fix: 优化后台数据

crawler/extractor_rules.py

@@ -55,16 +55,16 @@ def extract_from_news(text: str, timestamp: Optional[str] = None) -> Dict[str, A
     if v is not None:
         loss_us["civilian_wounded"] = v
 
-    # 基地损毁（美方基地居多）
-    v = _first_int(t, r"(\d+)[\s\w]*(?:base)[\s\w]*(?:destroyed|leveled)")
+    # 基地损毁（美方基地居多）+ 中文
+    v = _first_int(t, r"(\d+)[\s\w]*(?:base|基地)[\s\w]*(?:destroyed|leveled|摧毁|夷平)")
     if v is not None:
         loss_us["bases_destroyed"] = v
-    v = _first_int(t, r"(\d+)[\s\w]*(?:base)[\s\w]*(?:damaged|hit|struck)")
+    v = _first_int(t, r"(\d+)[\s\w]*(?:base|基地)[\s\w]*(?:damaged|hit|struck|受损|袭击)")
     if v is not None:
         loss_us["bases_damaged"] = v
-    if "base" in t and ("destroy" in t or "level" in t) and not loss_us.get("bases_destroyed"):
+    if ("base" in t or "基地" in t) and ("destroy" in t or "level" in t or "摧毁" in t or "夷平" in t) and not loss_us.get("bases_destroyed"):
         loss_us["bases_destroyed"] = 1
-    if "base" in t and ("damage" in t or "hit" in t or "struck" in t or "strike" in t) and not loss_us.get("bases_damaged"):
+    if ("base" in t or "基地" in t) and ("damage" in t or "hit" in t or "struck" in t or "strike" in t or "袭击" in t or "受损" in t) and not loss_us.get("bases_damaged"):
         loss_us["bases_damaged"] = 1
 
     # 战机 / 舰船（根据上下文判断阵营）
@@ -92,4 +92,32 @@ def extract_from_news(text: str, timestamp: Optional[str] = None) -> Dict[str, A
     if "wall street" in t or " dow " in t or "s&p" in t or "market slump" in t or "stock fall" in t or "美股" in t:
         out["wall_street"] = {"time": ts, "value": 55}
 
+    # key_location_updates：受袭基地（与 key_location.name 匹配）
+    # 新闻提及基地遭袭时，更新对应基地 status
+    base_attacked = ("base" in t or "基地" in t) and ("attack" in t or "hit" in t or "strike" in t or "damage" in t or "袭击" in t or "打击" in t)
+    if base_attacked:
+        updates: list = []
+        # 常见美军基地关键词 -> name_keywords（用于 db_merge 的 LIKE 匹配）
+        bases_us = [
+            ("阿萨德|阿因|asad|assad|ain", "us"),
+            ("巴格达|baghdad", "us"),
+            ("乌代德|udeid|卡塔尔|qatar", "us"),
+            ("阿克罗蒂里|akrotiri|塞浦路斯|cyprus", "us"),
+            ("巴格拉姆|bagram|阿富汗|afghanistan", "us"),
+            ("埃尔比勒|erbil", "us"),
+            ("因吉尔利克|incirlik|土耳其|turkey", "us"),
+            ("苏尔坦|sultan|沙特|saudi", "us"),
+            ("坦夫|tanf|叙利亚|syria", "us"),
+            ("达夫拉|dhafra|阿联酋|uae", "us"),
+            ("内瓦提姆|nevatim|拉蒙|ramon|以色列|israel", "us"),
+            ("赛利耶|sayliyah", "us"),
+            ("巴林|bahrain", "us"),
+            ("科威特|kuwait", "us"),
+        ]
+        for kws, side in bases_us:
+            if any(k in t for k in kws.split("|")):
+                updates.append({"name_keywords": kws, "side": side, "status": "attacked", "damage_level": 2})
+        if updates:
+            out["key_location_updates"] = updates
+
     return out