draftaholics data was very skewed towards masterpieces. which meant
decent cards were ranking very low relatively
I used the relative ranking from draftaholics and generated a normal
distribution mean60 std 20 for these rankings to smooth it out.
* when resources are missing, fails gracefully
* uses highest found rating, rather than most recently found for the
case of sets containing duplicates
* moved ratings files to their own directory
* thread safe reading of all files
the way stored ratings for cards has been reworked
1. each set has its own ratings file and it will be loaded dynamically
if the AI needs it
2. each ratings file for a set is normalized within itself, so the
rating systems don't need to be scaled to match eachother. essentially
cards are rated within the context of their set
3. changed the rarity rating booster to only apply to unrated cards
4. the file setsWithRatings.csv explains how to add additional ratings.
no code changes are required, but a recompilation will be required to
add the files to the jar.
these weren't being picked up as removal spells and now are:
* modal spells where one mode is a removal spell
* spells that do a dynamic amount of damage, such as fireball
* cards that reduce the toughness temporarily or permanently
* cards that keep a creature tapped down
* cards that exile a creature
* cards that damage creatures that are attacking or blocking
* enchantments that exile a creature when they etb
* fight cards, and one sided fight cards
the ai will now rate those types of cards higher as they are removal
this test uses ai to determine which card to imprint, with my other
changes to AI, the rarity of the cards changed the relative rating of
these two cards. I changed the merfolk to be a common merfolk so that
part of the card rating shouldn't affect it
addtarget was being ignored and AI was being used to determine a target.
the spell has a target so I think this is not a bug with add target but
rather the test itself
