Also noticed I made a mistake in some specs, would normally optimise spec code later on.
Added an extra test for when sell_in is 1 for safety as it depends on the order of operations in the extraction I make.
Without looking at the code I've written specs based on the requirements listed and the output of the texttest.
All specs are green except for the expected 2 failing 'quality' tests for the new 'Conjured' item.
With all models tested I feel more comfortable starting the refactor. The plan at the moment is to extract the code of 1 type of item at the time into it's own method.