Files
lc-mcp/scripts/build_theme_pack_systems.py
ywsandClaude Opus 5.5 96a31bc1ae 按台账修镜牢: 饰品弹窗兜底、按键扫描码、卡包按体系加分
按 docs/ISSUES.md 处理本机回归后待修/待数据的条目:

- I-06/R-04 开局「获得E.G.O饰品」弹窗死循环: 确认模板连续第 3 次命中起改点
  按钮中心(勾号 +76/1440), 连续 20 次还在就记 [ERROR]、抛 TaskAbort 结束任务
  (上游 continue 跳过了循环计数)。单测执行上游真实 select_init_ego_gift。
- I-05/R-03 键盘寻路 0 成功: 前台按键改发带扫描码和扩展键标志的 keybd_event,
  各模式按住 0.12 秒(上游注释: Unity 6 校验扫描码/扩展键); 连续 3 个节点进不去
  就本次任务停用键盘寻路(方案 B)。MCP key 工具前台同样处理。LCP_KEY_HOLD=0 关。
- I-07/R-02 事件表 6 -> 9 条: 蜘蛛巢、归乡路途两段(拿「归途」), 按 wiki.gg 核对结果;
  全表按本机装的零协 release 2026111301 重新核对唯一性。
- I-13 复查 5992827: 「进入」先模板确认再 OCR(原来每轮都跑一次整屏 OCR);
  选队页重试重新截图; 入场切队失败记 [ERROR] 且本局不再重翻列表。
- R-01/I-08 卡包按体系加分: wiki.gg 122 个卡包(楼层/专属饰品/Featured 分类) +
  零协中文名生成 lc_mcp/data/theme_pack_systems.json(scripts/build_theme_pack_systems.py);
  灰机 wiki 对云端 403、BWIKI 无卡包条目, 中文社区攻略用 codex CLI 联网检索,
  推荐/避开并入表的 community。选卡包时临时加到 AALC 权重上, 按当前难度只看会出现
  的卡包; 审计记 system_bonus。LCP_THEME_PACK_SYSTEMS=0 关。
- 文档: ISSUES/BACKLOG/README 更新, 新增 docs/feedback/2026-11-15/REGRESSION.md
  (本轮回归清单)和 theme-pack-research.md(检索记录)。

单测 703 个全过(Linux)。实机未验。

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-15 01:24:46 +00:00

158 lines
7.0 KiB
Python
Raw Permalink Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
# -*- coding: utf-8 -*-
"""生成 lc_mcp/data/theme_pack_systems.json: 镜牢卡包 → 体系(BACKLOG R-01)。
用法:
python scripts/build_theme_pack_systems.py <LLC_zh-CN 目录> <LLC EN 目录> [-o 输出文件]
两个目录分别是零协汉化的中文包(游戏里的 LimbusCompany_Data/Lang/LLC_zh-CN, 或 release 解压出来的同名目录)
和 LocalizeLimbusCompany 仓库里的 EN 目录, 只用其中的 MirrorDungeonTheme*.json(卡包 id → 名字)。
卡包数据来自 Limbus Company Wiki(https://limbuscompany.wiki.gg, CC BY-SA 4.0), 运行时在线读取:
- 各卡包条目的 {{ThemePackInfo}}: 出现楼层(normal/hard)、专属饰品(unique);
- 条目里「Featured Gifts」用的饰品分类: 是体系名(Burn/Charge/...)的就是体系包;
- Module:EgoGift/data: 每件饰品的体系(category)。
边狱巴士 BWIKI(wiki.biligame.com/limbuscompany)2026-11-15 查过: 699 个页面里没有卡包条目, 所以不用它。
"""
import argparse
import glob
import json
import os
import re
import sys
import time
import urllib.parse
import urllib.request
API = 'https://limbuscompany.wiki.gg/api.php'
RAW = 'https://limbuscompany.wiki.gg/index.php'
USER_AGENT = 'research-script/1.0 (python-urllib)'
CATEGORY = 'Category:Floor Theme Pack Pages'
SYSTEMS = {'Burn': 'burn', 'Bleed': 'bleed', 'Tremor': 'tremor', 'Rupture': 'rupture', 'Sinking': 'sinking',
'Poise': 'poise', 'Charge': 'charge', 'Slash': 'slash', 'Pierce': 'pierce', 'Blunt': 'blunt'}
# 零协 EN 文件里还没有英文名、或者写法和 wiki 条目名不一样的卡包
EN_OVERRIDES = {
1028: 'Storefront Campout', 1029: 'An Anarchy of Colors', 1030: 'Creed of Preservation',
1031: 'Readvent of Gluttony', 1111: 'Line 4 - Section 3', 1112: 'Line 4 - Section 4',
1129: 'The Wedge in the Wake of Dawnbreak',
}
def fetch(url, params):
request = urllib.request.Request(url + '?' + urllib.parse.urlencode(params), headers={'User-Agent': USER_AGENT})
with urllib.request.urlopen(request, timeout=60) as response:
return response.read().decode('utf-8')
def theme_pack_pages():
titles, cont = [], {}
while True:
data = json.loads(fetch(API, {'action': 'query', 'list': 'categorymembers', 'cmtitle': CATEGORY,
'cmlimit': '500', 'format': 'json', **cont}))
titles += [m['title'] for m in data['query']['categorymembers'] if m['title'].endswith(' Theme Pack')]
if 'continue' not in data:
break
cont = {'cmcontinue': data['continue']['cmcontinue']}
pages = {}
for i in range(0, len(titles), 40):
data = json.loads(fetch(API, {'action': 'query', 'prop': 'revisions', 'rvprop': 'content', 'rvslots': 'main',
'titles': '|'.join(titles[i:i + 40]), 'format': 'json', 'formatversion': '2'}))
for page in data['query']['pages']:
if 'revisions' in page:
pages[page['title'][:-len(' Theme Pack')]] = page['revisions'][0]['slots']['main']['content']
time.sleep(1)
return pages
def gift_categories():
lua = fetch(RAW, {'title': 'Module:EgoGift/data', 'action': 'raw'})
gifts = {}
for match in re.finditer(r'^\s*\["([^"]+)"\]\s*=\s*\{(.*)\},?\s*$', lua, re.M):
category = re.search(r'category\s*=\s*"([^"]*)"', match.group(2))
gifts[match.group(1)] = category.group(1) if category else None
return gifts
def parse_pack(text, gifts):
info = re.search(r'\{\{ThemePackInfo(.*?)\n\}\}', text, re.S)
fields = dict(re.findall(r'\|\s*(\w+)\s*=\s*([^\n|]*)', info.group(1))) if info else {}
unique = {}
for gift in (g.strip() for g in fields.get('unique', '').split(';')):
category = gifts.get(gift, gifts.get(gift.replace("'", '’'))) if gift else None
if category in SYSTEMS:
unique[SYSTEMS[category]] = unique.get(SYSTEMS[category], 0) + 1
featured = sorted({SYSTEMS[c] for c in re.findall(r'\{\{#invoke:GiftList\|Category\|([^}|]+)\}\}', text)
if c in SYSTEMS})
return {'normal': fields.get('normal', '').replace(' ', ''), 'hard': fields.get('hard', '').replace(' ', ''),
'featured': featured, 'unique': dict(sorted(unique.items()))}
def theme_names(directory):
names = {}
for path in glob.glob(os.path.join(directory, 'MirrorDungeonTheme*.json')):
with open(path, encoding='utf-8-sig') as fh:
for entry in json.load(fh).get('dataList', []):
names[entry['id']] = entry.get('name', '')
return names
def key(name):
return re.sub(r'[^0-9a-z]', '', name.lower())
def build(zh_dir, en_dir):
zh, en = theme_names(zh_dir), theme_names(en_dir)
en.update(EN_OVERRIDES)
gifts = gift_categories()
wiki = {key(title): (title, parse_pack(text, gifts)) for title, text in theme_pack_pages().items()}
packs, unmatched = [], []
for pack_id in sorted(zh):
title, data = wiki.get(key(en.get(pack_id, '')), (None, None))
if data is None:
unmatched.append('%d %s' % (pack_id, zh[pack_id]))
continue
packs.append({'id': pack_id, 'zh': zh[pack_id], 'en': title, **data})
return packs, unmatched
def _rows(items, indent):
return ['%s%s%s' % (indent, json.dumps(item, ensure_ascii=False), ',' if i < len(items) - 1 else '')
for i, item in enumerate(items)]
def dump_table(table):
"""写回 JSON: 卡包和社区条目每条一行, 好看 diff。"""
notes = json.dumps(table['_说明'], ensure_ascii=False, indent=1).replace('\n', '\n ')
lines = ['{', ' "_说明": %s,' % notes, ' "bonus": %s,' % json.dumps(table['bonus'], ensure_ascii=False)]
community = table.get('community')
if community:
lines += [' "community": {',
' "_说明": %s,' % json.dumps(community['_说明'], ensure_ascii=False, indent=1).replace('\n', '\n '),
' "recommend": ['] + _rows(community['recommend'], ' ') + \
[' ],', ' "avoid": ['] + _rows(community['avoid'], ' ') + [' ]', ' },']
lines += [' "packs": ['] + _rows(table['packs'], ' ') + [' ]', '}']
return '\n'.join(lines) + '\n'
def main(argv):
parser = argparse.ArgumentParser(description=__doc__, formatter_class=argparse.RawDescriptionHelpFormatter)
parser.add_argument('zh_dir')
parser.add_argument('en_dir')
parser.add_argument('-o', '--output', default=os.path.join(os.path.dirname(__file__), '..', 'lc_mcp', 'data',
'theme_pack_systems.json'))
args = parser.parse_args(argv)
packs, unmatched = build(args.zh_dir, args.en_dir)
with open(args.output, encoding='utf-8') as fh:
table = json.load(fh) # 保留手写的 _说明、bonus、community, 只换 packs
table['packs'] = packs
with open(args.output, 'w', encoding='utf-8') as fh:
fh.write(dump_table(table))
print('卡包 %d 个写入 %s' % (len(packs), args.output))
if unmatched:
print('wiki 上没对上的: %s' % ', '.join(unmatched))
return 0
if __name__ == '__main__':
sys.exit(main(sys.argv[1:]))