pokemon-color-search/anim_ingest.py

import io
import math

from PIL import Image
from bs4 import BeautifulSoup
from colorspacious import cspace_convert
from scipy.cluster import vq
import requests
import numpy as np

import ingest

cluster_seed = 20220326
cluster_attempts = 3
base = "https://play.pokemonshowdown.com/sprites/ani/"
back_base = "https://play.pokemonshowdown.com/sprites/ani-back/"

# removing all forms of a pokemon, and also pokestars
start_with_filters = [
  # no significant visual changes
  "arceus-", "silvally-", "genesect-", "pumpkaboo-", "gourgeist-", "unown-", "giratina-",
  # cannot start the battle in alternate form
  "castform-", "cherrim-", "aegislash-", "xerneas-", "wishiwashi-", "eiscue-", "mimikyu-",
  "cramorant-", "morepeko-",
  # weird event thing
  "greninja-", "eevee-", "pikachu-", "zarude-", "magearna-",
  # pokestars
  "pokestar",
]

# removing all forms of a type
end_with_filters = [
  "-mega.gif", "-megax.gif", "-megay.gif", "-primal.gif", "-ultra.gif",
  "-gmax.gif", "-eternamax.gif", "-totem.gif", "-f.gif", "-b.gif",
]

# removing pokemon entirely
full_filters = [
  # darmanitan zen forms (cannot start in zen)
  "darmanitan-galarzen.gif", "darmanitan-zen.gif",
  # minior core forms (cannot start in anything but -meteor, renamed below)
  "minior.gif", "minior-blue.gif", "minior-green.gif", "minior-indigo.gif", 
  "minior-orange.gif", "minior-violet.gif", "minior-yellow.gif",
  # because it is a create-a-pokemon
  "astrolotl.gif", "aurumoth.gif", "caribolt.gif", "cawmodore.gif", "chromera.gif", "crucibelle.gif",
  "equilibra.gif", "fidgit.gif", "jumbao.gif", "justyke.gif", "kerfluffle.gif", "kitsunoh.gif",
  "krilowatt.gif", "malaconda.gif", "miasmaw.gif", "mollux.gif", "naviathan.gif", "necturna.gif",
  "pajantom.gif", "plasmanta.gif", "pluffle.gif", "protowatt.gif", "scratchet.gif", "smogecko.gif",
  "smoguana.gif", "smokomodo.gif", "snaelstrom.gif", "stratagem.gif", "tomohawk.gif", "volkraken.gif", "voodoom.gif",
  # typos/duplicates
  "buffalant.gif", "klinklang-back.gif", "krikretot.gif", "pumpkabo-super.gif", "magcargo%20.gif",
  "ratatta-a.gif", "ratatta-alola.gif", "raticate-a.gif",
  "rotom-h.gif", "rotom-m.gif", "rotom-s.gif", "rotom-w.gif",
  # not a pokemon
  "substitute.gif",
]

# force certain pokemon to stay
force_keep = [ "meowstic-f.gif", "unfezant-f.gif", "pyroar-f.gif" ]

# rename certain pokemon after the fact
rename = {
  # dash consistency
  "nidoranm": "nidoran-m",
  "nidoranf": "nidoran-f",
  "porygonz": "porygon-z",
  "tapubulu": "tapu-bulu",
  "tapufini": "tapu-fini",
  "tapukoko": "tapu-koko",
  "tapulele": "tapu-lele",
  "hooh": "ho-oh",
  "mimejr": "mime-jr",
  "mrmime": "mr-mime",
  "mrmime-galar": "mr-mime-galar",
  "mrrime": "mr-rime",
  "jangmoo": "jangmo-o",
  "hakamoo": "hakamo-o",
  "kommoo": "kommo-o",
  "typenull": "type-null",
  "oricorio-pompom": "oricorio-pom-pom",
  "necrozma-duskmane": "necrozma-dusk-mane",
  "necrozma-dawnwings": "necrozma-dawn-wings",
  "toxtricity-lowkey": "toxtricity-low-key",

  # rename forms
  "shellos": "shellos-west",
  "shaymin": "shaymin-land",
  "meloetta": "meloetta-aria",
  "keldeo": "keldeo-ordinary",
  "hoopa": "hoopa-confined",
  "burmy": "burmy-plant",
  "wormadam": "wormadam-plant",
  "deerling": "deerling-spring",
  "sawsbuck": "sawsbuck-spring",
  "vivillon": "vivillon-meadow",
  "basculin": "basculin-redstriped",
  "meowstic": "meowstic-male",
  "meowstic-f": "meowstic-female",
  "pyroar-f": "pyroar-female",
  "flabebe": "flabebe-red",
  "floette": "floette-red",
  "florges": "florges-red",
  "minior-meteor": "minior",
  "sinistea": "sinistea-phony",
  "polteageist": "polteageist-phony",
  "gastrodon": "gastrodon-west",
  "furfrou": "furfrou-natural",
  "wishiwashi": "wishiwashi-school",
  "tornadus": "tornadus-incarnate",
  "landorus": "landorus-incarnate",
  "thundurus": "thundurus-incarnate",
  "calyrex-ice": "calyrex-ice-rider",
  "calyrex-shadow": "calyrex-shadow-rider",
  "urshifu-rapidstrike": "urshifu-rapid-strike",
  "zacian": "zacian-hero",
  "zamazenta": "zamazenta-hero",
}


def get_all_pokemon() -> list[str]:
  soup = BeautifulSoup(requests.get(back_base).text, "html.parser")
  gifs = [href for a in soup.find_all("a") if (href := a.get("href")).endswith("gif")]
  return [
    g[:-4] 
    for g in gifs 
    if g in force_keep or (
      g not in full_filters 
      and not any(g.startswith(f) for f in start_with_filters) 
      and not any(g.endswith(f) for f in end_with_filters)
    )
  ]


def load_image(base: str, name: str) -> Image:
  return Image.open(io.BytesIO(requests.get(base + name + ".gif").content))


def get_all_pixels(im: Image) -> list[tuple[int, int, int]]:
  rgb_pixels = []
  for fr in range(getattr(im, "n_frames", 1)):
    im.seek(fr)
    rgb_pixels += [
      (r, g, b) 
      for r, g, b, a in im.convert("RGBA").getdata() 
      if not ingest.is_outline(r, g, b, a)
    ]
  return rgb_pixels


def chromatic_dist_jab(p: np.array, q: np.array) -> float:
  _, pa, pb = p
  _, qa, qb = q
  return math.sqrt((pa - qa) ** 2 + (pb - qb) ** 2)


def chromatic_dist_rgb(p: np.array, q: np.array) -> float:
  return chromatic_dist_jab(*cspace_convert(np.array([p, q]), "sRGB255", "CAM02-UCS"))


def compute_stats(pixels: np.array, chroma_dist, verbose: bool = False) -> list[float]:
  # return [inertia, *mu (3 components), *nu (3 components), and all cluster data (8 * 4 = 32 components)]
  total_stats = [ingest.inertia(pixels), *ingest.mu(pixels), *ingest.nu(pixels)]

  # attempt all of k=2,3,4 and select the "best" k
  # for a normal k-means problem this involves using the cluster variance or similar
  # but for this, what we really want to detect is a tripartite vs bipartite pokemon
  # these means need to be *distinct* to avoid punishing a pokemon with two shades
  # of the same color
  # for that reason, the average distance between the means is used instead
  # rather, the distance between the 2 means for the k=2 case is compared against
  # the mean of the three distances for the k=3 case, and the mean of the six
  # differences for the k=4 case
  # also, since the actual color is most important, as opposed to shades/hues, a
  # "chromatic" distance is used (distance in the a-b plane)
  # finally, a few different clusterings are actually attempted, in the hopes of
  # avoiding local maxima
  # result is then padded to have 2 fake empty clusters which can then be ignored by the UI

  best_dist = 0
  best_means = None
  best_labels = None
  for i in range(cluster_attempts):
    kmeans2, labels2 = vq.kmeans2(pixels.astype(float), 2, minit="++", seed=cluster_seed + i)
    dist2 = chroma_dist(*kmeans2)
    if dist2 > best_dist:
      best_dist, best_means, best_labels = dist2, kmeans2, labels2
  for i in range(cluster_attempts):
    kmeans3, labels3 = vq.kmeans2(pixels.astype(float), 3, minit="++", seed=cluster_seed + i)
    dist3 = ( chroma_dist(kmeans3[0], kmeans3[1])
            + chroma_dist(kmeans3[1], kmeans3[2])
            + chroma_dist(kmeans3[2], kmeans3[0])
            ) / 3
    if dist3 > best_dist:
      best_dist, best_means, best_labels = dist3, kmeans3, labels3
  for i in range(cluster_attempts):
    kmeans4, labels4 = vq.kmeans2(pixels.astype(float), 4, minit="++", seed=cluster_seed + i)
    dist4 = ( chroma_dist(kmeans4[0], kmeans4[1])
            + chroma_dist(kmeans4[0], kmeans4[2])
            + chroma_dist(kmeans4[0], kmeans4[3])
            + chroma_dist(kmeans4[1], kmeans4[2])
            + chroma_dist(kmeans4[1], kmeans4[3])
            + chroma_dist(kmeans4[2], kmeans4[3])
            ) / 6
    if dist4 > best_dist:
      best_dist, best_means, best_labels = dist4, kmeans4, labels4

  cluster_stats = []
  for i in range(len(best_means)):
    cluster_pixels = pixels[best_labels == i]
    cluster_stats += [len(cluster_pixels), ingest.inertia(cluster_pixels), *best_means[i], *ingest.nu(cluster_pixels)]
  cluster_stats += [0] * (32 - len(cluster_stats))
  if verbose:
    print(f"    Selected k={len(best_means)}")

  return [*total_stats, *cluster_stats]


def get_stats(name: str, verbose: bool = False) -> list[float]:
  front = get_all_pixels(load_image(base, name))
  back = get_all_pixels(load_image(back_base, name))
  rgb_pixels = np.array(front + back)
  jab_pixels = cspace_convert(rgb_pixels, "sRGB255", "CAM02-UCS")
  if verbose:
    print(f"  Jab space...")
  jab_stats = compute_stats(jab_pixels, chromatic_dist_jab, verbose)
  if verbose:
    print(f"  RGB space...")
  rgb_stats = compute_stats(rgb_pixels, chromatic_dist_rgb, verbose)
  return [len(rgb_pixels), *jab_stats, *rgb_stats]


if __name__ == "__main__":
  pkmn = get_all_pokemon()
  print("Found", len(pkmn), "sprites...")
  errors = []
  with open("database-v3.js", "w") as outfile:
    outfile.write("const databaseV3 = [\n")
    for name in pkmn:
      print("Ingesting", name, "...")
      try:
        stats = get_stats(name, verbose=True)
        outfile.write(f'  [ "{rename.get(name, name)}", {", ".join(str(n) for n in stats)} ],\n')
      except Exception as e:
        print(e)
        errors.append((name, e))
    outfile.write("];\n")
  print("Errors:", errors)