PamGram/modules/wiki/character.py

201 lines
5.8 KiB
Python
Raw Normal View History

import re
from typing import List, Optional
from bs4 import BeautifulSoup
from httpx import URL
from modules.wiki.base import Model, HONEY_HOST
from modules.wiki.base import WikiModel
from modules.wiki.other import Association, Element, WeaponType
class Birth(Model):
"""生日
Attributes:
day:
month:
"""
day: int
month: int
class CharacterAscension(Model):
"""角色的突破材料
Attributes:
level: 等级突破材料
skill: 技能/天赋培养材料
"""
level: List[str] = []
skill: List[str] = []
class CharacterState(Model):
"""角色属性值
Attributes:
level: 等级
HP: 生命
ATK: 攻击力
DEF: 防御力
CR: 暴击率
CD: 暴击伤害
bonus: 突破属性
"""
level: str
HP: int
ATK: float
DEF: float
CR: str
CD: str
bonus: str
class CharacterIcon(Model):
icon: str
side: str
gacha: str
splash: Optional[str]
class Character(WikiModel):
"""角色
Attributes:
title: 称号
occupation: 所属
association: 地区
weapon_type: 武器类型
element: 元素
birth: 生日
constellation: 命之座
cn_cv: 中配
jp_cv: 日配
en_cv: 英配
kr_cv: 韩配
description: 描述
"""
id: str
title: str
occupation: str
association: Association
weapon_type: WeaponType
element: Element
birth: Optional[Birth]
constellation: str
cn_cv: str
jp_cv: str
en_cv: str
kr_cv: str
description: str
ascension: CharacterAscension
stats: List[CharacterState]
@classmethod
def scrape_urls(cls) -> List[URL]:
return [HONEY_HOST.join("fam_chars/?lang=CHS")]
@classmethod
async def _parse_soup(cls, soup: BeautifulSoup) -> "Character":
"""解析角色页"""
soup = soup.select(".wp-block-post-content")[0]
tables = soup.find_all("table")
table_rows = tables[0].find_all("tr")
def get_table_text(row_num: int) -> str:
"""一个快捷函数,用于返回表格对应行的最后一个单元格中的文本"""
return table_rows[row_num].find_all("td")[-1].text.replace("\xa0", "")
id_ = re.findall(r"img/(.*?_\d+)_.*", table_rows[0].find("img").attrs["src"])[0]
name = get_table_text(0)
if name != "旅行者": # 如果角色名不是 旅行者
title = get_table_text(1)
occupation = get_table_text(2)
association = Association.convert(get_table_text(3).lower().title())
rarity = len(table_rows[4].find_all("img"))
weapon_type = WeaponType[get_table_text(5)]
element = Element[get_table_text(6)]
birth = Birth(day=int(get_table_text(7)), month=int(get_table_text(8)))
constellation = get_table_text(10)
cn_cv = get_table_text(11)
jp_cv = get_table_text(12)
en_cv = get_table_text(13)
kr_cv = get_table_text(14)
else:
name = "" if id_.endswith("5") else ""
title = get_table_text(0)
occupation = get_table_text(1)
association = Association.convert(get_table_text(2).lower().title())
rarity = len(table_rows[3].find_all("img"))
weapon_type = WeaponType[get_table_text(4)]
element = Element[get_table_text(5)]
birth = None
constellation = get_table_text(7)
cn_cv = get_table_text(8)
jp_cv = get_table_text(9)
en_cv = get_table_text(10)
kr_cv = get_table_text(11)
description = get_table_text(-3)
ascension = CharacterAscension(
level=[
target[0]
for i in table_rows[-2].find_all("a")
if (target := re.findall(r"/(.*)/", i.attrs["href"])) # 过滤掉错误的材料(honey网页的bug)
],
skill=[re.findall(r"/(.*)/", i.attrs["href"])[0] for i in table_rows[-1].find_all("a")],
)
stats = []
for row in tables[2].find_all("tr")[1:]:
cells = row.find_all("td")
stats.append(
CharacterState(
level=cells[0].text,
HP=cells[1].text,
ATK=cells[2].text,
DEF=cells[3].text,
CR=cells[4].text,
CD=cells[5].text,
bonus=cells[6].text,
)
)
return Character(
id=id_,
name=name,
title=title,
occupation=occupation,
association=association,
weapon_type=weapon_type,
element=element,
birth=birth,
constellation=constellation,
cn_cv=cn_cv,
jp_cv=jp_cv,
rarity=rarity,
en_cv=en_cv,
kr_cv=kr_cv,
description=description,
ascension=ascension,
stats=stats,
)
@classmethod
async def get_url_by_name(cls, name: str) -> Optional[URL]:
# 重写此函数的目的是处理主角名字的 ID
_map = {"": "playergirl_007", "": "playerboy_005"}
if (id_ := _map.get(name)) is not None:
return await cls.get_url_by_id(id_)
return await super(Character, cls).get_url_by_name(name)
@property
def icon(self) -> CharacterIcon:
return CharacterIcon(
icon=str(HONEY_HOST.join(f"/img/{self.id}_icon.webp")),
side=str(HONEY_HOST.join(f"/img/{self.id}_side_icon.webp")),
gacha=str(HONEY_HOST.join(f"/img/{self.id}_gacha_card.webp")),
splash=str(HONEY_HOST.join(f"/img/{self.id}_gacha_splash.webp")),
)