Course:CPSC312-2021/Pokemon

From UBC Wiki

Pokemon Battle Similator

Authors: David, Johnson, Yifu

What is the problem?

For the project, we are constructing a simpiflied version of the battle system in the famous game Pokemon. The final result will be a one-on-one, console based battle simulator. User will be able to select different moves in their turn to attack and defend. The game ends if one of the pokemon's HP reaches 0 and the user runs out of usable pokemons.

Github (download version is outdated but info still exists) link

Google Drive (updated code) link

pokemon simulator

Video Demo: https://youtu.be/CXvIjlei0tE

What is the something extra?

Pokemon Database

  • Using a public dataset in CSV format for Pokemon on Kaggle (reference), we generated Haskell code from the CSV data using a python script.
  • We created a GUI table that allows the user to find the Pokemon's information by its name
  • We created a GUI CRUD table that allows user to add personalized notes about the Pokemon's data
  • We created an entire database for each of {Pokemon, Move, Type}:

Pokemon Database

  • Pokemon
    • index: the pokedex index (used to match Pokemon with the corresponding image)
      • int
    • name: pokemon's name
      • string
    • type1: first type attribute (used to calculate critical damage multiplier)
      • Type
    • type2: second type attribute (used to calculate critical damage multiplier)
      • Type
    • hp: base health
      • int
    • attack: base attack
      • int
    • defence: base defence
      • int
    • specialAttack: base attack for special Move
      • int
    • specialDefence: base defence against special Move
      • int
    • moves: contains all possible moves that can be unlocked in the game
      • list of Moves

Move Database

  • moveName: name of the move
    • string
  • moveDescription: a description about the move, including its effects
    • string
  • moveType: Type of move (used to calculate critical damage multiplier)
    • Type
  • power: base damage of the move (i.e. if the crit damage multiplier is 1.0)
    • int
  • accuracy: the probability that this move will be successful (otherwise skips a turn)
    • float
  • pp: the remaining number of charges for this move
    • int
  • moveEffect: the effects of the move, e.g., poison, stun, weakened, etc. (we left this for future work)
    • string

Type Database

  • Type
    • defenceType1: first type name of the defender
      • string
    • defenceType2: second type name of the defender
      • string
    • normal: the damage multiplier that will be applied if the attacker's type is Normal and the defender's type is defenceType1 and defenceType2
      • float, damage multiplier
    • fire: the damage multiplier that will be applied if the attacker's type is Fire and the defender's type is defenceType1 and defenceType2
      • float, damage multiplier
    • etc…… all other types listed in each Type object

Artificial Intelligence Agent

  • There are multiple levels of difficulties for the AI trainer:
Easy
  • Chose lowest attack moves every tern

Hard

  • When health is below 15, the trainer uses healing potion (max limit = 3)
  • Uses counter moves
    • Counters type
    • Counters multi-turn attack from enemy with defensive move
  • Makes optimal decisions based on game state (by a search algorithm) to do more tame to the player

Graphical User Interface

  • Colored Images for Pokemon and Background
  • Searchable list of pokemons
  • Controlled by terminal

Battle Options

  • Number of pokemons per trainer
    • 1 Vs 1
    • Multiple Vs Multiple
  • Pokemon option
    • Customized abilities

Battle Interactions

We create comprehensive logic around the interaction mechanisms

  • Damage calculation
    • Determine the amount of damage is done depending on he attacker's attack, target's defence, two pokemon's type and the power of the move.
  • Heal calculation
    • Determine the amount of healing is done

What did we learn from doing this?

Learned about Functional Programming

Functional programming language(referred to as FPL) is very different than the OOP languages frequently worked with. The formatting of FPL is straightforward to read as if they are English sentences. This characteristic makes it very easy to interpret and understand. Also, because of that, it is convenient to use FPL to do higher-level designing. FPL also provides fast ways to declare variables, methods and tests. Therefore it makes the work neat and concise. We are able to create objects that have a functional nature, and write modularized code that executes at fast speeds. Another important feature about FPL is that it is a strongly-typed language. This means the compiler will do a lot of work for you, including finding out more errors during the compiling stage. It makes our code safer, but it can also introduce type errors that are had to deal with.

Learned from GUI

We used the “threepenny-gui” library, which has built in functions that perform edits to the underlying HTML and CSS files using Haskell,

e.g.: canvas <- UI.canvas

# set UI.height canvasSizeHeight

# set UI.width canvasSizeWidth

# set style [("border", "solid black 1px"), ("background", "#eee")]

We think using Haskell is not the best idea because it is complicated passing variables around. We are not sure how to write asynchronous functions to perform the features we wanted. The strongly-typed nature also makes it difficult to pass data around. For example, the expected type is “UI(),” but the actual type is “IO(),” which has to be solved using the “liftIO” function. There are many built-in short-hands in Haskell, and the “threepenny-gui” used them a lot. It made modifying the sample code significantly harder.

e.g.: bSelectionPokemon <- stepper Nothing $ head <$> unions

[ eSelection

, Nothing <$ eDelete

, Just . nextKey <$> bDatabasePokemon <@ eDummy

, (\b s p -> b >>= \a -> if p (s a) then Just a else Nothing)

<$> bSelection <*> bShowDataItemPokemon <@> eFilter]

In the furture if we are going to deal with GUI with Haskell, we will consider an alternative library to work with.

Learned from AI

At first, we tried to use dynamic programming for this program. Yet, we find there are too many possibilities to calculate—hundreds of Pokemons with hundreds of moves. Also, there is no simple “winning-strategy” for this battle like the MagicSum game strategy. Instead, our AI now makes decisions based on the current status of the game. The strong-typed nature also makes it likely to have type mismatch. For example, when we are trying to calculate the best damage we could deal with in this stage, we had difficulties dealing with methods that take and produces Int, Integer, m Integer, IO Integer at the same time. That makes writing programs slower. However, due to the nature of functional programming, AI testing is surprisingly easy to do.

Is Functional Programming suitable for this program

As a terminal based game Haskell works fine (even though we need to check input validity), but when expanding to a game with GUI it is much more complicated (we had a hard time figuring out the shorthands and type checks).

Works Cited

Some Non-Pokemon Template Code is from the GUI library samples: https://github.com/HeinrichApfelmus/threepenny-gui

Pokemon Database csv data: http://kaggle.com/mylesoneill/pokemon-sun-and-moon-gen-7-stats

 Guidelines Create Your Wiki Page Past Projects Help and Resources 
Course title
Image:wiki.png
Course title
Instructor:
Email:
Office:
Office Hours:
Class Schedule:
Important Course Pages

Add a content here. You can also copy and paste content for each tabs from below: