Question

Ok so I’m starting on normalising my database. Currently I have one model “Products” which is populated with about 60,000 products via a data feed (XML), which contains a product with a category name and a merchant name. I want to split these into 3 models; products, categories and merchants.

Each product has one category and one merchant so the natural idea is to create these models:

category_id | category_name

merchant_id | merchant_name

I can work out the code to associate between the models i.e. has_one, belongs_to etc but I’m struggling to work out to automatically associate a new Product with a category and a merchant programatically.

I’ve seen examples in books where your start with an empty database and that seems pretty straightforward. However, I’m starting off with a full database and a list of Category names.

Here is my product creation statement which is working great:

Product.create(:name => node.xpath("./text/name/text()").inner_text.downcase, 
                   :description => node.xpath("./text/desc/text()").inner_text,
                   :brand => node.xpath("./brand/text()").inner_text,
                   :merchant => node.xpath("../@name").inner_text,
                   :category => node.xpath("./cat/text()").inner_text.downcase,
                   :price => "£" + node.xpath("./price/btext()").inner_text)

Would I need to do something like this, see the :category line, (i know the following is wrong btw!)…

Product.create(:name => node.xpath("./text/name/text()").inner_text.downcase, 
                   :description => node.xpath("./text/desc/text()").inner_text,
                   :brand => node.xpath("./brand/text()").inner_text,
                   :merchant => node.xpath("../@name").inner_text,
                   :category => << Category.find_by_name(node.xpath("./cat/text()").inner_text.downcase),
                   :price => "£" + node.xpath("./price/btext()").inner_text)

Any ideas? Does this even make sense!?

Answer 1

Assuming the columns are called category_name and merchant_name, and you’ve set up the associations on Category and Merchant, you could do something like this:

Product.all do |product|
  product.category = Category.find_or_create_by_category_name(product.category_name)
  product.merchant = Merchant.find_or_create_by_merchant_name(product.merchant_name)
  product.save!
end

It will take a while, so for large datasets you might need a better solution.

So would this actually set the :category value in the products table to a category_id or set the value to the category_name?

.find_or_create_by does a find on the attribute and returns the matching row, or creates one if it does not exist. When creating the association via `.category=, Rails will set the foreign key to match the id of the row in the categories table.

So to answer your question more directly:

Product.create(:category=>Category.find_or_create_by_name("Magic Beans"))

is like doing this:

category = Category.find_by_name("Magic Beans")
if category.nil?
  category = Category.create(:name=>"Magic Beans")
end
product = Product.new
product.category = category
product.save

where the penultimate step sets the foreign key category_id to the value category.id. By convention associations are set up such that the foreign key is the model name suffixed with _id, so your products table should have both category_id and merchant_id.