在前后端开发联调前 的提交20260223

scripts/ops/dataflow_analyzer.py

@@ -24,7 +24,8 @@ class AnalyzerConfig:
     date_to: date | None = None
     limit: int = 200
     tables: list[str] | None = None
-    output_dir: Path = field(default_factory=lambda: Path("docs/reports"))
+    # 调用方必须显式传入（从 SYSTEM_ANALYZE_ROOT 环境变量读取）
+    output_dir: Path = field(default_factory=lambda: Path(""))
     pg_dsn: str = ""
     api_base: str = ""
     api_token: str = ""
@@ -420,12 +421,41 @@ def dump_collection_results(
 
     返回 {类别: 目录路径} 的字典。
     """
-    json_trees_dir = output_dir / "json_trees"
-    db_schemas_dir = output_dir / "db_schemas"
-    field_mappings_dir = output_dir / "field_mappings"
-    json_trees_dir.mkdir(parents=True, exist_ok=True)
-    db_schemas_dir.mkdir(parents=True, exist_ok=True)
-    field_mappings_dir.mkdir(parents=True, exist_ok=True)
+    # CHANGE 2026-02-21 | 清理旧子目录后重建，避免 Windows 文件锁导致写入失败
+    import shutil as _shutil, time as _time
+    _sub_dirs = ["json_trees", "db_schemas", "field_mappings"]
+    for _name in _sub_dirs:
+        _d = output_dir / _name
+        if _d.exists():
+            try:
+                _shutil.rmtree(_d)
+            except (PermissionError, OSError):
+                # Windows 文件锁：无法删除也无法遍历，跳过（后面用备选名）
+                pass
+    # Windows rmtree 后句柄可能未释放，等待后再 mkdir
+    _time.sleep(1)
+
+    def _ensure_writable_dir(base: Path, name: str) -> Path:
+        """确保目录可写，如果被锁则用带后缀的备选名"""
+        d = base / name
+        for _attempt in range(3):
+            try:
+                d.mkdir(parents=True, exist_ok=True)
+                _test = d / ".write_test"
+                _test.write_text("ok", encoding="utf-8")
+                _test.unlink()
+                return d
+            except (FileNotFoundError, PermissionError, OSError):
+                _time.sleep(1)
+        # 旧目录不可用，用带后缀的新目录
+        d = base / f"{name}_new"
+        d.mkdir(parents=True, exist_ok=True)
+        print(f"  [警告] {name}/ 被锁定，使用备选目录 {d.name}/")
+        return d
+
+    json_trees_dir = _ensure_writable_dir(output_dir, "json_trees")
+    db_schemas_dir = _ensure_writable_dir(output_dir, "db_schemas")
+    field_mappings_dir = _ensure_writable_dir(output_dir, "field_mappings")
 
     # 解析 TABLE_MAP / FACT_MAPPINGS（用于构建字段映射）
     table_map = parse_table_map()
@@ -508,10 +538,18 @@ def dump_collection_results(
 
 def _write_json(path: Path, data: Any) -> None:
     """UTF-8 编码写入 JSON 文件，ensure_ascii=False，indent=2。"""
-    path.write_text(
-        json.dumps(data, ensure_ascii=False, indent=2, default=str),
-        encoding="utf-8",
-    )
+    content = json.dumps(data, ensure_ascii=False, indent=2, default=str)
+    try:
+        path.write_text(content, encoding="utf-8")
+    except PermissionError:
+        # CHANGE 2026-02-21 | Windows 文件锁重试：先删再写
+        import time
+        time.sleep(1)
+        try:
+            path.unlink(missing_ok=True)
+        except PermissionError:
+            pass
+        path.write_text(content, encoding="utf-8")
 
 
 # ══════════════════════════════════════════════════════════════════
@@ -571,18 +609,6 @@ ODS_SPECS: list[dict] = [
         "extra_params": {},
         "description": "助教服务流水",
     },
-    {
-        "code": "ODS_ASSISTANT_ABOLISH",
-        "table": "assistant_cancellation_records",
-        "dwd_table": "dwd_assistant_trash_event",
-        "endpoint": "/AssistantPerformance/GetAbolitionAssistant",
-        "data_path": ("data",),
-        "list_key": "abolitionAssistants",
-        "time_fields": ("startTime", "endTime"),
-        "requires_window": True,
-        "extra_params": {},
-        "description": "助教废除记录",
-    },
     {
         "code": "ODS_STORE_GOODS_SALES",
         "table": "store_goods_sales_records",
@@ -788,18 +814,6 @@ ODS_SPECS: list[dict] = [
         "extra_params": {},
         "description": "租户商品档案",
     },
-    {
-        "code": "ODS_SETTLEMENT_TICKET",
-        "table": "settlement_ticket_details",
-        "dwd_table": None,
-        "endpoint": "/Order/GetOrderSettleTicketNew",
-        "data_path": (),
-        "list_key": None,
-        "time_fields": None,
-        "requires_window": False,
-        "extra_params": {},
-        "description": "结账小票详情（按 orderSettleId 逐条获取，不走常规分页）",
-    },
 ]
 
 # 默认 list_key 候选（与 APIClient 一致）
@@ -896,7 +910,6 @@ def fetch_records(spec: dict, config: AnalyzerConfig) -> list[dict]:
 
     - 有时间字段的表：从今天往回 10 天一批，不够则继续扩展，最多 10 次重试
     - 无时间字段的表：单次请求
-    - 特殊表（settlement_ticket_details）：跳过
 
     参数:
         spec: ODS_SPECS 中的单项配置
@@ -912,10 +925,6 @@ def fetch_records(spec: dict, config: AnalyzerConfig) -> list[dict]:
     extra_params = _resolve_extra_params(spec.get("extra_params", {}), config)
     target_count = config.limit
 
-    # 结账小票是逐条获取的，跳过
-    if spec["table"] == "settlement_ticket_details":
-        return []
-
     tz = ZoneInfo("Asia/Shanghai")
     all_records: list[dict] = []
 
@@ -973,8 +982,10 @@ def fetch_records(spec: dict, config: AnalyzerConfig) -> list[dict]:
 
 import re
 
-# DWD 加载任务源码的默认路径
-_DWD_TASK_PY = Path("apps/etl/connectors/feiqiu/tasks/dwd/dwd_load_task.py")
+# DWD 加载任务源码的默认路径（使用绝对路径，避免 cwd 不在项目根时找不到）
+# CHANGE 2026-02-21 | 相对路径 → 绝对路径，与 _env_paths 同源
+_PROJECT_ROOT = Path(__file__).resolve().parents[2]
+_DWD_TASK_PY = _PROJECT_ROOT / "apps" / "etl" / "connectors" / "feiqiu" / "tasks" / "dwd" / "dwd_load_task.py"
 
 
 def parse_table_map(py_path: Path | None = None) -> dict[str, str]:
@@ -1059,8 +1070,9 @@ def parse_fact_mappings(py_path: Path | None = None) -> dict[str, list[tuple]]:
 # BD_manual 文档解析：提取字段级业务描述
 # ══════════════════════════════════════════════════════════════════
 
-# BD_manual 文档根目录
-_BD_DOCS_ROOT = Path("apps/etl/connectors/feiqiu/docs/database")
+# BD_manual 文档根目录（使用绝对路径，与 _DWD_TASK_PY 同源）
+# CHANGE 2026-02-21 | 相对路径 → 绝对路径，避免 cwd 不在项目根时找不到
+_BD_DOCS_ROOT = _PROJECT_ROOT / "apps" / "etl" / "connectors" / "feiqiu" / "docs" / "database"
 
 
 def parse_bd_manual_fields(doc_path: Path) -> dict[str, str]: