filter tags of django-taggit in Django's Queryset

Question

Having the following models:

class Post(models.Model):
    title = models.CharField(max_length=250)
    tags = TaggableManager()

and the data are：

**post.title**                          **post.tags**
Django By Example                       python,django,web
Who was Django Reinhardt                python,django,
Test-Driven Development with Python     python,web
Python for Data Analysis                python,data         
Learning Python                         python
Programming Python                      python
Automate the Boring Stuff with Python   python

I try to code below

>>> alist=Post.objects.filter(tags__name__in=["data","python"])
>>> for i in alist.annotate(sam_tags=Count('tags')):
...     print("\n---",i)
...     print("tags of it : ",i.tags.all())
...     print("value of sam_tags : ",i.sam_tags)
... 

--- Django By Example
tags of it :  [<Tag: django>, <Tag: python>, <Tag: web>]
value of sam_tags :  1

--- Who was Django Reinhardt
tags of it :  [<Tag: django>, <Tag: python>]
value of sam_tags :  1

--- Automate the Boring Stuff with Python
tags of it :  [<Tag: python>]
value of sam_tags :  1

--- Test-Driven Development with Python
tags of it :  [<Tag: python>, <Tag: web>]
value of sam_tags :  1

--- Learning Python
tags of it :  [<Tag: python>]
value of sam_tags :  1

--- Python for Data Analysis
tags of it :  [<Tag: data>, <Tag: python>]
value of sam_tags :  2

--- Programming Python
tags of it :  [<Tag: python>]
value of sam_tags :  1
>>> 

why is the value of slist[0].sam_tags (post:Django By Example) equal to 1?

I think the post object of (post:Django By Example) has three tags [python,django and web] after reading the Django's documentation.

https://docs.djangoproject.com/en/1.10/ref/models/querysets/#count

It said that Count(expression) returns the number of objects that are related through the provided expression. so the code

>>>alist[0].tags.all() 
[<Tag: django>, <Tag: python>, <Tag: web>]

shows there are three tags in alist[0].tags,

>>> slist=alist.annotate(sam_tags=Count('tags'))
>>> slist[0].tags.all()
[<Tag: django>, <Tag: python>, <Tag: web>]
>>> slist[0].sam_tags
1

but here I get the value 1,
why?

I understand that Django is counting only the "python" and "data" tags which I included in my filter clause - other tags are not counted.

The output of "slist[0].tags.all()" shows that slist[0] has three tags related with itself. Because the documentation of django says that Count(expression) returns the number of objects that are related through the provided expression, slist[0].sam_tags should be 3 according to the documentation, but django-taggit make slist[0].sam_tags to be 1.

So what I really want to know is how django-taggit lets Count(expression) in the filter clause only calculate the number of tags in the filter condition .

Answer 1

Django is counting only the python and data tags which you included in your filter clause - other tags are not counted. (Note that the only example with sam_tags of 2 is the one tagged both data and python.) This is probably unexpected behavior, but makes sense if you consider how the underlying SQL is executed. See this example from a similar schema to yours:

>>> a = Article.objects.filter(tags__slug__in=['python']).annotate(num_tags=Count('tags'))[0]
>>> a.num_tags
1
>>> a.tags.count()
2

If I change the filter clause to filter on somethign other than tags, it behaves as expected:

>>> Article.objects.filter(pk=a.pk).annotate(num_tags=Count('tags'))[0].num_tags
2
>>> Article.objects.filter(pk=a.pk).annotate(num_tags=Count('tags'))[0].tags.count()
2

Answer 2

it's relatively straight forward to look into the query by print(similar_post.query) we can see the code only counts the tag_id which belongs to the tags of the post is being shown.

SELECT "blog_post"."id", "blog_post"."title", "blog_post"."slug", "blog_post"."author_id", "blog_post"."body", "blog_post"."publish", "blog_post"."created", "blog_post"."updated", "blog_post"."status", COUNT("taggit_taggeditem"."tag_id") AS "same_tags" FROM "blog_post"

INNER JOIN "taggit_taggeditem" ON ("blog_post"."id" = "taggit_taggeditem"."object_id" AND ("taggit_taggeditem"."content_type_id" = 7))

WHERE ("blog_post"."status" = published

   AND "taggit_taggeditem"."tag_id" IN (SELECT DISTINCT U0."id" FROM "taggit_tag" 
                                                        U0 INNER JOIN "taggit_taggeditem" U1 ON (U0."id" = U1."tag_id") 
                                                        INNER JOIN "django_content_type" U2 ON (U1."content_type_id" = U2."id") 
                                                        WHERE (U2."app_label" = blog AND U2."model" = post AND U1."object_id" = 8)) 

   AND NOT ("blog_post"."id" = 8)) 

GROUP BY "blog_post"."id", "blog_post"."title", "blog_post"."slug", "blog_post"."author_id", "blog_post"."body", "blog_post"."publish", "blog_post"."created", "blog_post"."updated", "blog_post"."status"

ORDER BY "same_tags" DESC, "blog_post"."publish" DESC LIMIT 4