适合多个序列化器复用：

from rest_framework import serializers
from . import models

class FilteredSlugRelatedField(serializers.SlugRelatedField):
    def get_queryset(self):
        queryset = super().get_queryset()
        request = self.context.get('request')
        if not request or not request.data:
            return queryset
        
        # 针对subdivision字段：按organization过滤
        if self.field_name == 'subdivision':
            org_name = request.data.get('organization')
            if org_name:
                org = models.Organization.objects.get(name=org_name)
                return queryset.filter(organization=org)
        # 针对prediction_model字段：同时按organization和subdivision过滤
        elif self.field_name == 'prediction_model':
            org_name = request.data.get('organization')
            sub_name = request.data.get('subdivision')
            if org_name and sub_name:
                org = models.Organization.objects.get(name=org_name)
                sub = models.Subdivision.objects.get(name=sub_name, organization=org)
                return queryset.filter(organization=org, subdivision=sub)
        return queryset

class GenericReportSerializer(serializers.ModelSerializer):
    """
    GenericReportSerializer for POST.
    """
    organization = serializers.SlugRelatedField(slug_field='name', queryset=models.Organization.objects.all())
    subdivision = FilteredSlugRelatedField(slug_field='name', queryset=models.Subdivision.objects.all())
    prediction_model = FilteredSlugRelatedField(slug_field='name', queryset=models.PredictionModel.objects.all())

    class Meta:
        model = models.GenericReport
        fields = '__all__'

方案二：在序列化器__init__方法中动态设置查询集（针对性） #

直接在目标序列化器的初始化方法中，根据请求数据更新对应字段的查询集，逻辑更直观：

from rest_framework import serializers
from . import models

class GenericReportSerializer(serializers.ModelSerializer):
    """
    GenericReportSerializer for POST.
    """
    organization = serializers.SlugRelatedField(slug_field='name', queryset=models.Organization.objects.all())
    subdivision = serializers.SlugRelatedField(slug_field='name', queryset=models.Subdivision.objects.all())
    prediction_model = serializers.SlugRelatedField(slug_field='name', queryset=models.PredictionModel.objects.all())

    class Meta:
        model = models.GenericReport
        fields = '__all__'

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        request = self.context.get('request')
        # 仅在POST请求时处理动态过滤
        if request and request.method == 'POST' and request.data:
            # 过滤subdivision的查询集
            org_name = request.data.get('organization')
            if org_name:
                try:
                    org = models.Organization.objects.get(name=org_name)
                    self.fields['subdivision'].queryset = self.fields['subdivision'].queryset.filter(organization=org)
                except models.Organization.DoesNotExist:
                    pass  # 后续验证会自动返回名称不存在的错误

            # 过滤prediction_model的查询集
            sub_name = request.data.get('subdivision')
            if org_name and sub_name:
                try:
                    org = models.Organization.objects.get(name=org_name)
                    sub = models.Subdivision.objects.get(name=sub_name, organization=org)
                    self.fields['prediction_model'].queryset = self.fields['prediction_model'].queryset.filter(organization=org, subdivision=sub)
                except (models.Organization.DoesNotExist, models.Subdivision.DoesNotExist):
                    pass

关键注意点 #

• 确保视图将request传入序列化器的context，DRF默认视图已自动处理此逻辑。
• 异常无需额外捕获：如果提交的组织/分部名称不存在，DRF的默认验证流程会返回对应的错误提示。
• 字段顺序不影响：两种方案均直接从request.data中取值，无需依赖请求字段的提交顺序。

内容的提问来源于stack exchange，提问作者Scott