Django多对多字段查询：获取包含指定所有标签的Problem对象

Got it, let's break down how to get Problem objects that include every single tag in your specified list. Since the standard __in lookup only returns items with any of the tags, we need to adjust our approach to enforce an "all match" condition.

Approach 1: Chained Filter Calls (Simple & Intuitive) #

This method builds a query with consecutive filter() calls, each adding an AND condition for one of your target tags. Django will only return problems that satisfy all these conditions (i.e., have every tag in your list).

# Example using tag IDs
target_tag_ids = [2, 5, 7]  # Replace with your tag IDs
problems = Problem.objects.all()

for tag_id in target_tag_ids:
    problems = problems.filter(tags__id=tag_id)

# Or if you're working with Tag objects directly
target_tags = Tag.objects.filter(name__in=["Sorting", "Greedy"])
problems = Problem.objects.all()

for tag in target_tags:
    problems = problems.filter(tags=tag)

This works because each filter() narrows down the queryset to only problems that include the current tag. By the end, only problems with all your target tags remain.

Approach 2: Annotate with Count (Efficient for Large Datasets) #

If you're dealing with a lot of tags or a big database, this method is more performant—it runs a single database query instead of multiple joins. We'll annotate each problem with the number of matching target tags, then filter to keep only those where the count equals the total number of target tags.

from django.db.models import Count, Q

target_tag_ids = [2, 5, 7]
total_target_tags = len(target_tag_ids)

problems = Problem.objects.annotate(
    # Count how many of the target tags each problem has
    matched_tag_count=Count(
        'tags',
        filter=Q(tags__id__in=target_tag_ids),
        distinct=True  # Avoid counting duplicate tags (optional, safe to include)
    )
).filter(matched_tag_count=total_target_tags)

The distinct=True is optional but useful if your Problem-Tag relationship could have duplicates (though ManyToMany fields usually prevent this by default).

Quick Notes #

• If your target tag list is empty, decide whether to return all problems or an empty queryset based on your business logic.
• The chained filter approach is great for small tag lists and readability, while the annotate method scales better for larger datasets.

内容的提问来源于stack exchange，提问作者MD. Khairul Basar