Merge branch 'ji' of http://192.168.0.200/knowledge-base-bj/deerflow-server into ji

2026-06-05 21:17:31 +08:00
commit cbbde887d4
@@ -13,9 +13,12 @@ methods' ``user_id`` parameter is designed for.

 from __future__ import annotations

+import asyncio
 import json
 import logging
+import os
 import re
+import time
 from collections.abc import AsyncIterator
 from difflib import SequenceMatcher
 from hashlib import sha256
@@ -53,6 +56,27 @@ from deerflow.persistence.user.model import UserRow
 logger = logging.getLogger(__name__)
 router = APIRouter(prefix="/api/admin/users", tags=["admin-users"])

+_LEADERBOARD_MAX_CONCURRENT = max(
+    1,
+    int(os.getenv("ADMIN_LEADERBOARD_MAX_CONCURRENT", "1")),
+)
+_LEADERBOARD_TIMEOUT_SECONDS = max(
+    5.0,
+    float(os.getenv("ADMIN_LEADERBOARD_TIMEOUT_SECONDS", "45")),
+)
+_LEADERBOARD_ACQUIRE_TIMEOUT_SECONDS = max(
+    0.1,
+    float(os.getenv("ADMIN_LEADERBOARD_ACQUIRE_TIMEOUT_SECONDS", "0.5")),
+)
+_LEADERBOARD_CACHE_TTL_SECONDS = max(
+    5.0,
+    float(os.getenv("ADMIN_LEADERBOARD_CACHE_TTL_SECONDS", "300")),
+)
+_leaderboard_semaphore = asyncio.Semaphore(_LEADERBOARD_MAX_CONCURRENT)
+_leaderboard_jobs: dict[str, asyncio.Task] = {}
+_leaderboard_cache: dict[str, dict] = {}
+_leaderboard_jobs_lock = asyncio.Lock()
+

 async def _require_admin(request: Request) -> None:
    """Reject non-admin callers. In no-auth mode there is no user → allow."""
@@ -138,6 +162,9 @@ class AdminLeaderboardResponse(BaseModel):
    range_days: int
    since: str
    until: str
+    status: str = "ready"
+    generated_at: str = ""
+    message: str = ""
    overview: AdminLeaderboardOverview
    users_by_activity: list[AdminLeaderboardUser] = Field(default_factory=list)
    users_by_questions: list[AdminLeaderboardUser] = Field(default_factory=list)
@@ -965,31 +992,61 @@ async def _query_overview(
    )


-@router.get("/leaderboard", response_model=AdminLeaderboardResponse)
-async def get_user_leaderboard(
-    request: Request,
-    days: int = Query(default=30, ge=1, le=365),
-    since: str | None = Query(default=None, description="Inclusive range start, ISO datetime"),
-    until: str | None = Query(default=None, description="Inclusive range end, ISO datetime"),
-    exclude_user_ids: list[str] | None = Query(default=None, description="User ids excluded from this leaderboard aggregation"),
-    include_scheduled: bool | None = Query(default=None, description="Include runs created by scheduled tasks"),
-    include_admins: bool | None = Query(default=None, description="Include admin accounts in leaderboard aggregation"),
-    include_failed: bool | None = Query(default=None, description="Include failed runs and LLM calls"),
-    limit: int = Query(default=20, ge=5, le=100),
+def _empty_leaderboard_response(
+    since_dt: datetime,
+    until_dt: datetime,
+    range_days: int,
+    *,
+    status: str,
+    message: str,
 ) -> AdminLeaderboardResponse:
-    """Return admin analytics for user activity, Q&A, skills, questions, and trends."""
-    await _require_admin(request)
-    session_factory = get_session_factory()
-    if session_factory is None:
-        raise HTTPException(status_code=503, detail="Analytics requires database persistence")
+    return AdminLeaderboardResponse(
+        range_days=range_days,
+        since=_iso(since_dt),
+        until=_iso(until_dt),
+        status=status,
+        message=message,
+        overview=AdminLeaderboardOverview(),
+    )

-    settings = load_system_settings().leaderboard
-    since_dt, until_dt, range_days = _resolve_range(days, since, until)
-    excluded = {uid for uid in settings.exclude_user_ids if uid}
-    excluded.update({uid for uid in (exclude_user_ids or []) if uid})
-    include_scheduled = settings.include_scheduled if include_scheduled is None else include_scheduled
-    include_admins = settings.include_admins if include_admins is None else include_admins
-    include_failed = settings.include_failed if include_failed is None else include_failed
+
+def _leaderboard_cache_key(
+    *,
+    since_dt: datetime,
+    until_dt: datetime,
+    limit: int,
+    excluded: set[str],
+    include_scheduled: bool,
+    include_admins: bool,
+    include_failed: bool,
+    cleanup_system_questions: bool,
+) -> str:
+    payload = {
+        "version": 2,
+        "since": _iso(since_dt),
+        "until": _iso(until_dt),
+        "limit": limit,
+        "excluded": sorted(excluded),
+        "include_scheduled": include_scheduled,
+        "include_admins": include_admins,
+        "include_failed": include_failed,
+        "cleanup_system_questions": cleanup_system_questions,
+    }
+    return sha256(json.dumps(payload, sort_keys=True, ensure_ascii=False).encode("utf-8")).hexdigest()
+
+
+async def _build_user_leaderboard_response(
+    session_factory,
+    settings,
+    since_dt: datetime,
+    until_dt: datetime,
+    range_days: int,
+    limit: int,
+    excluded: set[str],
+    include_scheduled: bool,
+    include_admins: bool,
+    include_failed: bool,
+) -> AdminLeaderboardResponse:
    async with session_factory() as session:
        common = {
            "exclude_user_ids": excluded,
@@ -1023,6 +1080,8 @@ async def get_user_leaderboard(
        range_days=range_days,
        since=_iso(since_dt),
        until=_iso(until_dt),
+        status="ready",
+        generated_at=_iso(datetime.utcnow()),
        overview=overview,
        users_by_activity=sorted(users, key=lambda u: (u.run_count + u.tool_call_count, u.total_tokens), reverse=True)[:limit],
        users_by_questions=sorted(users, key=lambda u: (u.qa_count, u.run_count), reverse=True)[:limit],
@@ -1038,6 +1097,225 @@ async def get_user_leaderboard(
    )


+async def _refresh_leaderboard_cache(
+    cache_key: str,
+    *,
+    session_factory,
+    settings,
+    since_dt: datetime,
+    until_dt: datetime,
+    range_days: int,
+    limit: int,
+    excluded: set[str],
+    include_scheduled: bool,
+    include_admins: bool,
+    include_failed: bool,
+) -> None:
+    acquired = False
+    try:
+        await _leaderboard_semaphore.acquire()
+        acquired = True
+        response = await asyncio.wait_for(
+            _build_user_leaderboard_response(
+                session_factory,
+                settings,
+                since_dt,
+                until_dt,
+                range_days,
+                limit,
+                excluded,
+                include_scheduled,
+                include_admins,
+                include_failed,
+            ),
+            timeout=_LEADERBOARD_TIMEOUT_SECONDS,
+        )
+        _leaderboard_cache[cache_key] = {
+            "created_at": time.monotonic(),
+            "response": response,
+        }
+    except asyncio.TimeoutError:
+        logger.warning("Leaderboard refresh timed out for key=%s", cache_key)
+    except Exception:
+        logger.exception("Leaderboard refresh failed for key=%s", cache_key)
+    finally:
+        if acquired:
+            _leaderboard_semaphore.release()
+        async with _leaderboard_jobs_lock:
+            _leaderboard_jobs.pop(cache_key, None)
+
+
+async def _ensure_leaderboard_refresh(cache_key: str, **kwargs) -> None:
+    async with _leaderboard_jobs_lock:
+        task = _leaderboard_jobs.get(cache_key)
+        if task and not task.done():
+            return
+        _leaderboard_jobs[cache_key] = asyncio.create_task(
+            _refresh_leaderboard_cache(cache_key, **kwargs)
+        )
+
+
+@router.get("/leaderboard", response_model=AdminLeaderboardResponse)
+async def get_user_leaderboard(
+    request: Request,
+    days: int = Query(default=30, ge=1, le=365),
+    since: str | None = Query(default=None, description="Inclusive range start, ISO datetime"),
+    until: str | None = Query(default=None, description="Inclusive range end, ISO datetime"),
+    exclude_user_ids: list[str] | None = Query(default=None, description="User ids excluded from this leaderboard aggregation"),
+    include_scheduled: bool | None = Query(default=None, description="Include runs created by scheduled tasks"),
+    include_admins: bool | None = Query(default=None, description="Include admin accounts in leaderboard aggregation"),
+    include_failed: bool | None = Query(default=None, description="Include failed runs and LLM calls"),
+    limit: int = Query(default=20, ge=5, le=100),
+) -> AdminLeaderboardResponse:
+    """Return admin analytics for user activity, Q&A, skills, questions, and trends."""
+    await _require_admin(request)
+    session_factory = get_session_factory()
+    if session_factory is None:
+        raise HTTPException(status_code=503, detail="Analytics requires database persistence")
+
+    settings = load_system_settings().leaderboard
+    since_dt, until_dt, range_days = _resolve_range(days, since, until)
+    excluded = {uid for uid in settings.exclude_user_ids if uid}
+    excluded.update({uid for uid in (exclude_user_ids or []) if uid})
+    resolved_include_scheduled = settings.include_scheduled if include_scheduled is None else include_scheduled
+    resolved_include_admins = settings.include_admins if include_admins is None else include_admins
+    resolved_include_failed = settings.include_failed if include_failed is None else include_failed
+    cache_key = _leaderboard_cache_key(
+        since_dt=since_dt,
+        until_dt=until_dt,
+        limit=limit,
+        excluded=excluded,
+        include_scheduled=resolved_include_scheduled,
+        include_admins=resolved_include_admins,
+        include_failed=resolved_include_failed,
+        cleanup_system_questions=settings.cleanup_system_questions,
+    )
+
+    cached = _leaderboard_cache.get(cache_key)
+    if cached:
+        response = cached["response"].model_copy()
+        age = time.monotonic() - float(cached["created_at"])
+        if age <= _LEADERBOARD_CACHE_TTL_SECONDS:
+            response.status = "ready"
+            return response
+        response.status = "refreshing"
+        response.message = "排行榜缓存已返回，后台正在刷新最新统计。"
+        await _ensure_leaderboard_refresh(
+            cache_key,
+            session_factory=session_factory,
+            settings=settings,
+            since_dt=since_dt,
+            until_dt=until_dt,
+            range_days=range_days,
+            limit=limit,
+            excluded=excluded,
+            include_scheduled=resolved_include_scheduled,
+            include_admins=resolved_include_admins,
+            include_failed=resolved_include_failed,
+        )
+        return response
+
+    await _ensure_leaderboard_refresh(
+        cache_key,
+        session_factory=session_factory,
+        settings=settings,
+        since_dt=since_dt,
+        until_dt=until_dt,
+        range_days=range_days,
+        limit=limit,
+        excluded=excluded,
+        include_scheduled=resolved_include_scheduled,
+        include_admins=resolved_include_admins,
+        include_failed=resolved_include_failed,
+    )
+    return _empty_leaderboard_response(
+        since_dt,
+        until_dt,
+        range_days,
+        status="running",
+        message="排行榜统计已在后台运行。线上用户较多时会慢一些，页面会自动刷新结果。",
+    )
+
+    try:
+        await asyncio.wait_for(
+            _leaderboard_semaphore.acquire(),
+            timeout=_LEADERBOARD_ACQUIRE_TIMEOUT_SECONDS,
+        )
+    except TimeoutError as exc:
+        raise HTTPException(
+            status_code=429,
+            detail="排行榜正在统计中，请稍后再试。线上用户较多时统计时间可能较长，请耐心等待。",
+        ) from exc
+
+    async def build_response() -> AdminLeaderboardResponse:
+        settings = load_system_settings().leaderboard
+        since_dt, until_dt, range_days = _resolve_range(days, since, until)
+        excluded = {uid for uid in settings.exclude_user_ids if uid}
+        excluded.update({uid for uid in (exclude_user_ids or []) if uid})
+        resolved_include_scheduled = settings.include_scheduled if include_scheduled is None else include_scheduled
+        resolved_include_admins = settings.include_admins if include_admins is None else include_admins
+        resolved_include_failed = settings.include_failed if include_failed is None else include_failed
+        async with session_factory() as session:
+            common = {
+                "exclude_user_ids": excluded,
+                "include_scheduled": resolved_include_scheduled,
+                "include_admins": resolved_include_admins,
+                "include_failed": resolved_include_failed,
+            }
+            tool_common = {
+                "exclude_user_ids": excluded,
+                "include_scheduled": resolved_include_scheduled,
+                "include_admins": resolved_include_admins,
+            }
+            question_common = {
+                **common,
+                "cleanup_system_questions": settings.cleanup_system_questions,
+            }
+            overview = await _query_overview(session, since_dt, until_dt, **common)
+            users = await _query_user_leaderboard(session, since_dt, until_dt, **common)
+            skills = await _query_call_rankings(session, since_dt, until_dt, user_id=None, field="skill", limit=limit, **tool_common)
+            tools = await _query_call_rankings(session, since_dt, until_dt, user_id=None, field="tool", limit=limit, **tool_common)
+            models = await _query_model_rankings(session, since_dt, until_dt, user_id=None, limit=limit, **common)
+            if not models:
+                models = await _query_run_dimensions(session, since_dt, until_dt, user_id=None, field="model", limit=limit, **common)
+            agents = await _query_run_dimensions(session, since_dt, until_dt, user_id=None, field="agent", limit=limit, **common)
+            run_statuses = await _query_run_dimensions(session, since_dt, until_dt, user_id=None, field="status", limit=limit, **common)
+            recent_questions = await _query_recent_questions(session, since_dt, until_dt, user_id=None, limit=limit, **question_common)
+            frequent_questions = await _query_frequent_questions(session, since_dt, until_dt, limit=limit, **question_common)
+            trends = await _query_trends(session, since_dt, until_dt, **common)
+
+        return AdminLeaderboardResponse(
+            range_days=range_days,
+            since=_iso(since_dt),
+            until=_iso(until_dt),
+            overview=overview,
+            users_by_activity=sorted(users, key=lambda u: (u.run_count + u.tool_call_count, u.total_tokens), reverse=True)[:limit],
+            users_by_questions=sorted(users, key=lambda u: (u.qa_count, u.run_count), reverse=True)[:limit],
+            users_by_tokens=sorted(users, key=lambda u: u.total_tokens, reverse=True)[:limit],
+            skills=skills,
+            tools=tools,
+            models=models,
+            agents=agents,
+            run_statuses=run_statuses,
+            recent_questions=recent_questions,
+            frequent_questions=frequent_questions,
+            trends=trends,
+        )
+
+    try:
+        return await asyncio.wait_for(
+            build_response(),
+            timeout=_LEADERBOARD_TIMEOUT_SECONDS,
+        )
+    except TimeoutError as exc:
+        raise HTTPException(
+            status_code=504,
+            detail="排行榜统计时间过长，请缩短时间范围后重试，或稍后再查看。",
+        ) from exc
+    finally:
+        _leaderboard_semaphore.release()
+
+
 async def _build_leaderboard_report_prompt(
    session_factory,
    settings,