def scrape_dex_links(root: Element) -> Iterable[Tuple[str, str]]:
dex_table = root.find(".//*[@id='pokedex']")
dex_body = dex_table.find("tbody")
for row in (row for row in dex_body if row.tag == "tr"):
cols = row.findall("td")
link_tag = cols[1].find("a")
yield link_tag.text, f"https://pokemondb.net{link_tag.get('href')}"
def __init__(self, table: Element, headers: Sequence[str] = None) -> None:
self.current_row = 0
self.rows = table.findall("tr")
self.num_rows = len(self.rows)
self.headers = []
if headers:
self.headers = headers
return
thead = table.find("thead")
if thead is not None:
first_row = thead[0]
else:
first_row = self.rows[0]
self.rows = self.rows[1:] # first row will be taken as header
self.current_row = 0
self.num_rows -= 1
for td in first_row:
header = td.text.strip()
self.headers.append(header)
def parse_pokemon_from_tab(
root: Element,
tab_id: str,
tab_name: str,
pokemon_name: str,
default: Optional[PokemonData] = None) -> Optional[PokemonData]:
args = {
"move_list": MoveList(),
"misc_info": MiscInfo(),
"dex_entries": DexEntryCollection()
}
#
#
# Figure out name and variant
variant_str = tab_name.replace(pokemon_name, "").replace(" ", " ").strip()
variant_str_split = variant_str.split(" ")
first_word = variant_str_split[0]
remaining_words = " ".join(variant_str_split[1:])
if Region.is_region_descriptor(first_word):
region = Region.parse_descriptor(first_word)
variant_name = PrefixVariantName(first_word, {})
variant_str = remaining_words.strip()
else:
region = Region.NONE
variant_name = None
if variant_str == "":
args["variant"] = Variant(region=region)
elif variant_str in ["Male", "Female"]:
args["variant"] = Variant(gender=Gender(variant_str.upper()),
region=region)
variant_name = _merge_variant_names(variant_name,
SuffixVariantName(variant_str, {}))
elif variant_str.startswith("Mega"):
mega_type = MegaType.X if variant_str.endswith("X") else MegaType.Y if variant_str.endswith("Y") else \
MegaType.NORMAL
args["variant"] = Variant(mega_type=mega_type, region=region)
if mega_type == MegaType.NORMAL:
variant_name = _merge_variant_names(variant_name,
PrefixVariantName("Mega", {}))
else:
variant_name = _merge_variant_names(
variant_name,
CircumfixVariantName("Mega", variant_str[-1], {}, {}))
elif variant_str == "Primal":
args["variant"] = Variant(form="Primal", region=region)
variant_name = _merge_variant_names(variant_name,
PrefixVariantName("Primal", {}))
else:
args["variant"] = Variant(form=variant_str, region=region)
if pokemon_name in ["Kyurem", "Hoopa"]:
variant_name = _merge_variant_names(
variant_name, SuffixVariantName(variant_str, {}, comma=False))
elif pokemon_name == "Rotom":
variant_name = _merge_variant_names(
variant_name, SuffixVariantName(variant_str, {}, spacer="-"))
elif variant_str in ["Partner", "Own Tempo"
] or pokemon_name in ["Necrozma"]:
variant_name = _merge_variant_names(
variant_name, PrefixVariantName(variant_str, {}))
elif variant_str == "Ash-":
variant_name = _merge_variant_names(
variant_name, PrefixVariantName("Ash", {}, spacer="-"))
else:
variant_name = _merge_variant_names(
variant_name, SuffixVariantName(variant_str, {}, comma=True))
args["name"] = Name(pokemon_name, {}, variant_name=variant_name)
#
#
# Get typing, abilities, dex entries, and misc_info
tab_div = root.find(f".//*[@id='{tab_id}']")
tables = tab_div.findall(".//*[@class='vitals-table']")
pokedex_data = tables[0].find("tbody")
training = tables[1].find("tbody")
for row in pokedex_data:
header = row.find("th").text
#
if header == "Type":
types = [Type(a.text.upper()) for a in row.find("td").findall("a")]
if len(types) == 0:
return None
args["typing"] = Typing(*types)
#
elif header == "Abilities":
abilities = dict()
col = row.find("td")
for span in col.findall("span") + col.findall("small"):
ability_link = span.find("a")
if ability_link.tail == " (hidden ability)":
abilities["hidden_ability"] = Ability(ability_link.text)
elif span.text == "1. ":
abilities["primary_ability"] = Ability(ability_link.text)
elif span.text == "2. ":
abilities["secondary_ability"] = Ability(ability_link.text)
if len(abilities) == 0:
if default:
args["abilities"] = default.abilities
else:
return None
else:
args["abilities"] = AbilityList(**abilities)
#
elif header == "National №":
col = row.find("td").find("strong")
args["dex_entries"].add_entry(DexEntry(Dex.NATIONAL,
int(col.text)))
#
elif header == "Local №":
col = row.find("td")
if not col.text:
continue
curr_num = int(col.text)
for tag in col:
if tag.tag == "br":
curr_num = int(tag.tail) if tag.tail else None
else:
args["dex_entries"].add_entry(
DexEntry(Dex(tag.text[1:-1]), curr_num))
for row in training:
header = row.find("th").text
if header == "EV yield":
col = row.find("td")
if not col.text.strip():
continue
stats = (s.strip() for s in col.text.split(","))
ev_yield = EVYield(
*(({
"Attack": Stat.ATTACK,
"Defense": Stat.DEFENSE,
"Special Attack": Stat.SP_ATTACK,
"Special Defense": Stat.SP_DEFENSE,
"Speed": Stat.SPEED,
"HP": Stat.HP
}[stat], int(value)) for stat_info in stats
for value, stat in (stat_info.split(" ", maxsplit=1), )))
args["misc_info"].ev_yield = ev_yield
#
#
# Get evolution line
if args["name"].name in ["Nincada", "Ninjask", "Shedinja"]:
args["misc_info"].evolution_line = \
EvolutionLine("NINCADA",
(Evolution("NINCADA", "NINJASK", LevelUpEvolutionType(20)),
EvolutionLine("NINJASK")),
(Evolution("NINCADA", "SHEDINJA", UnknownEvolutionType()),
EvolutionLine("SHEDINJA")))
else:
evo_lines = []
for evo_line_info in root.findall(
f".//div[@class='infocard-list-evo']"):
evo_line = _scrape_evolution_line(evo_line_info)
if isinstance(evo_line, EvolutionLine):
evo_lines.append(evo_line)
evo_lines = list(EvolutionLine.merge(*evo_lines))
formatted_name = FormatUtils.format_name_as_id(args["name"],
args["variant"],
ignore_mega=True)
evo_lines = [
evl for evl in evo_lines
if formatted_name in evl.get_all_pokemon_ids_in_line()
]
if len(evo_lines) == 1:
args["misc_info"].evolution_line = evo_lines[0]
elif len(evo_lines) > 1:
print("MULTIPLE EVO LINES FOUND")
#
#
# Get stats
stats_div = tab_div.find(".//*[@id='dex-stats']...")
stats_table = stats_div.find(".//table")
if stats_table is None:
return None
stats_table = stats_table.find("tbody")
stats = dict()
for row in stats_table:
header = row.find("th").text
value = int(row.find("td").text)
stat = {
"HP": "hp",
"Attack": "attack",
"Defense": "defense",
"Sp. Atk": "special_attack",
"Sp. Def": "special_defense",
"Speed": "speed"
}[header]
stats[stat] = value
args["stats"] = BaseStats(**stats)
#
return PokemonData(**args)