如何在Django Rest Framework中正确序列化关联的Publisher字段

我有两个Django模型：

• Book（包含Wagtail StreamField）：

class Book(models.Model):
    title = models.CharField(max_length=255)
    cover_image = StreamField(...)
    ...

• Publisher（标准Django模型）：

class Publisher(models.Model):
    name = models.CharField(max_length=255)
    ...

为了在Wagtail后台用PublisherChooser选出版社，我给Book加了字段并配置了后台面板：

class Book(models.Model):
    title = models.CharField(max_length=255)
    cover_image = StreamField(...)
    ...
    publisher = models.CharField(max_length=10)

    panels = [
        FieldPanel("title"),
        FieldPanel("cover_image"),
        FieldPanel("publisher", widget=PublisherChooser),
    ]

现在后台创建Book时PublisherChooser能正常用，但序列化Book模型出问题：我写了PublisherSerializer和BookSerializer，其中BookSerializer用了PrimaryKeyRelatedField还改写了to_representation方法，结果返回的publisher字段数据异常：

"publisher": {
            "name": "<django.db.models.query_utils.DeferredAttribute object at 0x7f9ffa146310>",
}

需要解决这个序列化异常，输出正确的关联数据。

问题出在Book模型的publisher字段是CharField类型，不是ForeignKey，但序列化时却按关联模型的逻辑处理，导致没拿到实际的Publisher实例，反而直接引用了模型的字段属性（DeferredAttribute是Django模型字段的内部类，代表未实例化的字段）。

方案1：把Book的publisher字段改成ForeignKey（推荐） #

这是最规范的做法，符合Django关联模型的设计：

1. 更新Book模型：

from django.db import models

class Book(models.Model):
    title = models.CharField(max_length=255)
    cover_image = StreamField(...)
    # 替换原来的CharField为ForeignKey
    publisher = models.ForeignKey(
        Publisher,
        on_delete=models.SET_NULL,
        null=True,
        blank=True,
        related_name="books"
    )

    panels = [
        FieldPanel("title"),
        FieldPanel("cover_image"),
        # Wagtail的ChooserPanel支持ForeignKey，自定义的PublisherChooser也能继续用
        FieldPanel("publisher", widget=PublisherChooser),
    ]

2. 更新序列化器：
   直接用嵌套序列化器，不用手动改to_representation：

from rest_framework import serializers
from .models import Book, Publisher

class PublisherSerializer(serializers.ModelSerializer):
    class Meta:
        model = Publisher
        fields = ["id", "name"]

class BookSerializer(serializers.ModelSerializer):
    # 直接嵌套PublisherSerializer
    publisher = PublisherSerializer(read_only=True)

    class Meta:
        model = Book
        fields = ["title", "cover_image", "publisher"]

这样序列化后就能正确输出publisher的完整数据，不会再出现DeferredAttribute的错误。

方案2：不修改Book模型字段类型（兼容现有数据） #

如果因为已有数据不能改字段类型，就在序列化时手动通过ID查Publisher实例：

1. 修改BookSerializer的to_representation方法：
   确保先从publisher字段存的ID查到实际的Publisher实例，再序列化：

from rest_framework import serializers
from .models import Book, Publisher

class PublisherSerializer(serializers.ModelSerializer):
    class Meta:
        model = Publisher
        fields = ["id", "name"]

class BookSerializer(serializers.ModelSerializer):
    class Meta:
        model = Book
        fields = ["title", "cover_image", "publisher"]

    def to_representation(self, instance):
        representation = super().to_representation(instance)
        # 拿到publisher字段存的ID字符串
        publisher_id = representation.get("publisher")
        if publisher_id:
            try:
                # 查询对应的Publisher实例
                publisher = Publisher.objects.get(id=publisher_id)
                # 用PublisherSerializer序列化后替换原字段值
                representation["publisher"] = PublisherSerializer(publisher).data
            except Publisher.DoesNotExist:
                # 处理ID不存在的情况
                representation["publisher"] = {"name": "未知出版社"}
        else:
            representation["publisher"] = None
        return representation

注意：这种方案要保证CharField存的publisher值是有效的Publisher ID字符串，不然会查询失败，建议保留异常处理逻辑。

内容的提问来源于stack exchange，提问作者Faith Baghan