fix: 优化虫 机制，新增伊朗支援

crawler/db_merge.py

@@ -26,6 +26,12 @@ MAX_DELTA_PER_MERGE = {
     "civilian_ships": 20, "airport_port": 10,
 }
 
+# 反击情绪 / 华尔街：合理区间，避免爬虫单条提取 0 或 100 导致指标归零或打满
+RETALIATION_SMOOTH_WEIGHT = 0.6  # 当前值权重，1 - 此值为新值权重，使更新平滑
+RETALIATION_HISTORY_MAX_ROWS = 300  # 反击历史条数上限，供前端曲线与回放使用
+WALL_STREET_TREND_MAX_ROWS = 200  # 趋势表保留最近条数，避免无限增长
+VALUE_CLAMP_MIN, VALUE_CLAMP_MAX = 1, 99  # 0/100 视为异常，写入前夹在 [1,99]
+
 
 def _clamp_delta(key: str, value: int) -> int:
     """单次增量上限，避免误提「累计」导致波动"""
@@ -200,38 +206,69 @@ def merge(extracted: Dict[str, Any], db_path: Optional[str] = None) -> bool:
                             updated = True
                     except Exception:
                         pass
-        # retaliation
+        # retaliation：平滑更新，避免单条新闻 0/100 导致指标归零或打满
         if "retaliation" in extracted:
             r = extracted["retaliation"]
-            conn.execute("INSERT OR REPLACE INTO retaliation_current (id, value) VALUES (1, ?)", (r["value"],))
-            conn.execute("INSERT INTO retaliation_history (time, value) VALUES (?, ?)", (r["time"], r["value"]))
+            raw = max(VALUE_CLAMP_MIN, min(VALUE_CLAMP_MAX, int(r.get("value", 50))))
+            row = conn.execute("SELECT value FROM retaliation_current WHERE id = 1").fetchone()
+            current = int(row[0]) if row else 50
+            current = max(VALUE_CLAMP_MIN, min(VALUE_CLAMP_MAX, current))
+            new_val = round(
+                RETALIATION_SMOOTH_WEIGHT * current + (1 - RETALIATION_SMOOTH_WEIGHT) * raw
+            )
+            new_val = max(VALUE_CLAMP_MIN, min(VALUE_CLAMP_MAX, new_val))
+            ts = (r.get("time") or datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%S.000Z"))[:25]
+            conn.execute("INSERT OR REPLACE INTO retaliation_current (id, value) VALUES (1, ?)", (new_val,))
+            conn.execute("INSERT INTO retaliation_history (time, value) VALUES (?, ?)", (ts, new_val))
+            n_ret = conn.execute("SELECT COUNT(*) FROM retaliation_history").fetchone()[0]
+            if n_ret > RETALIATION_HISTORY_MAX_ROWS:
+                conn.execute(
+                    "DELETE FROM retaliation_history WHERE id IN (SELECT id FROM retaliation_history ORDER BY time ASC LIMIT ?)",
+                    (n_ret - RETALIATION_HISTORY_MAX_ROWS,),
+                )
             updated = True
-        # wall_street_trend
+        # wall_street_trend：限幅后写入，并保留最近 N 条避免表无限增长
         if "wall_street" in extracted:
             w = extracted["wall_street"]
-            conn.execute("INSERT INTO wall_street_trend (time, value) VALUES (?, ?)", (w["time"], w["value"]))
+            raw = int(w.get("value", 50))
+            val = max(VALUE_CLAMP_MIN, min(VALUE_CLAMP_MAX, raw))
+            ts = (w.get("time") or datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%S.000Z"))[:25]
+            conn.execute("INSERT INTO wall_street_trend (time, value) VALUES (?, ?)", (ts, val))
+            n = conn.execute("SELECT COUNT(*) FROM wall_street_trend").fetchone()[0]
+            if n > WALL_STREET_TREND_MAX_ROWS:
+                conn.execute(
+                    "DELETE FROM wall_street_trend WHERE id IN (SELECT id FROM wall_street_trend ORDER BY time ASC LIMIT ?)",
+                    (n - WALL_STREET_TREND_MAX_ROWS,),
+                )
             updated = True
-        # key_location：更新双方攻击地点（美军基地被打击 side=us，伊朗设施被打击 side=iran）的 status/damage_level
+        # key_location：更新双方攻击地点（美军基地被打击 side=us，伊朗设施被打击 side=iran）的 status/damage_level/attacked_at
+        event_time = extracted.get("_event_time") or datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%S.000Z")
         if "key_location_updates" in extracted:
             try:
                 for u in extracted["key_location_updates"]:
                     kw_raw = (u.get("name_keywords") or "").strip()
                     if not kw_raw:
                         continue
-                    # 支持 "a|b|c" 或 "a b c" 分隔
                     kw = [k.strip() for k in kw_raw.replace("|", " ").split() if k.strip()]
                     side = u.get("side")
                     status = (u.get("status") or "attacked")[:20]
                     dmg = u.get("damage_level", 2)
                     if not kw or side not in ("us", "iran"):
                         continue
-                    # 简化：name LIKE '%kw%' 对每个关键词 OR 连接，支持中英文
+                    attacked_at = (u.get("attacked_at") or event_time)[:25]
                     conditions = " OR ".join("name LIKE ?" for _ in kw)
-                    params = [status, dmg, side] + [f"%{k}%" for k in kw]
-                    cur = conn.execute(
-                        f"UPDATE key_location SET status=?, damage_level=? WHERE side=? AND ({conditions})",
-                        params,
-                    )
+                    params_with_at = [status, dmg, attacked_at, side] + [f"%{k}%" for k in kw]
+                    try:
+                        cur = conn.execute(
+                            f"UPDATE key_location SET status=?, damage_level=?, attacked_at=? WHERE side=? AND ({conditions})",
+                            params_with_at,
+                        )
+                    except sqlite3.OperationalError:
+                        params_no_at = [status, dmg, side] + [f"%{k}%" for k in kw]
+                        cur = conn.execute(
+                            f"UPDATE key_location SET status=?, damage_level=? WHERE side=? AND ({conditions})",
+                            params_no_at,
+                        )
                     if cur.rowcount > 0:
                         updated = True
             except Exception: