如何在DRF中将已验证标题传递给自定义SlugValidator类？

方案一：用序列化器的validate_slug方法替代字段级验证器 #

字段级验证器本身无法跨字段获取已验证值，直接把SlugsValidator的逻辑迁移到序列化器的validate_slug方法即可——这个方法会在字段自身的验证器执行完成后运行，此时已经能拿到通过TitleValidator验证后的title值：

from rest_framework import serializers
from rest_framework.exceptions import ValidationError

class TitleValidator:
    MIN_TITLE_LENGTH = 20
    
    def __call__(self, title: str):
        if len(title) < self.MIN_TITLE_LENGTH:
            raise ValidationError(f"Min title length is {self.MIN_TITLE_LENGTH}")
        return title

class PostSerializer(serializers.ModelSerializer):
    title = serializers.CharField(validators=[TitleValidator()])
    slug = serializers.CharField(max_length=100, required=False)

    def validate_slug(self, slug):
        # 直接获取已验证的title值
        validated_title = self.validated_data.get('title')
        # 执行原SlugsValidator的业务逻辑
        if not slug:
            # 示例：用title自动生成slug
            slug = validated_title.lower().replace(' ', '-')
        # 可添加其他slug格式验证
        if not slug.islower() or ' ' in slug:
            raise ValidationError("Slug必须为小写且不含空格")
        return slug

方案二：给自定义验证器传递序列化器上下文 #

如果坚持要用类形式的验证器，可以在序列化器初始化时，把自身实例传递给SlugsValidator的上下文，让验证器能访问到已验证的字段数据：

from rest_framework import serializers
from rest_framework.exceptions import ValidationError

class TitleValidator:
    MIN_TITLE_LENGTH = 20
    
    def __call__(self, title: str):
        if len(title) < self.MIN_TITLE_LENGTH:
            raise ValidationError(f"Min title length is {self.MIN_TITLE_LENGTH}")
        return title

class SlugsValidator:
    def __init__(self, dependent_field):
        self.dependent_field = dependent_field
        self.context = {}
    
    def __call__(self, slug):
        # 从上下文获取序列化器实例，进而拿到已验证的title
        serializer = self.context.get('serializer')
        validated_title = serializer.validated_data.get(self.dependent_field)
        # 执行你的验证逻辑
        if not slug:
            slug = validated_title.lower().replace(' ', '-')
        # 其他验证规则
        return slug

class PostSerializer(serializers.ModelSerializer):
    title = serializers.CharField(validators=[TitleValidator()])
    slug = serializers.CharField(
        max_length=100, 
        required=False,
        validators=[SlugsValidator(dependent_field='title')]
    )

    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        # 将序列化器实例传递给SlugsValidator的上下文
        for validator in self.fields['slug'].validators:
            if isinstance(validator, SlugsValidator):
                validator.context = {'serializer': self}

注：DRF会按字段声明顺序执行验证，title先于slug完成验证，所以validated_data中会存在已通过校验的title值。

补充说明 #

字段级验证器的设计初衷是独立验证单个字段，跨字段的验证逻辑更适合放在序列化器的validate_<field>或全局validate方法中。上述两种方案都无需使用Meta类的全局验证器，其中方案一的写法更简洁，符合DRF的常规开发习惯。

内容的提问来源于stack exchange，提问作者Andrew