支持验收项目独立启动爬虫采集

acceptance_service.py

@@ -950,7 +950,7 @@ def create_acceptance_task(data: dict[str, Any]) -> dict[str, Any]:
         raise TaskError("KEYWORD_LIMIT_EXCEEDED", "单次最多支持 20 个搜索词", 422)
     if method in {"screenshot", "combined"} and not keywords:
         raise TaskError("KEYWORD_REQUIRED", "截图取证请至少输入一个搜索词", 422)
-    if method in {"data-analysis", "combined"} and not links:
+    if method in {"data-analysis", "combined"} and not links and not keywords:
         raise TaskError("FEISHU_LINK_REQUIRED", "笔记收录请至少输入一个公开飞书链接", 422)
     resources = {
         "cookie_count": max(0, int(raw_resources.get("cookie_count") or 0)),
@@ -1194,6 +1194,35 @@ def collect_acceptance_now(public_id: str) -> dict[str, Any]:
     return _copy_acceptance_task(public_id, trigger_source="acceptance")
 
 
+def collect_acceptance_crawler_now(public_id: str) -> dict[str, Any]:
+    """按项目搜索词启动独立的 Cookie 爬虫采集批次，不下发截图设备。"""
+    with connection() as conn:
+        task = conn.execute("SELECT * FROM acceptance_tasks WHERE public_task_id=?", (public_id,)).fetchone()
+        if not task:
+            raise TaskError("ACCEPTANCE_TASK_NOT_FOUND", "验收任务不存在", 404)
+        root_id = task["scheduled_from_task_id"] or task["id"]
+        root = conn.execute("SELECT * FROM acceptance_tasks WHERE id=?", (root_id,)).fetchone()
+        keywords = [row["keyword"] for row in conn.execute("SELECT keyword FROM acceptance_keywords WHERE task_id=? ORDER BY keyword_order", (root_id,)).fetchall()]
+        matches = [row["keyword"] for row in conn.execute("SELECT keyword FROM acceptance_match_keywords WHERE task_id=? ORDER BY keyword_order", (root_id,)).fetchall()]
+    child = create_acceptance_task({
+        "name": f"{root['name']}｜爬虫数据采集",
+        "acceptance_method": "data-analysis",
+        "feishu_links": [],
+        "keywords": keywords,
+        "match_keywords": matches,
+        "created_by_id": root["created_by_id"],
+        "created_by_name": root["created_by_name"],
+        "trigger_source": "acceptance",
+    })
+    with connection() as conn:
+        conn.execute(
+            """UPDATE acceptance_tasks SET scheduled_from_task_id=?,dashboard_batch_key=?
+               WHERE public_task_id=?""",
+            (root_id, root["dashboard_batch_key"], child["task_id"]),
+        )
+    return get_acceptance_task(child["task_id"])
+
+
 def run_due_acceptance_schedules() -> int:
     current = datetime.now().astimezone()
     now = current.isoformat(timespec="seconds")