#!/usr/bin/env python3
"""
check_i18n.py — 产线平台前端国际化(i18n)检查器

框架事实（bricks，决定各违规分类）：
  - TextBase(Text/Title*)：只有 otext + i18n:true 才走翻译；裸 "text" 不翻译 → W1
  - Menu item label / Button label / Form field label / Tabular title /
    PopupWindow title / UiCode data.text / placeholder / Message.message：
    组件自动 i18n._()，不需要改文件，只需要词条进字典 → 缺词条记 W2
  - 字典链：模块 i18n/<lang>/msg.txt|i18n.json → scripts/merge_i18n.py →
    wwwroot/i18n/<lang>/i18n.json → /i18n_getmsgs 端点 → 前端 I18n.msgs
  - msg.txt 以 '#' 开头的词条被 appPublic.i18n 解析器当注释丢弃 → W6
  - git-ignored 的 wwwroot/<列表名>/ 是 xls2ui 生成物，禁止直接修改（pre-commit 铁律）
    → 生成物只收集 key，不报 W1（要改改 json/ 源 + 重新生成）

违规分类：
  W1  text/otext 硬编码中文且无 i18n 标记（git 跟踪的手写文件才报）
  W2  自动翻译组件的中文 key 不在合并字典（en/jp/ko）
  W3  手写 JS / dspy 内嵌 script 里硬编码中文提示（应 bricks.app.i18n._）
  W4  模块有 UI 但缺 i18n 目录或缺语种
  W5  merge_i18n.py 自动发现覆盖检查（动态调用 discover_modules() 验证）
  W6  msg.txt '#' 开头词条（静默丢弃）
  W7  动态拼接中文（% / + / f-string）作为可见文本 → 需 otext+i18n 传参重构（仅报告）

用法:
  python3 scripts/check_i18n.py [--repos-dir ~/work/repos] [--json out.json] [-v]
退出码: 0=无违规, 1=有违规
"""
import os, re, sys, json, subprocess, argparse
from collections import OrderedDict

SCRIPT_DIR = os.path.dirname(os.path.abspath(__file__))
APP_DIR = os.path.dirname(SCRIPT_DIR)
LANGS = ['zh', 'en', 'jp', 'ko']
CJK = re.compile(r'[\u4e00-\u9fff]')

SKIP_PATH = [
    re.compile(r'/i18n/'), re.compile(r'/docs?/'), re.compile(r'\.md$'),
    re.compile(r'node_modules'), re.compile(r'/\.git/'), re.compile(r'/py3/'),
    re.compile(r'/dist/'), re.compile(r'\.bak/'), re.compile(r'\.bak$'),
    re.compile(r'/wwwroot/bricks/'),  # 框架软链
]
AUTO_I18N_CTX = ('Menu', 'Button', 'Form', 'InlineForm', 'Tabular', 'PopupWindow',
                 'Title', 'Title1', 'Title2', 'Title3', 'Title4', 'Error')


def skip(path):
    return any(p.search(path) for p in SKIP_PATH)


def git_repo_root(path):
    d = path if os.path.isdir(path) else os.path.dirname(path)
    try:
        return subprocess.run(['git', '-C', d, 'rev-parse', '--show-toplevel'],
                              capture_output=True, text=True, timeout=10).stdout.strip()
    except Exception:
        return None


_ignored_cache = {}
def git_ignored(path, repo):
    key = (repo, path)
    if key in _ignored_cache:
        return _ignored_cache[key]
    try:
        r = subprocess.run(['git', '-C', repo, 'check-ignore', '-q', path],
                           capture_output=True, timeout=10)
        v = (r.returncode == 0)
    except Exception:
        v = False
    _ignored_cache[key] = v
    return v


def load_modules(repos_dir):
    mods = []
    build_sh = os.path.join(APP_DIR, 'build.sh')
    if os.path.exists(build_sh):
        content = open(build_sh, encoding='utf-8').read()
        for m in re.finditer(r'^for mod in ([^;]+); do\s*$', content, re.M):
            for name in m.group(1).split():
                if name not in mods:
                    mods.append(name)
    for name in ['pipeline_core', 'pipeline_ops']:
        if name not in mods:
            mods.insert(0, name)
    result = OrderedDict()
    result['app'] = APP_DIR
    for name in mods:
        for cand in (os.path.join(repos_dir, name), os.path.join(APP_DIR, 'pkgs', name),
                     os.path.join(APP_DIR, name)):
            if os.path.isdir(cand):
                result[name] = cand
                break
        else:
            result[name] = None
    return result


def iter_files(root, exts):
    for dirpath, dirnames, filenames in os.walk(root):
        dirnames[:] = [d for d in dirnames
                       if d not in ('.git', 'py3', 'node_modules', 'dist', '__pycache__', 'i18n')]
        for fn in filenames:
            if fn.endswith(exts):
                p = os.path.join(dirpath, fn)
                if not skip(p):
                    yield p


def merged_dict(lang):
    fp = os.path.join(APP_DIR, 'wwwroot', 'i18n', lang, 'i18n.json')
    if os.path.exists(fp):
        try:
            return json.load(open(fp, encoding='utf-8'))
        except Exception:
            return {}
    return {}


def in_code_data_array(raw, pos):
    """判断位置是否处于 code 选项 data 数组的 {"value":..,"text":..} 上下文。"""
    s = max(0, pos - 300)
    ctx = raw[s:pos]
    return '"value"' in ctx and ctx.rfind('"data"') >= 0 or \
           re.search(r'\{\s*"value"\s*:[^}]*$', ctx) is not None


def scan_ui(path, keys, viol, tracked):
    raw = open(path, encoding='utf-8', errors='replace').read()
    repo = git_repo_root(path)
    ign = repo and git_ignored(path, repo)
    for m in re.finditer(r'"(label|title|text|otext|placeholder|message|description)"\s*:\s*"([^"]*)"', raw):
        field, val = m.group(1), m.group(2)
        if not CJK.search(val) or val.startswith('{{'):
            continue
        s = max(0, m.start() - 400); e = min(len(raw), m.end() + 400)
        ctx = raw[s:e]
        has_i18n = re.search(r'"i18n"\s*:\s*true', ctx) is not None
        has_otext = '"otext"' in ctx
        if field == 'otext':
            if has_i18n:
                keys.add(val)
            else:
                viol.append((path, m.start(), 'W1', 'otext 无 i18n:true', val, ign))
        elif field == 'text':
            if has_i18n or has_otext:
                keys.add(val)
            elif in_code_data_array(raw, m.start()):
                keys.add(val)   # UiCode/Tabular 渲染 code 值自动 i18n._ (input.js:1232)
            else:
                # Text 裸 text 不翻译
                if ign:
                    keys.add(val)   # 生成物：改 json 源+重新生成，这里只收 key
                else:
                    viol.append((path, m.start(), 'W1', 'text 硬编码中文(应 otext+i18n:true)', val, ign))
        else:
            keys.add(val)   # label/title/placeholder/message/description 组件自动翻译
    # 内嵌 script 里的中文提示
    for m in re.finditer(r'(show_message|show_error|bricks\.Message|bricks\.Error)\s*\(\s*\{[^}]{0,200}?(title|message)\s*:\s*[\'"]([^\'"]*)[\'"]', raw):
        val = m.group(3)
        if CJK.search(val):
            keys.add(val)   # Message.message 自动 i18n(message.js:25)，title 走 PopupWindow 自动 i18n
    # 动态拼接
    for m in re.finditer(r'"text"\s*:\s*("[^"]*"\s*[%+][^,}]*)', raw):
        seg = m.group(1)
        if CJK.search(seg) and '{{' not in seg:
            viol.append((path, m.start(), 'W7', '动态拼接中文文本(需 otext+参数重构)', seg[:60], ign))


DSPY_UI_KEYS = re.compile(r"['\"](label|title|otext|placeholder|message|description)['\"]\s*:\s*(['\"])([^'\"]*)\2")
DSPY_TEXT = re.compile(r"['\"]text['\"]\s*:\s*(?:(['\"])([^'\"]*)\1|(\(?\s*['\"][^'\"]*['\"]\s*[%+][^)\n]*\)?))")

def scan_dspy(path, keys, viol):
    raw = open(path, encoding='utf-8', errors='replace').read()
    if 'widgettype' not in raw and 'mdtext' not in raw:
        return
    repo = git_repo_root(path)
    ign = repo and git_ignored(path, repo)
    for m in DSPY_UI_KEYS.finditer(raw):
        field, val = m.group(1), m.group(3)
        if not CJK.search(val) or val.startswith('{{'):
            continue
        s = max(0, m.start() - 500); e = min(len(raw), m.end() + 500)
        ctx = raw[s:e]
        has_i18n = re.search(r"['\"]i18n['\"]\s*:\s*True", ctx) is not None
        if field == 'otext':
            if has_i18n:
                keys.add(val)
            else:
                viol.append((path, m.start(), 'W1', 'dspy otext 无 i18n:True', val, ign))
        else:
            keys.add(val)
    for m in DSPY_TEXT.finditer(raw):
        if m.group(2) is not None:
            val = m.group(2)
            if not CJK.search(val):
                continue
            s = max(0, m.start() - 500); e = min(len(raw), m.end() + 500)
            ctx = raw[s:e]
            if re.search(r"['\"]i18n['\"]\s*:\s*True", ctx) or "'otext'" in ctx or '"otext"' in ctx \
               or in_code_data_array(raw, m.start()):
                keys.add(val)
            elif ign:
                keys.add(val)
            else:
                viol.append((path, m.start(), 'W1', 'dspy text 硬编码中文(应 otext+i18n:True)', val, ign))
        else:
            seg = m.group(3) or ''
            if CJK.search(seg):
                viol.append((path, m.start(), 'W7', 'dspy 动态拼接中文文本', seg[:60], ign))
    for m in re.finditer(r"(?:show_message|show_error)\s*\(\s*\{[^}]{0,200}?(?:title|message)['\"]?\s*[:=]\s*['\"]([^'\"]*)['\"]", raw):
        val = m.group(1)
        if CJK.search(val):
            keys.add(val)


def strip_js_comments(raw):
    """粗剥离 // 与 /* */ 注释（不处理字符串内含 // 的极端情形，够用）。"""
    out = []
    i, n = 0, len(raw)
    in_s = None
    while i < n:
        c = raw[i]
        if in_s:
            out.append(c)
            if c == '\\' and i + 1 < n:
                out.append(raw[i+1]); i += 2; continue
            if c == in_s:
                in_s = None
            i += 1; continue
        if c in ('"', "'", '`'):
            in_s = c; out.append(c); i += 1; continue
        if c == '/' and i + 1 < n and raw[i+1] == '/':
            while i < n and raw[i] != '\n':
                i += 1
            continue
        if c == '/' and i + 1 < n and raw[i+1] == '*':
            i += 2
            while i + 1 < n and not (raw[i] == '*' and raw[i+1] == '/'):
                i += 1
            i += 2
            continue
        out.append(c); i += 1
    return ''.join(out)


def scan_js(path, keys, viol):
    raw = open(path, encoding='utf-8', errors='replace').read()
    code = strip_js_comments(raw)
    repo = git_repo_root(path)
    ign = repo and git_ignored(path, repo)
    # i18n-aware 文件：定义了 _t()/i18n._ 辅助函数，对象字面量里的中文是「词条 key」
    # 而非硬编码显示文本（如 showcase.js MEDIA_LABEL_KEYS，经 mediaLabel→_t 翻译）
    has_i18n_helper = bool(re.search(r'function\s+_t\s*\(|(?:var|const|let)\s+_t\s*=|i18n\._', code))
    # 所有含中文的字符串字面量：不在 i18n._() 内、不在 Message/show_message 上下文 → W3
    for m in re.finditer(r"(['\"])([^'\"\n]*[\u4e00-\u9fff][^'\"\n]*)\1", code):
        val = m.group(2)
        pre = code[max(0, m.start() - 80):m.start()]
        if 'i18n._' in pre[-30:] or re.search(r'(?:^|[^.\w])_t\(\s*$', pre):
            keys.add(val)
            continue
        ctx = code[max(0, m.start() - 200):m.end()]
        if re.search(r'(show_message|show_error|bricks\.Message|bricks\.Error)\s*\(\s*\{[^}]*$', ctx):
            keys.add(val)   # Message title/message 自动翻译
            continue
        if re.search(r'(title|message|label)\s*:\s*$', pre):
            keys.add(val)   # 组件 option 自动翻译
            continue
        if has_i18n_helper and re.search(r'[\w\'\"]+\s*:\s*$', pre):
            keys.add(val)   # i18n-aware 文件的对象字面量中文值 = 词条 key
            continue
        viol.append((path, m.start(), 'W3', 'js 硬编码中文(应 bricks.app.i18n._ 或组件词条)', val, ign))


def check_infra(modules, merge_py):
    issues = []
    # W5 动态验证：直接调用 merge_i18n.discover_modules()，模块的 i18n 目录
    # 能被自动发现即通过（2026-09-14：merge 已重写为自动发现，旧版按源码文本
    # 扫描 MODULES 清单会把所有模块误报为漏）
    discovered = set()
    try:
        import importlib.util
        spec = importlib.util.spec_from_file_location('merge_i18n_mod', merge_py)
        if spec is not None and spec.loader is not None:
            mmod = importlib.util.module_from_spec(spec)
            spec.loader.exec_module(mmod)
            discovered = {n for n, _ in mmod.discover_modules()}
        else:
            print(f'!! merge_i18n.py 无法加载: {merge_py}', file=sys.stderr)
    except Exception as e:
        print(f'!! 无法调用 merge_i18n.discover_modules(): {e}', file=sys.stderr)
    for name, path in modules.items():
        if path is None or not os.path.isdir(path):
            continue
        ui_files = list(iter_files(path, ('.ui',)))
        has_ui = bool(ui_files)
        if not has_ui:
            n = 0
            for f in iter_files(path, ('.dspy',)):
                raw = open(f, encoding='utf-8', errors='replace').read(4000)
                if 'widgettype' in raw:
                    has_ui = True
                    break
                n += 1
                if n > 300:
                    break
        i18n_dir = os.path.join(path, 'i18n')
        if has_ui:
            if not os.path.isdir(i18n_dir):
                issues.append((name, 'W4', '有 UI 文件但无 i18n/ 目录'))
            else:
                missing = [l for l in LANGS
                           if not os.path.exists(os.path.join(i18n_dir, l, 'msg.txt'))
                           and not os.path.exists(os.path.join(i18n_dir, l, 'i18n.json'))]
                if missing:
                    issues.append((name, 'W4', 'i18n 缺语种: ' + ','.join(missing)))
            if name != 'app' and name not in discovered:
                issues.append((name, 'W5', 'i18n/ 未被 merge_i18n.discover_modules() 发现'))
        for lang in LANGS:
            fp = os.path.join(path, 'i18n', lang, 'msg.txt')
            if os.path.exists(fp):
                for i, line in enumerate(open(fp, encoding='utf-8', errors='replace'), 1):
                    ls = line.strip()
                    if ls and ls.startswith('#') and (':' in ls or '=' in ls):
                        issues.append((name, 'W6', f'i18n/{lang}/msg.txt:{i} #词条被丢弃: {ls[:50]}'))
    return issues


def main():
    ap = argparse.ArgumentParser()
    ap.add_argument('--repos-dir', default=os.path.expanduser('~/work/repos'))
    ap.add_argument('--json', default=None)
    ap.add_argument('-v', '--verbose', action='store_true')
    args = ap.parse_args()

    modules = load_modules(args.repos_dir)
    dicts = {l: merged_dict(l) for l in LANGS}
    all_keys = {}
    violations = []
    per_mod = OrderedDict()

    for name, path in modules.items():
        if path is None or not os.path.isdir(path):
            per_mod[name] = {'status': 'missing'}
            continue
        keys, viols = set(), []
        roots = [path]
        for f in iter_files(path, ('.ui',)):
            scan_ui(f, keys, viols, None)
        for f in iter_files(path, ('.dspy',)):
            scan_dspy(f, keys, viols)
        wr = os.path.join(path, 'wwwroot')
        if os.path.isdir(wr):
            for f in iter_files(wr, ('.js',)):
                scan_js(f, keys, viols)
        for k in keys:
            all_keys.setdefault(k, set()).add(name)
        violations.extend(viols)
        per_mod[name] = {'keys': len(keys), 'violations': len(viols)}

    missing_dict = {}
    for key, mods in sorted(all_keys.items()):
        for lang in ('en', 'jp', 'ko'):
            if key not in dicts[lang]:
                missing_dict.setdefault(lang, []).append({'key': key, 'modules': sorted(mods)})

    infra = check_infra(modules, os.path.join(APP_DIR, 'scripts', 'merge_i18n.py'))

    cnt = {}
    for v in violations:
        cnt[v[2]] = cnt.get(v[2], 0) + 1
    w4 = [i for i in infra if i[1] == 'W4']
    w5 = [i for i in infra if i[1] == 'W5']
    w6 = [i for i in infra if i[1] == 'W6']

    print('=' * 62)
    print('产线平台 i18n 检查报告')
    print('=' * 62)
    print(f"模块数: {len([1 for p in modules.values() if p])}")
    print(f"抽取可见中文 key(去重): {len(all_keys)}")
    print(f"W1 硬编码未标i18n: {cnt.get('W1',0)}  W3 js硬编码: {cnt.get('W3',0)}  W7 动态拼接: {cnt.get('W7',0)}")
    print(f"W2 字典缺失: en={len(missing_dict.get('en',[]))} jp={len(missing_dict.get('jp',[]))} ko={len(missing_dict.get('ko',[]))}")
    print(f"W4 缺i18n目录/语种: {len(w4)}  W5 merge漏模块: {len(w5)}  W6 #词条: {len(w6)}")
    if w5:
        print(f"   W5 漏: {[i[0] for i in w5]}")
    if w4:
        print(f"   W4: {[(i[0], i[2][:30]) for i in w4]}")
    print('-' * 62)
    for name, st in per_mod.items():
        if 'keys' in st:
            flag = ' <-- !' if st['violations'] else ''
            print(f"  {name:24s} keys={st['keys']:4d} viol={st['violations']:3d}{flag}")
    if args.verbose:
        print('-' * 62)
        for v in violations:
            tag = ' [git-ignored]' if v[5] else ''
            print(f"  [{v[2]}] {v[0]}@{v[1]}{tag}: {v[3]} | {v[4][:60]}")
    if args.json:
        json.dump({
            'summary': {'keys': len(all_keys), 'viol': cnt,
                        'W2': {l: len(missing_dict.get(l, [])) for l in ('en','jp','ko')},
                        'W4': len(w4), 'W5': len(w5), 'W6': len(w6)},
            'violations': [{'file': v[0], 'pos': v[1], 'code': v[2], 'why': v[3],
                            'text': v[4], 'git_ignored': bool(v[5])} for v in violations],
            'missing_dict': missing_dict,
            'keys': {k: sorted(v) for k, v in sorted(all_keys.items())},
            'infra': [{'module': i[0], 'code': i[1], 'detail': i[2]} for i in infra],
        }, open(args.json, 'w', encoding='utf-8'), ensure_ascii=False, indent=1)
        print(f'JSON 报告: {args.json}')
    total = sum(cnt.values()) + sum(len(v) for v in missing_dict.values()) + len(infra)
    return 1 if total else 0


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