#!/usr/bin/env python3
r"""doc_term_audit.py — GDD 退役术语扫描（半自动）

动机：机制换名/退役批（如 2026-09-20 伏击→警觉）之后，正文里容易残留旧措辞的
"现行语义"引用——修订记录与退役注记里的旧词是合法历史，混在一起肉眼难分。
本工具维护一份退役词表，全文档扫描命中并按上下文分两类：
  [疑似现行]  该行不含日期/退役/换装等白名单信号，也没落在版本历史节——人工判读
  [留档类]  行内带日期、退役/作废/换装/继任/留档/历史等信号，或位于修订记录节——默认忽略

用法:
  python tools/doc_term_audit.py                  # 扫默认范围，报告模式（exit 0）
  python tools/doc_term_audit.py --strict         # 有疑似现行命中时 exit 1（供门禁/CI）
  python tools/doc_term_audit.py <file.md> ...    # 只扫指定文件
  python tools/doc_term_audit.py --json           # 机器可读

口径：只扫 design/gdd/ design/registry/——design/research
（历史提案）、docs/（样例）与 production/ 不在扫描域；词表扩充 = 改 RETIRED_TERMS。
"""
import argparse
import json
import re
import sys
from pathlib import Path

# 退役词表：(正则词, 退役日期, 继任词/口径, 备注)。新增退役批时在此追加一行。
RETIRED_TERMS = [
    ("伏击", "2026-09-20", "警觉（blast 负向）", "伏击机制整体退役"),
    ("ambush_double|ambush_layers|flat_attach", "2026-09-20", "enemy_alert / effects 投影", "伏击数据键退役"),
    ("advantage_per_round_cap", "2026-09-20", "（已删字段——行动点天然封顶）", "单轮授予上限删除批"),
    ("溢出未生效", "2026-09-20", "（行类型退役；满层拒收走奖励未生效行）", "行协议 10→9 类"),
    ("m_amb|M_AMB", "2026-09-20", "（伏击暴露当量口径作废）", "挖掘风险不等式退役"),
    ("七型|五型×正负|五型 × 正负", "2026-09-20", "八型（十六格）", "格面资源八型批"),
    ("10 类行|行型封闭为 10 类|10 类封闭协议|行型 10 类", "2026-09-20", "9 类", "溢出未生效行下线"),
    ("拔旗|撤旗", "2026-09-19", "（旗无移除手段——插旗不可逆）", "插旗不可逆批"),
    ("三选一反噬|标错反噬|反噬三选一", "2026-09-15", "极性矩阵负向兑付", "格面资源批"),
    ("defense_up", "2026-09-16", "ac_up（护甲强化）", "护甲等级掷点批换装"),
    ("蓄力行|charge 行", "2026-09-16", "施放行（cast）", "敌人技能池批改型"),
    ("行为环", "2026-09-16", "技能池（池模板+摇号）", "固定出招剧本退役"),
    ("heavy_multiplier|重击倍率", "2026-09-16", "技能行 damage_mult", "行为环退役随批"),
    ("多段命中", "2026-09-19", "（退役，无防御磨损体系）", "技能池新效应词汇批"),
    ("充能", "2026-09-19", "（挖掘期/战斗段充能退役；蓄力 empower 仍在）", "充能退役批——注意误报"),
    ("英雄卡|敌人卡", "2026-09-20", "站桩立绘", "敌人卡形态裁决批"),
    ("道具窗口|可选类道具|已消耗未生效|使用被拒|窗口打开锚行|窗口跳过锚行", "2026-09-15", "（随可选类道具整体删除）", "格面资源批"),
    ("雾格", "2026-09-14", "（雾格移除）", "不做清单"),
    ("兄弟精", "2026-09-26", "L2 熊地精（bugbear 复用）", "变体名未经批准退役批——用户质问查证后全仓清除"),
]

DEFAULT_DIRS = ["design/gdd", "design/registry"]

# 行级白名单信号：命中任一即归"留档类"
WHITELIST_RE = re.compile(
    r"2026-\d{2}-\d{2}"
    r"|退役|作废|换装|继任|留档|历史口径|已下线|随批|随伏击|不再|已删除"
    r"|\bv0\.\d+"
    r"|修订记录|版本历史|履行回执|台账"
)
# 节级白名单：版本历史/修订记录节内整段豁免
SECTION_RE = re.compile(r"^#{1,6}\s*(版本历史|修订记录)")
COMMENT_RE = re.compile(r"^\s*[;#]")


def scan(paths):
    terms = [(re.compile(t), date, repl, note) for t, date, repl, note in RETIRED_TERMS]
    findings = []   # {term, file, line, text, klass, retired, replacement}
    for p in paths:
        text = Path(p).read_text(encoding="utf-8").replace("\r\n", "\n")
        in_version_section = False
        prev_lines = []   # 已处理行（供"上一 2 行留档标记继承"判读）
        for lineno, line in enumerate(text.split("\n"), 1):
            if SECTION_RE.match(line):
                in_version_section = True
            hits = [(m.group(0), t) for t in terms for m in [t[0].search(line)] if m]
            # 上一 2 行带留档/作废/历史标记（如"下表留档作废"头注）→ 本行视为留档
            prev = " ".join(prev_lines[-2:]) if prev_lines else ""
            prev_lines.append(line)
            if not hits:
                continue
            if in_version_section or WHITELIST_RE.search(line) or WHITELIST_RE.search(prev) or COMMENT_RE.match(line):
                klass = "留档类"
            else:
                klass = "疑似现行"
            findings.append({
                "term": "、".join(sorted({h[0] for h in hits})),
                "file": str(p).replace("\\", "/"),
                "line": lineno,
                "klass": klass,
                "text": line.strip()[:160],
                "retired": hits[0][1][1],
                "replacement": hits[0][1][2],
            })
    return findings


def main():
    ap = argparse.ArgumentParser(description="GDD 退役术语扫描（半自动）")
    ap.add_argument("files", nargs="*", help="指定文件；缺省扫默认目录")
    ap.add_argument("--strict", action="store_true", help="疑似现行命中时 exit 1")
    ap.add_argument("--json", action="store_true")
    args = ap.parse_args()

    if args.files:
        paths = args.files
    else:
        paths = []
        for d in DEFAULT_DIRS:
            paths += sorted(Path(d).rglob("*.md"))
            paths += sorted(Path(d).rglob("*.yaml"))
    paths = [p for p in paths if Path(p).exists()]

    findings = scan(paths)
    suspect = [f for f in findings if f["klass"] == "疑似现行"]
    kept = [f for f in findings if f["klass"] != "疑似现行"]

    if args.json:
        print(json.dumps({"suspect": suspect, "archived_count": len(kept)}, ensure_ascii=False, indent=1))
    else:
        print(f"扫描 {len(paths)} 个文件 · 退役词命中 {len(findings)} 处 · 疑似现行 {len(suspect)} 处（留档类 {len(kept)} 处默认忽略）")
        by_term = {}
        for f in suspect:
            by_term.setdefault(f["term"], []).append(f)
        for term, items in sorted(by_term.items()):
            print(f"\n【{term}】×{len(items)}（退役 {items[0]['retired']} → {items[0]['replacement']}）")
            for it in items[:8]:
                print(f"  {it['file']}:{it['line']}  {it['text']}")
            if len(items) > 8:
                print(f"  …另有 {len(items) - 8} 处")
    return 1 if (args.strict and suspect) else 0


if __name__ == "__main__":
    sys.exit(main())