#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""M5a 收口自证脚本（QC #1/#7/#8 退回项的**仓库内正式**取证工具）。

为什么存在
----------
上一轮把「新代码是否真在磁盘、证据日志是否齐备、临时补丁是否已清理」这套自证逻辑
写在工作空间根的 `tmp_patch_u7rb.py` 里，以一次性工具脚本充当 develop 交付件本体
（QC #8）。一次性脚本不入库 = 下一个人无从复跑，自证结论无法复核。本脚本把同一套
检查沉淀进模块仓库，与 `selfcheck_m5a.py` 互补：

* `selfcheck_m5a.py` —— 交付**门禁**（12 项，验模块结构/契约/幂等键/证据日志口径）
* 本脚本 —— **收口自证**（验「磁盘代码 == 门禁所声称的那份代码」+ 工作空间卫生 +
  README 与代码的门禁条数一致，防文档/代码漂移）

用法
----
    cd modules/pbl_evidence && python3 scripts/verify_m5a_closure.py
    # 全通过 → 逐条 [PASS] + `CLOSURE VERIFY ALL PASS (N/N)`，exit 0
    # 任一失败 → [FAIL] + 原因，exit 1

幂等契约：只读（不连库、不写文件、不取时间戳/随机数/绝对路径），所有明细 sorted()
后输出，重复执行逐字节一致。
"""

import ast
import json
import os
import re
import sys

ROOT = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
SELFCHK = os.path.join(ROOT, 'scripts', 'selfcheck_m5a.py')
HARNESS = os.path.join(ROOT, 'tests', 'm5a_live_db_harness.py')
README = os.path.join(ROOT, 'README.md')

# 机构工作空间根：modules/{模块}/scripts/ → up 3 级
WS_ROOT = os.path.abspath(os.path.join(ROOT, os.pardir, os.pardir))
DOCS = os.path.join(WS_ROOT, 'projects', 'pbls', 'docs', '02-develop')
ENV_TEST = os.path.join(WS_ROOT, 'projects', 'pbls', 'env', 'test.json')

# ① selfcheck 必须定义的 4 个新增门禁函数（QC #1 点名的改法）
GATE_FUNCS = (
    'check_harness_static',
    'check_no_hardcoded_credentials',
    'check_evidence_logs',
    'check_harness_matches_model',
)
# ② harness 里必须 0 命中的崩溃/吞参/明文凭据特征（QC #1/#2 点名的旧缺陷）
HARNESS_FORBIDDEN = ('test123', 'args or {}', 'params or {}', 'dict(params or {})')
# ③ 三份运行证据日志（与 selfcheck EVIDENCE_LOGS 同名单）
EVIDENCE_LOGS = ('m5a-live-db-harness.log', 'm5a-selfcheck.log', 'm5a-pytest-offline.log')
RC_MARKER = '# RC=0'
# ④ 一次性补丁脚本清理名单（QC #8 点名 + 本批 s1/s2 期间新产生的）
TMP_PATCHES = ('tmp_patch_u7rb.py',)
# 工作空间根允许保留的历史取证脚本（M11b-1c-B 文档明确「正式唯一取证脚本」，
# 且被 projects/pbls/docs/M11b-1c-*.md 逐条引用作复跑入口，删除会破坏历史验收可复现性）
TMP_KEPT = ('tmp_m11b1c_diff_check.py',)


def _read(path):
    with open(path, 'r', encoding='utf-8') as fh:
        return fh.read()


def _rel(path):
    try:
        return os.path.relpath(path, WS_ROOT).replace(os.sep, '/')
    except Exception:
        return os.path.basename(path)


# ── V1 selfcheck 四个门禁函数已落盘且被 CHECKS 装配（防「定义了但不可达」死代码）
def verify_gate_functions():
    if not os.path.isfile(SELFCHK):
        return False, '缺少 %s' % _rel(SELFCHK)
    src = _read(SELFCHK)
    try:
        tree = ast.parse(src, filename=SELFCHK)
    except Exception as exc:
        return False, 'selfcheck_m5a.py 语法错误: %s' % exc
    defined = {n.name for n in tree.body
               if isinstance(n, (ast.FunctionDef, ast.AsyncFunctionDef))}
    problems, detail = [], []
    missing = sorted(f for f in GATE_FUNCS if f not in defined)
    if missing:
        problems.append('selfcheck 缺门禁函数: %s' % ', '.join(missing))
    else:
        detail.append('4 个门禁函数已定义: %s' % ', '.join(sorted(GATE_FUNCS)))
    # 装配检查：函数名必须出现在 CHECKS 列表里，否则是永远不执行的死代码
    m = re.search(r'^CHECKS\s*=\s*\[(.*?)^\]', src, re.S | re.M)
    if not m:
        problems.append('未找到 CHECKS 装配表')
    else:
        unwired = sorted(f for f in GATE_FUNCS
                         if not re.search(r'\b' + re.escape(f) + r'\b(?!\s*def)', m.group(1)))
        if unwired:
            problems.append('门禁函数未装配进 CHECKS（死代码）: %s' % ', '.join(unwired))
        else:
            detail.append('均已装配进 CHECKS 装配表')
    if problems:
        return False, '; '.join(problems)
    return True, ' | '.join(detail)


# ── V2 harness 修复到位：崩溃版特征 0 命中
def verify_harness_clean():
    if not os.path.isfile(HARNESS):
        return False, '缺少 %s' % _rel(HARNESS)
    src = _read(HARNESS)
    hits = []
    for token in HARNESS_FORBIDDEN:
        n = src.count(token)
        if n:
            hits.append('%s×%d' % (token, n))
    if hits:
        return False, 'harness 仍含崩溃/吞参/明文凭据特征: %s' % ', '.join(sorted(hits))
    try:
        ast.parse(src, filename=HARNESS)
    except Exception as exc:
        return False, 'harness 语法错误: %s' % exc
    return True, ('%s 零命中 %s，且 ast.parse 通过（修复版已在磁盘）'
                  % (_rel(HARNESS), ' / '.join(HARNESS_FORBIDDEN)))


# ── V3 env/test.json 的 db.sandbox 段（凭据唯一事实源，代码内零明文）
def verify_env_sandbox():
    if not os.path.isfile(ENV_TEST):
        return False, '缺少 %s' % _rel(ENV_TEST)
    try:
        conf = json.loads(_read(ENV_TEST))
    except Exception as exc:
        return False, 'env/test.json 不是合法 JSON: %s' % exc
    db = conf.get('db') or {}
    sb = db.get('sandbox')
    if not isinstance(sb, dict):
        return False, 'env/test.json 缺 db.sandbox 段（harness 无凭据事实源）'
    problems, detail = [], []
    if sb.get('scope') != 'sandbox_only':
        problems.append('db.sandbox.scope != sandbox_only（会被误用于业务库）')
    else:
        detail.append('scope=sandbox_only')
    if not sb.get('sandbox_schema'):
        problems.append('db.sandbox 缺 sandbox_schema（沙箱库名未声明）')
    else:
        detail.append('sandbox_schema=%s' % sb['sandbox_schema'])
    # 沙箱库由 harness 运行时 CREATE DATABASE / DROP DATABASE，故不预置 dbname；
    # 连接必填四项 + 自建 schema 名，才是这段配置的真实语义。
    for key in ('host', 'port', 'user', 'password'):
        if key not in sb:
            problems.append('db.sandbox 缺连接字段 %s' % key)
    if problems:
        return False, '; '.join(problems)
    detail.append('连接必填字段齐备(host/port/user/password)，schema 由 harness 自建自删')
    return True, ' | '.join(detail)


# ── V4 三份证据日志齐备，且 harness 日志含 C-3 fail-closed + 1062 真防重原文
def verify_evidence_logs():
    if not os.path.isdir(DOCS):
        return False, '缺少证据目录 %s' % _rel(DOCS)
    top = sorted(f for f in os.listdir(DOCS) if os.path.isfile(os.path.join(DOCS, f)))
    problems, detail = [], []
    absent = sorted(set(EVIDENCE_LOGS) - set(top))
    if absent:
        problems.append('缺少证据日志: %s' % ', '.join(absent))
    else:
        detail.append('三份日志齐备: %s' % ', '.join(EVIDENCE_LOGS))
    for name in EVIDENCE_LOGS:
        if name not in top:
            continue
        text = _read(os.path.join(DOCS, name))
        if name == 'm5a-selfcheck.log':
            n = len(re.findall(r'^\[(?:PASS|FAIL)\] \d+\.', text, re.M))
            if n != 12:
                problems.append('m5a-selfcheck.log [PASS]/[FAIL] 行数 %d != 12' % n)
            elif 'ALL PASS (12/12 checks)' not in text:
                problems.append('m5a-selfcheck.log 结论行不是 ALL PASS (12/12 checks)')
            else:
                detail.append('自检日志 12 行齐备且结论为 ALL PASS (12/12 checks)')
        elif RC_MARKER not in text:
            problems.append('%s 缺 `%s` 收尾标记' % (name, RC_MARKER))
        if re.search(r'^Traceback \(most recent call last\)', text, re.M):
            problems.append('%s 含裸 Traceback 崩溃现场' % name)
    if not problems and 'm5a-live-db-harness.log' in top:
        hlog = _read(os.path.join(DOCS, 'm5a-live-db-harness.log'))
        need = ['C-3.1', 'C-3.2', 'C-3.3', 'C-3.4', 'IDM.6']
        miss = sorted(t for t in need
                      if not re.search(r'\[PASS\]\s*' + re.escape(t) + r'\b', hlog))
        if miss:
            problems.append('harness 日志缺 [PASS] 行: %s' % ', '.join(miss))
        elif '1062' not in hlog or 'uk_ev_dedup' not in hlog:
            problems.append('harness 日志缺 1062 Duplicate entry / uk_ev_dedup 真防重原文')
        else:
            detail.append('harness 日志含 C-3.1~C-3.4 全 [PASS] + IDM.6 1062(uk_ev_dedup) 原文')
    if problems:
        return False, '; '.join(problems)
    return True, ' | '.join(detail)


# ── V5 工作空间卫生：QC #8 点名的一次性补丁已删，历史取证脚本按文档保留
def verify_tmp_hygiene():
    problems, detail = [], []
    still = sorted(f for f in TMP_PATCHES if os.path.isfile(os.path.join(WS_ROOT, f)))
    if still:
        problems.append('一次性补丁脚本仍在磁盘: %s' % ', '.join(still))
    else:
        detail.append('QC #8 点名脚本已删除: %s' % ', '.join(TMP_PATCHES))
    left = sorted(f for f in os.listdir(WS_ROOT)
                  if f.startswith('tmp_') and f.endswith('.py'))
    unexpected = sorted(set(left) - set(TMP_KEPT))
    if unexpected:
        problems.append('工作空间根仍有未清理的一次性补丁: %s' % ', '.join(unexpected))
    kept = sorted(set(left) & set(TMP_KEPT))
    if kept:
        detail.append('按 M11b-1c-B 文档保留历史取证脚本: %s' % ', '.join(kept))
    if problems:
        return False, '; '.join(problems)
    return True, ' | '.join(detail)


# ── V6 README 门禁条数与代码实际条数一致（防文档/代码漂移）
def verify_readme_matches_code():
    problems, detail = [], []
    if not os.path.isfile(README):
        return False, '缺少 README.md'
    if not os.path.isfile(SELFCHK):
        return False, '缺少 scripts/selfcheck_m5a.py'
    rtxt = _read(README)
    stxt = _read(SELFCHK)
    try:
        tree = ast.parse(stxt, filename=SELFCHK)
    except Exception as exc:
        return False, 'selfcheck 语法错误: %s' % exc
    n_checks = None
    for node in tree.body:
        if isinstance(node, ast.Assign) and any(
                getattr(t, 'id', None) == 'CHECKS' for t in node.targets):
            if isinstance(node.value, ast.List):
                n_checks = len(node.value.elts)
    if n_checks is None:
        return False, '无法从 selfcheck 解析 CHECKS 条数'
    claimed = sorted(set(re.findall(r'ALL PASS \((\d+)/\d+ checks\)', rtxt)))
    if claimed != [str(n_checks)]:
        problems.append('README 声称的门禁条数 %s 与代码 CHECKS 条数 %d 不一致'
                        % (claimed, n_checks))
    else:
        detail.append('README 门禁条数 = 代码 CHECKS 条数 = %d' % n_checks)
    if problems:
        return False, '; '.join(problems)
    return True, ' | '.join(detail)


CHECKS = [
    ('selfcheck 四个新增门禁函数已落盘且装配进 CHECKS', verify_gate_functions),
    ('harness 修复版在磁盘：test123/args or {} 等 0 命中 + ast 可解析', verify_harness_clean),
    ('env/test.json db.sandbox 段齐备（scope=sandbox_only，凭据唯一事实源）', verify_env_sandbox),
    ('三份运行证据日志齐备（C-3.1~C-3.4 + 1062 真防重原文 + # RC=0）', verify_evidence_logs),
    ('工作空间根一次性补丁脚本已清理（QC #8）', verify_tmp_hygiene),
    ('README 门禁条数与 selfcheck CHECKS 条数一致', verify_readme_matches_code),
]


def main():
    print('verify_m5a_closure: target = modules/pbl_evidence')
    failed = []
    for i, (name, fn) in enumerate(CHECKS, 1):
        try:
            ok, detail = fn()
        except Exception as exc:
            ok, detail = False, '检查执行异常 %s: %s' % (type(exc).__name__, exc)
        print('[%s] V%d. %s' % ('PASS' if ok else 'FAIL', i, name))
        print('       %s' % detail)
        if not ok:
            failed.append(name)
    if failed:
        print('CLOSURE VERIFY FAILED (%d/%d)' % (len(failed), len(CHECKS)))
        for n in failed:
            print('  - %s' % n)
        return 1
    print('CLOSURE VERIFY ALL PASS (%d/%d)' % (len(CHECKS), len(CHECKS)))
    return 0


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