Clean up download script to be more maintainable

tools/download.py

@@ -210,114 +210,135 @@ async def load_pokedex(dex_file: Path, force_dex: bool) -> dict:
 
   return dex
 
-
+SHOWDOWN_STRIP_FORM = (
+  "burmy-plant",
+  "shellos-west", "gastrodon-west",
+  "deerling-spring", "sawsbuck-spring",
+  "flabébé-red", "floette-red", "florges-red",
+  "minior-red",
+  "vivillon-meadow",
+  "furfrou-natural",
+  "alcremie-vanilla-cream-strawberry",
+  "tatsugiri-curly",
+)
+# . for mime jr + mr mime + mr rime
+# ' for farfetch'd + sirfetch'd, also oricorio-pa'u
+# % for zygarde 10%
+# : for type: null
+# space mostly for tapus and paradox mons, but also some others
+SHOWDOWN_REMOVE_SYMBOLS = ".'%: "
 SHOWDOWN_REPLACEMENTS = [
   ("mega-", "mega"),  # charizard, mewtwo
   ("paldea-", "paldea"),  # tauros
-  ("mr. ", "mr"),  # mr mime + mr rime
-  (" jr.", "jr"),  # mime jr
-  ("'d", "d"),  # farfetch'd and sirfetch'd
   ("nidoran-m", "nidoranm"),
-  ("-f", "f"),  # gender diff forms TODO maybe only do this for some, clean up later fixes
-  (re.compile(r"-m$"), ""),  # gender diff forms
-  (re.compile(r"^ho-oh$"), "hooh"),
-  ("burmy-plant", "burmy"),
-  ("shellos-west", "shellos"),
-  ("gastrodon-west", "gastrodon"),
-  ("hippopotasf", "hippopotas-f"),
-  ("hippowdonf", "hippowdon-f"),
-  ("rotomf", "rotom-f"),  # rotom fan + frost special case
+  ("nidoran-f", "nidoranf"),
+  ("ho-oh", "hooh"),
   ("porygon-z", "porygonz"),
-  ("arceusf", "arceus-f"),  # fire, flying, fairy, fighting special case
-  ("unfezantf", "unfezant-f"),
   ("-striped", "striped"),  # basculin special cases
   ("-galar-zen", "-galarzen"),  # darmanitan special cases
-  ("-spring", ""),  # deerling, sawsbuck special cases
-  ("frillishf", "frillish-f"),
-  ("jellicentf", "jellicent-f"),
-  ("pyroarf", "pyroar-f"),
-  ("flabébé", "flabebe"),
-  ("flabebe-red", "flabebe"),
-  ("floette-red", "floette"),
-  ("florges-red", "florges"),
-  ("vivillonfancy", "vivillon-fancy"),
-  ("vivillon-meadow", "vivillon"),
+  ("é", "e"),  # flabebe
   ("vivillon-icy-snow", "vivillon-icysnow"),
   ("vivillon-high-plains", "vivillon-highplains"),
-  ("meowsticf", "meowstic-f"),
   ("furfrou-la-reine", "furfrou-lareine"),
-  ("furfrou-natural", "furfrou"),
-  ("%", ""),  # zygarde 10%,
   ("oricorio-pom-pom", "oricorio-pompom"),
-  ("oricorio-pa'u", "oricorio-pau"),
-  ("type: null", "typenull"),
-  ("silvallyf", "silvally-f"),  # fire, flying, fairy, fighting special case
-  ("minior-red", "minior"),
   ("mo-o", "moo"),  # jangmo-o, hakamo-o, kommo-o
-  ("tapu ", "tapu"),  # tapus
   ("dusk-mane", "duskmane"),  # necrozma
   ("dawn-wings", "dawnwings"),  # necrozma
   ("low-key", "lowkey"),  # toxtricity
-  ("swirlflower", "swirl-flower"),  # alcremie
-  ("creamflower", "cream-flower"),  # alcremie
   ("-swirl", "swirl"),  # alcremie
   ("-cream", "cream"),  # alcremie
-  ("-vanillacream-strawberry", ""),  # alcremie
   ("-strawberry", ""),  # alcremie
-  ("indeedeef", "indeedee-f"),
   ("rapid-strike", "rapidstrike"),  # Urshifu
   ("rapidstrike-gmax", "rapidstrikegmax"),  # Urshifu
-  ("basculegionf", "basculegion-f"),
-  ("oinkolognef", "oinkologne-f"),
-  ("mausholdfour", "maushold-four"),
-  ("tatsugiri-curly", "tatsugiri"),
   # dudunsparce (note - no visual diff on showdown as of 3/17)
   ("three-segment", "threesegment"),
   ("wo-chien", "wochien"),
   ("chien-pao", "chienpao"),
   ("ting-lu", "tinglu"),
   ("chi-yu", "chiyu"),
-  ("roaring moon", "roaringmoon"),
-  ("iron valiant", "ironvaliant"),
-  ("walking wake", "walkingwake"),
-  ("iron leaves", "ironleaves"),
 ]
 
 
 def get_showdown_urls(pkmn: Pokemon, form: Form) -> list[tuple[str, str]]:
   name = form.name.lower()
+  if name in SHOWDOWN_STRIP_FORM:
+    name = name.split("-", 1)[0]
+  for c in SHOWDOWN_REMOVE_SYMBOLS:
+    name = name.replace(c, "")
   for pat, ins in SHOWDOWN_REPLACEMENTS:
-    if isinstance(pat, re.Pattern):
-      name = re.sub(pat, ins, name)
-    else:
-      name = name.replace(pat, ins)
-  if 984 <= pkmn.num <= 995:
-    # paradox mons
-    name = name.replace(" ", "")
+    name = name.replace(pat, ins)
+  name = re.sub(r"-m$", "", name)  # gender diff forms
+  # lol they just reuse the male back sprite for wobbuffet
+  back_name = name if name != "wobbuffet-f" else "wobbuffet"
   return [
     (f"https://play.pokemonshowdown.com/sprites/ani/{name}.gif", "gif"),
-    (f"https://play.pokemonshowdown.com/sprites/ani-back/{name}.gif", "gif"),
+    (f"https://play.pokemonshowdown.com/sprites/ani-back/{back_name}.gif", "gif"),
     (f"https://play.pokemonshowdown.com/sprites/gen5/{name}.png", "png"),
-    (f"https://play.pokemonshowdown.com/sprites/gen5-back/{name}.png", "png"),
+    (f"https://play.pokemonshowdown.com/sprites/gen5-back/{back_name}.png", "png"),
   ]
 
 
+SEREBII_USE_SPECIES = (
+  "Wobbuffet-M",
+  "Shellos-West", "Gastrodon-West",
+  "Hippopotas-M", "Hippowdon-M",
+  "Unfezant-M",
+  "Deerling-Spring", "Sawsbuck-Spring",
+  "Frillish-M", "Jellicent-M",
+  "Pyroar-M",
+  "Furfrou-Natural",
+  "Minior-Meteor",
+  "Tatsugiri-Curly",
+)
+SEREBII_FORM_FIRST_LETTER = (
+  "Deoxys-Attack", "Deoxys-Defense", "Deoxys-Speed",
+  "Burmy-Plant", "Burmy-Sandy", "Burmy-Trash",
+  "Cherrim-Sunshine",
+  "Shellos-East", "Gastrodon-East",
+  "Dialga-Origin", "Palkia-Origin", "Giratina-Origin",
+  "Shaymin-Sky",
+  "Unfezant-F",
+  "Basculin-Blue-Striped", "Basculin-White-Striped",
+  "Deerling-Summer", "Deerling-Autumn", "Deerling-Winter",
+  "Sawsbuck-Summer", "Sawsbuck-Autumn", "Sawsbuck-Winter",
+  "Frillish-F", "Jellicent-F",
+  "Kyurem-Black", "Kyurem-White", "Keldeo-Resolute",
+  "Greninja-Ash",
+  "Pyroar-F",
+  "Flabébé-Blue", "Flabébé-Orange", "Flabébé-White", "Flabébé-Yellow", "Flabébé-Red",
+  "Floette-Blue", "Floette-Orange", "Floette-White", "Floette-Yellow", "Floette-Red",
+  "Florges-Blue", "Florges-Orange", "Florges-White", "Florges-Yellow", "Florges-Red",
+  "Meowstic-F",
+  "Aegislash-Blade",
+  "Hoopa-Unbound",
+  "Lycanroc-Midnight", "Lycanroc-Dusk",
+  "Wishiwashi-School",
+  "Minior-Blue", "Minior-Green", "Minior-Indigo", "Minior-Orange", "Minior-Red", "Minior-Violet", "Minior-Yellow",
+  "Mimikyu-Busted",
+  "Magearna-Original",
+  "Toxtricity-Low-Key",
+  "Eiscue-Noice",
+  "Indeedee-F",
+  "Morpeko-Hangry",
+  "Zacian-Crowned", "Zamazenta-Crowned",
+  "Eternatus-Eternamax",
+  "Zarude-Dada",
+  "Calyrex-Ice", "Calyrex-Shadow",
+  "Basculegion-F",
+  "Oinkologne-F",
+  "Maushold-Four",
+  "Palafin-Hero",
+  "Tatsugiri-Droopy", "Tatsugiri-Stretchy",
+  "Dudunsparce-Three-Segment",
+  "Gimmighoul-Roaming"
+)
 SEREBII_SPECIAL = {
   "Castform-Rainy": "r",
   "Castform-Snowy": "i",
   "Castform-Sunny": "s",
-  "Deoxys-Attack": "a",
-  "Deoxys-Defense": "d",
-  "Deoxys-Speed": "s",
-  "Burmy-Plant": "p",
-  "Burmy-Sandy": "s",
-  "Burmy-Trash": "t",
   # wormadam plant is default form
   "Wormadam-Sandy": "c",
   "Wormadam-Trash": "t",
-  "Cherrim-Sunshine": "s",
-  "Shellos-East": "e",
-  "Gastrodon-East": "e",
   "Tauros-Paldea-Blaze": "b",
   "Tauros-Paldea-Aqua": "a",
   "Rotom-Heat": "h",
@@ -325,32 +346,13 @@ SEREBII_SPECIAL = {
   "Rotom-Frost": "f",
   "Rotom-Fan": "s",
   "Rotom-Mow": "m",
-  "Dialga-Origin": "o",
-  "Palkia-Origin": "o",
-  "Giratina-Origin": "o",
-  "Shaymin-Sky": "s",
-  "Unfezant-F": "f",
-  "Basculin-Blue-Striped": "b",
-  "Basculin-White-Striped": "w",
   "Darmanitan-Zen": "z",
   "Darmanitan-Galar-Zen": "gz",
-  "Deerling-Summer": "s",
-  "Deerling-Autumn": "a",
-  "Deerling-Winter": "w",
-  "Sawsbuck-Summer": "s",
-  "Sawsbuck-Autumn": "a",
-  "Sawsbuck-Winter": "w",
-  "Frillish-F": "f",
-  "Jellicent-F": "f",
-  "Kyurem-Black": "b",
-  "Kyurem-White": "w",
-  "Keldeo-Resolute": "r",
   "Meloetta-Pirouette": "s",
   "Genesect-Douse": "w",
   "Genesect-Shock": "e",
   "Genesect-Burn": "f",
   "Genesect-Chill": "i",
-  "Greninja-Ash": "a",
   "Vivillon-Archipelago": "a",
   "Vivillon-Continental": "c",
   "Vivillon-Elegant": "e",
@@ -371,23 +373,6 @@ SEREBII_SPECIAL = {
   "Vivillon-Savanna": "sa",
   "Vivillon-Sun": "su",
   "Vivillon-Tundra": "t",
-  "Pyroar-F": "f",
-  "Flabébé-Blue": "b",
-  "Flabébé-Orange": "o",
-  "Flabébé-White": "w",
-  "Flabébé-Yellow": "y",
-  "Flabébé-Red": "r",
-  "Floette-Blue": "b",
-  "Floette-Orange": "o",
-  "Floette-White": "w",
-  "Floette-Yellow": "y",
-  "Floette-Red": "r",
-  "Florges-Blue": "b",
-  "Florges-Orange": "o",
-  "Florges-White": "w",
-  "Florges-Yellow": "y",
-  "Florges-Red": "r",
-  "Meowstic-F": "f",
   "Furfrou-Dandy": "da",
   "Furfrou-Debutante": "de",
   "Furfrou-Diamond": "d",
@@ -397,10 +382,8 @@ SEREBII_SPECIAL = {
   "Furfrou-Matron": "m",
   "Furfrou-Pharaoh": "p",
   "Furfrou-Star": "s",
-  "Aegislash-Blade": "b",
   "Zygarde-10%": "10",
   "Zygarde-Complete": "c",
-  "Hoopa-Unbound": "u",
   "Pumpkaboo-Small": "s",  # kinda dumb but w/e
   "Pumpkaboo-Large": "l",
   "Pumpkaboo-Super": "h",
@@ -410,59 +393,18 @@ SEREBII_SPECIAL = {
   "Oricorio-Pom-Pom": "p",
   "Oricorio-Pa'u": "pau",
   "Oricorio-Sensu": "s",
-  "Lycanroc-Midnight": "m",
-  "Lycanroc-Dusk": "d",
-  "Wishiwashi-School": "s",
-  "Minior-Blue": "b",
-  "Minior-Green": "g",
-  "Minior-Indigo": "i",
-  "Minior-Orange": "o",
-  "Minior-Red": "r",
-  "Minior-Violet": "v",
-  "Minior-Yellow": "y",
-  "Mimikyu-Busted": "b",
   "Necrozma-Dusk-Mane": "dm",
   "Necrozma-Dawn-Wings": "dw",
   "Necrozma-Ultra": "m",
-  "Magearna-Original": "o",
   "Cramorant-Gulping": "gu",
   "Cramorant-Gorging": "go",
-  "Toxtricity-Low-Key": "l",
   "Sinistea-Antique": "f",
   "Polteageist-Antique": "f",
-  "Eiscue-Noice": "n",
-  "Indeedee-F": "f",
-  "Morpeko-Hangry": "h",
-  "Zacian-Crowned": "c",
-  "Zamazenta-Crowned": "c",
-  "Eternatus-Eternamax": "e",
-  "Zarude-Dada": "d",
-  "Calyrex-Ice": "i",
-  "Calyrex-Shadow": "s",
   "Urshifu-Rapid-Strike": "r",
   "Urshifu-Rapid-Strike-Gmax": "rgi",
-  "Basculegion-F": "f",
-  "Oinkologne-F": "f",
-  "Maushold-Four": "f",
-  "Palafin-Hero": "h",
-  "Tatsugiri-Droopy": "d",
-  "Tatsugiri-Stretchy": "s",
-  "Dudunsparce-Three-Segment": "t",
-  "Gimmighoul-Roaming": "r",
 }
-SEREBII_USE_SPECIES = (
-  "Shellos-West", "Gastrodon-West",
-  "Hippopotas-M", "Hippowdon-M",
-  "Unfezant-M",
-  "Deerling-Spring", "Sawsbuck-Spring",
-  "Frillish-M", "Jellicent-M",
-  "Pyroar-M",
-  "Furfrou-Natural",
-  "Minior-Meteor",
-  "Tatsugiri-Curly",
-)
 SEREBII_IGNORE_MISSING = (
-  "Hippopotas-F", "Hippowdon-F", "Floette-Eternal",
+  "Wobbuffet-F", "Hippopotas-F", "Hippowdon-F", "Floette-Eternal",
   "Squawkabilly-Blue", "Squawkabilly-Yellow", "Squawkabilly-White",
 )
 
@@ -470,6 +412,9 @@ SEREBII_IGNORE_MISSING = (
 def get_serebii_url(pkmn: Pokemon, form: Form) -> str | None:
   if form.name == pkmn.species or form.name in SEREBII_USE_SPECIES:
     return f"https://www.serebii.net/pokemon/art/{pkmn.num:03d}.png"
+  if form.name in SEREBII_FORM_FIRST_LETTER:
+    letter = form.name.split("-", 1)[1][0].lower()
+    return f"https://www.serebii.net/pokemon/art/{pkmn.num:03d}-{letter}.png"
   if form.name in SEREBII_SPECIAL:
     return f"https://www.serebii.net/pokemon/art/{pkmn.num:03d}-{SEREBII_SPECIAL[form.name]}.png"
   if "gmax" in form.traits: