fix: 优化数据

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

crawler/README.md

@@ -18,12 +18,19 @@
 
 | 项目 | 说明 |
 |------|------|
-| 源 | Reuters、BBC World/MiddleEast、Al Jazeera、NYT World |
+| 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。
+**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）
 
 ---
 
@@ -69,6 +76,9 @@ GDELT API → 抓取(60s) → SQLite (gdelt_events, conflict_stats) → POST /ap
 - `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)
 

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"

crawler/config.py

@@ -13,14 +13,30 @@ 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（美伊/中东相关，多源保证实时事件脉络）
+# 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",
-    "https://rss.nytimes.com/services/xml/rss/nyt/World.xml",
 ]
 
 # 关键词过滤：至少匹配一个才会入库

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()

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

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

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()

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))

crawler/realtime_conflict_service.py

@@ -14,11 +14,13 @@ 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=["*"])
 
@@ -29,7 +31,7 @@ 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", "45"))  # 新闻抓取更频繁，优先保证事件脉络
+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）
@@ -77,7 +79,9 @@ def _parse_article(article: dict) -> Optional[dict]:
     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")
@@ -134,8 +138,8 @@ def fetch_gdelt_events() -> None:
         _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 as e:
+    except Exception:
-        print(f"GDELT 抓取失败: {e}")
+        pass
 
 
 def _ensure_table(conn: sqlite3.Connection) -> None:
@@ -213,38 +217,115 @@ def _notify_node() -> None:
 
 
 # ==========================
-# RSS 新闻抓取（补充 situation_update）
+# 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:
-            it["title"] = translate_to_chinese(it.get("title", "") or "")
+            raw_title = translate_to_chinese(it.get("title", "") or "")
-            it["summary"] = translate_to_chinese(it.get("summary", "") or it.get("title", ""))
+            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:
-            n = write_updates(items)
+            _extract_and_merge_panel_data(items)
             if n > 0:
                 _notify_node()
-                print(f"[{datetime.now().strftime('%H:%M:%S')}] 新闻入库 {n} 条")
+        print(f"[{datetime.now().strftime('%H:%M:%S')}] RSS 抓取 {len(items)} 条，新增入库 {n} 条")
     except Exception as e:
-        print(f"新闻抓取失败: {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)
+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)
+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 {

crawler/scrapers/rss_scraper.py

@@ -6,7 +6,7 @@ from datetime import datetime, timezone
 import feedparser
 
 from config import RSS_FEEDS, KEYWORDS
-from parser import classify, severity
+from parser_ai import classify_and_severity
 
 
 def _parse_date(entry) -> datetime:
@@ -62,8 +62,7 @@ def fetch_all() -> list[dict]:
                     continue
                 seen.add(key)
                 published = _parse_date(entry)
-                cat = classify(text)
+                cat, sev = classify_and_severity(text)
-                sev = severity(text, cat)
                 items.append({
                     "title": title,
                     "summary": summary[:400] if summary else title,

crawler/translate_utils.py

@@ -1,5 +1,6 @@
 # -*- coding: utf-8 -*-
 """英译中，入库前统一翻译"""
+import os
 import re
 from typing import Optional
 
@@ -12,17 +13,26 @@ def _is_mostly_chinese(text: str) -> bool:
 
 
 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。"""
+    """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)
-        return out if out else text
+            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

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",

package.json

@@ -4,11 +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 .",
@@ -25,6 +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"
   },

server/db.js

@@ -123,12 +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
+// 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

server/index.js

@@ -31,7 +31,7 @@ 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() {

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())

server/situationData.js

@@ -58,18 +58,23 @@ function getSituation() {
     if (statsRow) conflictStats = statsRow
   } catch (_) {}
 
-  // 根据爬虫 conflict_stats 实时合并平民伤亡估算（GDELT 数据）
+  // 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 est = conflictStats.estimated_casualties || 0
+  const usLosses = { ...usLossesBase, civilianCasualties: { killed: 0, wounded: 0 } }
-  const mergeCivilian = (base, share) => {
+  const irLosses = { ...irLossesBase, civilianCasualties: { killed: 0, wounded: 0 } }
-    if (est <= 0) return base.civilianCasualties || { killed: 0, wounded: 0 }
-    const gdeltKilled = Math.round(est * share)
-    const cur = base.civilianCasualties || { killed: 0, wounded: 0 }
-    return { killed: Math.max(cur.killed, gdeltKilled), wounded: cur.wounded }
-  }
-  const usLosses = { ...usLossesBase, civilianCasualties: mergeCivilian(usLossesBase, 0.35) }
-  const irLosses = { ...irLossesBase, civilianCasualties: mergeCivilian(irLossesBase, 0.65) }
 
   return {
     lastUpdated: meta?.updated_at || new Date().toISOString(),
@@ -135,6 +140,7 @@ function getSituation() {
       url: e.url,
     })),
     conflictStats,
+    civilianCasualtiesTotal,
   }
 }
 

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>
   )
 }

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()
 }

src/components/CombatLossesPanel.tsx

@@ -17,16 +17,15 @@ 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
 }
 
-export function CombatLossesPanel({ usLosses, iranLosses, conflictStats, className = '' }: CombatLossesPanelProps) {
+export function CombatLossesPanel({ usLosses, iranLosses, conflictStats, civilianTotal, className = '' }: CombatLossesPanelProps) {
-  const civUs = usLosses.civilianCasualties ?? { killed: 0, wounded: 0 }
+  const civ = civilianTotal ?? { killed: 0, wounded: 0 }
-  const civIr = iranLosses.civilianCasualties ?? { killed: 0, wounded: 0 }
-  const civTotal = { killed: (civUs.killed ?? 0) + (civIr.killed ?? 0), wounded: (civUs.wounded ?? 0) + (civIr.wounded ?? 0) }
 
   const otherRows = [
-    { label: '平民', icon: UserCircle, iconColor: 'text-amber-400', value: `${formatMillions(civTotal.killed)} / ${formatMillions(civTotal.wounded)}`, noSide: true },
     { label: '基地', icon: Building2, iconColor: 'text-amber-500', us: `${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 },
@@ -70,6 +69,25 @@ export function CombatLossesPanel({ usLosses, iranLosses, conflictStats, classNa
           </div>
         </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 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>
+        </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>

src/components/EventTimelinePanel.tsx

@@ -1,5 +1,7 @@
+import * as React from 'react'
 import type { SituationUpdate, ConflictEvent } from '@/data/mockData'
-import { History } from 'lucide-react'
+import { History, RefreshCw } from 'lucide-react'
+import { fetchAndSetSituation } from '@/store/situationStore'
 
 interface EventTimelinePanelProps {
   updates: SituationUpdate[]
@@ -29,6 +31,11 @@ type TimelineItem = {
 }
 
 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) => ({
@@ -49,7 +56,7 @@ export function EventTimelinePanel({ updates = [], conflictEvents = [], classNam
     })),
   ]
     .sort((a, b) => new Date(b.timestamp).getTime() - new Date(a.timestamp).getTime())
-    .slice(0, 6)
+    .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}`}>
@@ -58,7 +65,16 @@ export function EventTimelinePanel({ updates = [], conflictEvents = [], classNam
           <History className="h-3.5 w-3.5 shrink-0 text-military-accent" />
           事件脉络
         </span>
-        <span className="text-[8px] text-military-text-secondary/80">GDELT · Reuters · BBC · Al Jazeera · NYT</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 ? (

src/components/HeaderPanel.tsx

@@ -1,9 +1,10 @@
 import { useState, useEffect } from 'react'
+import { Link } from 'react-router-dom'
 import { StatCard } from './StatCard'
 import { useSituationStore } from '@/store/situationStore'
 import { useReplaySituation } from '@/hooks/useReplaySituation'
 import { usePlaybackStore } from '@/store/playbackStore'
-import { Wifi, WifiOff, Clock } from 'lucide-react'
+import { Wifi, WifiOff, Clock, Database } from 'lucide-react'
 
 export function HeaderPanel() {
   const situation = useReplaySituation()
@@ -58,7 +59,14 @@ export function HeaderPanel() {
           )}
         </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" />

src/data/mockData.ts

@@ -109,6 +109,8 @@ export interface MilitarySituation {
   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 = {
@@ -246,4 +248,5 @@ export const INITIAL_MOCK_DATA: MilitarySituation = {
   ],
   conflictEvents: [],
   conflictStats: { total_events: 0, high_impact_events: 0, estimated_casualties: 0, estimated_strike_count: 0 },
+  civilianCasualtiesTotal: { killed: 430, wounded: 1255 },
 }

src/hooks/useReplaySituation.ts

@@ -61,8 +61,7 @@ export function useReplaySituation(): MilitarySituation {
     const lerp = (a: number, b: number) => Math.round(a + progress * (b - a))
     const usLoss = situation.usForces.combatLosses
     const irLoss = situation.iranForces.combatLosses
-    const civUs = usLoss.civilianCasualties ?? { killed: 0, wounded: 0 }
+    const civTotal = situation.civilianCasualtiesTotal ?? { killed: 0, wounded: 0 }
-    const civIr = irLoss.civilianCasualties ?? { killed: 0, wounded: 0 }
     const usLossesAt = {
       bases: {
         destroyed: lerp(0, usLoss.bases.destroyed),
@@ -72,7 +71,7 @@ export function useReplaySituation(): MilitarySituation {
         killed: lerp(0, usLoss.personnelCasualties.killed),
         wounded: lerp(0, usLoss.personnelCasualties.wounded),
       },
-      civilianCasualties: { killed: lerp(0, civUs.killed), wounded: lerp(0, civUs.wounded) },
+      civilianCasualties: { killed: 0, wounded: 0 },
       aircraft: lerp(0, usLoss.aircraft),
       warships: lerp(0, usLoss.warships),
       armor: lerp(0, usLoss.armor),
@@ -87,7 +86,7 @@ export function useReplaySituation(): MilitarySituation {
         killed: lerp(0, irLoss.personnelCasualties.killed),
         wounded: lerp(0, irLoss.personnelCasualties.wounded),
       },
-      civilianCasualties: { killed: lerp(0, civIr.killed), wounded: lerp(0, civIr.wounded) },
+      civilianCasualties: { killed: 0, wounded: 0 },
       aircraft: lerp(0, irLoss.aircraft),
       warships: lerp(0, irLoss.warships),
       armor: lerp(0, irLoss.armor),
@@ -115,6 +114,10 @@ export function useReplaySituation(): MilitarySituation {
     return {
       ...situation,
       lastUpdated: playbackTime,
+      civilianCasualtiesTotal: {
+        killed: lerp(0, civTotal.killed),
+        wounded: lerp(0, civTotal.wounded),
+      },
       usForces: {
         ...situation.usForces,
         keyLocations: usLocsAt,

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;

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>
+    <BrowserRouter>
       <App />
+    </BrowserRouter>
   </StrictMode>
 )

src/pages/Dashboard.tsx

@@ -68,6 +68,7 @@ export function Dashboard() {
               usLosses={situation.usForces.combatLosses}
               iranLosses={situation.iranForces.combatLosses}
               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]" />

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>
+  )
+}

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)
 }

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