model relationship design for Blog app in Wagtail site - wagtail

Question

Below models.py to build a blog in wagtail site is from this post.
class BlogPage(Page):
description = models.CharField(max_length=255, blank=True,)
content_panels = Page.content_panels + [FieldPanel("description", classname="full")]
class PostPage(Page):
header_image = models.ForeignKey(
"wagtailimages.Image",
null=True,
blank=True,
on_delete=models.SET_NULL,
related_name="+",
)
tags = ClusterTaggableManager(through="blog.PostPageTag", blank=True)
content_panels = Page.content_panels + [
ImageChooserPanel("header_image"),
InlinePanel("categories", label="category"),
FieldPanel("tags"),
]
class PostPageBlogCategory(models.Model):
page = ParentalKey(
"blog.PostPage", on_delete=models.CASCADE, related_name="categories"
)
blog_category = models.ForeignKey(
"blog.BlogCategory", on_delete=models.CASCADE, related_name="post_pages"
)
panels = [
SnippetChooserPanel("blog_category"),
]
class Meta:
unique_together = ("page", "blog_category")
#register_snippet
class BlogCategory(models.Model):
name = models.CharField(max_length=255)
slug = models.SlugField(unique=True, max_length=80)
panels = [
FieldPanel("name"),
FieldPanel("slug"),
]
def __str__(self):
return self.name
class Meta:
verbose_name = "Category"
verbose_name_plural = "Categories"
class PostPageTag(TaggedItemBase):
content_object = ParentalKey("PostPage", related_name="post_tags")
#register_snippet
class Tag(TaggitTag):
class Meta:
proxy = True
I am wondering, what are the major reasons to introduce extra Intermediary model (class PostPageBlogCategory(models.Model): & class PostPageTag(TaggedItemBase):) to link PostPage to Category & Tag?
Why not just simply use ParentalForeignkey or ParentalManyToManyKey ?

Answer 1

The short answer is that Django's ManyToManyField has some limitations in how the relationship needs to be built when models on either end to not yet exist, adding an 'intermediary model' helps to work around this.
Longer Answer
As with anything in software, there are multiple ways to do something and each has its own pros & cons. Django's built in field is simple and lets you do lots of powerful things to represent relational data but that simplicity comes with some reduced flexibility in how these relationships are managed.
One of the main goals of the Wagtail CMS Admin interface is the ability to work as though the data has been created (including relationships) before actually clicking 'save'. While this may seem simple at first glance, getting to that point requires a bit of nuance under the hood once you start to consider relational data.
Wagtail comes built in with a very powerful library called django-modelcluster which has been purpose built for many of the cases where you want to work with relational data without having all the bits in the DB first.
Each Wagtail Page actually inherits the modelcluster.models.ClusterableModel, which is why some of the features in the blog post seem to work in the editor, even when the DB entries have not yet been saved.
On the blog post you linked, there is a section towards the end with the heading 'ParentalKey' that further explains this nuance and how just using Django's basic approach has some draw backs.
On the Django docs for many-to-many relationships, have a read through and note that each individual model instance must be in the database first and only then can you 'link' the two instances with a second update on each.