DWD完成

etl_billiards/tasks/init_dwd_schema_task.py

@@ -0,0 +1,36 @@
+# -*- coding: utf-8 -*-
+"""初始化 DWD Schema：执行 schema_dwd_doc.sql，可选先 DROP SCHEMA。"""
+from __future__ import annotations
+
+from pathlib import Path
+from typing import Any
+
+from .base_task import BaseTask, TaskContext
+
+
+class InitDwdSchemaTask(BaseTask):
+    """通过调度执行 DWD schema 初始化。"""
+
+    def get_task_code(self) -> str:
+        """返回任务编码。"""
+        return "INIT_DWD_SCHEMA"
+
+    def extract(self, context: TaskContext) -> dict[str, Any]:
+        """读取 DWD SQL 文件与参数。"""
+        base_dir = Path(__file__).resolve().parents[1] / "database"
+        dwd_path = Path(self.config.get("schema.dwd_file", base_dir / "schema_dwd_doc.sql"))
+        if not dwd_path.exists():
+            raise FileNotFoundError(f"未找到 DWD schema 文件: {dwd_path}")
+
+        drop_first = self.config.get("dwd.drop_schema_first", False)
+        return {"dwd_sql": dwd_path.read_text(encoding="utf-8"), "dwd_file": str(dwd_path), "drop_first": drop_first}
+
+    def load(self, extracted: dict[str, Any], context: TaskContext) -> dict:
+        """可选 DROP schema，再执行 DWD DDL。"""
+        with self.db.conn.cursor() as cur:
+            if extracted["drop_first"]:
+                cur.execute("DROP SCHEMA IF EXISTS billiards_dwd CASCADE;")
+                self.logger.info("已执行 DROP SCHEMA billiards_dwd CASCADE")
+            self.logger.info("执行 DWD schema 文件: %s", extracted["dwd_file"])
+            cur.execute(extracted["dwd_sql"])
+        return {"executed": 1, "files": [extracted["dwd_file"]]}