@hupan56/wlkj 2.6.0 → 2.7.1

package/templates/qoder/scripts/search_index.py

@@ -27,7 +27,7 @@ except (AttributeError, TypeError, OSError, IOError):
         pass
 
 sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
-from common.terms import CN_MAP, BUSINESS_PATH_MAP, CN_TO_EN, get_platform_map
+from common.terms import CN_MAP, BUSINESS_PATH_MAP, CN_TO_EN, get_platform_map, expand_chinese_query
 PLATFORM_MAP = get_platform_map()
 
 BASE = os.path.dirname(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
@@ -38,7 +38,20 @@ MIN_FUZZY_LEN = 4
 
 
 def load_index(filename, hint=''):
+    """Load an index. Prefers DuckDB (kg.duckdb), falls back to JSON file.
+
+    Returns the same dict structure as the JSON file would have, so all
+    callers work unchanged regardless of the backend.
+    """
+    # Try DuckDB first
+    result = _duckdb_load(filename)
+    if result is not None:
+        return result
+
+    # Fallback: read JSON file
     path = os.path.join(INDEX_DIR, filename)
+    if not os.path.isfile(path):
+        path = os.path.join(INDEX_DIR, '_build_cache', filename)
     if not os.path.isfile(path):
         print('{} not found.{}'.format(filename, ' ' + hint if hint else ''))
         return None
@@ -50,6 +63,150 @@ def load_index(filename, hint=''):
         return None
 
 
+# DuckDB path (lazy-connect)
+_duckdb_con = None
+
+
+def _get_duckdb():
+    global _duckdb_con
+    if _duckdb_con is not None:
+        return _duckdb_con
+    db_path = os.path.join(INDEX_DIR, 'kg.duckdb')
+    if not os.path.isfile(db_path):
+        return None
+    try:
+        import duckdb
+        _duckdb_con = duckdb.connect(db_path, read_only=True)
+        return _duckdb_con
+    except Exception:
+        return None
+
+
+def _load_keyword_with_cache(con):
+    """加载 code-keyword (11836 行), 带 pickle 缓存层。
+
+    性能: DuckDB SELECT + Python 循环重建 = ~180ms;
+          pickle.load = ~37ms (提速 5x, 实测数据一致)。
+    缓存失效: kg.duckdb 的 mtime 变了就重建 (和 context_pack 缓存逻辑一致)。
+    降级: pickle 损坏/版本不兼容 → 静默回退 DuckDB 重建, 不报错。
+    """
+    import pickle
+    duckdb_path = os.path.join(INDEX_DIR, 'kg.duckdb')
+    cache_path = os.path.join(BASE, '.qoder', '.runtime', 'code-keyword.pkl')
+    db_mtime = os.path.getmtime(duckdb_path) if os.path.isfile(duckdb_path) else 0
+
+    # ① 缓存有效 (存在 + mtime 匹配) → 直接 load, 跳过 DuckDB
+    if os.path.isfile(cache_path):
+        try:
+            if os.path.getmtime(cache_path) >= db_mtime:
+                with open(cache_path, 'rb') as f:
+                    return pickle.load(f)
+        except (pickle.PickleError, EOFError, OSError):
+            pass  # 缓存损坏 → 回退重建
+
+    # ② 缓存无效/缺失 → 从 DuckDB 重建
+    rows = con.execute("SELECT keyword, file_path FROM keywords").fetchall()
+    result = {}
+    for kw, fp in rows:
+        result.setdefault(kw, []).append(fp)
+    if not result:
+        return None
+
+    # ③ 写缓存 (best-effort, 失败不影响返回结果)
+    try:
+        os.makedirs(os.path.dirname(cache_path), exist_ok=True)
+        with open(cache_path, 'wb') as f:
+            pickle.dump(result, f, protocol=pickle.HIGHEST_PROTOCOL)
+    except OSError:
+        pass
+    return result
+
+
+def _duckdb_load(filename):
+    """Load an index from DuckDB, returning the same structure as the JSON.
+
+    Returns None if DuckDB unavailable or table doesn't have data,
+    triggering the JSON fallback.
+    """
+    con = _get_duckdb()
+    if con is None:
+        return None
+
+    try:
+        if filename == 'code-keyword.json':
+            # {keyword: [files]} — 11836 行重建耗时 180ms, 用 pickle 缓存提速到 37ms
+            return _load_keyword_with_cache(con)
+
+        elif filename == 'code-api.json':
+            # {endpoint: controller_file}
+            rows = con.execute("SELECT endpoint, controller_file FROM endpoints").fetchall()
+            return dict(rows) if rows else None
+
+        elif filename == 'ui-style.json':
+            # Reconstruct the full ui-style structure from pages/fields/style_meta
+            result = {'version': '1.0', 'projects': {}, 'common_components': {}, 'form_components': {}, 'field_map': {}}
+            # Pages
+            page_rows = con.execute("SELECT file, project, page_type, components, table_columns, form_fields, layout FROM pages").fetchall()
+            for f, proj, ptype, comps, cols, fields, layout in page_rows:
+                proj_data = result['projects'].setdefault(proj, {'page_types': {}, 'page_examples': {}})
+                proj_data['page_types'][ptype] = proj_data['page_types'].get(ptype, 0) + 1
+                examples = proj_data['page_examples'].setdefault(ptype, [])
+                examples.append({'file': f, 'components': json.loads(comps) if comps else [],
+                                 'table_columns': json.loads(cols) if cols else [],
+                                 'form_fields': json.loads(fields) if fields else [],
+                                 'layout': json.loads(layout) if layout else {}})
+            # Fields
+            field_rows = con.execute("SELECT field, title, count, sample_files FROM fields").fetchall()
+            for field, title, count, samples in field_rows:
+                titles = title.split('/') if title else []
+                result['field_map'][field] = {'field': field, 'titles': titles, 'count': count,
+                                              'sample_files': json.loads(samples) if samples else []}
+            # Style meta (layout_fingerprint per project)
+            meta_rows = con.execute("SELECT project, stack, layout, access_mode, sidebar_width, color_primary, color_success, color_warning, color_destructive, radius, vue_files, common_components, form_components FROM style_meta").fetchall()
+            for row in meta_rows:
+                proj = row[0]
+                proj_data = result['projects'].setdefault(proj, {'page_types': {}, 'page_examples': {}})
+                proj_data['layout_fingerprint'] = {
+                    'stack': row[1], 'layout': row[2], 'access_mode': row[3],
+                    'sidebar_width': row[4], 'color_primary': row[5], 'color_success': row[6],
+                    'color_warning': row[7], 'color_destructive': row[8], 'radius': row[9],
+                }
+                proj_data['vue_files'] = row[10]
+                if row[11]:
+                    result['common_components'] = json.loads(row[11])
+                if row[12]:
+                    result['form_components'] = json.loads(row[12])
+            return result if result['projects'] or result['field_map'] else None
+
+        elif filename == 'prd-index.json':
+            rows = con.execute("SELECT file, title, keywords, features, cn_terms FROM prds").fetchall()
+            result = {}
+            for f, title, kws, feats, cn in rows:
+                result[f] = {'title': title,
+                             'keywords': json.loads(kws) if kws else [],
+                             'features': json.loads(feats) if feats else [],
+                             'cn_terms': json.loads(cn) if cn else []}
+            return result if result else None
+
+        elif filename == 'ui-modules.json':
+            rows = con.execute("SELECT dir, cn, project, path FROM modules").fetchall()
+            result = {'projects': {}}
+            for d, cn, proj, path in rows:
+                result['projects'].setdefault(proj, []).append({'dir': d, 'cn': cn, 'path': path})
+            return result if result['projects'] else None
+
+        elif filename == 'code-stats.json':
+            rows = con.execute("SELECT project, vue_files FROM style_meta").fetchall()
+            result = {}
+            for proj, vf in rows:
+                result[proj] = {'files': vf or 0, 'classes': 0, 'apis': 0}
+            return result if result else None
+
+    except Exception:
+        return None
+    return None
+
+
 def filter_by_platform(files, platform=None):
     if not platform:
         return files
@@ -72,7 +229,7 @@ def _match_keyword(query_word, kw):
 def _search_cache_key(query, platform):
     """缓存 key: query|platform|keyword-index mtime。"""
     import hashlib
-    ki_path = os.path.join(BASE, 'data', 'index', 'keyword-index.json') if os.path.isdir(os.path.join(BASE, 'data', 'index')) else None
+    ki_path = os.path.join(BASE, 'data', 'index', 'code-keyword.json') if os.path.isdir(os.path.join(BASE, 'data', 'index')) else None
     mtime = '%d' % os.path.getmtime(ki_path) if ki_path and os.path.isfile(ki_path) else '0'
     raw = '{}|{}|{}'.format(query.lower(), platform or '', mtime)
     return hashlib.md5(raw.encode('utf-8')).hexdigest()[:16]
@@ -215,61 +372,18 @@ def _cleanup_cache_dir(cache_dir, prefix, max_files=50):
         pass
 
 
-def _expand_chinese_query(query):
-    """把中文查询拆成英文搜索词。
-
-    三级匹配:
-      1. 精确匹配: "异常" 在 CN_MAP → 直接扩展
-      2. 分词匹配: "异常记录" → 拆成 "异常" + "记录", 各自扩展
-         (贪心最长匹配: 从长到短扫 CN_MAP 的 key)
-      3. 兜底: 拆不出来的部分, 原样小写 (让英文子串索引做匹配)
-
-    同时合并 BUSINESS_PATH_MAP 的路径模式。
-    """
-    words = []
-    # 合并 CN_MAP + PRD 自动扩展
-    try:
-        from common.terms import get_cn_map_with_auto
-        cn_map = get_cn_map_with_auto()
-    except ImportError:
-        cn_map = CN_MAP
-
-    # 1. 精确匹配
-    if query in cn_map:
-        words.extend(cn_map[query].split())
-    if query in BUSINESS_PATH_MAP:
-        words.extend(p.strip('/-').lower() for p in BUSINESS_PATH_MAP[query])
-
-    # 2. 分词匹配 (查询比单个 CN_MAP key 长 → 需要拆词)
-    if not words and len(query) >= 3:
-        # 贪心最长匹配: 把 CN_MAP keys 按长度降序排, 逐个从 query 里找
-        remaining = query
-        matched_segments = []
-        for cn in sorted(cn_map.keys(), key=len, reverse=True):
-            if cn in remaining:
-                matched_segments.append(cn)
-                remaining = remaining.replace(cn, ' ')
-        if matched_segments:
-            for seg in matched_segments:
-                words.extend(cn_map[seg].split())
-                if seg in BUSINESS_PATH_MAP:
-                    words.extend(p.strip('/-').lower() for p in BUSINESS_PATH_MAP[seg])
-            # 剩余的英文/数字片段也加进去 (兜底)
-            for frag in remaining.split():
-                if frag and frag.isascii():
-                    words.append(frag.lower())
-
-    return words
+# _expand_chinese_query 已移至 common/terms.py (单一数据源, context_pack 共用)
+# import 见文件头部: from common.terms import expand_chinese_query
 
 
 def _search_keywords_impl(query, platform=None):
     """实际的搜索逻辑 (原 search_keywords 内容)。"""
-    ki = load_index('keyword-index.json', hint='Run: python git_sync.py')
+    ki = load_index('code-keyword.json', hint='Run: python git_sync.py')
     if ki is None:
         return
 
     # Expand Chinese query via the shared term maps (same maps used at build time)
-    words = _expand_chinese_query(query)
+    words = expand_chinese_query(query)
     if words:
         words = sorted(set(w.lower() for w in words))
         print('Chinese: {} -> English: {}'.format(query, ' '.join(words)))
@@ -370,7 +484,7 @@ def search_prds(query):
 
 def search_style(style_type, platform=None):
     """Find pages of a given UI type (table/form/detail/modal/dashboard)."""
-    si = load_index('style-index.json', hint='Run: python build_style_index.py')
+    si = load_index('ui-style.json', hint='Run: python build_style_index.py')
     if si is None:
         return
     q = style_type.lower()
@@ -389,7 +503,9 @@ def search_style(style_type, platform=None):
         found_any = True
         for pt in hit_types:
             print('[{}] {}: {} pages'.format(proj, pt, page_types[pt]))
-            for f in examples.get(pt, [])[:8]:
+            for item in examples.get(pt, [])[:8]:
+                # page_examples 元素可能是 dict (新格式, 含结构) 或 str (旧格式)
+                f = item['file'] if isinstance(item, dict) else item
                 print('  ' + f)
             if not examples.get(pt):
                 print('  (no examples stored - run: python build_style_index.py)')
@@ -403,8 +519,8 @@ def search_style(style_type, platform=None):
 
 
 def search_field(field_name):
-    """Search field usage in style-index.json field_map."""
-    si = load_index('style-index.json', hint='Run: python build_style_index.py')
+    """Search field usage in ui-style.json field_map."""
+    si = load_index('ui-style.json', hint='Run: python build_style_index.py')
     if si is None:
         return
     fm = si.get('field_map', {})
@@ -421,8 +537,8 @@ def search_field(field_name):
 
 
 def search_api(query):
-    """Search API endpoints in api-index.json."""
-    ai = load_index('api-index.json', hint='Run: python git_sync.py --index-only')
+    """Search API endpoints in code-api.json."""
+    ai = load_index('code-api.json', hint='Run: python git_sync.py --index-only')
     if ai is None:
         return
     q = query.lower()
@@ -438,8 +554,8 @@ def search_api(query):
 
 
 def show_modules():
-    """Show project overview from module-map.json."""
-    mm = load_index('module-map.json', hint='Run: python git_sync.py --index-only')
+    """Show project overview from code-stats.json."""
+    mm = load_index('code-stats.json', hint='Run: python git_sync.py --index-only')
     if mm is None:
         return
     print('=== Project Modules ===')
@@ -453,7 +569,7 @@ def show_modules():
 
 def show_top_keywords():
     """Show top keywords by file count."""
-    ki = load_index('keyword-index.json', hint='Run: python git_sync.py')
+    ki = load_index('code-keyword.json', hint='Run: python git_sync.py')
     if ki is None:
         return
     top = sorted(ki.items(), key=lambda x: len(x[1]), reverse=True)[:50]
@@ -464,7 +580,7 @@ def show_top_keywords():
 
 
 def show_components():
-    meta = load_index('style-meta.json', hint='Run: python build_style_index.py')
+    meta = load_index('ui-style-meta.json', hint='Run: python build_style_index.py')
     if meta is None:
         return
     print('=== UI Components Usage ===')
@@ -479,7 +595,7 @@ def show_components():
 
 
 def show_vben():
-    data = load_index('vben-style-reference.json')
+    data = load_index('ref-vben-style.json')
     if data is None:
         return
     print('=== Vben Admin Design Tokens ===')
@@ -538,6 +654,30 @@ def search_wiki_cli(query):
     print('\n  提示: 读 wiki 文档 (带行号引用) 比读源码更省 token')
 
 
+
+def search_impact(query):
+    """改这个接口/符号影响哪些页面/按钮/模块? (U1: 内存图遍历)"""
+    from common.graph_traverse import CodeGraph
+    g = CodeGraph(INDEX_DIR)
+    result = g.impact(query)
+    if result:
+        print(result)
+    else:
+        print('未找到匹配 "{}" 的端点。尝试用部分路径, 如 /veh 或 export'.format(query))
+        print('提示: trace-chain 当前覆盖率 31%, 未覆盖的调用不会被列出')
+
+
+def search_context360(query):
+    """符号的360度视图: 端点/函数/处理器/PRD (U1: 内存图遍历)"""
+    from common.graph_traverse import CodeGraph
+    g = CodeGraph(INDEX_DIR)
+    result = g.context360(query)
+    if result:
+        print(result)
+    else:
+        print('未找到匹配 "{}" 的符号。'.format(query))
+
+
 def print_usage():
     print('Usage:')
     print('  search_index.py <keyword> [--platform web|app]  - Search code')
@@ -545,6 +685,12 @@ def print_usage():
     print('  search_index.py --style <type> [--platform p]  - UI patterns (table/form/detail/modal/dashboard)')
     print('  search_index.py --field <name>                 - Field usage')
     print('  search_index.py --api <keyword>                - API endpoints')
+    print('  search_index.py --impact <endpoint>            - 改这个接口影响哪些页面/按钮/模块')
+    print('  search_index.py --context360 <symbol>          - 符号的360度视图(端点/函数/处理器/PRD)')
+    print('  search_index.py --coverage                     - 功能×测试覆盖矩阵(哪些功能没测试)')
+    print('  search_index.py --feature <名称>               - 功能画像(端点+按钮+测试+页面)')
+    print('  search_index.py --workflow <名称>              - 业务流程链(盘点有哪几步)')
+    print('  search_index.py --routes <关键词>              - 前端页面路由(保险->/veh/vehicle/vehAffair/insurance)')
     print('  search_index.py --components                   - UI components')
     print('  search_index.py --wiki <keyword>               - Repo Wiki 语义模块文档 (精准)')
     print('  search_index.py --modules                      - Project overview')
@@ -554,6 +700,95 @@ def print_usage():
     print('Platform: web/pc/管理端 -> fywl-ui, app/mobile/h5/移动端 -> Carmg-H5')
 
 
+def _search_routes(keyword):
+    """查前端页面路由: 优先用 menu-routes-tree.json(getRouters父子树, path准确),
+    回退 menu-routes.json(扁平, component猜的URL可能不准)。
+
+    例: _search_routes('保险') -> /veh/vehicle/vehAffair/insurance
+    ⚠️ 若结果URL打开404, 说明 tree 缓存过期, 重跑: python .qoder/scripts/extract_routes_tree.py
+    """
+    import os, json
+    kw = keyword.lower()
+    # ① 优先: 准确路由树 (path 直接来自后端父子拼接)
+    tree_file = os.path.join(INDEX_DIR, 'menu-routes-tree.json')
+    if os.path.isfile(tree_file):
+        data = json.load(open(tree_file, encoding='utf-8'))
+        routes = data.get('routes', [])
+        hits = [r for r in routes if kw in (r.get('path','') + r.get('name','') + r.get('component','')).lower()]
+        if hits:
+            print('=== 页面路由: %s (%d 条, 来源: 准确路由树) ===' % (keyword, len(hits)))
+            for r in hits:
+                print('  %s -> %s' % (r.get('name', '?'), r.get('path', '')))
+            return
+        # tree 里没有, 不再回退到不准的旧数据, 直接提示 (避免误导成404)
+        print('未找到 "%s" 的路由(已查准确路由树 %d 条)。' % (keyword, len(routes)))
+        print('提示: 试试英文关键词, 或重跑刷新: python .qoder/scripts/extract_routes_tree.py')
+        return
+    # ② 回退: 旧扁平数据 (component猜URL, 可能不准)
+    routes_file = os.path.join(INDEX_DIR, 'menu-routes.json')
+    if not os.path.isfile(routes_file):
+        print('无路由数据。先运行: python .qoder/scripts/extract_routes_tree.py')
+        return
+    data = json.load(open(routes_file, encoding='utf-8'))
+    routes = data.get('routes', [])
+    hits = []
+    for r in routes:
+        searchable = ' '.join([r.get('path',''), r.get('component',''), r.get('name','')]).lower()
+        if kw in searchable:
+            hits.append(r)
+    if not hits:
+        print('未找到 "%s" 的路由。' % keyword)
+        print('提示: 试试英文关键词, 或刷新路由树: python .qoder/scripts/extract_routes_tree.py')
+        return
+    print('=== 页面路由: %s (%d 条, 来源: 旧扁平数据, URL可能不准) ===' % (keyword, len(hits)))
+    for r in hits:
+        print('  %s -> path=%s component=%s' % (r.get('name', '?'), r.get('path', ''), r.get('component', '')))
+    print('⚠️ 以上URL可能不准(component猜的)。建议运行 extract_routes_tree.py 用准确路由树。')
+
+
+def _graph_cli(method, *args):
+    """CLI 入口: 调图谱方法(让纯 Qoder IDE/CLI 也能用, 不只 MCP)。
+
+    CodeGraph 方法(coverage_matrix/feature_overview/impact)走 JSON 索引;
+    get_workflow 走 DuckDB(业务操作链)。
+    """
+    try:
+        if method == 'get_workflow':
+            # DuckDB 路径
+            import kg_duckdb as K
+            q = (args[0] if args else '').strip()
+            if not q:
+                print('请提供模块名, 如: assets / cost / hr / maintenance')
+                return
+            con = K.get_db()
+            rows = con.execute(
+                "SELECT seq, state, endpoint, button FROM workflows WHERE module=? ORDER BY seq",
+                [q.lower()]
+            ).fetchall()
+            if not rows:
+                mods = [r[0] for r in con.execute("SELECT DISTINCT module FROM workflows ORDER BY module").fetchall()]
+                print("未找到模块 '%s'。现有: %s" % (q, ', '.join(mods)))
+                return
+            print('=== %s 业务操作链 (%d 步) ===' % (q, len(rows)))
+            for r in rows:
+                print('  %d. [%s] %s — %s' % (r[0], r[1] or '?', r[2] or '', r[3] or ''))
+            return
+        # JSON 索引路径 (CodeGraph)
+        from common.graph_traverse import CodeGraph
+        from common.paths import DATA_INDEX_DIR
+        g = CodeGraph(str(DATA_INDEX_DIR))
+        fn = getattr(g, method, None)
+        if not fn:
+            print('未知图谱方法: %s' % method)
+            return
+        result = fn(*args) if args else fn()
+        print(result or '(无结果)')
+    except ImportError as e:
+        print('依赖缺失: %s' % e)
+    except Exception as e:
+        print('图谱查询失败: %s' % e)
+
+
 if __name__ == '__main__':
     platform = None
     args = sys.argv[1:]
@@ -575,6 +810,18 @@ if __name__ == '__main__':
         search_field(args[1])
     elif args[0] == '--api' and len(args) >= 2:
         search_api(args[1])
+    elif args[0] == '--impact' and len(args) >= 2:
+        search_impact(args[1])
+    elif args[0] == '--context360' and len(args) >= 2:
+        search_context360(args[1])
+    elif args[0] == '--coverage':
+        _graph_cli('coverage_matrix')
+    elif args[0] == '--feature' and len(args) >= 2:
+        _graph_cli('feature_overview', args[1])
+    elif args[0] == '--workflow' and len(args) >= 2:
+        _graph_cli('get_workflow', args[1])
+    elif args[0] == '--routes' and len(args) >= 2:
+        _search_routes(args[1])
     elif args[0] == '--wiki' and len(args) >= 2:
         search_wiki_cli(args[1])
     elif args[0] == '--components':