guanlan-src/scripts/java_contract_parity.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-
r"""P12-A：Java 后端迁移的**裁判**——契约端点逐一比对 Java 与 Python 的响应。

用途（与前端对拍门同思路：门禁当裁判，逐条收敛）：
  · 读冻契约 `app_backEnd/.../contract/http_api_v1.json`（42 条）；
  · 对每条端点，分别请求 Python 侧与 Java 侧，做**深度相等**比较（JSON 归一化后）；
  · 输出计分板：`已一致 / 有差异 / Java 未实现`，并把"必须一致"的端点清单交给 `--require` 用于门禁。

用法：
  python scripts/java_contract_parity.py [--gateway URL] [--python-base /detail] [--java-base /java]
                                         [--require kpi,fleet,...] [--limit N]
                                         [--java-kinds detail_api]

★预热：比对前两侧各 GET 一次（丢弃结果）再取第二次 —— 带"按窗重算/进程内缓存"的端点（如 /api/fleet）
  冷启时点不同会报假差异（实逮 13≠15，预热后均 15）。
"""
from __future__ import annotations

import argparse
import json
import pathlib
import sys
import urllib.error
import urllib.request

ROOT = pathlib.Path(__file__).resolve().parents[1]
CONTRACT = ROOT / 'app_backEnd/app_backEnd_guanlan/contract/http_api_v1.json'


def get(url: str, timeout: float = 60.0):
    req = urllib.request.Request(url, headers={'Accept': 'application/json'})
    try:
        with urllib.request.urlopen(req, timeout=timeout) as r:
            body = r.read().decode('utf-8', 'replace')
            try:
                return r.status, json.loads(body)
            except Exception:
                return r.status, body[:400]
    except urllib.error.HTTPError as e:
        return e.code, (e.read().decode('utf-8', 'replace')[:200])
    except Exception as e:                                   # noqa: BLE001
        return 0, f'{type(e).__name__}: {e}'



def get_bytes(url: str, timeout: float = 900.0):
    """取原始字节（二进制端点用）；失败返回 None。"""
    import urllib.error
    import urllib.request
    try:
        with urllib.request.urlopen(urllib.request.Request(url), timeout=timeout) as r:
            return r.read()
    except Exception:                                                        # noqa: BLE001
        return None



def post_json(url: str, payload: dict, timeout: float = 300.0):
    """POST JSON（裁判的 POST 能力）：返回 (status, 解析后的 JSON 或错误文本)。"""
    import urllib.error
    import urllib.request
    data = json.dumps(payload).encode('utf-8')
    req = urllib.request.Request(url, data=data, headers={'Content-Type': 'application/json'})
    try:
        with urllib.request.urlopen(req, timeout=timeout) as r:
            body = r.read().decode('utf-8', 'replace')
            try:
                return r.status, json.loads(body)
            except Exception:                                                # noqa: BLE001
                return r.status, body[:300]
    except urllib.error.HTTPError as e:
        return e.code, e.read().decode('utf-8', 'replace')[:300]
    except Exception as e:                                                   # noqa: BLE001
        return 0, '%s: %s' % (type(e).__name__, e)


def norm(x):
    """比较用归一化：字典按键排序；列表保序（顺序也是行为）。"""
    if isinstance(x, dict):
        return {k: norm(v) for k, v in sorted(x.items())}
    if isinstance(x, list):
        return [norm(v) for v in x]
    return x



def doc_text(b: bytes) -> str | None:
    """Office(zip) 文档 → 归一化文本：剔除时间戳/文档属性后逐条目拼接（用于二进制端点的等价判定）。"""
    if not b[:2] == b'PK':
        return None
    import io
    import re as _re
    import zipfile
    out = []
    with zipfile.ZipFile(io.BytesIO(b)) as z:
        for name in sorted(z.namelist()):
            if name.endswith('.xml') or name.endswith('.rels'):
                s = z.read(name).decode('utf-8', 'replace')
                s = _re.sub(r'<dcterms:(created|modified)[^>]*>[^<]*</dcterms:\1>', '', s)
                s = _re.sub(r'<cp:lastModifiedBy>[^<]*</cp:lastModifiedBy>', '', s)
                s = _re.sub(r'<TotalTime>[^<]*</TotalTime>', '', s)
                out.append(name + '\n' + s)
    return '\n'.join(out)


def first_diff(a, b, path='$', depth=0):
    if depth > 6:
        return None
    if type(a) is not type(b) and not (isinstance(a, (int, float)) and isinstance(b, (int, float))):
        return f'{path}: 类型 {type(a).__name__} ≠ {type(b).__name__}'
    if isinstance(a, dict):
        for k in sorted(set(a) | set(b)):
            if k not in a:
                return f'{path}.{k}: Java 多键'
            if k not in b:
                return f'{path}.{k}: Java 缺键'
            d = first_diff(a[k], b[k], f'{path}.{k}', depth + 1)
            if d:
                return d
        return None
    if isinstance(a, list):
        if len(a) != len(b):
            return f'{path}: 长度 {len(a)} ≠ {len(b)}'
        for i, (x, y) in enumerate(zip(a, b)):
            d = first_diff(x, y, f'{path}[{i}]', depth + 1)
            if d:
                return d
        return None
    if a != b:
        sa, sb = str(a), str(b)
        return f'{path}: {sa[:60]!r} ≠ {sb[:60]!r}'
    return None


# 登记忽略字段（每条都要写理由；只忽略"有状态"字段，不忽略实现口径）。
#
# `/api/reload` 的 `old_stamp` = 清缓存**前**的产物指纹 —— 取决于比对顺序与两侧各自的缓存状态：
# 先被问的一侧缓存里有指纹（返回字符串），后被问的一侧可能已被其它探测清空（None）。
# 其余字段（ok / files / stamp）照常严格比对。
IGNORE_PATHS: dict[str, tuple[str, str]] = {
    '/api/reload': ('$.old_stamp', '清缓存前的指纹：取决于比对顺序与各自的缓存状态，非实现差异'),
}


def main() -> int:
    ap = argparse.ArgumentParser()
    ap.add_argument('--gateway', default='http://127.0.0.1:28084')
    ap.add_argument('--python-base', default='/detail')
    ap.add_argument('--java-base', default='/java')
    ap.add_argument('--require', default='', help='逗号分隔：这些端点必须一致，否则退出码 1')
    ap.add_argument('--limit', type=int, default=0)
    ap.add_argument('--java-kinds', default='detail_api',
                    help='哪些 kind 属于 Java 后端迁移范围（默认 detail_api；逗号分隔）')
    a = ap.parse_args()

    contract = json.loads(CONTRACT.read_text(encoding='utf-8'))
    eps = contract['endpoints']
    rows = eps if isinstance(eps, list) else [dict(path=k, **(v if isinstance(v, dict) else {})) for k, v in eps.items()]
    if a.limit:
        rows = rows[:a.limit]

    java_kinds = {x.strip() for x in a.java_kinds.split(',') if x.strip()}
    same, diff, missing, skipped, other, unsettled, ignored = [], [], [], [], [], [], []
    scope_total = 0
    for e in rows:
        path = str(e.get('path') or e.get('url'))
        method = str(e.get('method') or 'GET').upper()
        kind = str(e.get('kind') or '?')
        if kind not in java_kinds:
            other.append((path, kind))
            continue
        scope_total += 1
        # ★ 一律真的问两侧：Java 走 /java 前缀、Python 走 /detail 前缀（灰度期两个并存入口）
        jurl = a.gateway.rstrip('/') + a.java_base + path
        purl = a.gateway.rstrip('/') + a.python_base + path
        if method != 'GET':
            # ★POST 能力（P12）：按"契约抓取时的最小请求"发空体 {} —— 两侧都应回同样的错误语义
            #   （如 /api/ask 空问题 ⇒ {err: 问题为空}）。请求体若将来需要参数，再按契约补。
            js, jb = post_json(jurl, {})
            ps, pb = post_json(purl, {})
            if js in (0, 404):
                missing.append((path, 'Java HTTP %s' % js))
                continue
            d = first_diff(norm(jb), norm(pb))
            if d:
                diff.append((path, d))
            else:
                same.append(path)
            continue
        # ★二进制端点（契约 ctype 非 JSON，如 /api/rpt_export 的 docx）：走"字节 + 解包文本"比对
        ctype = str(e.get('ctype') or '')
        if ctype and 'json' not in ctype.lower():
            jb_b = get_bytes(jurl)
            pb_b = get_bytes(purl)
            if jb_b is None or pb_b is None:
                diff.append((path, 'Java 或 Python 未返回字节（HTTP %s/%s）' % (jb_b, pb_b)))
                continue
            tj, tp = doc_text(jb_b), doc_text(pb_b)
            if tj is not None and tp is not None:
                if tj == tp:
                    same.append(path)
                else:
                    d = first_diff({'t': tj}, {'t': tp})
                    diff.append((path, '解包文本不一致：' + str(d)[:120]))
            else:
                diff.append((path, '非 zip 字节：Java %d B / Python %d B' % (len(jb_b), len(pb_b))))
            continue

        # ★预热：该端点可能带"按窗重算 + 进程内缓存"（如 /api/fleet、/api/curves），
        #   冷启时点不同会报假差异。做法：各取一次丢弃结果 → 若仍有 *_pending=true 就轮询到落定 → 再比第二次。
        get(jurl)
        get(purl)
        # 再各取一次：对 /api/reload 这类**清缓存副作用**端点，连比两次会让"先比的一侧"持有旧指纹、
        # 后比的一侧为 None（实逮 old_stamp str ≠ None）⇒ 两次预热后两侧都处于"刚清过"的稳态。
        get(jurl)
        get(purl)

        def pending(x):
            if isinstance(x, dict):
                if x.get('building') is True:                     # 曲线等服务端后台算时的"建设中"标记
                    return True
                return any(str(k).endswith('_pending') and v is True for k, v in x.items()) \
                    or any(pending(v) for v in x.values())
            if isinstance(x, list):
                return any(pending(v) for v in x)
            return False

        settled = True
        for _ in range(20):
            js, jb = get(jurl)
            ps, pb = get(purl)
            if not pending(jb) and not pending(pb):
                settled = True
                break
            settled = False
            import time as _t
            _t.sleep(15)
        if not settled:
            unsettled.append((path, '两侧未在等待上限内落定（按窗重算仍在跑）—— 时点问题，不计为差异'))
            continue
        # ★只按 HTTP 状态判定（曾按正文里是否有 "404" 判 ⇒ 把"已实现但正文含该字样"的端点误判为未实现）
        if js in (0, 404):
            missing.append((path, f'Java HTTP {js}'))
            continue
        if js != 200 or ps != 200:
            diff.append((path, f'HTTP Java {js} / Python {ps}'))
            continue
        d = first_diff(norm(jb), norm(pb))
        if d and path in IGNORE_PATHS and d.startswith(IGNORE_PATHS[path][0]):
            ignored.append((path, IGNORE_PATHS[path][0], IGNORE_PATHS[path][1]))
            same.append(path)
        elif d:
            diff.append((path, d))
        else:
            same.append(path)

    print('Java ↔ Python 契约端点比对（裁判：冻契约 %d 条；Java 范围 kind=%s 共 %d 条）'
          % (len(rows), ','.join(sorted(java_kinds)), scope_total))
    print('  一致 %d/%d · 有差异 %d · 未实现 %d · 跳过 %d · 未落定 %d（范围外 %d 条：%s）\n'
          % (len(same), scope_total, len(diff), len(missing), len(skipped), len(unsettled), len(other),
             ', '.join(sorted({k for _, k in other})) or '-'))
    if same:
        print('  [一致] ' + ', '.join(same[:12]) + (' …' if len(same) > 12 else ''))
    for p, why in diff[:10]:
        print('  [差异] %-28s %s' % (p, why))
    for p, why in missing[:10]:
        print('  [未实现] %-26s %s' % (p, why))
    for p, why in skipped[:6]:
        print('  [跳过] %-28s %s' % (p, why))
    for p, why in unsettled[:6]:
        print('  [未落定] %-26s %s' % (p, why))
    for p, f, why in ignored[:6]:
        print('  [登记忽略] %-24s %s —— %s' % (p, f, why))

    need = [x.strip() for x in a.require.split(',') if x.strip()]
    bad = [p for p in need if p not in same]
    if need:
        print('\n门禁要求一致 %d 条 → %s' % (len(need), '通过' if not bad else '未通过: ' + ', '.join(bad)))
    print('\n结论: %s' % ('Java 范围端点全部一致' if not diff and not missing else
                          '迁移进行中（%d/%d 已一致）' % (len(same), scope_total)))
    return 0 if not bad else 1


if __name__ == '__main__':
    raise SystemExit(main())