count() 方法在多对多条件查询distinct()后，会返回所有的数，不是distinct()后实际的数量

[FatXiaoBai]

Describe the bug
多条件复合查询，含多对多查询，调用distinct()后再使用count()统计数

To Reproduce
应该为distinct()后打印出来的实列数

Expected behavior
实际返回了所有多对多的匹配结果，没有去处重复

Additional context
python3.12
tortoise-orm[asyncpg]==0.25.3

[waketzheng]

Show your code

[FatXiaoBai]

    filters = {}
    if query.username:
        filters["username__icontains"] = query.username
    if query.nickname:
        filters["nickname__icontains"] = query.nickname
    filterUser = User.filter(**filters)
    if query.role:
        filterUser = filterUser.filter(role__urid__in=query.role).distinct() #这里多对多去重
    filterUser = filterUser.order_by("createtime")
    if query.page and query.pageSize:
        offset = (query.page - 1) * query.pageSize
        data = await filterUser.offset(offset).limit(query.pageSize)
        total = await filterUser.count() #统计的数据不对，之前有distinct()，数据返回时对的，统计不对
        return data, total

[seladb]

@FatXiaoBai it'd be helpful if you can translate it to English, for people who don't understand Chinese 🙏

[miss85246]

@FatXiaoBai it'd be helpful if you can translate it to English, for people who don't understand Chinese 🙏

Describe the bug
Multi-conditional compound queries, including many-to-many queries, call distinct() and then use the count() statistic

To Reproduce
It should be the number of columns printed after distinct().

Expected behavior
All many-to-many matches are actually returned, with no duplicates going anywhere

Additional context
python3.12
tortoise-orm[asyncpg]==0.25.3

[seladb]

Thank you @miss85246 for the translation! 🙏

Multi-conditional compound queries, including many-to-many queries, call distinct() and then use the count() statistic

I'm not sure I understand the bug, does it mean that when using distinct() and count() then distinct() is not honored?

It should be the number of columns printed after distinct().

Do you mean the number of rows maybe? 🤔

[miss85246]

@seladb Yes, that’s likely what he meant. The workflow he described is: query based on multiple conditions -> call distinct() to remove duplicates -> apply sorting -> perform pagination. Finally, he claimed that the resulting count is incorrect.

I don't have more specific details either. However, looking at his code, I have a feeling it might be an issue with how he chained the methods. Perhaps he should be calling total = await filterUser.distinct().count() instead.

[seladb]

Thanks @miss85246 ! I will check out the code example provided and see if I can reproduce the issue