refactor battle
This commit is contained in:
bridge
@@ -2,7 +2,7 @@ from __future__ import annotations
|
||||
|
||||
from src.classes.action import InstantAction
|
||||
from src.classes.event import Event
|
||||
from src.classes.battle import decide_battle
|
||||
from src.classes.battle import decide_battle, get_effective_strength_pair
|
||||
from src.classes.story_teller import StoryTeller
|
||||
|
||||
|
||||
@@ -35,7 +35,9 @@ class Battle(InstantAction):
|
||||
def start(self, avatar_name: str) -> Event:
|
||||
target = self._get_target(avatar_name)
|
||||
target_name = target.name if target is not None else avatar_name
|
||||
event = Event(self.world.month_stamp, f"{self.avatar.name} 对 {target_name} 发起战斗")
|
||||
# 展示双方折算战斗力（基于对手、含克制）
|
||||
s_att, s_def = get_effective_strength_pair(self.avatar, target)
|
||||
event = Event(self.world.month_stamp, f"{self.avatar.name} 对 {target_name} 发起战斗（战斗力：{self.avatar.name} {int(s_att)} vs {target_name} {int(s_def)}）")
|
||||
# 记录开始事件内容，供故事生成使用
|
||||
self._start_event_content = event.content
|
||||
return event
|
||||
|
||||
@@ -1,120 +1,150 @@
|
||||
from __future__ import annotations
|
||||
|
||||
import math
|
||||
import random
|
||||
from typing import Tuple, TYPE_CHECKING
|
||||
import random
|
||||
|
||||
from src.classes.cultivation import Realm
|
||||
from src.classes.technique import get_suppression_bonus, get_grade_advantage_bonus
|
||||
from src.classes.technique import TechniqueGrade, get_suppression_bonus
|
||||
|
||||
if TYPE_CHECKING:
|
||||
from src.classes.avatar import Avatar
|
||||
|
||||
|
||||
def _realm_order(realm: Realm) -> int:
|
||||
# 战斗力参数（参考文明6思想，但适配本项目数值体系）
|
||||
_STRENGTH_LOG_SCALE: float = 10.0 # 修为强度的对数缩放：10×ln(1+level)
|
||||
_GRADE_POINTS = {
|
||||
TechniqueGrade.LOWER: 0.0,
|
||||
TechniqueGrade.MIDDLE: 3.0,
|
||||
TechniqueGrade.UPPER: 6.0,
|
||||
}
|
||||
_SUPPRESSION_POINTS: float = 3.0 # 属性克制即加固定战斗力点数
|
||||
_CIV6_K: float = 0.04 # 伤害指数系数：e^(K×差值)
|
||||
_WIN_BETA: float = 0.15 # 胜率逻辑函数斜率
|
||||
_BASE_DAMAGE_LOW: int = 24 # 基础伤害下限（按 defender.maxHP/100 缩放）
|
||||
_BASE_DAMAGE_HIGH: int = 36 # 基础伤害上限（按 defender.maxHP/100 缩放）
|
||||
_MIN_RATIO: float = 1.05 # 最小相对优势比，确保赢家伤害严格更低
|
||||
_PAIR_BIAS: float = 1.1 # 成对偏置：让败者再多一点、赢家再少一点
|
||||
|
||||
|
||||
def _combat_strength_vs(opponent: "Avatar", self_avatar: "Avatar") -> float:
|
||||
"""
|
||||
将境界映射为数值顺序，用于胜率计算。
|
||||
计算对某个对手的有效战斗力：
|
||||
= 10×ln(1+修为等级) + 品阶点数(0/3/6) + 克制点数(若克制则+3)
|
||||
说明：
|
||||
- 修为使用总等级（1..120）并做对数缩放，避免过大；
|
||||
- 品阶加点为线性，避免与修为重复放大；
|
||||
- 克制只在“对某个对手”时生效，因此放在此函数处理。
|
||||
"""
|
||||
order_map = {
|
||||
Realm.Qi_Refinement: 1,
|
||||
Realm.Foundation_Establishment: 2,
|
||||
Realm.Core_Formation: 3,
|
||||
Realm.Nascent_Soul: 4,
|
||||
}
|
||||
return order_map.get(realm, 1)
|
||||
level = max(1, self_avatar.cultivation_progress.level)
|
||||
strength_from_level = _STRENGTH_LOG_SCALE * math.log1p(level)
|
||||
|
||||
grade_points = 0.0
|
||||
if self_avatar.technique is not None:
|
||||
grade_points = _GRADE_POINTS.get(self_avatar.technique.grade, 0.0)
|
||||
|
||||
suppression_points = 0.0
|
||||
if self_avatar.technique is not None and opponent.technique is not None:
|
||||
# 仅需“是否克制”的布尔，不引入倍率。
|
||||
if get_suppression_bonus(self_avatar.technique.attribute, opponent.technique.attribute) > 0.0:
|
||||
suppression_points = _SUPPRESSION_POINTS
|
||||
|
||||
return strength_from_level + grade_points + suppression_points
|
||||
|
||||
|
||||
def _strength_diff(attacker: "Avatar", defender: "Avatar") -> float:
|
||||
return _combat_strength_vs(defender, attacker) - _combat_strength_vs(attacker, defender)
|
||||
|
||||
|
||||
def get_effective_strength(self_avatar: "Avatar", opponent: "Avatar") -> float:
|
||||
"""
|
||||
对外公开：返回 self_avatar 面对 opponent 时的折算战斗力。
|
||||
用于事件展示与调试，不参与状态修改。
|
||||
"""
|
||||
return _combat_strength_vs(opponent, self_avatar)
|
||||
|
||||
|
||||
def get_effective_strength_pair(a: "Avatar", b: "Avatar") -> tuple[float, float]:
|
||||
"""
|
||||
一次性返回双方（a 面对 b，b 面对 a）的折算战斗力。
|
||||
顺序：(a_strength, b_strength)
|
||||
"""
|
||||
return _combat_strength_vs(b, a), _combat_strength_vs(a, b)
|
||||
|
||||
|
||||
def calc_win_rate(attacker: "Avatar", defender: "Avatar") -> float:
|
||||
"""
|
||||
胜率计算（返回进攻方胜率 p ∈ [0.1, 0.9]）：
|
||||
- 基准：50%
|
||||
- 境界差：每高一大境界 +15%
|
||||
- 功法品阶差：按品阶差的相对加成（可正可负）
|
||||
- 属性克制：若进攻方克制防守方，再 +10%
|
||||
最后夹紧到 [0.1, 0.9]
|
||||
胜率 = sigmoid(β×战斗力差)，并夹紧到 [0.1, 0.9]
|
||||
- 战斗力差 = 有效战斗力(att) - 有效战斗力(def)
|
||||
- β 默认 0.15，使差值≈10时胜率≈0.82
|
||||
"""
|
||||
atk_order = _realm_order(attacker.cultivation_progress.realm)
|
||||
def_order = _realm_order(defender.cultivation_progress.realm)
|
||||
delta = atk_order - def_order
|
||||
base = 0.5 + 0.15 * delta
|
||||
# 功法品阶差相对加成
|
||||
atk_grade = getattr(getattr(attacker, "technique", None), "grade", None)
|
||||
def_grade = getattr(getattr(defender, "technique", None), "grade", None)
|
||||
base += get_grade_advantage_bonus(atk_grade, def_grade)
|
||||
# 属性克制：若进攻方克制防守方，再+10%
|
||||
atk_attr = getattr(getattr(attacker, "technique", None), "attribute", None)
|
||||
def_attr = getattr(getattr(defender, "technique", None), "attribute", None)
|
||||
if atk_attr is not None and def_attr is not None:
|
||||
base += get_suppression_bonus(atk_attr, def_attr)
|
||||
return max(0.1, min(0.9, base))
|
||||
diff = _strength_diff(attacker, defender)
|
||||
p = 1.0 / (1.0 + math.exp(-_WIN_BETA * diff))
|
||||
if p < 0.1:
|
||||
return 0.1
|
||||
if p > 0.9:
|
||||
return 0.9
|
||||
return p
|
||||
|
||||
|
||||
def _base_damage_scale(defender: "Avatar") -> float:
|
||||
# 以 100 HP 为基准，将 Civ6 的 24~36 损伤映射到不同境界的 HP 档位
|
||||
max_hp = defender.hp.max
|
||||
return max(1.0, max_hp / 100.0)
|
||||
|
||||
|
||||
def _damage_from_to(attacker: "Avatar", defender: "Avatar") -> int:
|
||||
"""
|
||||
使用 Civ6 风格伤害：damage = U(24,36)×scale × e^(K×差值)
|
||||
- scale = defender.maxHP / 100，使不同境界下伤害相对一致
|
||||
- 差值 = strength(att) - strength(def)
|
||||
"""
|
||||
diff = _strength_diff(attacker, defender)
|
||||
base = random.randint(_BASE_DAMAGE_LOW, _BASE_DAMAGE_HIGH) * _base_damage_scale(defender)
|
||||
dmg = base * math.exp(_CIV6_K * diff)
|
||||
return max(1, int(dmg))
|
||||
|
||||
|
||||
def _damage_pair(winner: "Avatar", loser: "Avatar") -> tuple[int, int]:
|
||||
"""
|
||||
成对伤害：使用同一基础与对称比值，保证赢家伤害严格小于败者伤害。
|
||||
- ratio = max(exp(K×|diff|), MIN_RATIO)
|
||||
- 中间尺度 = 几何均值 sqrt(scale_winner × scale_loser)
|
||||
- 败者伤害 = base × 中间尺度 × ratio
|
||||
- 赢家伤害 = base × 中间尺度 ÷ ratio
|
||||
"""
|
||||
abs_diff = abs(_strength_diff(winner, loser))
|
||||
ratio = math.exp(_CIV6_K * abs_diff)
|
||||
ratio *= _PAIR_BIAS
|
||||
if ratio < _MIN_RATIO:
|
||||
ratio = _MIN_RATIO
|
||||
|
||||
base = random.randint(_BASE_DAMAGE_LOW, _BASE_DAMAGE_HIGH)
|
||||
scale_w = _base_damage_scale(winner)
|
||||
scale_l = _base_damage_scale(loser)
|
||||
mid_scale = math.sqrt(scale_w * scale_l)
|
||||
|
||||
loser_damage = max(1, int(base * mid_scale * ratio))
|
||||
winner_damage = max(1, int(base * mid_scale / ratio))
|
||||
return loser_damage, winner_damage
|
||||
|
||||
|
||||
def decide_battle(attacker: "Avatar", defender: "Avatar") -> Tuple["Avatar", "Avatar", int, int]:
|
||||
"""
|
||||
结算一场战斗，返回(胜者, 败者, 败者掉血, 赢家掉血)。
|
||||
规则：
|
||||
- 先按 calc_win_rate 判定胜负；
|
||||
- 以 get_damage 计算基准伤害，再让败者“多掉一点血”（适度上调，例如 +15%）；
|
||||
- 赢家也会受伤，但伤害不超过败者伤害的一半（随机 15%~40% 区间）。
|
||||
结算战斗，返回 (胜者, 败者, 败者掉血, 赢家掉血)。
|
||||
- 胜率由战斗力差的逻辑函数给出；
|
||||
- 双方伤害均按 Civ6 风格由同一差值决定（对称公式），HP 与战斗力独立。
|
||||
"""
|
||||
p = calc_win_rate(attacker, defender)
|
||||
print(f"胜率: {p}")
|
||||
if random.random() < p:
|
||||
winner, loser = attacker, defender
|
||||
else:
|
||||
winner, loser = defender, attacker
|
||||
|
||||
base_damage = get_damage(winner, loser)
|
||||
# 败者多掉一点血：适度上调，保持上限由 HP.reduce 自然处理
|
||||
loser_damage = max(1, int(base_damage * 1.15))
|
||||
|
||||
# 赢家也掉血，但不超过败者的一半：在 15%~40% 的范围取随机值
|
||||
rnd_ratio = random.uniform(0.15, 0.40)
|
||||
winner_damage = int(loser_damage * rnd_ratio)
|
||||
winner_damage = max(0, min(winner_damage, loser_damage // 2))
|
||||
|
||||
loser_damage, winner_damage = _damage_pair(winner, loser)
|
||||
return winner, loser, loser_damage, winner_damage
|
||||
|
||||
|
||||
def get_escape_success_rate(attacker: "Avatar", defender: "Avatar") -> float:
|
||||
"""
|
||||
逃跑成功率：临时返回常量值，后续可基于双方能力细化。
|
||||
attacker: 追击方（通常为进攻者）
|
||||
defender: 逃跑方（通常为被攻击者）
|
||||
"""
|
||||
return 0.1
|
||||
|
||||
def get_damage(winner: "Avatar", loser: "Avatar") -> int:
|
||||
"""
|
||||
伤害计算（返回单次战斗伤害值，整数）：
|
||||
1) 先计算“期望伤害” expected：
|
||||
- 境界差：base = 100 + 80 × gap，其中 gap = max(0, winnerRealmOrder - loserRealmOrder)
|
||||
- 功法品阶差：按品阶差 bonus 调整 expected *= (1 + bonus)
|
||||
- 属性克制：若胜者克制败者，再乘 1.15
|
||||
- 夹紧：期望伤害最终限制在 [30, 500]
|
||||
2) 再生成随机区间：[low, high] = [0.85×expected, 1.15×expected]
|
||||
3) 下限与比例保护：不低于败者最大HP的 10%，并至少为 15 的硬下限
|
||||
4) 返回区间内的随机整数
|
||||
"""
|
||||
gap = max(0, _realm_order(winner.cultivation_progress.realm) - _realm_order(loser.cultivation_progress.realm))
|
||||
expected = 100 + 80 * gap
|
||||
win_grade = getattr(getattr(winner, "technique", None), "grade", None)
|
||||
lose_grade = getattr(getattr(loser, "technique", None), "grade", None)
|
||||
expected *= (1.0 + get_grade_advantage_bonus(win_grade, lose_grade))
|
||||
win_attr = getattr(getattr(winner, "technique", None), "attribute", None)
|
||||
lose_attr = getattr(getattr(loser, "technique", None), "attribute", None)
|
||||
if win_attr is not None and lose_attr is not None:
|
||||
if get_suppression_bonus(win_attr, lose_attr) > 0:
|
||||
expected *= 1.15
|
||||
# 期望伤害夹紧
|
||||
expected = max(30.0, min(500.0, expected))
|
||||
|
||||
# 设定伤害区间并随机
|
||||
low = int(expected * 0.85)
|
||||
high = int(expected * 1.15)
|
||||
|
||||
# 与最大HP挂钩的最低保护
|
||||
loser_max_hp = getattr(getattr(loser, "hp", None), "max", 0) or 0
|
||||
hp_floor = int(max(15, loser_max_hp * 0.10))
|
||||
low = max(low, hp_floor)
|
||||
high = max(high, low + 1)
|
||||
|
||||
return random.randint(low, high)
|
||||
"""逃跑成功率：后续可基于双方能力细化。"""
|
||||
return 0.1
|
||||
@@ -38,10 +38,7 @@ class StoryTeller:
|
||||
"""
|
||||
infos: Dict[str, dict] = {}
|
||||
for av in avatars:
|
||||
try:
|
||||
infos[av.name] = av.get_info(detailed=True)
|
||||
except Exception:
|
||||
infos[av.name] = getattr(av, "name", "未知角色")
|
||||
infos[av.name] = av.get_info(detailed=True)
|
||||
return infos
|
||||
|
||||
@staticmethod
|
||||
|
||||
Reference in New Issue
Block a user