#!/usr/bin/env python3
"""词库自检：JSON 合法 / 正则能编译 / 占位符与命名组一一对应 / 候选条数。

用法：python3 validate_phrases.py [phrases.json]
每次改完词库都跑一遍——占位符写错（候选用 {n} 而正则组叫 code）不会报错，
只会在运行时把 {n} 字面量发给用户，肉眼很难发现。
"""
import json, re, sys, pathlib

path = pathlib.Path(sys.argv[1] if len(sys.argv) > 1 else "phrases.json")
data = json.loads(path.read_text(encoding="utf-8"))
cats = data.get("categories", [])
problems, total, minv = [], 0, 10 ** 9
for c in cats:
    cid = c.get("id", "?")
    groups = set()
    for pat in c.get("patterns", []):
        try:
            groups |= set(re.compile(pat, re.IGNORECASE).groupindex)
        except re.error as exc:
            problems.append(f"{cid}: 正则编译失败 {pat!r} -> {exc}")
    variants = c.get("variants") or []
    total += len(variants)
    minv = min(minv, len(variants))
    for v in variants:
        missing = set(re.findall(r"\{(\w+)\}", v)) - groups
        if missing:
            problems.append(f"{cid}: 候选用了 {sorted(missing)}，正则里没有这些组（{sorted(groups)}）")
    if not c.get("patterns"):
        problems.append(f"{cid}: 没有 patterns")
    if not variants:
        problems.append(f"{cid}: 没有 variants")

print(f"类别 {len(cats)} | 候选 {total} | 最少的一类 {minv} 条")
if problems:
    print(f"\n✗ 发现 {len(problems)} 个问题：")
    for p in problems:
        print("   -", p)
    sys.exit(1)
print("✓ 全部通过")