Nested Sets model or some other model for representing hierarchy?

I have a project where I hope to store hierarchical information (specifically, categories and subcategories) in what is basically a flat database system (in short, it's a collection of XML records). I'd like to store information about categories and subcategories in the system:

• Animals
  • Invertebrates
  • Vertebrates
• Weather
• Buildings
  • Skyscrapers
  • Historic buildings

...and so forth.

Everything in the system, for better or worse, is stored as an XML record; this is just how the storage system works.

Which means that each category in the system is also stored as an XML record, like so:

<record id="12345">
   <attribute name="Skyscrapers" />
   <attribute type="Category" />
</record>

So I am wondering how to implement a hierarchy under these constraints.

I'm used to data storage in a relational database. In those cases I pretty much always use the nested set model. It seems like this wouldn't be a good choice in this case because:

1. Each time you insert an item, you have to change the right and/or left values for many of the nodes. I can't do a bulk update on the XML files, so I'd have to update each one individually.
2. Although there are search functions that let me filter by "less-than" or "greater-than" (so I could in theory pull only the relevant child nodes or parent nodes of a given category), I can't order the XML records by attributes. So I'd have to retrieve all of the documents, transform them into a list of objects that can be sorted (in this case with Python) and then sort them using a lambda function.

Since my data storage model isn't significantly different than storing data using NoSQL I was wondering if anyone using that storage mechanism has come up with a good t开发者_C百科rick for handling and storing hierarchical data.

---

This class (based on the Peewee ORM) allows you to handle hierarchical data with a flat relational database (PostgreSQL, MySQL and SQLite are supported):

https://github.com/mathieurodic/peewee-tree/blob/master/node.py

You can make a few changes in the class methods so the changes are also applied to the XML file you are manipulating.

---

Not sure how applicable this would be for your use case, but as a thought, maybe using Beautiful Soup will help. Perhaps it's default hierarchical representation will be sufficient for your needs.