fix: 优化数据

fix:优化数据来源
fix:优化整个大屏界面
2026-03-02 11:28:13 +08:00 · 2026-03-02 01:00:04 +08:00 · 2026-03-02 00:59:40 +08:00
56 changed files with 2732 additions and 108 deletions
--- a/=1.11.0
+++ b/=1.11.0
--- a/README.md
+++ b/README.md
@@ -41,7 +41,13 @@ npm run api:seed
 npm run api
 ```

-开发时需同时运行前端与 API：
+开发时可用一键启动（推荐）：
+
+```bash
+npm start
+```
+
+或分终端分别运行：

 ```bash
 # 终端 1
@@ -53,6 +59,13 @@ npm run dev

 API 会由 Vite 代理到 `/api`，前端通过 `/api/situation` 获取完整态势数据。数据库文件位于 `server/data.db`，可通过修改表数据实现动态调整。

+### 爬虫不生效时
+
+1. 测试 RSS 抓取：`npm run crawler:test`（需网络，返回抓取条数）
+2. 单独启动爬虫查看日志：`npm run gdelt`（另开终端）
+3. 查看爬虫状态：`curl http://localhost:8000/crawler/status`（需爬虫服务已启动）
+4. 数据库面板 `/db` 每 30 秒自动刷新，可观察 situation_update 条数是否增加
+
 ## Development

 ```bash
--- a/crawler/README.md
+++ b/crawler/README.md
@@ -0,0 +1,105 @@
+# GDELT 实时冲突服务 + 新闻爬虫
+
+## 数据来源梳理
+
+### 1. GDELT Project (gdelt_events)
+
+| 项目 | 说明 |
+|------|------|
+| API | `https://api.gdeltproject.org/api/v2/doc/doc` |
+| 查询 | `query=United States Iran military`（可配 `GDELT_QUERY`） |
+| 模式 | `mode=ArtList`，`format=json`，`maxrecords=30` |
+| 时间范围 | **未指定时默认最近 3 个月**，按相关性排序，易返回较旧文章 |
+| 更新频率 | GDELT 约 15 分钟级，爬虫 60 秒拉一次 |
+
+**数据偏老原因**：未传 `timespan` 和 `sort=datedesc`，API 返回 3 个月内“最相关”文章，不保证最新。
+
+### 2. RSS 新闻 (situation_update) — 主事件脉络来源
+
+| 项目 | 说明 |
+|------|------|
+| 源 | 多国主流媒体：美(Reuters/NYT)、英(BBC)、法(France 24)、俄(TASS/RT)、中(Xinhua/CGTN)、伊(Press TV)、卡塔尔(Al Jazeera) |
+| 过滤 | 标题/摘要需含 `KEYWORDS` 之一（iran、usa、strike、military 等） |
+| 更新 | 爬虫 45 秒拉一次（`RSS_INTERVAL_SEC`），优先保证事件脉络 |
+| 优先级 | 启动时先拉 RSS，再拉 GDELT |
+
+**GDELT 无法访问时**：设置 `GDELT_DISABLED=1`，仅用 RSS 新闻即可维持事件脉络。部分境外源可能受网络限制。
+
+### 3. AI 新闻清洗与分类（可选）
+
+- **清洗**：`cleaner_ai.py` 用 Ollama 提炼新闻为简洁摘要，供面板展示
+- **分类**：`parser_ai.py` 用 Ollama 替代规则做 category/severity 判定
+- 需先安装并运行 Ollama：`ollama run llama3.1`
+- 环境变量：`OLLAMA_MODEL=llama3.1`、`PARSER_AI_DISABLED=1`、`CLEANER_AI_DISABLED=1`（禁用对应 AI）
+
+---
+
+**事件脉络可实时更新**：爬虫抓取后 → 写入 SQLite → 调用 Node 通知 → WebSocket 广播 → 前端自动刷新。
+
+## 依赖
+
+```bash
+pip install -r requirements.txt
+```
+
+新增 `deep-translator`：GDELT 与 RSS 新闻入库前自动翻译为中文。
+
+## 运行（需同时启动 3 个服务）
+
+| 终端 | 命令 | 说明 |
+|------|------|------|
+| 1 | `npm run api` | Node API + WebSocket（必须） |
+| 2 | `npm run gdelt` | GDELT + RSS 爬虫（**事件脉络数据来源**） |
+| 3 | `npm run dev` | 前端开发 |
+
+**事件脉络不更新时**：多半是未启动 `npm run gdelt`。只跑 `npm run api` 时，事件脉络会显示空或仅有缓存。
+
+## 数据流
+
+```
+GDELT API → 抓取(60s) → SQLite (gdelt_events, conflict_stats) → POST /api/crawler/notify
+                                                                    ↓
+                                              Node 更新 situation.updated_at + WebSocket 广播
+                                                                    ↓
+                                                              前端实时展示
+```
+
+## 配置
+
+环境变量：
+
+- `DB_PATH`: SQLite 路径，默认 `../server/data.db`
+- `API_BASE`: Node API 地址，默认 `http://localhost:3001`
+- `GDELT_QUERY`: 搜索关键词，默认 `United States Iran military`
+- `GDELT_MAX_RECORDS`: 最大条数，默认 30
+- `GDELT_TIMESPAN`: 时间范围，`1h` / `1d` / `1week`，默认 `1d`（近日资讯）
+- `GDELT_DISABLED`: 设为 `1` 则跳过 GDELT，仅用 RSS 新闻（GDELT 无法访问时用）
+- `FETCH_INTERVAL_SEC`: GDELT 抓取间隔（秒），默认 60
+- `RSS_INTERVAL_SEC`: RSS 抓取间隔（秒），默认 45（优先保证事件脉络）
+- `OLLAMA_MODEL`: AI 分类模型，默认 `llama3.1`
+- `PARSER_AI_DISABLED`: 设为 `1` 则禁用 AI 分类，仅用规则
+- `CLEANER_AI_DISABLED`: 设为 `1` 则禁用 AI 清洗，仅用规则截断
+
+## 冲突强度 (impact_score)
+
+| 分数 | 地图效果   |
+|------|------------|
+| 1–3  | 绿色点     |
+| 4–6  | 橙色闪烁   |
+| 7–10 | 红色脉冲扩散 |
+
+## API
+
+- `GET http://localhost:8000/events`：返回事件列表与冲突统计（Python 服务直连）
+- `GET http://localhost:3001/api/events`：从 Node 读取（推荐，含 WebSocket 同步）
+
+## 故障排查
+
+| 现象 | 可能原因 | 排查 |
+|------|----------|------|
+| 事件脉络始终为空 | 未启动 GDELT 爬虫 | 另开终端运行 `npm run gdelt`，观察是否有 `GDELT 更新 X 条事件` 输出 |
+| 事件脉络不刷新 | WebSocket 未连上 | 确认 `npm run api` 已启动，前端需通过 `npm run dev` 访问（Vite 会代理 /ws） |
+| GDELT 抓取失败 | 系统代理超时 / ProxyError | 爬虫默认直连，不走代理；若需代理请设 `CRAWLER_USE_PROXY=1` |
+| GDELT 抓取失败 | 网络 / GDELT API 限流 | 检查 Python 终端报错；GDELT 在国外，国内网络可能较慢或超时 |
+| 新闻条数为 0 | RSS 源被墙或关键词不匹配 | 检查 crawler/config.py 中 RSS_FEEDS、KEYWORDS；国内需代理 |
+| **返回数据偏老** | GDELT 默认 3 个月内按相关性 | 设置 `GDELT_TIMESPAN=1d` 限制为近日；加 `sort=datedesc` 最新优先 |
--- a/crawler/pycache/cleaner_ai.cpython-39.pyc
+++ b/crawler/pycache/cleaner_ai.cpython-39.pyc
--- a/crawler/pycache/config.cpython-39.pyc
+++ b/crawler/pycache/config.cpython-39.pyc
--- a/crawler/pycache/db_writer.cpython-39.pyc
+++ b/crawler/pycache/db_writer.cpython-39.pyc
--- a/crawler/pycache/parser.cpython-39.pyc
+++ b/crawler/pycache/parser.cpython-39.pyc
--- a/crawler/pycache/parser_ai.cpython-39.pyc
+++ b/crawler/pycache/parser_ai.cpython-39.pyc
--- a/crawler/pycache/realtime_conflict_service.cpython-39.pyc
+++ b/crawler/pycache/realtime_conflict_service.cpython-39.pyc
--- a/crawler/pycache/translate_utils.cpython-39.pyc
+++ b/crawler/pycache/translate_utils.cpython-39.pyc
--- a/crawler/cleaner_ai.py
+++ b/crawler/cleaner_ai.py
@@ -0,0 +1,87 @@
+# -*- coding: utf-8 -*-
+"""
+AI 清洗新闻数据，严格按面板字段约束输出
+面板 EventTimelinePanel 所需：summary(≤120字)、category(枚举)、severity(枚举)
+"""
+import os
+import re
+from typing import Optional
+
+CLEANER_AI_DISABLED = os.environ.get("CLEANER_AI_DISABLED", "0") == "1"
+OLLAMA_MODEL = os.environ.get("OLLAMA_MODEL", "llama3.1")
+
+# 面板 schema：必须与 EventTimelinePanel / SituationUpdate 一致
+SUMMARY_MAX_LEN = 120  # 面板 line-clamp-2 展示
+CATEGORIES = ("deployment", "alert", "intel", "diplomatic", "other")
+SEVERITIES = ("low", "medium", "high", "critical")
+
+
+def _sanitize_summary(text: str, max_len: int = SUMMARY_MAX_LEN) -> str:
+    """确保 summary 符合面板：纯文本、无换行、限制长度"""
+    if not text or not isinstance(text, str):
+        return ""
+    s = re.sub(r"\s+", " ", str(text).strip())
+    s = re.sub(r"[\x00-\x08\x0b\x0c\x0e-\x1f]", "", s)  # 去除控制字符
+    return s[:max_len].rstrip()
+
+
+def _rule_clean(text: str, max_len: int = SUMMARY_MAX_LEN) -> str:
+    """规则清洗：去空白、去控制符、截断"""
+    return _sanitize_summary(text, max_len)
+
+
+def _call_ollama_summary(text: str, max_len: int, timeout: int = 6) -> Optional[str]:
+    """调用 Ollama 提炼摘要，输出须为纯文本、≤max_len 字"""
+    if CLEANER_AI_DISABLED or not text or len(str(text).strip()) < 5:
+        return None
+    try:
+        import requests
+        prompt = f"""将新闻提炼为1-2句简洁中文事实，直接输出纯文本，不要标号、引号、解释。限{max_len}字内。
+
+原文：{str(text)[:350]}
+
+输出："""
+        r = requests.post(
+            "http://localhost:11434/api/chat",
+            json={
+                "model": OLLAMA_MODEL,
+                "messages": [{"role": "user", "content": prompt}],
+                "stream": False,
+                "options": {"num_predict": 150},
+            },
+            timeout=timeout,
+        )
+        if r.status_code != 200:
+            return None
+        out = (r.json().get("message", {}).get("content", "") or "").strip()
+        out = re.sub(r"^[\d\.\-\*\s]+", "", out)  # 去编号
+        out = re.sub(r"^['\"\s]+|['\"\s]+$", "", out)
+        out = _sanitize_summary(out, max_len)
+        if out and len(out) > 3:
+            return out
+        return None
+    except Exception:
+        return None
+
+
+def clean_news_for_panel(text: str, max_len: int = SUMMARY_MAX_LEN) -> str:
+    """清洗 summary 字段，供 EventTimelinePanel 展示。输出必为≤max_len 的纯文本"""
+    if not text or not isinstance(text, str):
+        return ""
+    t = str(text).strip()
+    if not t:
+        return ""
+    res = _call_ollama_summary(t, max_len, timeout=6)
+    if res:
+        return res
+    return _rule_clean(t, max_len)
+
+
+def ensure_category(cat: str) -> str:
+    """确保 category 在面板枚举内"""
+    return cat if cat in CATEGORIES else "other"
+
+
+def ensure_severity(sev: str) -> str:
+    """确保 severity 在面板枚举内"""
+    return sev if sev in SEVERITIES else "medium"
--- a/crawler/config.py
+++ b/crawler/config.py
@@ -0,0 +1,49 @@
+# -*- coding: utf-8 -*-
+"""爬虫配置"""
+import os
+from pathlib import Path
+
+# 数据库路径（与 server 共用 SQLite）
+PROJECT_ROOT = Path(__file__).resolve().parent.parent
+DB_PATH = os.environ.get("DB_PATH", str(PROJECT_ROOT / "server" / "data.db"))
+
+# Node API 地址（用于通知推送）
+API_BASE = os.environ.get("API_BASE", "http://localhost:3001")
+
+# 抓取间隔（秒）
+CRAWL_INTERVAL = int(os.environ.get("CRAWL_INTERVAL", "300"))
+
+# RSS 源：世界主流媒体，覆盖美伊/中东多视角
+RSS_FEEDS = [
+    # 美国
+    "https://feeds.reuters.com/reuters/topNews",
+    "https://rss.nytimes.com/services/xml/rss/nyt/World.xml",
+    # 英国
+    "https://feeds.bbci.co.uk/news/world/rss.xml",
+    "https://feeds.bbci.co.uk/news/world/middle_east/rss.xml",
+    "https://www.theguardian.com/world/rss",
+    # 法国
+    "https://www.france24.com/en/rss",
+    # 德国
+    "https://rss.dw.com/xml/rss-en-world",
+    # 俄罗斯
+    "https://tass.com/rss/v2.xml",
+    "https://www.rt.com/rss/",
+    # 中国
+    "https://english.news.cn/rss/world.xml",
+    "https://www.cgtn.com/rss/world",
+    # 伊朗
+    "https://www.presstv.ir/rss",
+    # 卡塔尔（中东）
+    "https://www.aljazeera.com/xml/rss/all.xml",
+    "https://www.aljazeera.com/xml/rss/middleeast.xml",
+]
+
+# 关键词过滤：至少匹配一个才会入库
+KEYWORDS = [
+    "iran", "iranian", "tehran", "以色列", "israel",
+    "usa", "us ", "american", "美军", "美国",
+    "middle east", "中东", "persian gulf", "波斯湾",
+    "strike", "attack", "military", "missile", "核", "nuclear",
+    "carrier", "航母", "houthi", "胡塞", "hamas",
+]
--- a/crawler/db_merge.py
+++ b/crawler/db_merge.py
@@ -0,0 +1,126 @@
+# -*- coding: utf-8 -*-
+"""
+将 AI 提取的结构化数据合并到 SQLite
+与 panel schema 及 situationData.getSituation 对齐，支持回放
+"""
+import os
+import sqlite3
+from datetime import datetime
+from pathlib import Path
+from typing import Any, Dict, Optional
+
+PROJECT_ROOT = Path(__file__).resolve().parent.parent
+DB_PATH = os.environ.get("DB_PATH", str(PROJECT_ROOT / "server" / "data.db"))
+
+
+def _ensure_tables(conn: sqlite3.Connection) -> None:
+    """确保所需表存在（与 db.js 一致）"""
+    conn.execute("""
+        CREATE TABLE IF NOT EXISTS situation_update (
+            id TEXT PRIMARY KEY, timestamp TEXT NOT NULL, category TEXT NOT NULL,
+            summary TEXT NOT NULL, severity TEXT NOT NULL
+        )
+    """)
+    conn.execute("""
+        CREATE TABLE IF NOT EXISTS combat_losses (
+            side TEXT PRIMARY KEY CHECK (side IN ('us', 'iran')),
+            bases_destroyed INTEGER NOT NULL, bases_damaged INTEGER NOT NULL,
+            personnel_killed INTEGER NOT NULL, personnel_wounded INTEGER NOT NULL,
+            aircraft INTEGER NOT NULL, warships INTEGER NOT NULL, armor INTEGER NOT NULL, vehicles INTEGER NOT NULL
+        )
+    """)
+    try:
+        conn.execute("ALTER TABLE combat_losses ADD COLUMN civilian_killed INTEGER NOT NULL DEFAULT 0")
+    except sqlite3.OperationalError:
+        pass
+    try:
+        conn.execute("ALTER TABLE combat_losses ADD COLUMN civilian_wounded INTEGER NOT NULL DEFAULT 0")
+    except sqlite3.OperationalError:
+        pass
+    try:
+        conn.execute("ALTER TABLE combat_losses ADD COLUMN updated_at TEXT DEFAULT (datetime('now'))")
+    except sqlite3.OperationalError:
+        pass
+    conn.execute("CREATE TABLE IF NOT EXISTS wall_street_trend (id INTEGER PRIMARY KEY AUTOINCREMENT, time TEXT NOT NULL, value INTEGER NOT NULL)")
+    conn.execute("CREATE TABLE IF NOT EXISTS retaliation_current (id INTEGER PRIMARY KEY CHECK (id = 1), value INTEGER NOT NULL)")
+    conn.execute("CREATE TABLE IF NOT EXISTS retaliation_history (id INTEGER PRIMARY KEY AUTOINCREMENT, time TEXT NOT NULL, value INTEGER NOT NULL)")
+    conn.execute("CREATE TABLE IF NOT EXISTS situation (id INTEGER PRIMARY KEY CHECK (id = 1), data TEXT NOT NULL, updated_at TEXT NOT NULL)")
+    conn.commit()
+
+
+def merge(extracted: Dict[str, Any], db_path: Optional[str] = None) -> bool:
+    """将提取数据合并到 DB，返回是否有更新"""
+    path = db_path or DB_PATH
+    if not os.path.exists(path):
+        return False
+    conn = sqlite3.connect(path, timeout=10)
+    try:
+        _ensure_tables(conn)
+        updated = False
+        # situation_update
+        if "situation_update" in extracted:
+            u = extracted["situation_update"]
+            uid = f"ai_{hash(u.get('summary','')+u.get('timestamp','')) % 10**10}"
+            conn.execute(
+                "INSERT OR IGNORE INTO situation_update (id, timestamp, category, summary, severity) VALUES (?, ?, ?, ?, ?)",
+                (uid, u.get("timestamp", ""), u.get("category", "other"), u.get("summary", "")[:500], u.get("severity", "medium")),
+            )
+            if conn.total_changes > 0:
+                updated = True
+        # combat_losses：增量叠加到当前值
+        if "combat_losses_delta" in extracted:
+            for side, delta in extracted["combat_losses_delta"].items():
+                if side not in ("us", "iran"):
+                    continue
+                try:
+                    row = conn.execute(
+                        "SELECT personnel_killed,personnel_wounded,civilian_killed,civilian_wounded,bases_destroyed,bases_damaged,aircraft,warships,armor,vehicles FROM combat_losses WHERE side = ?",
+                        (side,),
+                    ).fetchone()
+                    if not row:
+                        continue
+                    cur = {
+                        "personnel_killed": row[0], "personnel_wounded": row[1], "civilian_killed": row[2] or 0,
+                        "civilian_wounded": row[3] or 0, "bases_destroyed": row[4], "bases_damaged": row[5],
+                        "aircraft": row[6], "warships": row[7], "armor": row[8], "vehicles": row[9],
+                    }
+                    pk = max(0, (cur["personnel_killed"] or 0) + delta.get("personnel_killed", 0))
+                    pw = max(0, (cur["personnel_wounded"] or 0) + delta.get("personnel_wounded", 0))
+                    ck = max(0, (cur["civilian_killed"] or 0) + delta.get("civilian_killed", 0))
+                    cw = max(0, (cur["civilian_wounded"] or 0) + delta.get("civilian_wounded", 0))
+                    bd = max(0, (cur["bases_destroyed"] or 0) + delta.get("bases_destroyed", 0))
+                    bm = max(0, (cur["bases_damaged"] or 0) + delta.get("bases_damaged", 0))
+                    ac = max(0, (cur["aircraft"] or 0) + delta.get("aircraft", 0))
+                    ws = max(0, (cur["warships"] or 0) + delta.get("warships", 0))
+                    ar = max(0, (cur["armor"] or 0) + delta.get("armor", 0))
+                    vh = max(0, (cur["vehicles"] or 0) + delta.get("vehicles", 0))
+                    ts = datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%S.000Z")
+                    conn.execute(
+                        """UPDATE combat_losses SET personnel_killed=?, personnel_wounded=?, civilian_killed=?, civilian_wounded=?,
+                        bases_destroyed=?, bases_damaged=?, aircraft=?, warships=?, armor=?, vehicles=?, updated_at=? WHERE side=?""",
+                        (pk, pw, ck, cw, bd, bm, ac, ws, ar, vh, ts, side),
+                    )
+                    if conn.total_changes > 0:
+                        updated = True
+                except Exception:
+                    pass
+        # retaliation
+        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"]))
+            updated = True
+        # wall_street_trend
+        if "wall_street" in extracted:
+            w = extracted["wall_street"]
+            conn.execute("INSERT INTO wall_street_trend (time, value) VALUES (?, ?)", (w["time"], w["value"]))
+            updated = True
+        if updated:
+            conn.execute("INSERT OR REPLACE INTO situation (id, data, updated_at) VALUES (1, '{}', ?)", (datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%S.000Z"),))
+        conn.commit()
+        return updated
+    except Exception as e:
+        conn.rollback()
+        raise e
+    finally:
+        conn.close()
--- a/crawler/db_writer.py
+++ b/crawler/db_writer.py
@@ -0,0 +1,110 @@
+# -*- coding: utf-8 -*-
+"""写入 SQLite 并确保 situation_update 表存在"""
+import sqlite3
+import hashlib
+import os
+from datetime import datetime, timezone
+
+from config import DB_PATH
+
+CATEGORIES = ("deployment", "alert", "intel", "diplomatic", "other")
+SEVERITIES = ("low", "medium", "high", "critical")
+
+
+def _ensure_table(conn: sqlite3.Connection) -> None:
+    conn.execute("""
+        CREATE TABLE IF NOT EXISTS situation_update (
+            id TEXT PRIMARY KEY,
+            timestamp TEXT NOT NULL,
+            category TEXT NOT NULL,
+            summary TEXT NOT NULL,
+            severity TEXT NOT NULL
+        )
+    """)
+    conn.commit()
+
+
+def _make_id(title: str, url: str, published: str) -> str:
+    raw = f"{title}|{url}|{published}"
+    return "nw_" + hashlib.sha256(raw.encode("utf-8")).hexdigest()[:16]
+
+
+def _to_utc_iso(dt: datetime) -> str:
+    if dt.tzinfo:
+        dt = dt.astimezone(timezone.utc)
+    return dt.strftime("%Y-%m-%dT%H:%M:%S.000Z")
+
+
+def insert_update(
+    conn: sqlite3.Connection,
+    title: str,
+    summary: str,
+    url: str,
+    published: datetime,
+    category: str = "other",
+    severity: str = "medium",
+) -> bool:
+    """插入一条更新，若 id 已存在则跳过。返回是否插入了新记录。"""
+    _ensure_table(conn)
+    ts = _to_utc_iso(published)
+    uid = _make_id(title, url, ts)
+    if category not in CATEGORIES:
+        category = "other"
+    if severity not in SEVERITIES:
+        severity = "medium"
+    try:
+        conn.execute(
+            "INSERT OR IGNORE INTO situation_update (id, timestamp, category, summary, severity) VALUES (?, ?, ?, ?, ?)",
+            (uid, ts, category, summary[:500], severity),
+        )
+        conn.commit()
+        return conn.total_changes > 0
+    except Exception:
+        conn.rollback()
+        return False
+
+
+def touch_situation_updated_at(conn: sqlite3.Connection) -> None:
+    """更新 situation 表的 updated_at"""
+    conn.execute(
+        "INSERT OR REPLACE INTO situation (id, data, updated_at) VALUES (1, '{}', ?)",
+        (datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%S.000Z"),),
+    )
+    conn.commit()
+
+
+def write_updates(updates: list[dict]) -> int:
+    """
+    updates: [{"title","summary","url","published","category","severity"}, ...]
+    返回新增条数。
+    """
+    if not os.path.exists(DB_PATH):
+        return 0
+    conn = sqlite3.connect(DB_PATH, timeout=10)
+    try:
+        count = 0
+        for u in updates:
+            pub = u.get("published")
+            if isinstance(pub, str):
+                try:
+                    pub = datetime.fromisoformat(pub.replace("Z", "+00:00"))
+                except ValueError:
+                    pub = datetime.utcnow()
+            elif pub is None:
+                pub = datetime.utcnow()
+            ok = insert_update(
+                conn,
+                title=u.get("title", "")[:200],
+                summary=u.get("summary", "") or u.get("title", ""),
+                url=u.get("url", ""),
+                published=pub,
+                category=u.get("category", "other"),
+                severity=u.get("severity", "medium"),
+            )
+            if ok:
+                count += 1
+        if count > 0:
+            touch_situation_updated_at(conn)
+        return count
+    finally:
+        conn.close()
--- a/crawler/extractor_ai.py
+++ b/crawler/extractor_ai.py
@@ -0,0 +1,100 @@
+# -*- coding: utf-8 -*-
+"""
+从新闻文本中 AI 提取结构化数据，映射到面板 schema
+输出符合 panel_schema 的字段，供 db_merge 写入
+"""
+import json
+import os
+import re
+from datetime import datetime, timezone
+from typing import Any, Dict, List, Optional
+
+from panel_schema import validate_category, validate_severity, validate_summary
+
+CLEANER_AI_DISABLED = os.environ.get("CLEANER_AI_DISABLED", "0") == "1"
+OLLAMA_MODEL = os.environ.get("OLLAMA_MODEL", "llama3.1")
+
+
+def _call_ollama_extract(text: str, timeout: int = 10) -> Optional[Dict[str, Any]]:
+    """调用 Ollama 提取结构化数据。输出 JSON，仅包含新闻中可明确推断的字段"""
+    if CLEANER_AI_DISABLED or not text or len(str(text).strip()) < 10:
+        return None
+    try:
+        import requests
+        prompt = f"""从以下美伊/中东新闻中提取可推断的数值，输出 JSON，仅包含有明确依据的字段。无依据则省略该字段。
+
+要求：
+- summary: 1-2句中文事实，≤80字
+- category: deployment|alert|intel|diplomatic|other
+- severity: low|medium|high|critical
+- us_personnel_killed, iran_personnel_killed 等：仅当新闻明确提及具体数字时填写
+- retaliation_sentiment: 0-100，仅当新闻涉及伊朗报复情绪时
+- wall_street_value: 0-100，仅当新闻涉及美股/市场反应时
+
+原文：{str(text)[:500]}
+
+直接输出 JSON，不要解释："""
+        r = requests.post(
+            "http://localhost:11434/api/chat",
+            json={
+                "model": OLLAMA_MODEL,
+                "messages": [{"role": "user", "content": prompt}],
+                "stream": False,
+                "options": {"num_predict": 256},
+            },
+            timeout=timeout,
+        )
+        if r.status_code != 200:
+            return None
+        raw = (r.json().get("message", {}).get("content", "") or "").strip()
+        raw = re.sub(r"^```\w*\s*|\s*```$", "", raw)
+        return json.loads(raw)
+    except Exception:
+        return None
+
+
+def extract_from_news(text: str, timestamp: Optional[str] = None) -> Dict[str, Any]:
+    """
+    从新闻文本提取结构化数据，严格符合面板 schema
+    返回: { situation_update?, combat_losses_delta?, retaliation?, wall_street?, ... }
+    """
+    ts = timestamp or datetime.now(timezone.utc).strftime("%Y-%m-%dT%H:%M:%S.000Z")
+    out: Dict[str, Any] = {}
+    parsed = _call_ollama_extract(text)
+    if not parsed:
+        return out
+    # situation_update
+    if parsed.get("summary"):
+        out["situation_update"] = {
+            "summary": validate_summary(str(parsed["summary"])[:120], 120),
+            "category": validate_category(str(parsed.get("category", "other")).lower()),
+            "severity": validate_severity(str(parsed.get("severity", "medium")).lower()),
+            "timestamp": ts,
+        }
+    # combat_losses 增量（仅数字字段）
+    loss_us = {}
+    loss_ir = {}
+    for k in ["personnel_killed", "personnel_wounded", "civilian_killed", "civilian_wounded", "bases_destroyed", "bases_damaged", "aircraft", "warships", "armor", "vehicles"]:
+        uk = f"us_{k}"
+        ik = f"iran_{k}"
+        if uk in parsed and isinstance(parsed[uk], (int, float)):
+            loss_us[k] = max(0, int(parsed[uk]))
+        if ik in parsed and isinstance(parsed[ik], (int, float)):
+            loss_ir[k] = max(0, int(parsed[ik]))
+    if loss_us or loss_ir:
+        out["combat_losses_delta"] = {}
+        if loss_us:
+            out["combat_losses_delta"]["us"] = loss_us
+        if loss_ir:
+            out["combat_losses_delta"]["iran"] = loss_ir
+    # retaliation
+    if "retaliation_sentiment" in parsed:
+        v = parsed["retaliation_sentiment"]
+        if isinstance(v, (int, float)) and 0 <= v <= 100:
+            out["retaliation"] = {"value": int(v), "time": ts}
+    # wall_street
+    if "wall_street_value" in parsed:
+        v = parsed["wall_street_value"]
+        if isinstance(v, (int, float)) and 0 <= v <= 100:
+            out["wall_street"] = {"time": ts, "value": int(v)}
+    return out
--- a/crawler/extractor_rules.py
+++ b/crawler/extractor_rules.py
@@ -0,0 +1,49 @@
+# -*- coding: utf-8 -*-
+"""
+基于规则的新闻数据提取（无需 Ollama）
+从新闻文本中提取战损、报复情绪等数值，供 db_merge 写入
+"""
+import re
+from datetime import datetime, timezone
+from typing import Any, Dict, Optional
+
+
+def _first_int(text: str, pattern: str) -> Optional[int]:
+    m = re.search(pattern, text, re.I)
+    if m and m.group(1) and m.group(1).replace(",", "").isdigit():
+        return max(0, int(m.group(1).replace(",", "")))
+    return None
+
+
+def extract_from_news(text: str, timestamp: Optional[str] = None) -> Dict[str, Any]:
+    """
+    规则提取：匹配数字+关键词，输出符合 panel schema 的字段（无需 Ollama）
+    """
+    ts = timestamp or datetime.now(timezone.utc).strftime("%Y-%m-%dT%H:%M:%S.000Z")
+    out: Dict[str, Any] = {}
+    t = (text or "").lower()
+
+    loss_us, loss_ir = {}, {}
+    v = _first_int(t, r"(?:us|american|u\.?s\.?)[\s\w]*(?:say|report)[\s\w]*(\d+)[\s\w]*(?:troop|soldier|killed|dead)")
+    if v is not None:
+        loss_us["personnel_killed"] = v
+    v = _first_int(t, r"(\d+)[\s\w]*(?:us|american)[\s\w]*(?:troop|soldier|killed|dead)")
+    if v is not None:
+        loss_us["personnel_killed"] = v
+    v = _first_int(t, r"(?:iran|iranian)[\s\w]*(?:say|report)[\s\w]*(\d+)[\s\w]*(?:troop|soldier|killed|dead)")
+    if v is not None:
+        loss_ir["personnel_killed"] = v
+    v = _first_int(t, r"(\d+)[\s\w]*(?:iranian|iran)[\s\w]*(?:troop|soldier|killed|dead)")
+    if v is not None:
+        loss_ir["personnel_killed"] = v
+
+    if loss_us:
+        out.setdefault("combat_losses_delta", {})["us"] = loss_us
+    if loss_ir:
+        out.setdefault("combat_losses_delta", {})["iran"] = loss_ir
+    if "retaliat" in t or "revenge" in t or "报复" in t:
+        out["retaliation"] = {"value": 75, "time": ts}
+    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}
+
+    return out
--- a/crawler/main.py
+++ b/crawler/main.py
@@ -0,0 +1,57 @@
+# -*- coding: utf-8 -*-
+"""爬虫入口：定时抓取 → 解析 → 入库 → 通知 API"""
+import time
+import sys
+from pathlib import Path
+
+# 确保能导入 config
+sys.path.insert(0, str(Path(__file__).resolve().parent))
+
+from config import DB_PATH, API_BASE, CRAWL_INTERVAL
+from scrapers.rss_scraper import fetch_all
+from db_writer import write_updates
+
+
+def notify_api() -> bool:
+    """调用 Node API 触发立即广播"""
+    try:
+        import urllib.request
+        req = urllib.request.Request(
+            f"{API_BASE}/api/crawler/notify",
+            method="POST",
+            headers={"Content-Type": "application/json"},
+        )
+        with urllib.request.urlopen(req, timeout=5) as resp:
+            return resp.status == 200
+    except Exception as e:
+        print(f"  [warn] notify API failed: {e}")
+        return False
+
+
+def run_once() -> int:
+    items = fetch_all()
+    if not items:
+        return 0
+    n = write_updates(items)
+    if n > 0:
+        notify_api()
+    return n
+
+
+def main() -> None:
+    print("Crawler started. DB:", DB_PATH)
+    print("API:", API_BASE, "| Interval:", CRAWL_INTERVAL, "s")
+    while True:
+        try:
+            n = run_once()
+            if n > 0:
+                print(f"[{time.strftime('%H:%M:%S')}] Inserted {n} new update(s)")
+        except KeyboardInterrupt:
+            break
+        except Exception as e:
+            print(f"[{time.strftime('%H:%M:%S')}] Error: {e}")
+        time.sleep(CRAWL_INTERVAL)
+
+
+if __name__ == "__main__":
+    main()
--- a/crawler/panel_schema.py
+++ b/crawler/panel_schema.py
@@ -0,0 +1,42 @@
+# -*- coding: utf-8 -*-
+"""
+前端面板完整数据 schema，与 DB / situationData / useReplaySituation 对齐
+爬虫 + AI 清洗后的数据必须符合此 schema 才能正确更新前端
+"""
+from typing import Any, Dict, List, Literal, Optional, Tuple
+
+# 事件脉络
+SITUATION_UPDATE_CATEGORIES = ("deployment", "alert", "intel", "diplomatic", "other")
+SITUATION_UPDATE_SEVERITIES = ("low", "medium", "high", "critical")
+SUMMARY_MAX_LEN = 120
+
+# 战损
+CombatLossesRow = Dict[str, Any]  # bases_destroyed, bases_damaged, personnel_killed, ...
+
+# 时间序列（回放用）
+TimeSeriesPoint = Tuple[str, int]  # (ISO time, value)
+
+# AI 可从新闻中提取的字段
+EXTRACTABLE_FIELDS = {
+    "situation_update": ["summary", "category", "severity", "timestamp"],
+    "combat_losses": ["personnel_killed", "personnel_wounded", "civilian_killed", "civilian_wounded", "bases_destroyed", "bases_damaged", "aircraft", "warships", "armor", "vehicles"],
+    "retaliation": ["value"],  # 0-100
+    "wall_street_trend": ["time", "value"],  # 0-100
+    "conflict_stats": ["estimated_casualties", "estimated_strike_count"],
+}
+
+
+def validate_category(cat: str) -> str:
+    return cat if cat in SITUATION_UPDATE_CATEGORIES else "other"
+
+
+def validate_severity(sev: str) -> str:
+    return sev if sev in SITUATION_UPDATE_SEVERITIES else "medium"
+
+
+def validate_summary(s: str, max_len: int = SUMMARY_MAX_LEN) -> str:
+    import re
+    if not s or not isinstance(s, str):
+        return ""
+    t = re.sub(r"\s+", " ", str(s).strip())[:max_len]
+    return re.sub(r"[\x00-\x1f]", "", t).rstrip()
--- a/crawler/parser.py
+++ b/crawler/parser.py
@@ -0,0 +1,52 @@
+# -*- coding: utf-8 -*-
+"""新闻分类与严重度判定"""
+import re
+from typing import Literal
+
+Category = Literal["deployment", "alert", "intel", "diplomatic", "other"]
+Severity = Literal["low", "medium", "high", "critical"]
+
+# 分类关键词
+CAT_DEPLOYMENT = ["deploy", "carrier", "航母", "military build", "troop", "forces"]
+CAT_ALERT = ["strike", "attack", "fire", "blast", "hit", "爆炸", "袭击", "打击"]
+CAT_INTEL = ["satellite", "intel", "image", "surveillance", "卫星", "情报"]
+CAT_DIPLOMATIC = ["talk", "negotiation", "diplomat", "sanction", "谈判", "制裁"]
+
+
+def _match(text: str, words: list[str]) -> bool:
+    t = (text or "").lower()
+    for w in words:
+        if w.lower() in t:
+            return True
+    return False
+
+
+def classify(text: str) -> Category:
+    if _match(text, CAT_ALERT):
+        return "alert"
+    if _match(text, CAT_DEPLOYMENT):
+        return "deployment"
+    if _match(text, CAT_INTEL):
+        return "intel"
+    if _match(text, CAT_DIPLOMATIC):
+        return "diplomatic"
+    return "other"
+
+
+def severity(text: str, category: Category) -> Severity:
+    t = (text or "").lower()
+    critical = [
+        "nuclear", "核", "strike", "attack", "killed", "dead", "casualty",
+        "war", "invasion", "袭击", "打击", "死亡",
+    ]
+    high = [
+        "missile", "drone", "bomb", "explosion", "blasted", "fire",
+        "导弹", "无人机", "爆炸", "轰炸",
+    ]
+    if _match(t, critical):
+        return "critical"
+    if _match(t, high) or category == "alert":
+        return "high"
+    if category == "deployment":
+        return "medium"
+    return "low"
--- a/crawler/parser_ai.py
+++ b/crawler/parser_ai.py
@@ -0,0 +1,101 @@
+# -*- coding: utf-8 -*-
+"""
+AI 新闻分类与严重度判定
+优先使用 Ollama 本地模型（免费），失败则回退到规则
+设置 PARSER_AI_DISABLED=1 可只用规则（更快）
+"""
+import os
+from typing import Literal, Optional, Tuple
+
+Category = Literal["deployment", "alert", "intel", "diplomatic", "other"]
+Severity = Literal["low", "medium", "high", "critical"]
+
+PARSER_AI_DISABLED = os.environ.get("PARSER_AI_DISABLED", "0") == "1"
+OLLAMA_MODEL = os.environ.get("OLLAMA_MODEL", "llama3.1")  # 或 qwen2.5:7b
+
+_CATEGORIES = ("deployment", "alert", "intel", "diplomatic", "other")
+_SEVERITIES = ("low", "medium", "high", "critical")
+
+
+def _parse_ai_response(text: str) -> Tuple[Category, Severity]:
+    """从 AI 回复解析 category:severity"""
+    t = (text or "").strip().lower()
+    cat, sev = "other", "low"
+    for c in _CATEGORIES:
+        if c in t:
+            cat = c
+            break
+    for s in _SEVERITIES:
+        if s in t:
+            sev = s
+            break
+    return cat, sev  # type: ignore
+
+
+def _call_ollama(text: str, timeout: int = 5) -> Optional[Tuple[Category, Severity]]:
+    """调用 Ollama 本地模型。需先运行 ollama run llama3.1 或 qwen2.5:7b"""
+    if PARSER_AI_DISABLED:
+        return None
+    try:
+        import requests
+        prompt = f"""Classify this news about US-Iran/middle east (one line only):
+- category: deployment|alert|intel|diplomatic|other
+- severity: low|medium|high|critical
+
+News: {text[:300]}
+
+Reply format: category:severity (e.g. alert:high)"""
+        r = requests.post(
+            "http://localhost:11434/api/chat",
+            json={
+                "model": OLLAMA_MODEL,
+                "messages": [{"role": "user", "content": prompt}],
+                "stream": False,
+                "options": {"num_predict": 32},
+            },
+            timeout=timeout,
+        )
+        if r.status_code != 200:
+            return None
+        out = r.json().get("message", {}).get("content", "")
+        return _parse_ai_response(out)
+    except Exception:
+        return None
+
+
+def _rule_classify(text: str) -> Category:
+    from parser import classify
+    return classify(text)
+
+
+def _rule_severity(text: str, category: Category) -> Severity:
+    from parser import severity
+    return severity(text, category)
+
+
+def classify(text: str) -> Category:
+    """分类。AI 失败时回退规则"""
+    res = _call_ollama(text)
+    if res:
+        return res[0]
+    return _rule_classify(text)
+
+
+def severity(text: str, category: Category) -> Severity:
+    """严重度。AI 失败时回退规则"""
+    res = _call_ollama(text)
+    if res:
+        return res[1]
+    return _rule_severity(text, category)
+
+
+def classify_and_severity(text: str) -> Tuple[Category, Severity]:
+    """一次调用返回分类和严重度（减少 AI 调用）"""
+    if PARSER_AI_DISABLED:
+        from parser import classify, severity
+        c = classify(text)
+        return c, severity(text, c)
+    res = _call_ollama(text)
+    if res:
+        return res
+    return _rule_classify(text), _rule_severity(text, _rule_classify(text))
--- a/crawler/realtime_conflict_service.py
+++ b/crawler/realtime_conflict_service.py
@@ -0,0 +1,367 @@
+# -*- coding: utf-8 -*-
+"""
+GDELT 实时冲突抓取 + API 服务
+核心数据源：GDELT Project，约 15 分钟级更新，含经纬度、事件编码、参与方、事件强度
+"""
+import os
+# 直连外网，避免系统代理导致 ProxyError / 超时（需代理时设置 CRAWLER_USE_PROXY=1）
+if os.environ.get("CRAWLER_USE_PROXY") != "1":
+    os.environ.setdefault("NO_PROXY", "*")
+
+import hashlib
+import sqlite3
+from datetime import datetime
+from pathlib import Path
+from typing import List, Optional
+
+import logging
+import requests
+from fastapi import FastAPI
+from fastapi.middleware.cors import CORSMiddleware
+from apscheduler.schedulers.background import BackgroundScheduler
+
+logging.getLogger("apscheduler.scheduler").setLevel(logging.ERROR)
+app = FastAPI(title="GDELT Conflict Service")
+app.add_middleware(CORSMiddleware, allow_origins=["*"], allow_methods=["*"])
+
+# 配置
+PROJECT_ROOT = Path(__file__).resolve().parent.parent
+DB_PATH = os.environ.get("DB_PATH", str(PROJECT_ROOT / "server" / "data.db"))
+API_BASE = os.environ.get("API_BASE", "http://localhost:3001")
+QUERY = os.environ.get("GDELT_QUERY", "United States Iran military")
+MAX_RECORDS = int(os.environ.get("GDELT_MAX_RECORDS", "30"))
+FETCH_INTERVAL_SEC = int(os.environ.get("FETCH_INTERVAL_SEC", "60"))
+RSS_INTERVAL_SEC = int(os.environ.get("RSS_INTERVAL_SEC", "60"))  # 每分钟抓取世界主流媒体
+# 时间范围：1h=1小时 1d=1天 1week=1周；不设则默认 3 个月（易返回旧文）
+GDELT_TIMESPAN = os.environ.get("GDELT_TIMESPAN", "1d")
+# 设为 1 则跳过 GDELT，仅用 RSS 新闻作为事件脉络（GDELT 国外可能无法访问）
+GDELT_DISABLED = os.environ.get("GDELT_DISABLED", "0") == "1"
+
+# 伊朗攻击源（无经纬度时默认）
+IRAN_COORD = [51.3890, 35.6892]  # Tehran [lng, lat]
+
+# 请求直连，不经过系统代理（避免 ProxyError / 代理超时）
+_REQ_KW = {"timeout": 15, "headers": {"User-Agent": "US-Iran-Dashboard/1.0"}}
+if os.environ.get("CRAWLER_USE_PROXY") != "1":
+    _REQ_KW["proxies"] = {"http": None, "https": None}
+
+EVENT_CACHE: List[dict] = []
+
+
+# ==========================
+# 冲突强度评分 (1–10)
+# ==========================
+def calculate_impact_score(title: str) -> int:
+    score = 1
+    t = (title or "").lower()
+    if "missile" in t:
+        score += 3
+    if "strike" in t:
+        score += 2
+    if "killed" in t or "death" in t or "casualt" in t:
+        score += 4
+    if "troops" in t or "soldier" in t:
+        score += 2
+    if "attack" in t or "attacked" in t:
+        score += 3
+    if "nuclear" in t or "核" in t:
+        score += 4
+    if "explosion" in t or "blast" in t or "bomb" in t:
+        score += 2
+    return min(score, 10)
+
+
+# ==========================
+# 获取 GDELT 实时事件
+# ==========================
+def _parse_article(article: dict) -> Optional[dict]:
+    title_raw = article.get("title") or article.get("seendate") or ""
+    if not title_raw:
+        return None
+    from translate_utils import translate_to_chinese
+    from cleaner_ai import clean_news_for_panel
+    title = translate_to_chinese(str(title_raw)[:500])
+    title = clean_news_for_panel(title, max_len=150)
+    url = article.get("url") or article.get("socialimage") or ""
+    seendate = article.get("seendate") or datetime.utcnow().isoformat()
+    lat = article.get("lat")
+    lng = article.get("lng")
+    # 无经纬度时使用伊朗坐标（攻击源）
+    if lat is None or lng is None:
+        lat, lng = IRAN_COORD[1], IRAN_COORD[0]
+    try:
+        lat, lng = float(lat), float(lng)
+    except (TypeError, ValueError):
+        lat, lng = IRAN_COORD[1], IRAN_COORD[0]
+    impact = calculate_impact_score(title_raw)
+    event_id = hashlib.sha256(f"{url}{seendate}".encode()).hexdigest()[:24]
+    return {
+        "event_id": event_id,
+        "event_time": seendate,
+        "title": title[:500],
+        "lat": lat,
+        "lng": lng,
+        "impact_score": impact,
+        "url": url,
+    }
+
+
+def fetch_gdelt_events() -> None:
+    if GDELT_DISABLED:
+        return
+    url = (
+        "https://api.gdeltproject.org/api/v2/doc/doc"
+        f"?query={QUERY}"
+        "&mode=ArtList"
+        "&format=json"
+        f"&maxrecords={MAX_RECORDS}"
+        f"&timespan={GDELT_TIMESPAN}"
+        "&sort=datedesc"
+    )
+    try:
+        resp = requests.get(url, **_REQ_KW)
+        resp.raise_for_status()
+        data = resp.json()
+        articles = data.get("articles", data) if isinstance(data, dict) else (data if isinstance(data, list) else [])
+        if not isinstance(articles, list):
+            articles = []
+        new_events = []
+        for a in articles:
+            ev = _parse_article(a) if isinstance(a, dict) else None
+            if ev:
+                new_events.append(ev)
+        # 按 event_time 排序，最新在前
+        new_events.sort(key=lambda e: e.get("event_time", ""), reverse=True)
+        global EVENT_CACHE
+        EVENT_CACHE = new_events
+        # 写入 SQLite 并通知 Node
+        _write_to_db(new_events)
+        _notify_node()
+        print(f"[{datetime.now().strftime('%H:%M:%S')}] GDELT 更新 {len(new_events)} 条事件")
+    except Exception:
+        pass
+
+
+def _ensure_table(conn: sqlite3.Connection) -> None:
+    conn.execute("""
+        CREATE TABLE IF NOT EXISTS gdelt_events (
+            event_id TEXT PRIMARY KEY,
+            event_time TEXT NOT NULL,
+            title TEXT NOT NULL,
+            lat REAL NOT NULL,
+            lng REAL NOT NULL,
+            impact_score INTEGER NOT NULL,
+            url TEXT,
+            created_at TEXT DEFAULT (datetime('now'))
+        )
+    """)
+    conn.execute("""
+        CREATE TABLE IF NOT EXISTS conflict_stats (
+            id INTEGER PRIMARY KEY CHECK (id = 1),
+            total_events INTEGER NOT NULL,
+            high_impact_events INTEGER NOT NULL,
+            estimated_casualties INTEGER NOT NULL,
+            estimated_strike_count INTEGER NOT NULL,
+            updated_at TEXT NOT NULL
+        )
+    """)
+    conn.commit()
+
+
+def _write_to_db(events: List[dict]) -> None:
+    if not os.path.exists(DB_PATH):
+        return
+    conn = sqlite3.connect(DB_PATH, timeout=10)
+    try:
+        _ensure_table(conn)
+        for e in events:
+            conn.execute(
+                "INSERT OR REPLACE INTO gdelt_events (event_id, event_time, title, lat, lng, impact_score, url) VALUES (?, ?, ?, ?, ?, ?, ?)",
+                (
+                    e["event_id"],
+                    e.get("event_time", ""),
+                    e.get("title", ""),
+                    e.get("lat", 0),
+                    e.get("lng", 0),
+                    e.get("impact_score", 1),
+                    e.get("url", ""),
+                ),
+            )
+        # 战损统计模型（展示用）
+        high = sum(1 for x in events if x.get("impact_score", 0) >= 7)
+        strikes = sum(1 for x in events if "strike" in (x.get("title") or "").lower() or "attack" in (x.get("title") or "").lower())
+        casualties = min(5000, high * 80 + len(events) * 10)  # 估算
+        conn.execute(
+            "INSERT OR REPLACE INTO conflict_stats (id, total_events, high_impact_events, estimated_casualties, estimated_strike_count, updated_at) VALUES (1, ?, ?, ?, ?, ?)",
+            (len(events), high, casualties, strikes, datetime.utcnow().isoformat()),
+        )
+        conn.execute(
+            "INSERT OR REPLACE INTO situation (id, data, updated_at) VALUES (1, '{}', ?)",
+            (datetime.utcnow().strftime("%Y-%m-%dT%H:%M:%S.000Z"),),
+        )
+        conn.commit()
+    except Exception as e:
+        print(f"写入 DB 失败: {e}")
+        conn.rollback()
+    finally:
+        conn.close()
+
+
+def _notify_node() -> None:
+    try:
+        r = requests.post(f"{API_BASE}/api/crawler/notify", timeout=5, proxies={"http": None, "https": None})
+        if r.status_code != 200:
+            print("  [warn] notify API 失败")
+    except Exception as e:
+        print(f"  [warn] notify API: {e}")
+
+
+# ==========================
+# RSS 新闻抓取（补充 situation_update + AI 提取面板数据）
+# ==========================
+LAST_FETCH = {"items": 0, "inserted": 0, "error": None}
+
+
+def fetch_news() -> None:
+    try:
+        from scrapers.rss_scraper import fetch_all
+        from db_writer import write_updates
+        from translate_utils import translate_to_chinese
+        from cleaner_ai import clean_news_for_panel
+        from cleaner_ai import ensure_category, ensure_severity
+        LAST_FETCH["error"] = None
+        items = fetch_all()
+        for it in items:
+            raw_title = translate_to_chinese(it.get("title", "") or "")
+            raw_summary = translate_to_chinese(it.get("summary", "") or it.get("title", ""))
+            it["title"] = clean_news_for_panel(raw_title, max_len=80)
+            it["summary"] = clean_news_for_panel(raw_summary or raw_title, max_len=120)
+            it["category"] = ensure_category(it.get("category", "other"))
+            it["severity"] = ensure_severity(it.get("severity", "medium"))
+        n = write_updates(items) if items else 0
+        LAST_FETCH["items"] = len(items)
+        LAST_FETCH["inserted"] = n
+        if items:
+            _extract_and_merge_panel_data(items)
+            if n > 0:
+                _notify_node()
+        print(f"[{datetime.now().strftime('%H:%M:%S')}] RSS 抓取 {len(items)} 条，新增入库 {n} 条")
+    except Exception as e:
+        LAST_FETCH["error"] = str(e)
+        print(f"[{datetime.now().strftime('%H:%M:%S')}] 新闻抓取失败: {e}")
+
+
+def _extract_and_merge_panel_data(items: list) -> None:
+    """对新闻做 AI/规则 提取，合并到 combat_losses / retaliation / wall_street_trend 等表"""
+    if not items or not os.path.exists(DB_PATH):
+        return
+    try:
+        from db_merge import merge
+        if os.environ.get("CLEANER_AI_DISABLED", "0") == "1":
+            from extractor_rules import extract_from_news
+        else:
+            from extractor_ai import extract_from_news
+        from datetime import timezone
+        merged_any = False
+        # 只对前几条有足够文本的新闻做提取，避免 Ollama 调用过多
+        for it in items[:5]:
+            text = (it.get("title", "") or "") + " " + (it.get("summary", "") or "")
+            if len(text.strip()) < 20:
+                continue
+            pub = it.get("published")
+            ts = None
+            if pub:
+                try:
+                    if isinstance(pub, str):
+                        pub_dt = datetime.fromisoformat(pub.replace("Z", "+00:00"))
+                    else:
+                        pub_dt = pub
+                    if pub_dt.tzinfo:
+                        pub_dt = pub_dt.astimezone(timezone.utc)
+                    ts = pub_dt.strftime("%Y-%m-%dT%H:%M:%S.000Z")
+                except Exception:
+                    pass
+            extracted = extract_from_news(text, timestamp=ts)
+            if extracted:
+                if merge(extracted, db_path=DB_PATH):
+                    merged_any = True
+        if merged_any:
+            _notify_node()
+    except Exception as e:
+        print(f"  [warn] AI 面板数据提取/合并: {e}")
+
+
+# ==========================
+# 定时任务（RSS 更频繁，优先保证事件脉络实时）
+# ==========================
+scheduler = BackgroundScheduler()
+scheduler.add_job(fetch_news, "interval", seconds=RSS_INTERVAL_SEC, max_instances=2, coalesce=True)
+scheduler.add_job(fetch_gdelt_events, "interval", seconds=FETCH_INTERVAL_SEC, max_instances=2, coalesce=True)
+scheduler.start()
+
+
+# ==========================
+# API 接口
+# ==========================
+@app.get("/crawler/status")
+def crawler_status():
+    """爬虫状态：用于排查数据更新链路"""
+    import os
+    db_ok = os.path.exists(DB_PATH)
+    total = 0
+    if db_ok:
+        try:
+            conn = sqlite3.connect(DB_PATH, timeout=3)
+            total = conn.execute("SELECT COUNT(*) FROM situation_update").fetchone()[0]
+            conn.close()
+        except Exception:
+            pass
+    return {
+        "db_path": DB_PATH,
+        "db_exists": db_ok,
+        "situation_update_count": total,
+        "last_fetch_items": LAST_FETCH.get("items", 0),
+        "last_fetch_inserted": LAST_FETCH.get("inserted", 0),
+        "last_fetch_error": LAST_FETCH.get("error"),
+    }
+
+
+@app.get("/events")
+def get_events():
+    return {
+        "updated_at": datetime.utcnow().isoformat(),
+        "count": len(EVENT_CACHE),
+        "events": EVENT_CACHE,
+        "conflict_stats": _get_conflict_stats(),
+    }
+
+
+def _get_conflict_stats() -> dict:
+    if not os.path.exists(DB_PATH):
+        return {"total_events": 0, "high_impact_events": 0, "estimated_casualties": 0, "estimated_strike_count": 0}
+    try:
+        conn = sqlite3.connect(DB_PATH, timeout=5)
+        row = conn.execute("SELECT total_events, high_impact_events, estimated_casualties, estimated_strike_count FROM conflict_stats WHERE id = 1").fetchone()
+        conn.close()
+        if row:
+            return {
+                "total_events": row[0],
+                "high_impact_events": row[1],
+                "estimated_casualties": row[2],
+                "estimated_strike_count": row[3],
+            }
+    except Exception:
+        pass
+    return {"total_events": 0, "high_impact_events": 0, "estimated_casualties": 0, "estimated_strike_count": 0}
+
+
+@app.on_event("startup")
+def startup():
+    # 新闻优先启动，确保事件脉络有数据
+    fetch_news()
+    fetch_gdelt_events()
+
+
+if __name__ == "__main__":
+    import uvicorn
+    uvicorn.run(app, host="0.0.0.0", port=8000)
--- a/crawler/requirements.txt
+++ b/crawler/requirements.txt
@@ -0,0 +1,6 @@
+requests>=2.31.0
+feedparser>=6.0.0
+fastapi>=0.109.0
+uvicorn>=0.27.0
+apscheduler>=3.10.0
+deep-translator>=1.11.0
--- a/crawler/scrapers/init.py
+++ b/crawler/scrapers/init.py
@@ -0,0 +1 @@
+# -*- coding: utf-8 -*-
--- a/crawler/scrapers/pycache/init.cpython-311.pyc
+++ b/crawler/scrapers/pycache/init.cpython-311.pyc
--- a/crawler/scrapers/pycache/init.cpython-39.pyc
+++ b/crawler/scrapers/pycache/init.cpython-39.pyc
--- a/crawler/scrapers/pycache/rss_scraper.cpython-311.pyc
+++ b/crawler/scrapers/pycache/rss_scraper.cpython-311.pyc
--- a/crawler/scrapers/pycache/rss_scraper.cpython-39.pyc
+++ b/crawler/scrapers/pycache/rss_scraper.cpython-39.pyc
--- a/crawler/scrapers/rss_scraper.py
+++ b/crawler/scrapers/rss_scraper.py
@@ -0,0 +1,76 @@
+# -*- coding: utf-8 -*-
+"""RSS 抓取"""
+import re
+from datetime import datetime, timezone
+
+import feedparser
+
+from config import RSS_FEEDS, KEYWORDS
+from parser_ai import classify_and_severity
+
+
+def _parse_date(entry) -> datetime:
+    for attr in ("published_parsed", "updated_parsed"):
+        val = getattr(entry, attr, None)
+        if val:
+            try:
+                return datetime(*val[:6], tzinfo=timezone.utc)
+            except (TypeError, ValueError):
+                pass
+    return datetime.now(timezone.utc)
+
+
+def _strip_html(s: str) -> str:
+    return re.sub(r"<[^>]+>", "", s) if s else ""
+
+
+def _matches_keywords(text: str) -> bool:
+    t = (text or "").lower()
+    for k in KEYWORDS:
+        if k.lower() in t:
+            return True
+    return False
+
+
+def fetch_all() -> list[dict]:
+    import socket
+    items: list[dict] = []
+    seen: set[str] = set()
+    # 单源超时 10 秒，避免某源卡住
+    old_timeout = socket.getdefaulttimeout()
+    socket.setdefaulttimeout(10)
+    try:
+        for url in RSS_FEEDS:
+            try:
+                feed = feedparser.parse(
+                    url,
+                    request_headers={"User-Agent": "US-Iran-Dashboard/1.0"},
+                    agent="US-Iran-Dashboard/1.0",
+                )
+            except Exception:
+                continue
+            for entry in feed.entries:
+                title = getattr(entry, "title", "") or ""
+                raw_summary = getattr(entry, "summary", "") or getattr(entry, "description", "") or ""
+                summary = _strip_html(raw_summary)
+                link = getattr(entry, "link", "") or ""
+                text = f"{title} {summary}"
+                if not _matches_keywords(text):
+                    continue
+                key = (title[:80], link)
+                if key in seen:
+                    continue
+                seen.add(key)
+                published = _parse_date(entry)
+                cat, sev = classify_and_severity(text)
+                items.append({
+                    "title": title,
+                    "summary": summary[:400] if summary else title,
+                    "url": link,
+                    "published": _parse_date(entry),
+                    "category": cat,
+                    "severity": sev,
+                })
+    finally:
+        socket.setdefaulttimeout(old_timeout)
+    return items
--- a/crawler/translate_utils.py
+++ b/crawler/translate_utils.py
@@ -0,0 +1,38 @@
+# -*- coding: utf-8 -*-
+"""英译中，入库前统一翻译"""
+import os
+import re
+from typing import Optional
+
+
+def _is_mostly_chinese(text: str) -> bool:
+    if not text or len(text.strip()) < 2:
+        return False
+    chinese = len(re.findall(r"[\u4e00-\u9fff]", text))
+    return chinese / max(len(text), 1) > 0.3
+
+
+def translate_to_chinese(text: str) -> str:
+    """将文本翻译成中文，失败或已是中文则返回原文。Google 失败时尝试 MyMemory。"""
+    if not text or not text.strip():
+        return text
+    if os.environ.get("TRANSLATE_DISABLED", "0") == "1":
+        return text
+    s = str(text).strip()
+    if len(s) > 2000:
+        s = s[:2000]
+    if _is_mostly_chinese(s):
+        return text
+    for translator in ["google", "mymemory"]:
+        try:
+            if translator == "google":
+                from deep_translator import GoogleTranslator
+                out = GoogleTranslator(source="auto", target="zh-CN").translate(s)
+            else:
+                from deep_translator import MyMemoryTranslator
+                out = MyMemoryTranslator(source="auto", target="zh-CN").translate(s)
+            if out and out.strip() and out != s:
+                return out
+        except Exception:
+            continue
+    return text
--- a/docs/BACKEND_MODULES.md
+++ b/docs/BACKEND_MODULES.md
@@ -0,0 +1,91 @@
+# 后端模块说明
+
+## 一、现有模块结构
+
+```
+server/
+├── index.js         # HTTP + WebSocket 入口
+├── routes.js        # REST API 路由
+├── db.js            # SQLite schema 与连接
+├── situationData.js # 态势数据聚合 (从 DB 读取)
+├── seed.js          # 初始数据填充
+├── data.db          # SQLite 数据库
+└── package.json
+
+crawler/
+├── realtime_conflict_service.py  # GDELT 实时冲突服务 (核心)
+├── requirements.txt
+├── config.py, db_writer.py       # 旧 RSS 爬虫（可保留）
+├── main.py
+└── README.md
+```
+
+### 1. server/index.js
+- Express + CORS
+- WebSocket (`/ws`)，每 5 秒广播 `situation`
+- `POST /api/crawler/notify`：爬虫写入后触发立即广播
+
+### 2. server/routes.js
+- `GET /api/situation`：完整态势
+- `GET /api/events`：GDELT 事件 + 冲突统计
+- `GET /api/health`：健康检查
+
+### 3. server/db.js
+- 表：`situation`、`force_summary`、`power_index`、`force_asset`、
+  `key_location`、`combat_losses`、`wall_street_trend`、
+  `retaliation_current`、`retaliation_history`、`situation_update`、
+  **`gdelt_events`**、**`conflict_stats`**
+
+---
+
+## 二、GDELT 核心数据源
+
+**GDELT Project**：全球冲突数据库，约 15 分钟级更新，含经纬度、事件编码、参与方、事件强度。
+
+### realtime_conflict_service.py
+
+- 定时（默认 60 秒）从 GDELT API 抓取
+- 冲突强度评分：missile +3, strike +2, killed +4 等
+- 无经纬度时默认攻击源：`IRAN_COORD = [51.3890, 35.6892]`
+- 写入 `gdelt_events`、`conflict_stats`
+- 调用 `POST /api/crawler/notify` 触发 Node 广播
+
+### 冲突强度 → 地图效果
+
+| impact_score | 效果       |
+|--------------|------------|
+| 1–3          | 绿色点     |
+| 4–6          | 橙色闪烁   |
+| 7–10         | 红色脉冲扩散 |
+
+### 战损统计模型（展示用）
+
+- `total_events`
+- `high_impact_events` (impact ≥ 7)
+- `estimated_casualties`
+- `estimated_strike_count`
+
+---
+
+## 三、数据流
+
+```
+GDELT API → Python 服务(60s) → gdelt_events, conflict_stats
+                                    ↓
+              POST /api/crawler/notify → situation.updated_at
+                                    ↓
+              WebSocket 广播 getSituation() → 前端
+```
+
+---
+
+## 四、运行方式
+
+```bash
+# 1. 启动 Node API
+npm run api
+
+# 2. 启动 GDELT 服务
+npm run gdelt
+# 或: cd crawler && uvicorn realtime_conflict_service:app --port 8000
+```
--- a/package-lock.json
+++ b/package-lock.json
@@ -18,6 +18,7 @@
        "react": "^18.3.1",
        "react-dom": "^18.3.1",
        "react-map-gl": "^7.1.7",
+        "react-router-dom": "^7.13.1",
        "ws": "^8.19.0",
        "zustand": "^5.0.0"
      },
@@ -4396,6 +4397,54 @@
        "node": ">=0.10.0"
      }
    },
+    "node_modules/react-router": {
+      "version": "7.13.1",
+      "resolved": "https://registry.npmmirror.com/react-router/-/react-router-7.13.1.tgz",
+      "integrity": "sha512-td+xP4X2/6BJvZoX6xw++A2DdEi++YypA69bJUV5oVvqf6/9/9nNlD70YO1e9d3MyamJEBQFEzk6mbfDYbqrSA==",
+      "dependencies": {
+        "cookie": "^1.0.1",
+        "set-cookie-parser": "^2.6.0"
+      },
+      "engines": {
+        "node": ">=20.0.0"
+      },
+      "peerDependencies": {
+        "react": ">=18",
+        "react-dom": ">=18"
+      },
+      "peerDependenciesMeta": {
+        "react-dom": {
+          "optional": true
+        }
+      }
+    },
+    "node_modules/react-router-dom": {
+      "version": "7.13.1",
+      "resolved": "https://registry.npmmirror.com/react-router-dom/-/react-router-dom-7.13.1.tgz",
+      "integrity": "sha512-UJnV3Rxc5TgUPJt2KJpo1Jpy0OKQr0AjgbZzBFjaPJcFOb2Y8jA5H3LT8HUJAiRLlWrEXWHbF1Z4SCZaQjWDHw==",
+      "dependencies": {
+        "react-router": "7.13.1"
+      },
+      "engines": {
+        "node": ">=20.0.0"
+      },
+      "peerDependencies": {
+        "react": ">=18",
+        "react-dom": ">=18"
+      }
+    },
+    "node_modules/react-router/node_modules/cookie": {
+      "version": "1.1.1",
+      "resolved": "https://registry.npmmirror.com/cookie/-/cookie-1.1.1.tgz",
+      "integrity": "sha512-ei8Aos7ja0weRpFzJnEA9UHJ/7XQmqglbRwnf2ATjcB9Wq874VKH9kfjjirM6UhU2/E5fFYadylyhFldcqSidQ==",
+      "engines": {
+        "node": ">=18"
+      },
+      "funding": {
+        "type": "opencollective",
+        "url": "https://opencollective.com/express"
+      }
+    },
    "node_modules/read-cache": {
      "version": "1.0.0",
      "resolved": "https://registry.npmmirror.com/read-cache/-/read-cache-1.0.0.tgz",
@@ -4645,6 +4694,11 @@
        "node": ">= 0.8.0"
      }
    },
+    "node_modules/set-cookie-parser": {
+      "version": "2.7.2",
+      "resolved": "https://registry.npmmirror.com/set-cookie-parser/-/set-cookie-parser-2.7.2.tgz",
+      "integrity": "sha512-oeM1lpU/UvhTxw+g3cIfxXHyJRc/uidd3yK1P242gzHds0udQBYzs3y8j4gCCW+ZJ7ad0yctld8RYO+bdurlvw=="
+    },
    "node_modules/set-value": {
      "version": "2.0.1",
      "resolved": "https://registry.npmmirror.com/set-value/-/set-value-2.0.1.tgz",
--- a/package.json
+++ b/package.json
@@ -4,9 +4,13 @@
  "version": "1.0.0",
  "type": "module",
  "scripts": {
+    "start": "./start.sh",
    "dev": "vite",
    "api": "node server/index.js",
    "api:seed": "node server/seed.js",
+    "crawler": "cd crawler && python main.py",
+    "gdelt": "cd crawler && uvicorn realtime_conflict_service:app --host 0.0.0.0 --port 8000",
+    "crawler:test": "cd crawler && python3 -c \"import sys; sys.path.insert(0,'.'); from scrapers.rss_scraper import fetch_all; n=len(fetch_all()); print('RSS 抓取:', n, '条' if n else '(0 条，检查网络或关键词过滤)')\"",
    "build": "vite build",
    "typecheck": "tsc --noEmit",
    "lint": "eslint .",
@@ -23,6 +27,7 @@
    "react": "^18.3.1",
    "react-dom": "^18.3.1",
    "react-map-gl": "^7.1.7",
+    "react-router-dom": "^7.13.1",
    "ws": "^8.19.0",
    "zustand": "^5.0.0"
  },
--- a/server/data.db-shm
+++ b/server/data.db-shm
--- a/server/data.db-wal
+++ b/server/data.db-wal
--- a/server/db.js
+++ b/server/db.js
@@ -92,6 +92,26 @@ db.exec(`
    summary TEXT NOT NULL,
    severity TEXT NOT NULL
  );
+
+  CREATE TABLE IF NOT EXISTS gdelt_events (
+    event_id TEXT PRIMARY KEY,
+    event_time TEXT NOT NULL,
+    title TEXT NOT NULL,
+    lat REAL NOT NULL,
+    lng REAL NOT NULL,
+    impact_score INTEGER NOT NULL,
+    url TEXT,
+    created_at TEXT DEFAULT (datetime('now'))
+  );
+
+  CREATE TABLE IF NOT EXISTS conflict_stats (
+    id INTEGER PRIMARY KEY CHECK (id = 1),
+    total_events INTEGER NOT NULL DEFAULT 0,
+    high_impact_events INTEGER NOT NULL DEFAULT 0,
+    estimated_casualties INTEGER NOT NULL DEFAULT 0,
+    estimated_strike_count INTEGER NOT NULL DEFAULT 0,
+    updated_at TEXT NOT NULL
+  );
 `)

 // 迁移：为已有 key_location 表添加 type、region、status、damage_level 列
@@ -103,5 +123,28 @@ try {
  if (!names.includes('status')) db.exec('ALTER TABLE key_location ADD COLUMN status TEXT DEFAULT "operational"')
  if (!names.includes('damage_level')) db.exec('ALTER TABLE key_location ADD COLUMN damage_level INTEGER')
 } catch (_) {}
+// 迁移：combat_losses 添加平民伤亡、updated_at
+try {
+  const lossCols = db.prepare('PRAGMA table_info(combat_losses)').all()
+  const lossNames = lossCols.map((c) => c.name)
+  if (!lossNames.includes('civilian_killed')) db.exec('ALTER TABLE combat_losses ADD COLUMN civilian_killed INTEGER NOT NULL DEFAULT 0')
+  if (!lossNames.includes('civilian_wounded')) db.exec('ALTER TABLE combat_losses ADD COLUMN civilian_wounded INTEGER NOT NULL DEFAULT 0')
+  if (!lossNames.includes('updated_at')) db.exec('ALTER TABLE combat_losses ADD COLUMN updated_at TEXT DEFAULT (datetime("now"))')
+} catch (_) {}
+
+// 迁移：所有表添加 updated_at 用于数据回放
+const addUpdatedAt = (table) => {
+  try {
+    const cols = db.prepare(`PRAGMA table_info(${table})`).all()
+    if (!cols.some((c) => c.name === 'updated_at')) {
+      db.exec(`ALTER TABLE ${table} ADD COLUMN updated_at TEXT DEFAULT (datetime("now"))`)
+    }
+  } catch (_) {}
+}
+addUpdatedAt('force_summary')
+addUpdatedAt('power_index')
+addUpdatedAt('force_asset')
+addUpdatedAt('key_location')
+addUpdatedAt('retaliation_current')

 module.exports = db
--- a/server/index.js
+++ b/server/index.js
@@ -12,6 +12,10 @@ app.use(cors())
 app.use(express.json())
 app.use('/api', routes)
 app.get('/api/health', (_, res) => res.json({ ok: true }))
+app.post('/api/crawler/notify', (_, res) => {
+  notifyCrawlerUpdate()
+  res.json({ ok: true })
+})

 const server = http.createServer(app)

@@ -27,7 +31,16 @@ function broadcastSituation() {
    })
  } catch (_) {}
 }
-setInterval(broadcastSituation, 5000)
+setInterval(broadcastSituation, 3000)
+
+// 供爬虫调用：更新 situation.updated_at 并立即广播
+function notifyCrawlerUpdate() {
+  try {
+    const db = require('./db')
+    db.prepare("INSERT OR REPLACE INTO situation (id, data, updated_at) VALUES (1, '{}', ?)").run(new Date().toISOString())
+    broadcastSituation()
+  } catch (_) {}
+}

 server.listen(PORT, () => {
  console.log(`API + WebSocket running at http://localhost:${PORT}`)
--- a/server/routes.js
+++ b/server/routes.js
@@ -1,8 +1,58 @@
 const express = require('express')
 const { getSituation } = require('./situationData')
+const db = require('./db')

 const router = express.Router()

+// 数据库 Dashboard：返回各表原始数据
+router.get('/db/dashboard', (req, res) => {
+  try {
+    const tables = [
+      'situation',
+      'force_summary',
+      'power_index',
+      'force_asset',
+      'key_location',
+      'combat_losses',
+      'wall_street_trend',
+      'retaliation_current',
+      'retaliation_history',
+      'situation_update',
+      'gdelt_events',
+      'conflict_stats',
+    ]
+    const data = {}
+    const timeSort = {
+      situation: 'updated_at DESC',
+      situation_update: 'timestamp DESC',
+      gdelt_events: 'event_time DESC',
+      wall_street_trend: 'time DESC',
+      retaliation_history: 'time DESC',
+      conflict_stats: 'updated_at DESC',
+    }
+    for (const name of tables) {
+      try {
+        const order = timeSort[name]
+        let rows
+        try {
+          rows = order
+            ? db.prepare(`SELECT * FROM ${name} ORDER BY ${order}`).all()
+            : db.prepare(`SELECT * FROM ${name}`).all()
+        } catch (qerr) {
+          rows = db.prepare(`SELECT * FROM ${name}`).all()
+        }
+        data[name] = rows
+      } catch (e) {
+        data[name] = { error: e.message }
+      }
+    }
+    res.json(data)
+  } catch (err) {
+    console.error(err)
+    res.status(500).json({ error: err.message })
+  }
+})
+
 router.get('/situation', (req, res) => {
  try {
    res.json(getSituation())
@@ -12,4 +62,19 @@ router.get('/situation', (req, res) => {
  }
 })

+router.get('/events', (req, res) => {
+  try {
+    const s = getSituation()
+    res.json({
+      updated_at: s.lastUpdated,
+      count: (s.conflictEvents || []).length,
+      events: s.conflictEvents || [],
+      conflict_stats: s.conflictStats || {},
+    })
+  } catch (err) {
+    console.error(err)
+    res.status(500).json({ error: err.message })
+  }
+})
+
 module.exports = router
--- a/server/seed.js
+++ b/server/seed.js
@@ -140,11 +140,19 @@ function seed() {
  ]
  iranLocs.forEach((r) => insertLoc.run(...r))

-  db.exec(`
-    INSERT OR REPLACE INTO combat_losses (side, bases_destroyed, bases_damaged, personnel_killed, personnel_wounded, aircraft, warships, armor, vehicles) VALUES
-    ('us', 0, 27, 127, 384, 2, 0, 0, 8),
-    ('iran', 3, 8, 2847, 5620, 24, 12, 18, 42);
-  `)
+  try {
+    db.exec(`
+      INSERT OR REPLACE INTO combat_losses (side, bases_destroyed, bases_damaged, personnel_killed, personnel_wounded, civilian_killed, civilian_wounded, aircraft, warships, armor, vehicles) VALUES
+      ('us', 0, 27, 127, 384, 18, 52, 2, 0, 0, 8),
+      ('iran', 3, 8, 2847, 5620, 412, 1203, 24, 12, 18, 42);
+    `)
+  } catch (_) {
+    db.exec(`
+      INSERT OR REPLACE INTO combat_losses (side, bases_destroyed, bases_damaged, personnel_killed, personnel_wounded, aircraft, warships, armor, vehicles) VALUES
+      ('us', 0, 27, 127, 384, 2, 0, 0, 8),
+      ('iran', 3, 8, 2847, 5620, 24, 12, 18, 42);
+    `)
+  }

  db.exec('DELETE FROM wall_street_trend')
  const trendRows = [['2025-03-01T00:00:00', 82], ['2025-03-01T03:00:00', 85], ['2025-03-01T06:00:00', 88], ['2025-03-01T09:00:00', 90], ['2025-03-01T12:00:00', 92], ['2025-03-01T15:00:00', 94], ['2025-03-01T18:00:00', 95], ['2025-03-01T21:00:00', 96], ['2025-03-01T23:00:00', 98]]
--- a/server/situationData.js
+++ b/server/situationData.js
@@ -15,6 +15,7 @@ function toLosses(row) {
  return {
    bases: { destroyed: row.bases_destroyed, damaged: row.bases_damaged },
    personnelCasualties: { killed: row.personnel_killed, wounded: row.personnel_wounded },
+    civilianCasualties: { killed: row.civilian_killed ?? 0, wounded: row.civilian_wounded ?? 0 },
    aircraft: row.aircraft,
    warships: row.warships,
    armor: row.armor,
@@ -25,6 +26,7 @@ function toLosses(row) {
 const defaultLosses = {
  bases: { destroyed: 0, damaged: 0 },
  personnelCasualties: { killed: 0, wounded: 0 },
+  civilianCasualties: { killed: 0, wounded: 0 },
  aircraft: 0,
  warships: 0,
  armor: 0,
@@ -45,9 +47,35 @@ function getSituation() {
  const trend = db.prepare('SELECT time, value FROM wall_street_trend ORDER BY time').all()
  const retaliationCur = db.prepare('SELECT value FROM retaliation_current WHERE id = 1').get()
  const retaliationHist = db.prepare('SELECT time, value FROM retaliation_history ORDER BY time').all()
-  const updates = db.prepare('SELECT * FROM situation_update ORDER BY timestamp DESC').all()
+  const updates = db.prepare('SELECT * FROM situation_update ORDER BY timestamp DESC LIMIT 50').all()
  const meta = db.prepare('SELECT updated_at FROM situation WHERE id = 1').get()

+  let conflictEvents = []
+  let conflictStats = { total_events: 0, high_impact_events: 0, estimated_casualties: 0, estimated_strike_count: 0 }
+  try {
+    conflictEvents = db.prepare('SELECT event_id, event_time, title, lat, lng, impact_score, url FROM gdelt_events ORDER BY event_time DESC LIMIT 30').all()
+    const statsRow = db.prepare('SELECT total_events, high_impact_events, estimated_casualties, estimated_strike_count FROM conflict_stats WHERE id = 1').get()
+    if (statsRow) conflictStats = statsRow
+  } catch (_) {}
+
+  // 平民伤亡：合计显示，不区分阵营
+  const civUsK = lossesUs?.civilian_killed ?? 0
+  const civUsW = lossesUs?.civilian_wounded ?? 0
+  const civIrK = lossesIr?.civilian_killed ?? 0
+  const civIrW = lossesIr?.civilian_wounded ?? 0
+  const dbKilled = civUsK + civIrK
+  const dbWounded = civUsW + civIrW
+  const est = conflictStats.estimated_casualties || 0
+  const civilianCasualtiesTotal = {
+    killed: est > 0 ? Math.max(dbKilled, est) : dbKilled,
+    wounded: dbWounded,
+  }
+
+  const usLossesBase = lossesUs ? toLosses(lossesUs) : defaultLosses
+  const irLossesBase = lossesIr ? toLosses(lossesIr) : defaultLosses
+  const usLosses = { ...usLossesBase, civilianCasualties: { killed: 0, wounded: 0 } }
+  const irLosses = { ...irLossesBase, civilianCasualties: { killed: 0, wounded: 0 } }
+
  return {
    lastUpdated: meta?.updated_at || new Date().toISOString(),
    usForces: {
@@ -69,7 +97,7 @@ function getSituation() {
      },
      assets: (assetsUs || []).map(toAsset),
      keyLocations: locUs || [],
-      combatLosses: lossesUs ? toLosses(lossesUs) : defaultLosses,
+      combatLosses: usLosses,
      wallStreetInvestmentTrend: trend || [],
    },
    iranForces: {
@@ -91,7 +119,7 @@ function getSituation() {
      },
      assets: (assetsIr || []).map(toAsset),
      keyLocations: locIr || [],
-      combatLosses: lossesIr ? toLosses(lossesIr) : defaultLosses,
+      combatLosses: irLosses,
      retaliationSentiment: retaliationCur?.value ?? 0,
      retaliationSentimentHistory: retaliationHist || [],
    },
@@ -102,6 +130,17 @@ function getSituation() {
      summary: u.summary,
      severity: u.severity,
    })),
+    conflictEvents: conflictEvents.map((e) => ({
+      event_id: e.event_id,
+      event_time: e.event_time,
+      title: e.title,
+      lat: e.lat,
+      lng: e.lng,
+      impact_score: e.impact_score,
+      url: e.url,
+    })),
+    conflictStats,
+    civilianCasualtiesTotal,
  }
 }

--- a/src/App.tsx
+++ b/src/App.tsx
@@ -1,4 +1,6 @@
+import { Routes, Route } from 'react-router-dom'
 import { Dashboard } from '@/pages/Dashboard'
+import { DbDashboard } from '@/pages/DbDashboard'

 function App() {
  return (
@@ -6,7 +8,10 @@ function App() {
      className="min-h-screen w-full bg-military-dark overflow-hidden"
      style={{ background: '#0A0F1C' }}
    >
-      <Dashboard />
+      <Routes>
+        <Route path="/" element={<Dashboard />} />
+        <Route path="/db" element={<DbDashboard />} />
+      </Routes>
    </div>
  )
 }
--- a/src/api/situation.ts
+++ b/src/api/situation.ts
@@ -1,7 +1,7 @@
 import type { MilitarySituation } from '@/data/mockData'

 export async function fetchSituation(): Promise<MilitarySituation> {
-  const res = await fetch('/api/situation')
+  const res = await fetch(`/api/situation?t=${Date.now()}`, { cache: 'no-store' })
  if (!res.ok) throw new Error(`API error: ${res.status}`)
  return res.json()
 }
--- a/src/components/CombatLossesPanel.tsx
+++ b/src/components/CombatLossesPanel.tsx
@@ -1,6 +1,5 @@
 import {
  Building2,
-  Users,
  Skull,
  Bandage,
  Plane,
@@ -8,108 +7,110 @@ import {
  Shield,
  Car,
  TrendingDown,
+  UserCircle,
+  Activity,
 } from 'lucide-react'
 import { formatMillions } from '@/utils/formatNumber'
-import type { CombatLosses } from '@/data/mockData'
+import type { CombatLosses, ConflictStats } from '@/data/mockData'

 interface CombatLossesPanelProps {
  usLosses: CombatLosses
  iranLosses: CombatLosses
+  conflictStats?: ConflictStats | null
+  /** 平民伤亡合计（不区分阵营） */
+  civilianTotal?: { killed: number; wounded: number }
  className?: string
 }

-const LOSS_ITEMS: {
-  key: keyof Omit<CombatLosses, 'bases' | 'personnelCasualties'>
-  label: string
-  icon: typeof Plane
-  iconColor: string
-}[] = [
-  { key: 'aircraft', label: '战机', icon: Plane, iconColor: 'text-sky-400' },
-  { key: 'warships', label: '战舰', icon: Ship, iconColor: 'text-blue-500' },
-  { key: 'armor', label: '装甲', icon: Shield, iconColor: 'text-emerald-500' },
-  { key: 'vehicles', label: '车辆', icon: Car, iconColor: 'text-slate-400' },
-]
+export function CombatLossesPanel({ usLosses, iranLosses, conflictStats, civilianTotal, className = '' }: CombatLossesPanelProps) {
+  const civ = civilianTotal ?? { killed: 0, wounded: 0 }
+
+  const otherRows = [
+    { label: '基地', icon: Building2, iconColor: 'text-amber-500', us: `${usLosses.bases.destroyed}/${usLosses.bases.damaged}`, ir: `${iranLosses.bases.destroyed}/${iranLosses.bases.damaged}` },
+    { label: '战机', icon: Plane, iconColor: 'text-sky-400', us: usLosses.aircraft, ir: iranLosses.aircraft },
+    { label: '战舰', icon: Ship, iconColor: 'text-blue-500', us: usLosses.warships, ir: iranLosses.warships },
+    { label: '装甲', icon: Shield, iconColor: 'text-emerald-500', us: usLosses.armor, ir: iranLosses.armor },
+    { label: '车辆', icon: Car, iconColor: 'text-slate-400', us: usLosses.vehicles, ir: iranLosses.vehicles },
+  ]

-export function CombatLossesPanel({ usLosses, iranLosses, className = '' }: CombatLossesPanelProps) {
  return (
-    <div
-      className={`
-        min-w-0 shrink-0 overflow-x-auto overflow-y-hidden border-t border-military-border scrollbar-thin bg-military-panel/95 px-4 py-2 font-orbitron
-        ${className}
-      `}
-    >
-      <div className="mb-2 flex items-center gap-1 text-[10px] uppercase tracking-wider text-military-text-secondary">
+    <div className={`flex min-h-[220px] max-h-[240px] min-w-0 flex-1 flex-col overflow-hidden rounded border border-military-border bg-military-panel/95 font-orbitron ${className}`}>
+      <div className="mb-1.5 flex shrink-0 items-center justify-center gap-2 text-[10px] uppercase tracking-wider text-military-text-secondary">
        <TrendingDown className="h-2.5 w-2.5 shrink-0 text-amber-400" />
-        战损数据
+        战损
+        {conflictStats && conflictStats.total_events > 0 && (
+          <span className="flex items-center gap-0.5 rounded bg-cyan-950/50 px-1 py-0.5 text-[9px] text-cyan-400">
+            <Activity className="h-2 w-2" />
+            {conflictStats.total_events}
+          </span>
+        )}
      </div>
-      <div className="flex min-w-0 flex-wrap gap-x-4 gap-y-3 overflow-x-auto text-xs">
-        {/* 基地 */}
-        <div className="flex shrink-0 min-w-0 flex-col gap-0.5 overflow-visible">
-          <span className="flex shrink-0 items-center gap-1 text-military-text-secondary">
-            <Building2 className="h-3 w-3 shrink-0 text-amber-500" />
-            基地
-          </span>
-          <div className="flex flex-col gap-0.5 tabular-nums">
-            <div className="flex min-w-0 items-baseline gap-1 overflow-hidden" title={`美 毁${usLosses.bases.destroyed} 损${usLosses.bases.damaged}`}>
-              <span className="shrink-0 text-military-us">美</span>
-              <span className="truncate">
-                毁<strong className="text-amber-400">{usLosses.bases.destroyed}</strong>
-                损<strong className="text-amber-300">{usLosses.bases.damaged}</strong>
-              </span>
+
+      <div className="flex min-h-0 flex-1 flex-col gap-2 overflow-hidden px-3 pb-2">
+        {/* 人员伤亡 - 单独容器 */}
+        <div className="flex shrink-0 flex-col justify-center overflow-hidden rounded-lg border border-red-900/50 bg-red-950/40 px-3 py-2">
+          <div className="mb-1 flex shrink-0 items-center justify-center gap-3 text-[9px] text-military-text-secondary">
+            <span className="flex items-center gap-0.5"><Skull className="h-3 w-3 text-red-500" /> 阵亡</span>
+            <span className="flex items-center gap-0.5"><Bandage className="h-3 w-3 text-amber-500" /> 受伤</span>
+            <span className="text-military-text-secondary/70">|</span>
+            <span>美 : 伊</span>
+          </div>
+          <div className="grid min-w-0 grid-cols-2 gap-x-2 gap-y-1 overflow-hidden text-center tabular-nums sm:gap-x-4">
+            <div className="min-w-0 truncate text-military-us" title={`美: ${formatMillions(usLosses.personnelCasualties.killed)} / ${formatMillions(usLosses.personnelCasualties.wounded)}`}>
+              <span className="text-base font-bold text-red-500">{formatMillions(usLosses.personnelCasualties.killed)}</span>
+              <span className="mx-0.5 text-military-text-secondary">/</span>
+              <span className="text-base font-semibold text-amber-500">{formatMillions(usLosses.personnelCasualties.wounded)}</span>
            </div>
-            <div className="flex min-w-0 items-baseline gap-1 overflow-hidden" title={`伊 毁${iranLosses.bases.destroyed} 损${iranLosses.bases.damaged}`}>
-              <span className="shrink-0 text-military-iran">伊</span>
-              <span className="truncate">
-                毁<strong className="text-amber-400">{iranLosses.bases.destroyed}</strong>
-                损<strong className="text-amber-300">{iranLosses.bases.damaged}</strong>
-              </span>
+            <div className="min-w-0 truncate text-military-iran" title={`伊: ${formatMillions(iranLosses.personnelCasualties.killed)} / ${formatMillions(iranLosses.personnelCasualties.wounded)}`}>
+              <span className="text-base font-bold text-red-500">{formatMillions(iranLosses.personnelCasualties.killed)}</span>
+              <span className="mx-0.5 text-military-text-secondary">/</span>
+              <span className="text-base font-semibold text-amber-500">{formatMillions(iranLosses.personnelCasualties.wounded)}</span>
            </div>
          </div>
        </div>

-        {/* 人员伤亡 */}
-        <div className="flex shrink-0 min-w-0 flex-col gap-0.5 overflow-visible">
-          <span className="flex shrink-0 items-center gap-1 text-military-text-secondary">
-            <Users className="h-3 w-3 shrink-0 text-slate-400" />
-            人员伤亡
-          </span>
-          <div className="flex flex-col gap-0.5 tabular-nums">
-            <div className="flex min-w-0 flex-wrap items-baseline gap-x-1" title={`美 阵亡${formatMillions(usLosses.personnelCasualties.killed)} 受伤${formatMillions(usLosses.personnelCasualties.wounded)}`}>
-              <span className="shrink-0 text-military-us">美</span>
-              <Skull className="h-2.5 w-2.5 shrink-0 text-red-500" />
-              <strong className="text-red-500">{formatMillions(usLosses.personnelCasualties.killed)}</strong>
-              <Bandage className="h-2.5 w-2.5 shrink-0 text-amber-500" />
-              <strong className="text-amber-500">{formatMillions(usLosses.personnelCasualties.wounded)}</strong>
-            </div>
-            <div className="flex min-w-0 flex-wrap items-baseline gap-x-1" title={`伊 阵亡${formatMillions(iranLosses.personnelCasualties.killed)} 受伤${formatMillions(iranLosses.personnelCasualties.wounded)}`}>
-              <span className="shrink-0 text-military-iran">伊</span>
-              <Skull className="h-2.5 w-2.5 shrink-0 text-red-500" />
-              <strong className="text-red-500">{formatMillions(iranLosses.personnelCasualties.killed)}</strong>
-              <Bandage className="h-2.5 w-2.5 shrink-0 text-amber-500" />
-              <strong className="text-amber-500">{formatMillions(iranLosses.personnelCasualties.wounded)}</strong>
-            </div>
+        {/* 平民伤亡：合计显示，不区分阵营 */}
+        <div className="flex shrink-0 flex-col justify-center overflow-hidden rounded-lg border border-amber-900/50 bg-amber-950/40 px-3 py-2">
+          <div className="mb-1 flex shrink-0 items-center justify-center gap-2 text-[9px] text-military-text-secondary">
+            <UserCircle className="h-3 w-3 text-amber-400" />
+            平民伤亡（合计）
          </div>
-        </div>
-
-        {/* 战机 / 战舰 / 装甲 / 车辆 */}
-        {LOSS_ITEMS.map(({ key, label, icon: Icon, iconColor }) => (
-          <div key={key} className="flex shrink-0 min-w-0 flex-col gap-0.5 overflow-visible">
-            <span className="flex shrink-0 items-center gap-1 text-military-text-secondary">
-              <Icon className={`h-3 w-3 shrink-0 ${iconColor}`} />
-              {label}
+          <div className="flex items-center justify-center gap-3 text-center tabular-nums">
+            <span className="flex items-center gap-0.5">
+              <Skull className="h-3 w-3 text-red-500" />
+              <span className="text-base font-bold text-red-500">{formatMillions(civ.killed)}</span>
+            </span>
+            <span className="text-military-text-secondary/60">/</span>
+            <span className="flex items-center gap-0.5">
+              <Bandage className="h-3 w-3 text-amber-500" />
+              <span className="text-base font-semibold text-amber-500">{formatMillions(civ.wounded)}</span>
            </span>
-            <div className="flex flex-col gap-0.5 tabular-nums">
-              <div className="flex min-w-0 items-baseline gap-1">
-                <span className="shrink-0 text-military-us">美</span>
-                <strong>{usLosses[key]}</strong>
-              </div>
-              <div className="flex min-w-0 items-baseline gap-1">
-                <span className="shrink-0 text-military-iran">伊</span>
-                <strong>{iranLosses[key]}</strong>
-              </div>
-            </div>
          </div>
-        ))}
+        </div>
+
+        {/* 其它 - 标签+图标+数字，单独容器 */}
+        <div className="min-h-0 min-w-0 flex-1 overflow-hidden rounded border border-military-border/50 bg-military-dark/30 px-2 py-1.5">
+          <div className="mb-1 text-[8px] text-military-text-secondary">美:伊</div>
+          <div className="grid grid-cols-2 gap-x-2 gap-y-0.5 overflow-hidden text-[11px] tabular-nums lg:grid-cols-3">
+            {otherRows.map(({ label, icon: Icon, iconColor, ...rest }, i) => (
+              <div key={i} className="flex min-w-0 items-center justify-between gap-1 overflow-hidden">
+                <span className="flex shrink-0 items-center gap-0.5 text-military-text-primary">
+                  <Icon className={`h-3 w-3 ${iconColor}`} />
+                  {label}
+                </span>
+                {'value' in rest ? (
+                  <span className="min-w-0 truncate text-right text-amber-400">{rest.value}</span>
+                ) : (
+                  <span className="min-w-0 truncate text-right">
+                    <span className="text-military-us">{rest.us}</span>
+                    <span className="text-military-text-secondary/60">:</span>
+                    <span className="text-military-iran">{rest.ir}</span>
+                  </span>
+                )}
+              </div>
+            ))}
+          </div>
+        </div>
      </div>
    </div>
  )
--- a/src/components/EventTimelinePanel.tsx
+++ b/src/components/EventTimelinePanel.tsx
@@ -0,0 +1,105 @@
+import * as React from 'react'
+import type { SituationUpdate, ConflictEvent } from '@/data/mockData'
+import { History, RefreshCw } from 'lucide-react'
+import { fetchAndSetSituation } from '@/store/situationStore'
+
+interface EventTimelinePanelProps {
+  updates: SituationUpdate[]
+  conflictEvents?: ConflictEvent[]
+  className?: string
+}
+
+const CAT_LABELS: Record<string, string> = {
+  deployment: '部署',
+  alert: '警报',
+  intel: '情报',
+  diplomatic: '外交',
+  other: '其他',
+}
+function formatTime(iso: string): string {
+  const d = new Date(iso)
+  return d.toLocaleString('zh-CN', { month: '2-digit', day: '2-digit', hour: '2-digit', minute: '2-digit', hour12: false })
+}
+
+type TimelineItem = {
+  id: string
+  summary: string
+  timestamp: string
+  source: 'gdelt' | 'rss'
+  category?: string
+  severity?: string
+}
+
+export function EventTimelinePanel({ updates = [], conflictEvents = [], className = '' }: EventTimelinePanelProps) {
+  const [refreshing, setRefreshing] = React.useState(false)
+  const handleRefresh = React.useCallback(async () => {
+    setRefreshing(true)
+    await fetchAndSetSituation().finally(() => setRefreshing(false))
+  }, [])
+  // 合并 GDELT + RSS，按时间倒序（最新在前）
+  const merged: TimelineItem[] = [
+    ...(conflictEvents || []).map((e) => ({
+      id: e.event_id,
+      summary: e.title,
+      timestamp: e.event_time,
+      source: 'gdelt' as const,
+      category: 'alert',
+      severity: e.impact_score >= 7 ? 'high' : e.impact_score >= 4 ? 'medium' : 'low',
+    })),
+    ...(updates || []).map((u) => ({
+      id: u.id,
+      summary: u.summary,
+      timestamp: u.timestamp,
+      source: 'rss' as const,
+      category: u.category,
+      severity: u.severity,
+    })),
+  ]
+    .sort((a, b) => new Date(b.timestamp).getTime() - new Date(a.timestamp).getTime())
+    .slice(0, 8)
+
+  return (
+    <div className={`flex min-w-0 max-w-[280px] shrink flex-col overflow-hidden rounded border border-military-border bg-military-panel/80 ${className}`}>
+      <div className="flex shrink-0 items-center justify-between border-b border-military-border px-3 py-1.5">
+        <span className="flex items-center gap-1.5 text-[10px] font-semibold uppercase tracking-wider text-military-text-secondary">
+          <History className="h-3.5 w-3.5 shrink-0 text-military-accent" />
+          事件脉络
+        </span>
+        <button
+          type="button"
+          onClick={handleRefresh}
+          disabled={refreshing}
+          className="flex items-center gap-1 rounded p-0.5 text-[8px] text-military-text-secondary/80 hover:bg-military-border/30 hover:text-cyan-400 disabled:opacity-50"
+          title="刷新"
+        >
+          <RefreshCw className={`h-3 w-3 ${refreshing ? 'animate-spin' : ''}`} />
+          刷新
+        </button>
+      </div>
+      <div className="min-h-0 max-h-[140px] flex-1 overflow-y-auto overflow-x-hidden px-2 py-1">
+        {merged.length === 0 ? (
+          <p className="py-4 text-center text-[11px] text-military-text-secondary">暂无事件</p>
+        ) : (
+          <ul className="space-y-2">
+            {merged.map((item) => (
+              <li key={item.id} className="flex gap-2 border-b border-military-border/50 pb-2 last:border-0 last:pb-0">
+                <span className="shrink-0 pt-0.5">
+                  <span className={`inline-block h-1.5 w-1.5 rounded-full ${item.source === 'gdelt' ? 'bg-cyan-500' : 'bg-amber-500'}`} />
+                </span>
+                <div className="min-w-0 flex-1 text-right">
+                  <p className="text-[11px] leading-tight text-military-text-primary line-clamp-2">{item.summary}</p>
+                  <span className="mt-0.5 flex items-center justify-end gap-1 text-[9px] text-military-text-secondary">
+                    {formatTime(item.timestamp)}
+                    <span className="text-military-text-secondary/60">
+                      {item.source === 'gdelt' ? 'GDELT' : CAT_LABELS[item.category ?? ''] ?? '新闻'}
+                    </span>
+                  </span>
+                </div>
+              </li>
+            ))}
+          </ul>
+        )}
+      </div>
+    </div>
+  )
+}
--- a/src/components/HeaderPanel.tsx
+++ b/src/components/HeaderPanel.tsx
@@ -1,10 +1,15 @@
 import { useState, useEffect } from 'react'
+import { Link } from 'react-router-dom'
 import { StatCard } from './StatCard'
 import { useSituationStore } from '@/store/situationStore'
-import { Wifi, WifiOff, Clock } from 'lucide-react'
+import { useReplaySituation } from '@/hooks/useReplaySituation'
+import { usePlaybackStore } from '@/store/playbackStore'
+import { Wifi, WifiOff, Clock, Database } from 'lucide-react'

 export function HeaderPanel() {
-  const { situation, isConnected } = useSituationStore()
+  const situation = useReplaySituation()
+  const isConnected = useSituationStore((s) => s.isConnected)
+  const isReplayMode = usePlaybackStore((s) => s.isReplayMode)
  const { usForces, iranForces } = situation
  const [now, setNow] = useState(() => new Date())

@@ -47,14 +52,21 @@ export function HeaderPanel() {
            <Clock className="h-4 w-4 shrink-0" />
            <span className="min-w-[11rem] tabular-nums">{formatDateTime(now)}</span>
          </div>
-          {isConnected && (
-            <span className="text-[10px] text-green-500/90">
-              {formatDataTime(situation.lastUpdated)} (实时更新)
+          {(isConnected || isReplayMode) && (
+            <span className={`text-[10px] ${isReplayMode ? 'text-military-accent' : 'text-green-500/90'}`}>
+              {formatDataTime(situation.lastUpdated)} {isReplayMode ? '(回放)' : '(实时更新)'}
            </span>
          )}
        </div>
      </div>
-      <div className="flex shrink-0 items-center gap-2">
+      <div className="flex shrink-0 items-center gap-3">
+        <Link
+          to="/db"
+          className="flex items-center gap-1 rounded border border-military-border px-2 py-1 text-[10px] text-military-text-secondary hover:bg-military-border/30 hover:text-cyan-400"
+        >
+          <Database className="h-3 w-3" />
+          数据库
+        </Link>
        {isConnected ? (
          <>
            <Wifi className="h-3.5 w-3.5 text-green-500" />
--- a/src/components/RecentUpdatesPanel.tsx
+++ b/src/components/RecentUpdatesPanel.tsx
@@ -0,0 +1,71 @@
+import type { SituationUpdate, ConflictEvent } from '@/data/mockData'
+import { Newspaper, AlertTriangle } from 'lucide-react'
+
+interface RecentUpdatesPanelProps {
+  updates: SituationUpdate[]
+  conflictEvents?: ConflictEvent[]
+  className?: string
+}
+
+const CAT_LABELS: Record<string, string> = {
+  deployment: '部署',
+  alert: '警报',
+  intel: '情报',
+  diplomatic: '外交',
+  other: '其他',
+}
+const SEV_COLORS: Record<string, string> = {
+  low: 'text-military-text-secondary',
+  medium: 'text-amber-400',
+  high: 'text-orange-500',
+  critical: 'text-red-500',
+}
+
+function formatTime(iso: string): string {
+  const d = new Date(iso)
+  const now = new Date()
+  const diff = now.getTime() - d.getTime()
+  if (diff < 60000) return '刚刚'
+  if (diff < 3600000) return `${Math.floor(diff / 60000)}分钟前`
+  if (diff < 86400000) return `${Math.floor(diff / 3600000)}小时前`
+  return d.toLocaleString('zh-CN', { month: '2-digit', day: '2-digit', hour: '2-digit', minute: '2-digit' })
+}
+
+export function RecentUpdatesPanel({ updates, conflictEvents = [], className = '' }: RecentUpdatesPanelProps) {
+  // 优先展示 GDELT 冲突事件（最新 10 条），无则用 updates
+  const fromConflict = (conflictEvents || [])
+    .slice(0, 10)
+    .map((e) => ({ id: e.event_id, summary: e.title, timestamp: e.event_time, category: 'alert' as const, severity: (e.impact_score >= 7 ? 'high' : e.impact_score >= 4 ? 'medium' : 'low') as const }))
+  const list = fromConflict.length > 0 ? fromConflict : (updates || []).slice(0, 8)
+
+  return (
+    <div className={`flex min-w-0 flex-1 flex-col overflow-hidden rounded border border-military-border bg-military-panel/80 ${className}`}>
+      <div className="flex shrink-0 items-center gap-1.5 border-b border-military-border px-3 py-1.5">
+        <Newspaper className="h-3.5 w-3.5 shrink-0 text-military-accent" />
+        <span className="truncate text-[10px] font-semibold uppercase tracking-wider text-military-text-secondary">
+          近期动态
+        </span>
+      </div>
+      <div className="min-h-0 flex-1 overflow-y-auto overflow-x-hidden px-2 py-1">
+        {list.length === 0 ? (
+          <p className="py-4 text-center text-[11px] text-military-text-secondary">暂无动态</p>
+        ) : (
+          <ul className="space-y-2">
+            {list.map((u) => (
+              <li key={u.id} className="flex gap-2 border-b border-military-border/50 pb-2 last:border-0 last:pb-0">
+                <span className={`shrink-0 text-[9px] ${SEV_COLORS[u.severity] ?? 'text-military-text-secondary'}`}>
+                  {u.severity === 'critical' && <AlertTriangle className="inline h-2.5 w-2.5" />}
+                  {CAT_LABELS[u.category] ?? u.category}
+                </span>
+                <div className="min-w-0 flex-1 text-right">
+                  <p className="text-[11px] leading-tight text-military-text-primary line-clamp-2">{u.summary}</p>
+                  <span className="mt-0.5 block text-[9px] text-military-text-secondary">{formatTime(u.timestamp)}</span>
+                </div>
+              </li>
+            ))}
+          </ul>
+        )}
+      </div>
+    </div>
+  )
+}
--- a/src/components/TimelinePanel.tsx
+++ b/src/components/TimelinePanel.tsx
@@ -0,0 +1,142 @@
+import { useEffect, useRef } from 'react'
+import { Play, Pause, SkipBack, SkipForward, History } from 'lucide-react'
+import { usePlaybackStore, REPLAY_TICKS, REPLAY_START, REPLAY_END } from '@/store/playbackStore'
+
+function formatTick(iso: string): string {
+  const d = new Date(iso)
+  return d.toLocaleString('zh-CN', {
+    month: '2-digit',
+    day: '2-digit',
+    hour: '2-digit',
+    minute: '2-digit',
+    hour12: false,
+  })
+}
+
+export function TimelinePanel() {
+  const {
+    isReplayMode,
+    playbackTime,
+    isPlaying,
+    speedSecPerTick,
+    setReplayMode,
+    setPlaybackTime,
+    setIsPlaying,
+    stepForward,
+    stepBack,
+    setSpeed,
+  } = usePlaybackStore()
+
+  const timerRef = useRef<ReturnType<typeof setInterval> | null>(null)
+
+  useEffect(() => {
+    if (!isPlaying || !isReplayMode) {
+      if (timerRef.current) {
+        clearInterval(timerRef.current)
+        timerRef.current = null
+      }
+      return
+    }
+    timerRef.current = setInterval(() => {
+      const current = usePlaybackStore.getState().playbackTime
+      const i = REPLAY_TICKS.indexOf(current)
+      if (i >= REPLAY_TICKS.length - 1) {
+        setIsPlaying(false)
+        return
+      }
+      setPlaybackTime(REPLAY_TICKS[i + 1])
+    }, speedSecPerTick * 1000)
+    return () => {
+      if (timerRef.current) clearInterval(timerRef.current)
+    }
+  }, [isPlaying, isReplayMode, speedSecPerTick, setPlaybackTime, setIsPlaying])
+
+  const index = REPLAY_TICKS.indexOf(playbackTime)
+  const value = index >= 0 ? index : REPLAY_TICKS.length - 1
+
+  const handleSliderChange = (e: React.ChangeEvent<HTMLInputElement>) => {
+    const i = parseInt(e.target.value, 10)
+    setPlaybackTime(REPLAY_TICKS[i])
+  }
+
+  return (
+    <div className="shrink-0 border-b border-military-border bg-military-panel/95 px-3 py-2">
+      <div className="flex flex-wrap items-center gap-3">
+        <button
+          type="button"
+          onClick={() => setReplayMode(!isReplayMode)}
+          className={`flex items-center gap-1.5 rounded px-2 py-1 text-xs font-medium transition-colors ${
+            isReplayMode
+              ? 'bg-military-accent/30 text-military-accent'
+              : 'bg-military-border/50 text-military-text-secondary hover:bg-military-border hover:text-military-text-primary'
+          }`}
+        >
+          <History className="h-3.5 w-3.5" />
+          数据回放
+        </button>
+
+        {isReplayMode && (
+          <>
+            <div className="flex items-center gap-1">
+              <button
+                type="button"
+                onClick={stepBack}
+                disabled={index <= 0}
+                className="rounded p-1 text-military-text-secondary hover:bg-military-border hover:text-military-text-primary disabled:opacity-40 disabled:hover:bg-transparent disabled:hover:text-military-text-secondary"
+                title="上一步"
+              >
+                <SkipBack className="h-4 w-4" />
+              </button>
+              <button
+                type="button"
+                onClick={() => setIsPlaying(!isPlaying)}
+                className="rounded p-1 text-military-text-secondary hover:bg-military-border hover:text-military-text-primary"
+                title={isPlaying ? '暂停' : '播放'}
+              >
+                {isPlaying ? <Pause className="h-4 w-4" /> : <Play className="h-4 w-4" />}
+              </button>
+              <button
+                type="button"
+                onClick={stepForward}
+                disabled={index >= REPLAY_TICKS.length - 1}
+                className="rounded p-1 text-military-text-secondary hover:bg-military-border hover:text-military-text-primary disabled:opacity-40 disabled:hover:bg-transparent disabled:hover:text-military-text-secondary"
+                title="下一步"
+              >
+                <SkipForward className="h-4 w-4" />
+              </button>
+            </div>
+
+            <div className="flex min-w-0 flex-1 items-center gap-2 lg:min-w-[320px]">
+              <input
+                type="range"
+                min={0}
+                max={REPLAY_TICKS.length - 1}
+                value={value}
+                onChange={handleSliderChange}
+                className="h-1.5 flex-1 cursor-pointer appearance-none rounded-full bg-military-border [&::-webkit-slider-thumb]:h-3 [&::-webkit-slider-thumb]:w-3 [&::-webkit-slider-thumb]:appearance-none [&::-webkit-slider-thumb]:cursor-pointer [&::-webkit-slider-thumb]:rounded-full [&::-webkit-slider-thumb]:bg-military-accent"
+              />
+            </div>
+
+            <div className="flex items-center gap-2 text-[11px] tabular-nums text-military-text-secondary">
+              <span>{formatTick(REPLAY_START)}</span>
+              <span className="font-medium text-military-accent">{formatTick(playbackTime)}</span>
+              <span>{formatTick(REPLAY_END)}</span>
+            </div>
+
+            <select
+              value={speedSecPerTick}
+              onChange={(e) => setSpeed(Number(e.target.value))}
+              className="rounded border border-military-border bg-military-dark/80 px-2 py-1 text-[11px] text-military-text-secondary focus:border-military-accent focus:outline-none"
+            >
+              <option value={0.5}>0.5 秒/刻度</option>
+              <option value={1}>1 秒/刻度</option>
+              <option value={2}>2 秒/刻度</option>
+              <option value={3}>3 秒/刻度</option>
+              <option value={5}>5 秒/刻度</option>
+            </select>
+          </>
+        )}
+      </div>
+    </div>
+  )
+}
--- a/src/components/WarMap.tsx
+++ b/src/components/WarMap.tsx
@@ -3,7 +3,7 @@ import Map, { Source, Layer } from 'react-map-gl'
 import type { MapRef } from 'react-map-gl'
 import type { Map as MapboxMap } from 'mapbox-gl'
 import 'mapbox-gl/dist/mapbox-gl.css'
-import { useSituationStore } from '@/store/situationStore'
+import { useReplaySituation } from '@/hooks/useReplaySituation'
 import {
  ATTACKED_TARGETS,
  ALLIED_STRIKE_LOCATIONS,
@@ -129,8 +129,8 @@ export function WarMap() {
  const lincolnPathsRef = useRef<[number, number][][]>([])
  const fordPathsRef = useRef<[number, number][][]>([])
  const israelPathsRef = useRef<[number, number][][]>([])
-  const { situation } = useSituationStore()
-  const { usForces, iranForces } = situation
+  const situation = useReplaySituation()
+  const { usForces, iranForces, conflictEvents = [] } = situation

  const usLocs = (usForces.keyLocations || []) as KeyLoc[]
  const irLocs = (iranForces.keyLocations || []) as KeyLoc[]
@@ -239,6 +239,29 @@ export function WarMap() {
    [attackPaths]
  )

+  // GDELT 冲突事件：1–3 绿, 4–6 橙闪, 7–10 红脉
+  const { conflictEventsGreen, conflictEventsOrange, conflictEventsRed } = useMemo(() => {
+    const green: GeoJSON.Feature<GeoJSON.Point>[] = []
+    const orange: GeoJSON.Feature<GeoJSON.Point>[] = []
+    const red: GeoJSON.Feature<GeoJSON.Point>[] = []
+    for (const e of conflictEvents) {
+      const score = e.impact_score ?? 1
+      const f: GeoJSON.Feature<GeoJSON.Point> = {
+        type: 'Feature',
+        properties: { event_id: e.event_id, impact_score: score },
+        geometry: { type: 'Point', coordinates: [e.lng, e.lat] },
+      }
+      if (score <= 3) green.push(f)
+      else if (score <= 6) orange.push(f)
+      else red.push(f)
+    }
+    return {
+      conflictEventsGreen: { type: 'FeatureCollection' as const, features: green },
+      conflictEventsOrange: { type: 'FeatureCollection' as const, features: orange },
+      conflictEventsRed: { type: 'FeatureCollection' as const, features: red },
+    }
+  }, [conflictEvents])
+
  const hideNonBelligerentLabels = (map: MapboxMap) => {
    const labelLayers = [
      'country-label',
@@ -362,6 +385,20 @@ export function WarMap() {
          map.setPaintProperty('allied-strike-targets-pulse', 'circle-radius', r)
          map.setPaintProperty('allied-strike-targets-pulse', 'circle-opacity', opacity)
        }
+        // GDELT 橙色 (4–6)：闪烁
+        if (map.getLayer('gdelt-events-orange')) {
+          const blink = 0.5 + 0.5 * Math.sin(elapsed * 0.004)
+          map.setPaintProperty('gdelt-events-orange', 'circle-opacity', blink)
+        }
+        // GDELT 红色 (7–10)：脉冲扩散
+        if (map.getLayer('gdelt-events-red-pulse')) {
+          const cycle = 2200
+          const phase = (elapsed % cycle) / cycle
+          const r = 30 * phase
+          const opacity = Math.max(0, 1 - phase * 1.1)
+          map.setPaintProperty('gdelt-events-red-pulse', 'circle-radius', r)
+          map.setPaintProperty('gdelt-events-red-pulse', 'circle-opacity', opacity)
+        }
      } catch (_) {}
      animRef.current = requestAnimationFrame(tick)
    }
@@ -376,7 +413,10 @@ export function WarMap() {
        (map.getSource('attack-dots') && attackPathsRef.current.length > 0) ||
        (map.getSource('allied-strike-dots-lincoln') && lincolnPathsRef.current.length > 0) ||
        (map.getSource('allied-strike-dots-ford') && fordPathsRef.current.length > 0) ||
-        (map.getSource('allied-strike-dots-israel') && israelPathsRef.current.length > 0)
+        (map.getSource('allied-strike-dots-israel') && israelPathsRef.current.length > 0) ||
+        map.getSource('gdelt-events-green') ||
+        map.getSource('gdelt-events-orange') ||
+        map.getSource('gdelt-events-red')
      if (hasAnim) {
        animRef.current = requestAnimationFrame(tick)
      } else {
@@ -439,6 +479,15 @@ export function WarMap() {
        <span className="flex items-center gap-1">
          <span className="h-1.5 w-1.5 rounded-full bg-[#22D3EE]" /> 以色列打击
        </span>
+        <span className="flex items-center gap-1">
+          <span className="h-1.5 w-1.5 rounded-full bg-[#22C55E]" /> 低烈度
+        </span>
+        <span className="flex items-center gap-1">
+          <span className="h-1.5 w-1.5 rounded-full bg-[#F97316]" /> 中烈度
+        </span>
+        <span className="flex items-center gap-1">
+          <span className="h-1.5 w-1.5 rounded-full bg-[#EF4444]" /> 高烈度
+        </span>
      </div>
      <Map
        ref={mapRef}
@@ -537,6 +586,52 @@ export function WarMap() {
          />
        </Source>

+        {/* GDELT 冲突事件：1–3 绿点, 4–6 橙闪, 7–10 红脉 */}
+        <Source id="gdelt-events-green" type="geojson" data={conflictEventsGreen}>
+          <Layer
+            id="gdelt-events-green"
+            type="circle"
+            paint={{
+              'circle-radius': ['interpolate', ['linear'], ['zoom'], 4, 2, 8, 4, 12, 6],
+              'circle-color': '#22C55E',
+              'circle-stroke-width': 0.5,
+              'circle-stroke-color': '#fff',
+            }}
+          />
+        </Source>
+        <Source id="gdelt-events-orange" type="geojson" data={conflictEventsOrange}>
+          <Layer
+            id="gdelt-events-orange"
+            type="circle"
+            paint={{
+              'circle-radius': ['interpolate', ['linear'], ['zoom'], 4, 3, 8, 5, 12, 8],
+              'circle-color': '#F97316',
+              'circle-opacity': 0.8,
+            }}
+          />
+        </Source>
+        <Source id="gdelt-events-red" type="geojson" data={conflictEventsRed}>
+          <Layer
+            id="gdelt-events-red-dot"
+            type="circle"
+            paint={{
+              'circle-radius': ['interpolate', ['linear'], ['zoom'], 4, 2, 8, 4, 12, 6],
+              'circle-color': '#EF4444',
+              'circle-stroke-width': 0.5,
+              'circle-stroke-color': '#fff',
+            }}
+          />
+          <Layer
+            id="gdelt-events-red-pulse"
+            type="circle"
+            paint={{
+              'circle-radius': 0,
+              'circle-color': 'rgba(239, 68, 68, 0.5)',
+              'circle-opacity': 0,
+            }}
+          />
+        </Source>
+
        {/* 美以联军打击伊朗：路径线 */}
        <Source id="allied-strike-lines-lincoln" type="geojson" data={lincolnLinesGeoJson}>
          <Layer
--- a/src/data/mockData.ts
+++ b/src/data/mockData.ts
@@ -31,6 +31,8 @@ export interface PowerIndex {
 export interface CombatLosses {
  bases: { destroyed: number; damaged: number }
  personnelCasualties: { killed: number; wounded: number }
+  /** 平民伤亡 */
+  civilianCasualties?: { killed: number; wounded: number }
  aircraft: number
  warships: number
  armor: number
@@ -45,6 +47,23 @@ export interface SituationUpdate {
  severity: 'low' | 'medium' | 'high' | 'critical'
 }

+export interface ConflictEvent {
+  event_id: string
+  event_time: string
+  title: string
+  lat: number
+  lng: number
+  impact_score: number
+  url?: string
+}
+
+export interface ConflictStats {
+  total_events: number
+  high_impact_events: number
+  estimated_casualties: number
+  estimated_strike_count: number
+}
+
 export interface MilitarySituation {
  lastUpdated: string
  usForces: {
@@ -86,6 +105,12 @@ export interface MilitarySituation {
    retaliationSentimentHistory: { time: string; value: number }[]
  }
  recentUpdates: SituationUpdate[]
+  /** GDELT 冲突事件 (1–3 绿点, 4–6 橙闪, 7–10 红脉) */
+  conflictEvents?: ConflictEvent[]
+  /** 战损统计（展示用） */
+  conflictStats?: ConflictStats
+  /** 平民伤亡合计（不区分阵营） */
+  civilianCasualtiesTotal?: { killed: number; wounded: number }
 }

 export const INITIAL_MOCK_DATA: MilitarySituation = {
@@ -122,6 +147,7 @@ export const INITIAL_MOCK_DATA: MilitarySituation = {
    combatLosses: {
      bases: { destroyed: 0, damaged: 2 },
      personnelCasualties: { killed: 127, wounded: 384 },
+      civilianCasualties: { killed: 18, wounded: 52 },
      aircraft: 2,
      warships: 0,
      armor: 0,
@@ -171,6 +197,7 @@ export const INITIAL_MOCK_DATA: MilitarySituation = {
    combatLosses: {
      bases: { destroyed: 3, damaged: 8 },
      personnelCasualties: { killed: 2847, wounded: 5620 },
+      civilianCasualties: { killed: 412, wounded: 1203 },
      aircraft: 24,
      warships: 12,
      armor: 18,
@@ -219,4 +246,7 @@ export const INITIAL_MOCK_DATA: MilitarySituation = {
      severity: 'low',
    },
  ],
+  conflictEvents: [],
+  conflictStats: { total_events: 0, high_impact_events: 0, estimated_casualties: 0, estimated_strike_count: 0 },
+  civilianCasualtiesTotal: { killed: 430, wounded: 1255 },
 }
--- a/src/hooks/useReplaySituation.ts
+++ b/src/hooks/useReplaySituation.ts
@@ -0,0 +1,146 @@
+import { useMemo } from 'react'
+import type { MilitarySituation } from '@/data/mockData'
+import { useSituationStore } from '@/store/situationStore'
+import { usePlaybackStore } from '@/store/playbackStore'
+
+/** 将系列时间映射到回放日 (2026-03-01) 以便按当天时刻插值 */
+function toReplayDay(iso: string, baseDay: string): string {
+  const d = new Date(iso)
+  const [y, m, day] = baseDay.slice(0, 10).split('-').map(Number)
+  return new Date(y, (m || 1) - 1, day || 1, d.getUTCHours(), d.getUTCMinutes(), 0, 0).toISOString()
+}
+
+function interpolateAt(
+  series: { time: string; value: number }[],
+  at: string,
+  baseDay = '2026-03-01'
+): number {
+  if (series.length === 0) return 0
+  const t = new Date(at).getTime()
+  const mapped = series.map((p) => ({
+    time: toReplayDay(p.time, baseDay),
+    value: p.value,
+  }))
+  const sorted = [...mapped].sort((a, b) => new Date(a.time).getTime() - new Date(b.time).getTime())
+  const before = sorted.filter((p) => new Date(p.time).getTime() <= t)
+  const after = sorted.filter((p) => new Date(p.time).getTime() > t)
+  if (before.length === 0) return sorted[0].value
+  if (after.length === 0) return sorted[sorted.length - 1].value
+  const a = before[before.length - 1]
+  const b = after[0]
+  const ta = new Date(a.time).getTime()
+  const tb = new Date(b.time).getTime()
+  const f = tb === ta ? 1 : (t - ta) / (tb - ta)
+  return a.value + f * (b.value - a.value)
+}
+
+function linearProgress(start: string, end: string, at: string): number {
+  const ts = new Date(start).getTime()
+  const te = new Date(end).getTime()
+  const ta = new Date(at).getTime()
+  if (ta <= ts) return 0
+  if (ta >= te) return 1
+  return (ta - ts) / (te - ts)
+}
+
+/** 根据回放时刻派生态势数据 */
+export function useReplaySituation(): MilitarySituation {
+  const situation = useSituationStore((s) => s.situation)
+  const { isReplayMode, playbackTime } = usePlaybackStore()
+
+  return useMemo(() => {
+    if (!isReplayMode) return situation
+
+    const progress = linearProgress('2026-03-01T02:00:00.000Z', '2026-03-01T11:45:00.000Z', playbackTime)
+
+    // 华尔街趋势、反击情绪：按时间插值
+    const wsValue = interpolateAt(situation.usForces.wallStreetInvestmentTrend, playbackTime)
+    const retValue = interpolateAt(situation.iranForces.retaliationSentimentHistory, playbackTime)
+
+    // 战斗损失：从 0 线性增长到当前值
+    const lerp = (a: number, b: number) => Math.round(a + progress * (b - a))
+    const usLoss = situation.usForces.combatLosses
+    const irLoss = situation.iranForces.combatLosses
+    const civTotal = situation.civilianCasualtiesTotal ?? { killed: 0, wounded: 0 }
+    const usLossesAt = {
+      bases: {
+        destroyed: lerp(0, usLoss.bases.destroyed),
+        damaged: lerp(0, usLoss.bases.damaged),
+      },
+      personnelCasualties: {
+        killed: lerp(0, usLoss.personnelCasualties.killed),
+        wounded: lerp(0, usLoss.personnelCasualties.wounded),
+      },
+      civilianCasualties: { killed: 0, wounded: 0 },
+      aircraft: lerp(0, usLoss.aircraft),
+      warships: lerp(0, usLoss.warships),
+      armor: lerp(0, usLoss.armor),
+      vehicles: lerp(0, usLoss.vehicles),
+    }
+    const irLossesAt = {
+      bases: {
+        destroyed: lerp(0, irLoss.bases.destroyed),
+        damaged: lerp(0, irLoss.bases.damaged),
+      },
+      personnelCasualties: {
+        killed: lerp(0, irLoss.personnelCasualties.killed),
+        wounded: lerp(0, irLoss.personnelCasualties.wounded),
+      },
+      civilianCasualties: { killed: 0, wounded: 0 },
+      aircraft: lerp(0, irLoss.aircraft),
+      warships: lerp(0, irLoss.warships),
+      armor: lerp(0, irLoss.armor),
+      vehicles: lerp(0, irLoss.vehicles),
+    }
+
+    // 被袭基地：按 damage_level 排序，高损毁先出现；根据 progress 决定显示哪些为 attacked
+    const usLocs = situation.usForces.keyLocations || []
+    const attackedBases = usLocs
+      .filter((loc) => loc.status === 'attacked')
+      .sort((a, b) => (b.damage_level ?? 0) - (a.damage_level ?? 0))
+    const totalAttacked = attackedBases.length
+    const shownAttackedCount = Math.round(progress * totalAttacked)
+    const attackedNames = new Set(
+      attackedBases.slice(0, shownAttackedCount).map((l) => l.name)
+    )
+
+    const usLocsAt = usLocs.map((loc) => {
+      if (loc.status === 'attacked' && !attackedNames.has(loc.name)) {
+        return { ...loc, status: 'operational' as const }
+      }
+      return { ...loc }
+    })
+
+    return {
+      ...situation,
+      lastUpdated: playbackTime,
+      civilianCasualtiesTotal: {
+        killed: lerp(0, civTotal.killed),
+        wounded: lerp(0, civTotal.wounded),
+      },
+      usForces: {
+        ...situation.usForces,
+        keyLocations: usLocsAt,
+        combatLosses: usLossesAt,
+        wallStreetInvestmentTrend: [
+          ...situation.usForces.wallStreetInvestmentTrend.filter((p) => new Date(p.time).getTime() <= new Date(playbackTime).getTime()),
+          { time: playbackTime, value: wsValue },
+        ].slice(-20),
+      },
+      iranForces: {
+        ...situation.iranForces,
+        combatLosses: irLossesAt,
+        retaliationSentiment: retValue,
+        retaliationSentimentHistory: [
+          ...situation.iranForces.retaliationSentimentHistory.filter((p) => new Date(p.time).getTime() <= new Date(playbackTime).getTime()),
+          { time: playbackTime, value: retValue },
+        ].slice(-20),
+      },
+      recentUpdates: (situation.recentUpdates || []).filter(
+        (u) => new Date(u.timestamp).getTime() <= new Date(playbackTime).getTime()
+      ),
+      conflictEvents: situation.conflictEvents || [],
+      conflictStats: situation.conflictStats || { total_events: 0, high_impact_events: 0, estimated_casualties: 0, estimated_strike_count: 0 },
+    }
+  }, [situation, isReplayMode, playbackTime])
+}
--- a/src/index.css
+++ b/src/index.css
@@ -31,6 +31,11 @@ body,
  font-family: 'Orbitron', sans-serif;
 }

+/* 数据库面板：易读字体 */
+.font-db {
+  font-family: 'Noto Sans SC', system-ui, -apple-system, sans-serif;
+}
+
 /* Tabular numbers for aligned stat display */
 .tabular-nums {
  font-variant-numeric: tabular-nums;
--- a/src/main.tsx
+++ b/src/main.tsx
@@ -1,10 +1,13 @@
 import { StrictMode } from 'react'
 import { createRoot } from 'react-dom/client'
+import { BrowserRouter } from 'react-router-dom'
 import App from './App.tsx'
 import './index.css'

 createRoot(document.getElementById('root')!).render(
  <StrictMode>
-    <App />
+    <BrowserRouter>
+      <App />
+    </BrowserRouter>
  </StrictMode>
 )
--- a/src/pages/Dashboard.tsx
+++ b/src/pages/Dashboard.tsx
@@ -1,17 +1,20 @@
 import { useEffect } from 'react'
 import { HeaderPanel } from '@/components/HeaderPanel'
+import { TimelinePanel } from '@/components/TimelinePanel'
 import { ForcePanel } from '@/components/ForcePanel'
 import { WarMap } from '@/components/WarMap'
 import { CombatLossesPanel } from '@/components/CombatLossesPanel'
+import { EventTimelinePanel } from '@/components/EventTimelinePanel'
 import { BaseStatusPanel } from '@/components/BaseStatusPanel'
 import { PowerChart } from '@/components/PowerChart'
 import { InvestmentTrendChart } from '@/components/InvestmentTrendChart'
 import { RetaliationGauge } from '@/components/RetaliationGauge'
 import { useSituationStore } from '@/store/situationStore'
+import { useReplaySituation } from '@/hooks/useReplaySituation'
 import { fetchAndSetSituation, startSituationWebSocket, stopSituationWebSocket } from '@/store/situationStore'

 export function Dashboard() {
-  const situation = useSituationStore((s) => s.situation)
+  const situation = useReplaySituation()
  const isLoading = useSituationStore((s) => s.isLoading)
  const lastError = useSituationStore((s) => s.lastError)

@@ -28,6 +31,7 @@ export function Dashboard() {
        </div>
      )}
      <HeaderPanel />
+      <TimelinePanel />

      <div className="flex min-h-0 flex-1 flex-col overflow-auto lg:flex-row lg:overflow-hidden">
        <aside className="flex min-h-0 min-w-0 shrink-0 flex-col gap-2 overflow-y-auto overflow-x-visible border-b border-military-border p-3 lg:min-w-[320px] lg:max-w-[340px] lg:border-b-0 lg:border-r lg:p-4">
@@ -63,8 +67,11 @@ export function Dashboard() {
            <CombatLossesPanel
              usLosses={situation.usForces.combatLosses}
              iranLosses={situation.iranForces.combatLosses}
-              className="min-w-0 flex-1 border-t-0"
+              conflictStats={situation.conflictStats}
+              civilianTotal={situation.civilianCasualtiesTotal}
+              className="min-w-0 flex-1 py-1"
            />
+            <EventTimelinePanel updates={situation.recentUpdates} conflictEvents={situation.conflictEvents} className="min-w-0 shrink-0 min-h-[80px] overflow-hidden lg:min-w-[240px]" />
          </div>
        </main>

--- a/src/pages/DbDashboard.tsx
+++ b/src/pages/DbDashboard.tsx
@@ -0,0 +1,161 @@
+import { useEffect, useState } from 'react'
+import { Database, Table, ArrowLeft, RefreshCw } from 'lucide-react'
+import { Link } from 'react-router-dom'
+
+interface TableData {
+  [table: string]: Record<string, unknown>[] | { error: string }
+}
+
+export function DbDashboard() {
+  const [data, setData] = useState<TableData | null>(null)
+  const [loading, setLoading] = useState(true)
+  const [error, setError] = useState<string | null>(null)
+  const [expanded, setExpanded] = useState<Set<string>>(new Set(['situation_update', 'combat_losses', 'conflict_stats']))
+
+  useEffect(() => {
+    fetchData()
+    const t = setInterval(fetchData, 30000)
+    return () => clearInterval(t)
+  }, [])
+
+  const fetchData = async () => {
+    setLoading(true)
+    setError(null)
+    try {
+      const res = await fetch('/api/db/dashboard')
+      if (!res.ok) throw new Error(res.statusText)
+      const json = await res.json()
+      setData(json)
+    } catch (e) {
+      setError(e instanceof Error ? e.message : '加载失败')
+    } finally {
+      setLoading(false)
+    }
+  }
+
+  const toggle = (name: string) => {
+    setExpanded((s) => {
+      const next = new Set(s)
+      if (next.has(name)) next.delete(name)
+      else next.add(name)
+      return next
+    })
+  }
+
+  if (loading && !data) {
+    return (
+      <div className="flex min-h-screen items-center justify-center bg-military-dark text-military-text-secondary">
+        <RefreshCw className="h-6 w-6 animate-spin" />
+      </div>
+    )
+  }
+
+  return (
+    <div className="min-h-screen bg-military-dark font-db text-military-text-primary">
+      <header className="sticky top-0 z-10 flex items-center justify-between border-b border-military-border bg-military-panel/95 px-4 py-3">
+        <div className="flex items-center gap-4">
+          <Link
+            to="/"
+            className="flex items-center gap-2 rounded border border-military-border px-3 py-1.5 text-sm text-military-text-secondary hover:bg-military-border/30 hover:text-military-text-primary"
+          >
+            <ArrowLeft className="h-4 w-4" />
+            返回主面板
+          </Link>
+          <span className="flex items-center gap-2 text-lg">
+            <Database className="h-5 w-5 text-cyan-400" />
+            数据库内容
+          </span>
+        </div>
+        <button
+          onClick={fetchData}
+          disabled={loading}
+          className="flex items-center gap-2 rounded border border-military-border px-3 py-1.5 text-sm text-military-text-secondary hover:bg-military-border/30 disabled:opacity-50"
+        >
+          <RefreshCw className={`h-4 w-4 ${loading ? 'animate-spin' : ''}`} />
+          刷新
+        </button>
+      </header>
+
+      {error && (
+        <div className="mx-4 mt-4 rounded border border-amber-600/50 bg-amber-950/30 px-4 py-2 text-amber-400">
+          {error}（请确保 API 已启动：npm run api）
+        </div>
+      )}
+
+      <main className="max-w-6xl space-y-4 p-4">
+        {data &&
+          Object.entries(data).map(([name, rows]) => {
+            const isExpanded = expanded.has(name)
+            const isError = rows && typeof rows === 'object' && 'error' in rows
+            const arr = Array.isArray(rows) ? rows : []
+            return (
+              <section
+                key={name}
+                className="rounded border border-military-border bg-military-panel/80 overflow-hidden"
+              >
+                <button
+                  onClick={() => toggle(name)}
+                  className="flex w-full items-center justify-between px-4 py-3 text-left hover:bg-military-border/20"
+                >
+                  <span className="flex items-center gap-2 font-medium">
+                    <Table className="h-4 w-4 text-cyan-400" />
+                    {name}
+                    <span className="text-military-text-secondary font-normal">
+                      {isError ? '(错误)' : `(${arr.length} 条)`}
+                    </span>
+                  </span>
+                  <span className="text-military-text-secondary">{isExpanded ? '▼' : '▶'}</span>
+                </button>
+                {isExpanded && (
+                  <div className="border-t border-military-border overflow-x-auto">
+                    {isError ? (
+                      <pre className="p-4 text-sm text-amber-400">{(rows as { error: string }).error}</pre>
+                    ) : arr.length === 0 ? (
+                      <p className="p-4 text-sm text-military-text-secondary">无数据</p>
+                    ) : (
+                      <table className="w-full text-left text-sm">
+                        <thead>
+                          <tr className="border-b border-military-border bg-military-dark/50">
+                            {Object.keys(arr[0] as object).map((col) => (
+                              <th
+                                key={col}
+                                className="whitespace-nowrap px-3 py-2 font-medium text-cyan-400"
+                              >
+                                {col}
+                              </th>
+                            ))}
+                          </tr>
+                        </thead>
+                        <tbody>
+                          {arr.map((row, i) => (
+                            <tr
+                              key={i}
+                              className="border-b border-military-border/50 hover:bg-military-border/10"
+                            >
+                              {Object.values(row as object).map((v, j) => (
+                                <td
+                                  key={j}
+                                  className="max-w-xs truncate px-3 py-2 text-military-text-secondary"
+                                  title={String(v ?? '')}
+                                >
+                                  {v === null || v === undefined
+                                    ? '—'
+                                    : typeof v === 'object'
+                                      ? JSON.stringify(v)
+                                      : String(v)}
+                                </td>
+                              ))}
+                            </tr>
+                          ))}
+                        </tbody>
+                      </table>
+                    )}
+                  </div>
+                )}
+              </section>
+            )
+          })}
+      </main>
+    </div>
+  )
+}
--- a/src/store/playbackStore.ts
+++ b/src/store/playbackStore.ts
@@ -0,0 +1,80 @@
+import { create } from 'zustand'
+
+const REPLAY_DAY = '2026-03-01'
+const TICK_MS = 30 * 60 * 1000 // 30 minutes
+
+export const REPLAY_START = `${REPLAY_DAY}T00:00:00.000Z`
+export const REPLAY_END = `${REPLAY_DAY}T23:30:00.000Z`
+
+function parseTime(iso: string): number {
+  return new Date(iso).getTime()
+}
+
+export function getTicks(): string[] {
+  const ticks: string[] = []
+  let t = parseTime(REPLAY_START)
+  const end = parseTime(REPLAY_END)
+  while (t <= end) {
+    ticks.push(new Date(t).toISOString())
+    t += TICK_MS
+  }
+  return ticks
+}
+
+export const REPLAY_TICKS = getTicks()
+
+export interface PlaybackState {
+  /** 是否开启回放模式 */
+  isReplayMode: boolean
+  /** 当前回放时刻 (ISO) */
+  playbackTime: string
+  /** 是否正在自动播放 */
+  isPlaying: boolean
+  /** 播放速度 (秒/刻度) */
+  speedSecPerTick: number
+  setReplayMode: (v: boolean) => void
+  setPlaybackTime: (iso: string) => void
+  setIsPlaying: (v: boolean) => void
+  stepForward: () => void
+  stepBack: () => void
+  setSpeed: (sec: number) => void
+}
+
+export const usePlaybackStore = create<PlaybackState>((set, get) => ({
+  isReplayMode: false,
+  playbackTime: REPLAY_END,
+  isPlaying: false,
+  speedSecPerTick: 2,
+
+  setReplayMode: (v) => set({ isReplayMode: v, isPlaying: false }),
+
+  setPlaybackTime: (iso) => {
+    const ticks = REPLAY_TICKS
+    if (ticks.includes(iso)) {
+      set({ playbackTime: iso })
+      return
+    }
+    const idx = ticks.findIndex((t) => t >= iso)
+    const clamp = Math.max(0, Math.min(idx < 0 ? ticks.length - 1 : idx, ticks.length - 1))
+    set({ playbackTime: ticks[clamp] })
+  },
+
+  setIsPlaying: (v) => set({ isPlaying: v }),
+
+  stepForward: () => {
+    const { playbackTime } = get()
+    const ticks = REPLAY_TICKS
+    const i = ticks.indexOf(playbackTime)
+    if (i < ticks.length - 1) set({ playbackTime: ticks[i + 1] })
+    else set({ isPlaying: false })
+  },
+
+  stepBack: () => {
+    const { playbackTime } = get()
+    const ticks = REPLAY_TICKS
+    const i = ticks.indexOf(playbackTime)
+    if (i > 0) set({ playbackTime: ticks[i - 1] })
+  },
+
+  setSpeed: (sec) => set({ speedSecPerTick: sec }),
+}))
--- a/src/store/situationStore.ts
+++ b/src/store/situationStore.ts
@@ -47,20 +47,36 @@ export function fetchAndSetSituation(): Promise<void> {
 }

 let disconnectWs: (() => void) | null = null
+let pollInterval: ReturnType<typeof setInterval> | null = null
+
+const POLL_INTERVAL_MS = 5000
+
+function pollSituation() {
+  fetchSituation()
+    .then((situation) => useSituationStore.getState().setSituation(situation))
+    .catch(() => {})
+}

 export function startSituationWebSocket(): () => void {
-  useSituationStore.getState().setConnected(true)
  useSituationStore.getState().setLastError(null)

  disconnectWs = connectSituationWebSocket((data) => {
+    useSituationStore.getState().setConnected(true)
    useSituationStore.getState().setSituation(data as MilitarySituation)
  })

+  pollSituation()
+  pollInterval = setInterval(pollSituation, POLL_INTERVAL_MS)
+
  return stopSituationWebSocket
 }

 export function stopSituationWebSocket(): void {
  disconnectWs?.()
  disconnectWs = null
+  if (pollInterval) {
+    clearInterval(pollInterval)
+    pollInterval = null
+  }
  useSituationStore.getState().setConnected(false)
 }
--- a/start.sh
+++ b/start.sh
@@ -0,0 +1,48 @@
+#!/usr/bin/env bash
+# 一键启动 US-Iran 态势面板：API + 前端 + 爬虫服务
+set -e
+cd "$(dirname "$0")"
+
+# 无 Ollama 时禁用 AI；GDELT 国内常超时，仅用 RSS 更新
+export CLEANER_AI_DISABLED=1
+export PARSER_AI_DISABLED=1
+export GDELT_DISABLED=1
+export RSS_INTERVAL_SEC=60
+
+echo "==> Checking dependencies..."
+[ ! -d node_modules ] && npm install
+
+echo "==> Checking crawler Python deps..."
+pip install -q -r crawler/requirements.txt 2>/dev/null || true
+
+echo "==> Seeding database (if needed)..."
+[ ! -f server/data.db ] && npm run api:seed
+
+echo "==> Starting API (http://localhost:3001)..."
+npm run api &
+API_PID=$!
+
+# 等待 API 就绪后再启动爬虫
+sleep 2
+
+echo "==> Starting GDELT/RSS crawler (http://localhost:8000)..."
+npm run gdelt &
+GDELT_PID=$!
+
+echo "==> Starting frontend (Vite dev server)..."
+npm run dev &
+DEV_PID=$!
+
+cleanup() {
+  echo ""
+  echo "==> Shutting down..."
+  kill $API_PID $GDELT_PID $DEV_PID 2>/dev/null || true
+  exit 0
+}
+trap cleanup SIGINT SIGTERM
+
+echo ""
+echo "==> All services running. Frontend: http://localhost:5173  |  API: http://localhost:3001"
+echo "    Press Ctrl+C to stop all."
+echo ""
+wait
Author	SHA1	Message	Date
Daniel	004d10b283	fix: 优化数据	2026-03-02 11:28:13 +08:00
Daniel	4a8fff5a00	fix:优化数据来源	2026-03-02 01:00:04 +08:00
Daniel	91d9e48e1e	fix:优化整个大屏界面	2026-03-02 00:59:40 +08:00