初始化项目版本

scripts/cursor-usage-snapshot.py

@@ -0,0 +1,118 @@
+#!/usr/bin/env python3
+"""本地用量快照：不消耗 Cursor Agent token，直接读官方 API。"""
+from __future__ import annotations
+
+import datetime as dt
+import json
+import sqlite3
+import urllib.request
+from collections import defaultdict
+
+STATE_DB = "/Users/sylvawong/Library/Application Support/Cursor/User/globalStorage/state.vscdb"
+TZ = dt.timezone(dt.timedelta(hours=8))
+
+
+def token() -> str:
+    conn = sqlite3.connect(STATE_DB)
+    try:
+        row = conn.execute(
+            "SELECT value FROM ItemTable WHERE key='cursorAuth/accessToken'"
+        ).fetchone()
+    finally:
+        conn.close()
+    if not row:
+        raise SystemExit("未找到 Cursor accessToken，请先在 Cursor 登录。")
+    return row[0]
+
+
+def post(path: str, body: dict | None = None) -> dict:
+    req = urllib.request.Request(
+        f"https://api2.cursor.sh/{path}",
+        data=json.dumps(body or {}).encode(),
+        headers={
+            "Authorization": f"Bearer {token()}",
+            "Content-Type": "application/json",
+            "Connect-Protocol-Version": "1",
+            "User-Agent": "oneos-token-budget/1.0",
+        },
+        method="POST",
+    )
+    with urllib.request.urlopen(req, timeout=60) as resp:
+        return json.loads(resp.read().decode() or "{}")
+
+
+def ms_to_local(ms: str | int) -> str:
+    return dt.datetime.fromtimestamp(int(ms) / 1000, TZ).strftime("%Y-%m-%d %H:%M")
+
+
+def main() -> None:
+    usage = post("aiserver.v1.DashboardService/GetCurrentPeriodUsage")
+    plan = post("aiserver.v1.DashboardService/GetPlanInfo")
+    pu = usage.get("planUsage") or {}
+    info = plan.get("planInfo") or {}
+
+    print("=== Cursor 用量快照 ===")
+    print(f"套餐: {info.get('planName')} ({info.get('price')})")
+    print(f"账期: {ms_to_local(usage.get('billingCycleStart'))} → {ms_to_local(usage.get('billingCycleEnd'))}")
+    print(f"Auto+Composer: {float(pu.get('autoPercentUsed') or 0):.2f}%")
+    print(f"API:            {float(pu.get('apiPercentUsed') or 0):.2f}%")
+    print(f"综合 total:     {float(pu.get('totalPercentUsed') or 0):.2f}%")
+    print(
+        f"等价花费: ${float(pu.get('totalSpend') or 0)/100:.2f} / ${float(pu.get('limit') or 0)/100:.2f}"
+    )
+
+    # Today since 00:00
+    now = dt.datetime.now(TZ)
+    start = now.replace(hour=0, minute=0, second=0, microsecond=0)
+    data = post(
+        "aiserver.v1.DashboardService/GetFilteredUsageEvents",
+        {
+            "startDate": str(int(start.timestamp() * 1000)),
+            "endDate": str(int(now.timestamp() * 1000)),
+            "page": 1,
+            "pageSize": 200,
+        },
+    )
+    events = list(data.get("usageEventsDisplay") or [])
+    total = int(data.get("totalUsageEventsCount") or 0)
+    pages = max(1, (total + 199) // 200)
+    for page in range(2, pages + 1):
+        more = post(
+            "aiserver.v1.DashboardService/GetFilteredUsageEvents",
+            {
+                "startDate": str(int(start.timestamp() * 1000)),
+                "endDate": str(int(now.timestamp() * 1000)),
+                "page": page,
+                "pageSize": 200,
+            },
+        )
+        events.extend(more.get("usageEventsDisplay") or [])
+
+    tot = defaultdict(int)
+    by_model = defaultdict(lambda: defaultdict(int))
+    for e in events:
+        tu = e.get("tokenUsage") or {}
+        inp = int(tu.get("inputTokens") or 0)
+        out = int(tu.get("outputTokens") or 0)
+        cache = int(tu.get("cacheReadTokens") or 0)
+        model = e.get("model") or "unknown"
+        tot["events"] += 1
+        tot["input"] += inp
+        tot["output"] += out
+        tot["cache"] += cache
+        tot["all"] += inp + out + cache
+        tot["io"] += inp + out
+        by_model[model]["all"] += inp + out + cache
+        by_model[model]["events"] += 1
+
+    print(f"\n=== 今日（自 {start.strftime('%H:%M')}）===")
+    print(f"请求: {tot['events']:,}")
+    print(f"输入+输出: {tot['io']:,}")
+    print(f"含缓存合计: {tot['all']:,}")
+    print("按模型:")
+    for model, b in sorted(by_model.items(), key=lambda kv: -kv[1]["all"]):
+        print(f"  - {model}: {b['events']} 次, 含缓存 {b['all']:,}")
+
+
+if __name__ == "__main__":
+    main()