Django按子查询内带注释的计数筛选

下面是一个抽象和简化的示例

假设我想要获得作者，并在类别中注释最小数量的书籍，如果它大于三本

书籍和作者模型与外键字段无关（记住，抽象和简化，这是有原因的）：

下面是我可以复制的最简单的查询：

(Author.objects
 .annotate(min_valuable_count=Subquery(
    Book.objects
        .filter(author_name=OuterRef('name'))
        .annotate(cnt=Count('book_category'))
        .filter(cnt__gt=3)
        .order_by('cnt')
        .values('cnt')[:1],
    output_field=models.IntegerField()
 ))
)

我得到一个错误：

psycopg2.ProgrammingError: missing FROM-clause entry for table "U0"
LINE 1: ... "core_author" GROUP BY "core_author"."id", "U0"."id" ...
                                                       ^

下面是SQL：

SELECT "core_author"."id", "core_author"."name", (
    SELECT COUNT(U0."book_category") AS "cnt" 
    FROM "core_book" U0 WHERE U0."id" = ("core_author"."chat_id") 
    GROUP BY U0."id" HAVING COUNT(U0."book_category") > 3 
    ORDER BY "cnt" ASC  LIMIT 1) 
AS "min_valuable_count" 
FROM "core_author" 
GROUP BY "core_author"."id", "U0"."id"

更新#1

我发现删除

.filter（cnt_ugt=3）

会删除无法访问

U0

的最后一个

groupby

：

SELECT "core_author"."id", "core_author"."name", (
    SELECT COUNT(U0."book_category") AS "cnt" 
    FROM "core_book" U0 WHERE U0."id" = ("core_author"."chat_id") 
    GROUP BY U0."id"
    ORDER BY "cnt" ASC  LIMIT 1) 
AS "min_valuable_count" 
FROM "core_author" 

---

是否有任何方法可以在不删除子查询中的

.filter（cnt_ugt=3）

的情况下删除外部查询中的

GROUP BY

。以下是目前的工作：

min_valuable_count_qs = Subquery(
    Book.objects
        .filter(author_name=OuterRef('name'))
        .annotate(cnt=Count('book_category'))
        .filter(cnt__gt=3)
        .order_by('cnt')
        .values('cnt')[:1],
    output_field=models.IntegerField()
)

min_valuable_count_qs.contains_aggregate = False

qs = Author.objects.annotate(min_valuable_count=min_valuable_count_qs))

min_valuable_count_qs = Subquery(
    Book.objects
        .filter(author_name=OuterRef('name'))
        .annotate(cnt=Count('book_category'))
        .filter(cnt__gt=3)
        .order_by('cnt')
        .values('cnt')[:1],
    output_field=models.IntegerField()
)

min_valuable_count_qs.contains_aggregate = False

qs = Author.objects.annotate(min_valuable_count=min_valuable_count_qs))